The application of in vitro cell-free conversion systems to human prion diseases
- First Online:
- Cite this article as:
- Jones, M., Peden, A.H., Head, M.W. et al. Acta Neuropathol (2011) 121: 135. doi:10.1007/s00401-010-0708-8
A key event in the pathogenesis of prion diseases is the conversion of the normal cellular isoform of the prion protein into the disease-associated isoform, but the mechanisms operating in this critical event are not yet fully understood. A number of novel approaches have recently been developed to study factors influencing this process. One of these, the protein misfolding cyclical amplification (PMCA) technique, has been used to explore defined factors influencing the conversion of cellular prion protein in a cell-free model system. Although initially developed in animal models, this technique has been increasingly applied to human prion diseases. Recent studies have focused on the role of different isoforms of the disease-associated human prion protein and the effects of the naturally occurring polymorphism at codon 129 in the human prion protein gene on the conversion process, improving our understanding of the interaction between host and agent factors that influence the wide range of phenotypes in human prion diseases. This technique also allows a greatly enhanced sensitivity of detection of disease-associated prion protein in human tissues and fluids, which is potentially applicable to disease screening, particularly for variant Creutzfeldt–Jakob disease. The PMCA technique can also be used to model human susceptibility to a range of prions of non-human origin, which is likely to prove of considerable future interest as more novel and potentially pathogenic prion diseases are identified in animal species that form part of the human food chain.
KeywordsPMCACJDvCJDPrion protein genePrPCPrPSc
Phenotypic, genetic and molecular classification of human prion diseases
PrPres description (and nomenclature)
~21 kDa non-glycosylated fragment and diglycosylated does not predominate (Type 1A) and/or ~19 kDa non-glycosylated fragment and diglycosylated does not predominate (Type 2A)
Sporadic fatal insomnia
~19 kDa non-glycosylated fragment and diglycosylated does not predominate (Type 2A)
Poorly protease-resistant, resulting in defined low molecular weight fragments ranging from ~8 to 30 kDa
~19 kDa non-glycosylated fragment and diglycosylated does predominate (Type 2B)
Iatrogenic CJD (growth hormone, and dura mater associated)
~21 kDa non-glycosylated fragment and diglycosylated does not predominate (Type 1A) and/or ~19 kDa non-glycosylated fragment and diglycosylated does not predominate (Type 2A)
Familial or genetic
Type 1 (unglycosylated underrepresented)
Type 2 (unglycosylated underrepresented)
Type 1 (unglycosylated underrepresented)
Fatal familial insomnia
Type 2 (unglycosylated underrepresented)
Type 1 (unglycosylated underrepresented) and/or low molecular weight fragments of ~8 kDa
Animal transmission studies have been invaluable in developing our understanding of the biology of human prion diseases. Indeed, these models can be considered amongst the best for any human neurodegenerative disorder. From their beginnings in the experimental transmission of the scrapie agent, a wide range of models have evolved, ranging from inbred lines of laboratory mice to transgenic mice, other small rodents and a range of primates . These studies have allowed the biological characterisation of the agent strain in scrapie, bovine spongiform encephalopathy (BSE) and some forms of human prion disease, particularly vCJD, where the close resemblance to the BSE agent was critical in confirming vCJD as a zoonosis . The development of gene targeted “humanised” transgenic mice (with the human PRNP gene inserted in place of the corresponding murine gene) has allowed modelling of the effects of the PRNP codon 129 polymorphism  and PRNP mutations on disease transmission while the development of transgenic mice with multiple copies of PRNP and over-expression of PrPC has allowed transmission experiments with substantially reduced incubation periods . Other transgenic mice in which the murine prion protein gene has been ablated, with no detectable PrPC, have also been valuable in demonstrating the fundamental role of this host gene in disease transmission . It is clear that the pathogenesis of prion diseases is complex, and these complexities cannot necessarily all be addressed within a single experiment, or indeed within a single model.
The challenges of modelling these complexities are formidable, but a number of novel approaches have recently been developed, including cell culture systems and cell-free systems to study the conversion of PrPC to PrPSc. Cell culture systems have been developed for some strains of the scrapie agent , but have not been widely developed for the study of human prion diseases, although this challenge is the subject of intense current activity. However, a cell-free system of PrPC conversion, protein misfolding cyclical amplification (PMCA), is applicable to both animal and human prion diseases [49, 84]. The flexibility of this technique to explore defined factors influencing PrPC conversion allows remarkable opportunities for focused studies in a cell- and animal-free model system.
