PrP aggregation can be seeded by pre-formed recombinant PrP amyloid fibrils without the replication of infectious prions

The misfolding and aggregation of host protein in the brain is a pathological characteristic of several neurodegenerative diseases, such as Alzheimer’s disease (AD), Parkinson’s disease (PD) and transmissible spongiform encephalopathy (TSE). However, the actual role of these protein aggregates in the neurodegenerative process is currently unclear. TSEs differ from other neurodegenerative diseases, since they affect several mammalian species other than humans and are infectious. TSEs such as scrapie in sheep and goats, bovine spongiform encephalopathy (BSE) in cattle and Creutzfeldt–Jakob disease (CJD) in humans can be transmitted between individuals of the same species and, in some cases, can spread between different species. TSEs exist as a large number of strains/isolates that show specific, reproducible clinical and pathological characteristics on transmission in animals. They can be contagious and can spread horizontally between animals in the field, such as scrapie in sheep and chronic wasting disease (CWD) in deer [60]. Other TSEs, such as BSE in cattle and CJD in humans, are not contagious, but are infectious, and can be transmitted indirectly between individuals following ingestion of contaminated tissues [3, 4, 19, 73], inoculation of such tissues into the CNS or periphery [1, 2, 7, 36], or transfusion of blood [26, 27, 38, 57, 74]. The infectious agent responsible for TSE is not a conventional pathogen such as a bacterium or virus. However, the disease is caused by a titratable infectious agent, which maintains specific characteristics on transmission in animals. It is thought that the infectious agent is a misfolded form of the host prion protein (PrP^C) [59]. This misfolded protein propagates by binding to and converting PrP^C into the abnormal isoform (PrP^Sc), and this autocatalytic conversion results in the deposition and spread of PrP^Sc through the central and peripheral nervous system, and also some viscera such as the lymphoreticular system, in patterns characteristic of each individual TSE strain/isolate.

The central role of the prion protein in TSE disease has led to the infectious agent being termed a prion [59] and the disease being referred to as prion disease. However, the term “prion” or “prion-like” has more recently been used to describe the observed spread of protein aggregates and amyloid accumulation in the brains of transgenic mice expressing either wild-type or mutant [51] human amyloid precursor protein (APP), following intracerebral [47, 49, 70, 71] and peripheral [17] inoculation with pre-formed aggregates of beta-amyloid peptide (Aβ). Similar models of induced protein aggregation have also been described for PD (inoculation of α-synuclein aggregates) [50], amyotrophic lateral sclerosis (ALS) (inoculation of misfolded SOD-1) [52] and tauopathy (inoculation of filamentous tau) [11]. “Prion-like” spread occurs when pre-formed protein aggregates are introduced into a host, which has the effect of accelerating the misfolding and aggregation of the endogenous host protein in the brain. On this basis, it is entirely possible that these pre-formed aggregates of Aβ, α-synuclein, SOD-1 and tau are acting as “prions” and are providing a template driving the misfolding of normal cellular forms of the protein into amyloid fibrils. However, in APP transgenic mouse models, acceleration of protein aggregation has been observed only following direct intracerebral inoculation [47, 50, 71] or peripheral inoculation [17]. Protein aggregation is not accelerated by oral, intravenous, intraocular or intranasal inoculation [15]. Several groups have demonstrated the potential for existence of “strains” of Aβ and α-synuclein [5, 25], and other reports described Aβ amyloid plaques in human growth hormone recipients [28, 29], and dura matter graft recipients [21, 35]. However, no definitive evidence of direct or indirect transmission has been described for either AD or PD in humans [28].

The generic use of the terms “prion” and “prion-like” to describe various aspects of all neurodegenerative diseases does, therefore, cause confusion. AD, PD, ALS and tauopathy are not transmissible diseases, and there is no epidemiological evidence to support an infectious aetiology [28]. These diseases are, therefore, distinctly different in their aetiology from TSEs. It has been proposed that the extended term “propagon” could be used to simply describe that a misfolded protein is being propagated, and avoid confusion over prion transmissibility and infection [16]. In this paper, we will use the terms “prion” and “prion-like” to describe the general mechanism of protein aggregation and spread, and not as a surrogate for TSE.

Although the TSE agent is thought to be PrP^Sc, previous work from our laboratory has shown that PrP aggregation does not always lead to replication of an infectious agent and subsequent TSE [55, 56]. There exists a subset of cases in which PrP aggregates are formed in the brain in the absence of TSE agent replication [9, 10, 24, 55, 56]. One such disorder that exemplifies this phenomenon is a variant of Gerstmann–Sträussler–Scheinker (GSS) disease, a familial human TSE disease characterised by PrP amyloid plaque deposition. The most common form of GSS is associated with a proline-to-leucine mutation at PrP codon 102 (GSS P102L) [54]. Inoculation of knock-in transgenic mice homozygous for the equivalent mutation in murine PrP (101LL) with classical forms of GSS P102L (associated with spongiform degeneration and diffuse PrP deposition) resulted in the development of clinical and pathological signs of TSE disease. When 101LL mice were inoculated with atypical forms of GSS P102L (no spongiform change and PrP amyloid plaques), inefficient disease transmission was observed with most animals surviving for full lifespan with no clinical signs or spongiform degeneration. However, on postmortem analysis, many of the inoculated mice were found to have large PrP amyloid plaques in the brain [56]. Similarly, when 101LL mice were inoculated with brain homogenate from sick GSS22 mice [53] (which overexpress 101L murine PrP and spontaneously develop neurological signs and large PrP amyloid plaques), no clinical signs of TSE disease or spongiform degeneration were evident, but large PrP amyloid plaques were again identified in the brains of these mice postmortem [55]. Hence, extracts from both humans with an atypical form of GSS P102L and transgenic mice (GSS22) overexpressing murine 101L-PrP (both characterised by PrP amyloid plaque disposition) do not transmit TSE disease to 101LL mice. Components of the brain inoculum are instead capable of seeding the aggregation of PrP amyloid plaques in recipient mice expressing 101L but not wild-type PrP. However, disease may be transmissible from such atypical TSE isolates in other model systems [58].

Although the above data argue for a direct “seeding” of PrP amyloid deposition by pre-formed PrP aggregates in the inoculum, our experiments performed to date have utilised brain material harvested from individual patients or mice with neurological symptoms, which may contain other components, or possibly low levels of an infectious agent, in addition to pre-formed PrP amyloid seeds. Therefore, in this study, we aimed to investigate directly whether misfolded forms of PrP, in the absence of other components of brain homogenate, can seed amyloid plaque formation, or cause TSE disease, following intracerebral inoculation in recipient mice. To this end, wild-type murine recombinant PrP (WT-recPrP) and 101L murine recPrP (101L-recPrP) preparations were refolded into different conformations in vitro and inoculated into groups of wild-type 129/Ola and 101LL mice to examine the ability of the different recombinant PrP (recPrP) conformers to either seed PrP amyloid plaques, or cause the development of TSE disease in 101LL mice. In contrast to other studies describing inoculation of recPrP fibrils [12, 13, 37, 39, 40, 42, 43, 61, 66], we saw no TSE disease in recipient mice, but did reproduce the seeding of PrP amyloid plaques observed in our previous work following inoculation of brain extract containing in vivo-derived amyloid fibrils.

Brain sections from all 378 mice were analysed for PrP deposition by immunostaining with anti-PrP Mab 6H4. All mice inoculated with α-monomeric recPrP, oligomeric recPrP or saline control inocula were negative for any forms of PrP deposition in brain tissue by immunohistochemistry (IHC) and showed no thioflavin-s fluorescence regardless of sequence of recPrP or recipient mouse genotype (Figs. 2, 3). Hence, preparations containing only monomeric and oligomeric isoforms, generated under our experimental conditions, were unable to elicit neurological disease, spongiform degeneration, or induce PrP aggregation in wild-type or 101LL mice. However, some 101LL mice inoculated with either WT-recPrP or 101L-recPrP amyloid fibrils showed large thioflavin-s fluorescent PrP amyloid plaques that were absent in wild-type mice that received the same inocula (Figs. 2, 3). Large multicentric plaques were observed in 10/21 101LL mice that received WT-recPrP amyloid fibrils, and in 14/19 101LL mice that received 101L-recPrP fibrils (Table 1). Plaques were present mainly in the corpus callosum and vicinity, as well as in some areas of the stratum lacunosum and stratum moleculare of the hippocampus. In a few animals, plaques were also visible in the subventricular area. Plaque morphology was variable, and composed of unicentric, stellate and multicentric deposits. Bilateral distribution of plaques (in the corpus callosum, hippocampus and striatum) was observed in animals for which whole brain slices were available (Figure S1). Analysis by histoblot showed plaques to be formed of proteinase K-resistant PrP (Figure S2). Gliosis and glial cell activation was similar to that observed in aged matched un-inoculated control 101LL mice (Fig. 3). No obvious difference in plaque morphology or distribution was observed in recipient 101LL mice following challenge with either WT-recPrP or 101L-recPrP preparation. Overall, despite the absence of TSE disease, PrP amyloid plaques were induced in 101LL mice following inoculation of recPrP fibrils.