A central role for PrPSc formation
The prion hypothesis holds that these diseases have a central common pathogenic process resulting from the conversion of PrPC to PrPSc, and that PrPSc itself constitutes a novel potentially infectious agent . Whether neurotoxicity results from a loss of PrPC function, acquisition of toxic properties following conversion, the generation of a toxic intermediate or by-product or a combination of all three mechanisms remains open to debate [2, 22, 29, 33, 95]. The failure of PrP null brain grafts to develop pathology when residing in prion infected experimental animal brain [1, 12, 13] and the apparent arrest or even reversal of neurodegenerative change when PrPC expression is interfered with [67, 68, 96] both point firmly towards the centrality of the conversion process itself.
PrPSc and disease phenotype
Although the prion hypothesis may be the dominant paradigm of the field, it is important to note that it is a working hypothesis, currently incomplete in its ability to explain all aspects of these diseases. Nevertheless, PrPSc accumulation marks all cases that receive a pathological diagnosis of CJD, or indeed any of the other human prion diseases. Partial resistance to proteolytic degradation is the most commonly utilised operational distinction between PrPC and PrPSc (but not the only one) and a combination of digestion of brain tissue specimens with proteinase-K and the analysis of the pattern of protease-resistant fragments by Western blotting has been shown to have diagnostic value in distinguishing between prion diseases and other neurodegenerative conditions, and in correctly identifying the form of prion disease in individual patients (reviewed in ). The current situation with respect to these phenotype–genotype–prion protein isotype correlations, in the most commonly occurring human prion diseases, is summarised in Table 1. This includes the tight association of PRNP mutations with familial prion diseases and the profound effect of the M/V polymorphism at codon 129. The association of PrPSc with different biophysical properties marking different human prion disease phenotypes is a necessary corollary to the prion hypothesis, providing a possible explanation of how a prion agent or agents (in combination with host prion protein genetics) can cause different disease phenotypes.
If PrPSc formation is both necessary and sufficient to generate prion infectivity, then prions should be able to be generated synthetically from defined reagents, specifically PrPC, whether purified from tissue or in a more stringent test made using recombinant means. Although initial attempts proved unsuccessful (see for example ), more recent studies suggest that under very specific experimental conditions (sometimes novel) prion strains can be generated de novo from PrPC, both ex vivo [6, 35, 63] and in vivo [53, 65, 87].
The development of cell-free conversion systems
Recombinant prion protein or PrPC can be induced to refold by a variety of chemical or biophysical treatments. Typically, these include an element of denaturation using detergents, acidic pH, chaotropic salts and reduction of the disulphide bond, alone or in combination [48, 78, 98, 99]. The resultant structures acquire higher β-sheet content, an enhanced degree of protease resistance and typically show progressive aggregation, in some cases producing fibrillar structure with the properties of amyloid. Other potential influences on this fibrillisation process have been reported to include copper binding, deamidation, glycosylation and the presence of nucleic acid [30, 72, 81]. To what extent the prion protein (PrP) based structures formed by these chemical and biophysical treatments mimic PrPSc deposition in vivo remains an open question, but at the very least they provide insights into the structural repertoire of PrP. The most recent work, however, appears to show that PrP amyloids made under different conditions not only have different structures that are capable of self-propagation, but remarkably these structures can correlate with subtle biological properties commonly regarded as aspects of prion agent strain [31, 64, 66].
Template directed in vitro conversion
The original in vitro prion protein conversion assay was developed in the early 1990s. The methodology employed partially purified and guanidine denatured brain PrPSc (or PrPres) to convert protease-sensitive PrPC (or PrPsen), metabolically labelled with 35S and isolated from cultured cells. The conversion reactions were allowed to proceed for 48 h in the presence of guanidine and detergent and conversion was assayed for by the presence of radioactively labelled protease-resistant PrP by SDS-PAGE and autoradiography [23, 58]. In a series of landmark papers, Caughey et al.  showed that this in vitro conversion assay could be used to address issues of the molecular basis of agent strain, intrinsic barriers to conversion [10, 11, 59, 79, 82] factors that may promote [34, 85] or inhibit [26–28, 36] conversion and the molecular details of the conversion reaction itself [24, 25, 41, 43]. Although key in establishing the principle of in vitro conversion assays and highly informative in gaining a better understanding the mechanisms by which PrPsen can become converted to PrPres, the original in vitro conversion assay and its various improvements  suffered from a number of limitations, particularly the relative inefficiency of the reaction. Two relatively minor, but inspired, modifications to the reaction were incorporated in a new method, termed PMCA, which considerably broadened the applicability of the original in vitro conversion assay.