Previous work in our laboratory has demonstrated that misfolding and aggregation of PrP can occur in the absence of TSE agent replication and infectious TSE disease [30, 55, 56]. These experiments involved inoculation of brain material derived from cases of neurological disease (GSS patients and sick GSS22 mice overexpressing 101L-PrP), which contained other tissue components in addition to amyloid protein seeds. Using refolded WT-recPrP and 101L-recPrP fibrils as inocula, we have now demonstrated seeding of PrP amyloid plaques in 101LL mice (but not wild type mice), proving that the misfolded protein seed alone can initiate plaque formation. No such seeding was produced following inoculation of recPrP oligomers or control α-helical PrP isoforms, indicating that amyloid fibrils, or specific, smaller protofibrils are required to initiate seeding. Aggregation of 101L-PrP in the host was also seeded using both WT-recPrP and 101L-recPrP fibrils as inoculum, indicating that it is the structure of the inoculated seed, not the amino acid sequence, that is important in initiating aggregation. Hence, both the macromolecular structure of PrP in the inoculum and the expression of mutant 101L-PrP in the host are required for efficient amyloid plaque formation. Whether seeding would have been observed in wild-type mice if lifespan had been extended is unknown, but possible. The 101L mutation in murine PrP may alter the kinetics of aggregation, or cellular processes involved in the clearance of protein aggregates, resulting in an acceleration of the process. These issues will be examined in future studies of this model.

The creation of “synthetic prions” has been described previously by others following inoculation of amyloid fibrils derived from refolded recPrP sources [12, 13, 37, 39, 40, 42, 43, 61, 66]. In contrast to our study, these preparations were shown to cause a neurological disease/TSE on primary inoculation or subpassage in mice. One recent study has also shown development of neurological disease following inoculation of recPrP oligomers rather than fibrils, but only following refolding in the presence of RNA molecules extracted from purified 263K scrapie fibrils [66]. It is difficult to draw general conclusions from these experiments as in each case, the conditions used to refold recPrP prior to inoculation were different, with varying concentrations of denaturant, different buffering, and variations in speed of agitation [12, 37, 39, 42, 61, 66]. Inoculations were performed in different mouse lines [12, 13, 37, 61], and hamsters [39, 40, 42, 43, 66], using amounts of recPrP ranging from 0.2 µg [66] to 30 µg [12]. Following recPrP fibril inoculation, some studies [39, 40, 42, 61] did show evidence of PrP aggregation in brain in the absence of TSE disease, agreeing with the data presented here. However, in contrast to our study where only plaque formation was observed on subpassage, neurological disease/TSE was observed in mice following subpassage of these other models [39‐41, 61]. A recent study which examined 19,468 unique refolding conditions and assayed by infection in cell culture concluded that none of the conditions reproducibly created high-titre infectious synthetic prions, and that creation of synthetic infectivity is a rare event which cannot be efficiently reproduced in vitro [64].

EM analysis of subpassage tissue from 101LL mice with seeded plaques confirmed that amyloid plaques (including multicentric plaques) seeded by recPrP shared the same origin on cell membranes, the same growth by conversion of native cell-membrane PrP^C on cellular processes and seed dispersal through the interstitial spaces as those previously described in 101LL mice inoculated with atypical GSS [30]. Similarly, 101LL mice showed no abnormal membrane pathology and intra-lysosomal PrP labelling, both associated with TSE disease [30, 31]. The generation of abnormal PrP predominantly by oligodendroglial cells is also atypical of that of classical, naturally occurring TSEs, where most abnormal PrP is formed on the surface of neuronal dendrites. These data suggest that recPrP fibrils and GSS inocula seed plaque formation by a common mechanism, and that plaques in GSS-inoculated 101LL mice were formed from amyloid seeds in the atypical GSS inoculum, and not from other components of the inoculum. Although no TSE disease was observed in atypical GSS-inoculated 101LL mice, it should be noted that these isolates do appear to be transmissible in different animal models [58]. Interestingly, the lack of abnormal membrane pathology and intra-lysosomal PrP staining was also observed following pathological analysis of hamsters with SSLOW (Synthetic Strain Leading to Overweight) initiated with refolded recPrP fibrils [39‐41]. In addition, the localisation of plaques in SSLOW hamsters was also distinct from most TSE disease models, being found predominantly in the glial limitans and microfolded astrocytic processes. The pattern of deposition of amyloid plaques in SSLOW hamsters and recPrP fibril-inoculated 101LL mice may, therefore, be due to distribution of amyloid seeds through the extracellular space following high-volume intracerebral inoculation. Indeed, distribution of plaques in 101LL mice mimics the distribution of India ink observed by others when inoculated into murine brain [71]. Other studies have shown that following injection into the brain, inocula are drained via perivascular pathways or leak into the CSF and enter the blood [72]. Once in the blood, seeds may re-enter the brain through the circumventricular organs (CVOs) due to the incomplete blood brain barrier at this site [67, 68]. The variation in pattern and intensity of plaque deposition between different recPrP preparations could, therefore, be determined by the aggregate size (due to variations in the refolding conditions) and ability to spread via the CSF, interstitial fluid or blood giving rise to deposition in further areas of the brain via the CVOs.