Application of PMCA to human prion diseases
Whilst PMCA has been successfully applied to the amplification of PrPSc associated with a number of animal TSEs in a range of tissues, including blood [62, 71, 84, 89, 90], with the amplified products retaining the biochemical characteristics, structural properties and infectivity associated with the original seeds [18, 19, 39], the application of PMCA to human prion diseases is still in its infancy. Initial results have shown that substrates prepared from human brain tissue, human platelets and humanised PRNP transgenic mouse brain tissue can support efficient amplification of PrPSc associated with vCJD, sCJD and some animal prion diseases [49–52, 88] with a high degree of fidelity even after multiple rounds of sPMCA . However, it was evident that the PRNP-129 genotype of both the seed and substrate had a major influence on amplification efficiency [18, 49–52].
PrPSc associated with vCJD (type MM2B) and sCJD types MM1 and MM2 was efficiently amplified in PRNP-129 MM substrate to a lesser degree in PRNP-129MV substrate and weakly if at all in PRNP-129VV substrate . In contrast, PrPSc associated with sCJD types VV1 and VV2 could only be amplified in a PRNP-129VV substrate. A surprising observation was that the substrate specificity of PrPSc associated with sCJD types MV1 and MV2 appeared to be determined by the PrPSc type . sCJD MV1 PrPSc resembled sCJD MM1 and MM2 PrPSc in its preference for PRNP-129MM substrate, whereas sCJD MV2 PrPSc resembled sCJD VV1 and VV2 PrPSc in its preference for PRNP-129VV substrate. Although it is not known whether both PRNP alleles are equally expressed in PRNP-129MV heterozygous individuals, the results from our PMCA experiments were consistent with the propagation of PRNP-129M PrPSc in sCJD MV1 cases and PRNP-129V PrPSc in sCJD MV2 cases. One possible explanation is that the presence of methionine at codon 129 favours a type 1 conformation, whereas valine at the same position predisposes to a type 2 conformation . Based on these PMCA results, it might be proposed that MM1 and MV1 sCJD could be grouped into one subgroup and VV2 and MV2 sCJD grouped into another subgroup and a recent report examining sCJD subtypes according to PrPSc proteinase-K resistance came to a similar conclusion . MM1 and MV1 sCJD do indeed form a single clinicopathological phenotype, whereas sCJD VV2 and MV2 do not [47, 75]. This issue is complicated by the presence of both types 1 and 2 PrPSc in up to 40% of sCJD cases  and it will be interesting to determine whether both PrPSc types or one single PrPSc type can be amplified from such cases.
Why distinct PrPSc types accumulate in specific areas in a CJD brain and why more than one PrPSc type can be found in some CJD brains has yet to be explained. One possible explanation is that the differential levels of PrPC expression, endogenous proteolytic processing of PrPC and PrPC post-translational modification, such as the degree of glycosylation that have been shown to occur in different regions of the human brain  could favour the propagation and accumulation of distinct PrPSc types in certain regions. Animal PMCA experiments have indeed shown that the amount of PrPC  and the stoichiometry of PrPC glycoforms  in a substrate can modulate PrPSc amplification efficiency. Whether or not PrPC expression levels, PrPC proteolytic processing and PrPC post-translation modification in different regions of the human brain can influence PrPSc propagation could be investigated by carrying out PMCA reactions in substrates prepared from the different regions.
The only known animal TSE to have transmitted to humans is bovine BSE, the accepted cause of vCJD [14, 97]. To date, all genotyped confirmed cases of vCJD have occurred in PRNP-129MM individuals. However, a possible case of vCJD in PRNP-129MV individual based on clinical symptoms  and evidence of possible asymptomatic pre-clinical disease, based on the detection of PrPSc in peripheral tissues in two PRNP-129MV individuals [76, 77] and two PRNP-129VV individuals  have been reported. This would suggest that all three PRNP-129 genotypes are susceptible to infection, as previously predicted by transmission studies in transgenic mice expressing human PrP of the three possible PRNP-129 genotypes, which concluded that whilst the presence of methionine at codon 129 was associated with efficient disease transmission and short disease incubation all three genotypes were susceptible . The results of these transmission studies were mirrored by the results of analogous PMCA experiments which attempted to amplify PrPSc associated with bovine BSE, ovine BSE and human vCJD in substrates containing PrPC of each PRNP-129 genotype . All three PrPSc seeds, regardless of the species amino acid sequence, displayed similar substrate preferences with the most efficient amplification achieved in the PRNP-129MM substrate and not much, if any, evident in the PRNP-129VV substrate. In contrast, PrPSc associated with ovine scrapie, an animal TSE not thought to present a threat to humans, could not be amplified in any of the human PrPC containing substrates, but could be amplified in substrate prepared from healthy sheep brain. Given that the PrPSc associated with both ovine BSE and ovine scrapie was comprised of PrP of the same amino acid sequence, it would appear that it was the strain-specific conformation of PrPSc rather than its species-specific amino acid sequence that determined the susceptibility of the host’s PrPC to conversion. PMCA in substrates containing human PrPC could provide a rapid method of assessing the risks posed to human health by newly identified animal TSEs  which at present relies on lengthy, expensive in vivo bioassays in transgenic mice expressing human PrPC [3, 9, 60].