Our observations of plaque formation in 101LL mice in the absence of TSE disease show similarities to work performed, by others, in transgenic mice expressing human amyloid precursor protein (huAPP). Plaques can be induced in transgenic mice expressing wild-type huAPP and huAPP with familial AD mutations following inoculation with AD brain homogenate or aged huAPP Tg brain homogenate [47, 49, 71]. Initial attempts to induce seeding in AD Tgs using recombinant Aβ were unsuccessful [47]. However, recent studies have demonstrated accelerated seeding in APP23 mice expressing a GFAP-luc transgene using high levels (7.5 μg) of aggregated recombinant Aβ 1–40 monomers or mutant Aβ 1–40 dimers (AβS26C)₂ [69, 70]. Seeding of Aβ plaques may, therefore, occur via a mechanism similar to seeding of non-pathogenic PrP amyloid plaques in 101LL mice. Recent studies examining archival material from cases of iatrogenic CJD (iCJD) have also shown the probable seeding of Aβ plaques in humans following the administration of human pituitary-derived growth hormone [29] or the implantation of dura matter grafts [21, 35]. Whilst these materials appear to be the source of Aβ seeds, and responsible for the formation of Aβ plaques in brains of the recipients, other signs of AD pathology were not present. In particular, in the detailed pathological analysis of two cases of iCJD that developed following dura matter grafting described by Kovacs et al. [35], the authors state that Aβ seeding is unable to reproduce the full clinicopathological phenotype of AD.

We have shown that PrP amyloid can be seeded in the brains of 101LL mice in the absence of neurological signs of TSE disease, spongiform degeneration of the brain and TSE infectious agent replication. Importantly, we have established that such seeding can occur not only following inoculation of brain extracts from human (atypical P102L GSS) and murine (GSS-22) sources, but also from non-brain-derived recPrP fibrils. Our data provide no evidence for the generation of “infectious prions” following inoculation and subpassage of synthetic PrP amyloid fibrils; instead, we observe the induction and maintenance of a seeded proteinopathy [30, 55, 56]. The development of neurological disease in some synthetic prion models may, therefore, depend on the severity and distribution of PrP aggregates in the brain, due to the concentration, volume and distribution of seeds inoculated. Indeed, GSS22 Tg mice, which overexpress 101L PrP ~12-fold, develop neurological signs, spongiform degeneration and PrP amyloid plaques, but do not transmit TSE disease on subpassage [53]. Material from the brains of these mice can seed plaques in low-expressing (~2-fold) 101L Tg mice and knock-in 101LL mice [53, 55], indicating that high levels of plaque deposition may, indeed, lead to neurological signs, but not to an infectious TSE disease. These data have led us to hypothesise that there are three potential pathways associated with protein aggregation in the CNS; (a) resulting in an infectious TSE disease and replication of the TSE infectious agent; (b) the induction of a proteinopathy in which large accumulations of amyloid lead to toxic effects in the brain, but are not infectious; (c) the seeding of protein accumulations in the brain that are neither infectious nor toxic (as determined by absence of neurological signs and spongiform degeneration of the brain). On this basis, protein misfolding and aggregation in the brain does not invariably lead to replication of an “infectious prion”, even when the protein in question is PrP. This may explain why some prion diseases (particularly in humans) appear to be non-transmissible, such as atypical P102L GSS, A117V GSS and P105L GSS [34, 56] (P105L GSS: Barron, unpublished data). However, transmission of atypical P102L GSS has recently been demonstrated in bank voles [58] indicating that transmission of such cases may be possible in the right host species. Overall, our results and those of others where plaque formation is seeded by inoculation of pre-formed protein fibrils (or propagons [16]) are consistent with a seeded proteinopathy, in which native proteins can be converted into misfolded aggregates, but does not result in a contagious, naturally transmissible neurodegenerative disease. Hence, it may be preferable to refer to the formation of misfolded protein aggregates in APP transgenic mice and 101LL mice as “seeded proteinopathy”, rather than “prion-like transmission”, to distinguish this mechanism from a truly infectious TSE disease which is transmissible from one individual to another.