PMCA and de novo/spontaneous generation of prions: a model for sCJD?
The prion hypothesis assumes that PrPSc formation is both necessary and sufficient to generate prion infectivity. If this is true then prions should be able to be generated from either purified PrPC or bacterially expressed recombinant PrP (recPrP). Using the hamster scrapie 263K model, it has been shown that the minimal components required for efficient amplification of PrPSc by PMCA are purified PrPSc, purified PrPC which also contains lipid and polyA+ RNA . Amplification of purified PrPC supplemented with polyA+ RNA, in the absence of PrPSc seed, also resulted in the de novo/spontaneous generation of PrPSc. This de novo/spontaneously produced PrPSc was infectious when inoculated into hamsters. Kim et al. subsequently demonstrated that PMCA seeded initially with hamster PrPSc, then propagated extensively by serial PMCA using highly purified bacterially expressed hamster recPrP as a substrate in the absence of synthetic or mammalian cofactors (other than buffer salts and detergents) yields a final product that is infectious, but apparently containing none of the original seed . More recently Wang et al.  have described the creation of infectious recombinant prions with the hallmarks of PrPSc following the serial PMCA of bacterially expressed recombinant murine PrP in the presence of the synthetic anion phospholipid POPG and RNA. Whilst these models offer evidence in support of the prion hypothesis, could similar results be obtained using normal brain tissue as a substrate source and if so, could it serve as a model for sCJD in humans?
A recent report described attempts to generate de novo/spontaneous PrPSc from normal brain tissue from a range of species hamster, mouse, human and transgenic mouse over-expressing human PrP, by the sPMCA procedure . Using standard sPMCA conditions, and taking precautions to avoid cross-contamination, generation of de novo/spontaneous PrPSc was never observed validating the use of PMCA for diagnostic applications such as those proposed for a confirmatory human blood screening assay . However, sPMCA could be modified by extending the number of cycles per round to generate de novo/spontaneous PrPSc in hamster and mouse brain homogenate, but not in human nor in transgenic mice over-expressing human PrPC. The de novo/spontaneous PrPSc generated from normal hamster brain tissue was infectious when inoculated into wild-type hamsters producing a new disease phenotype with unique clinical, neuropathological and biochemical features. Given that sCJD is thought to arise from the low frequency, spontaneous misfolding of the PrPC, which then propagates to other PrPC molecules in a manner similar to infectious cases , the authors of this study expressed surprise that spontaneous PrPSc was never obtained from human nor humanised transgenic mouse brain tissue under the experimental conditions used . Their interpretation of the results was that human PrPC has a low propensity to initiate misfolding and that the appearance of sCJD in humans may simply reflect the long life span that provides greater chance for stochastic processes of spontaneous misfolding to occur. Extrapolation from a relatively simple cell-free system to a transgenic animal model and onto a disease (which is poorly understood) is fraught with difficulties. Nevertheless in the absence of any other model for this idiopathic form of CJD, PMCA may offer insights into spontaneous PrPSc formation and the conditions that might promote or inhibit it.
Technological developments, QuIC, and the problem of de novo PrPSc formation
As described above, sPMCA provides a way of detecting extremely low levels of PrPSc, but some potential drawbacks are the time taken, the complexity of the substrate and reliance on sonication, which is difficult to standardise [35, 86]. Although human PrPC would be predicted to be the most appropriate substrate for amplification of human PrPSc, the most obvious sources, such as normal human brain tissue, or transgenic mouse brains expressing human PrPC, can be difficult to obtain or standardise.
These disadvantages have prompted research groups to investigate ways of improving and simplifying PMCA. Purified cellular PrPC can be used as substrate (see above), but is expensive to produce in sufficient quantities and polyanionic cofactors are necessary to promote amplification . A technique known as rPrP-PMCA, recently developed by Caughey et al.  may circumvent these problems using recombinant PrP (rPrP) expressed in bacteria as a substrate. When rPrP is seeded with scrapie-infected hamster brain and subjected to PMCA cycles of sonication, it converts to a protease-resistant form, called rPrP-resSc, in yields far greater than those obtained by other groups investigating the PrPSc-induced conversion of rPrP [37, 44, 56]. Unlike conventional PMCA, a high concentration of rPrP substrate (~0.1 mg/ml) and a low concentration of the detergent sodium dodecyl sulphate, which destabilises PrP were found to be necessary for conversion. The banding pattern for rPrP-res(Sc) on Western blots following digestion with proteinase-K consists of a 17-kDa protease-resistant core fragment combined with shorter C-terminal fragments at 11–13 kDa .
Recombinant PrP has the advantage of being easily generated and modified by sequence changes or the addition of epitope tags, facilitating a range of molecular studies. As with PMCA, the detection sensitivity of rPrP-PMCA can be increased by performing serial rounds with fresh rPrP substrate. Two rounds could detect attogram levels of hamster scrapie PrPSc (~0.01 lethal dose) making serial rPrP-PMCA significantly more rapid than standard sPMCA [4, 19, 83].
Caughey et al.  showed that under specific conditions, automated shaking at elevated temperatures promoted conversion as effectively as sonication and they named this method quaking induced conversion (QuIC). Two QuIC rounds were sufficient to detect PrPSc in cerebrospinal fluid from scrapie-infected hamsters and sheep, but not healthy control animals [5, 74]. The use of shaking has the potential to make QuIC easier to standardise than PMCA. However, when optimising the conditions for both QuIC and rPrP-PMCA, a careful balance needs to be struck to maximise sensitivity whilst avoiding the spontaneous conversion of rPrP in unseeded controls [4, 5, 74]. The use of rPrP and shaking in QuIC shares certain similarities to the microplate-based amyloid seeding assay (ASA) developed by Colby et al. in which the conversion of rPrP to an amyloidal state is monitored in real-time by a change in the excitation/emission maxima of thioflavin T, included in the reaction . However, with ASA, a partial purification of the PrPSc seed was needed to effect conversion of rPrP.
An important question is whether QuIC developed using a hamster model can be applied to human prion diseases. Interestingly, hamster rPrP was a more effective substrate than human rPrP, enabling the detection of femtogram amounts of vCJD PrPSc . Further work may be required to increase the sensitivity of QuIC for the detection of human PrPSc and to optimise the precise conditions to completely avoid de novo conversion. However, at this point in time QuIC appears to offer the promise of a PrPSc amplification assay in which some of the difficulties associated with PMCA can be avoided.
It is clear that the development of cell-free systems to model the conversion of PrPC to PrPSc has enormous potential for the study of key aspects of the pathogenesis of human prion diseases. The capacity for greatly enhanced sensitivity of PrPSc detection in human tissues and fluids is of considerable interest both in terms of disease screening and for risk assessments of infection following iatrogenic exposure, but the development of these applications must address the potential problem of false-positive reactions relating to the spontaneous conversion of substrate to PrPSc [6, 35, 90]. However, this potential for spontaneous conversion may prove to be an advantage when studying factors that relate to the apparently spontaneous occurrence of CJD (in the form of sporadic CJD). This possibility, combined with more sophisticated techniques to investigate the effects of subtle modifications of PrPSc structure and aggregation on the biological properties of the transmissible agent [31, 64, 66], is likely to aid our understanding of the host and agent factors that determine the wide range of phenotypes in human prion diseases. Finally, the capacity to use cell-free systems to model human susceptibility to a range of non-human prions is likely to prove considerable future interest as more novel prion diseases (such as atypical forms of scrapie and BSE) [16, 93] are identified in species that form a part of the human food chain and in which widespread screening for prions is either being withdrawn or has never been performed.
The UK National Creutzfeldt–Jakob Disease Surveillance Unit is funded by the Department of Health and the Scottish Government. The development of PMCA and QuIC within this Unit has been externally funded by the European network of excellence NeuroPrion (FOOD-CT-2004-506579) and the Chief Scientist’s Office of the Scottish Government (CZB/4/357 and CZB/4/688), in addition to financial support from the UK Blood Forum and to close collaborative arrangements with the Scottish National Blood Transfusion Services. The Brain Bank in the National Creutzfeldt–Jakob Disease Surveillance Unit is funded by the Medical Research Council (G0900580). We are indebted to the Neuropathologists, Neurologists and their staff across UK for their continuing support. We are most grateful to the families of CJD patients for giving consent to use tissues for research.