Animal and human prion diseases
The term prion was originally coined by S.B. Prusiner to denote a small proteinaceous infectious particle, which is resistant to most procedures that inactivate nucleic acids (Ref. Reference Prusiner1). Prion diseases or transmissible spongiform encephalopathies (TSEs) are fatal neurodegenerative disorders affecting humans and animals (Fig. 1). Human TSEs are often categorised with other protein misfolding neurodegenerative diseases, including Alzheimer's disease (AD), Parkinson's disease, Huntington's disease, fronto-temporal dementia and amyotrophic lateral sclerosis (Ref. Reference Prusiner2). These diseases share a common mechanism that involves a conformational change in the structure of the disease-implicated protein, leading to self-replicating propagation and subsequent pathological changes within the central nervous system (CNS). However, only TSEs are known to cause infections of epidemic proportions in humans (kuru) and animals [bovine spongiform encephalopathy (BSE)], and be endemically present in domestic (scrapie) and wild animals [chronic wasting disease (CWD)] (Fig. 1).
Animal and human TSEs.
Human TSEs can be subdivided into three aetiological groups: sporadic, genetic and environmentally acquired (i.e. infectious). Sporadic Creutzfeldt–Jakob disease (sCJD) is the most common form, occurring without any obvious cause with a frequency of 1 case per million people per year worldwide (Ref. Reference Puoti3). Sporadic fatal insomnia (sFI) is a very rare disease described in only approximately 2 dozen cases so far (Refs Reference Puoti3, Reference Moody4). The spectrum has been recently expanded to include variably protease-sensitive prionopathy, a rare sporadic condition recognised in a small number of patients, with certain features resembling Gerstmann–Sträussler–Scheinker syndrome (GSS) (Refs Reference Notari5, Reference Diack6, Reference Gambetti7, Reference Liberski and Budka8). The genetic forms represent 10–15% of all TSE cases and include familial CJD (fCJD), GSS and fatal familial insomnia (FFI) (Refs Reference Brown and Calne9, Reference Gambetti and Prusiner10). These diseases are inherited in an autosomal dominant fashion and are associated with more than 30 pathogenic mutations in the prion protein gene (PRNP) (Ref. Reference Mastrianni, Pagon, Adam, Ardinger, Wallace, Amemiya, Bean, Bird, Fong, Mefford, Smith and Stephens11). The infectious forms include variant CJD (vCJD), presumably resulting from dietary exposure to BSE (Ref. Reference Will12), iatrogenic CJD (iCJD) (Ref. Reference Brown13), and kuru, an almost extinct disease described in cannibalistic tribes of New Guinea (Refs Reference Liberski14, Reference Collinge15, Reference Gajdusek16, Reference Alpers, Prusiner and McKinley17). iCJD has been linked to therapeutic treatments with human pituitary hormones (growth hormone and gonadotropin), dura mater, cornea or pericardial grafts unknowingly sourced from CJD-afflicted individuals, and to rare neurosurgical procedures performed with inadequately decontaminated instruments previously used on CJD patients (Refs Reference Brown13, Reference Tange, Troost and Limburg18). Recently, the transmission of vCJD has been reported in four instances through the therapeutic use of nonleukoreduced red blood cell concentrates (Refs Reference Brown13, Reference Llewelyn19, Reference Gillies20, Reference Peden21, Reference Wroe22).
Each form of human TSE results in a distinctive phenotype characterised by differences in the age of onset, variability in clinical symptoms, brain pathology and disease duration, and by different PrPTSE distribution and deposition patterns [reviewed in (Refs Reference Puoti3, Reference Liberski and Budka8, Reference Gambetti and Prusiner10, Reference Collinge15, Reference Gajdusek16, Reference Alpers, Prusiner and McKinley17, Reference Montagna23)]. The molecular mechanisms of TSE phenotypic heterogeneity are not fully understood, although polymorphisms in the host PRNP gene influence the phenotypic differences and individual susceptibility to certain forms of the disease (Refs Reference Hunter24, Reference Mead25, Reference Kovacs and Budka26, Reference Collinge, Palmer and Dryden27, Reference Palmer28, Reference Palmer and Collinge29, Reference Belt30, Reference Lee31). These polymorphisms also partially explain the origin of biochemically distinct PrPTSE conformers substantiating the existence of prion strains in humans (Refs Reference Gambetti32, Reference Hill33, Reference Collinge and Clarke34) and animals (Refs Reference Sabuncu35, Reference Rigter and Bossers36, Reference Thackray37).
In animals, TSEs manifest, among other forms, as scrapie in sheep and goat, BSE (known to the general public as ‘mad cow disease’) in cattle, and CWD in cervids (Fig. 1). Animal diseases are relatively easily transmitted within the same species, but cross-species transmissions have been documented as well. BSE is the only animal form of TSE that has been causatively linked to a disease in humans, vCJD (Refs Reference Will12, Reference Hill38). For a complete review on animal prion disease, see Collinge (Ref. Reference Collinge39).
It is widely accepted that the causative agent of TSEs is the prion, which is mainly composed of a misfolded, insoluble, and proteinase K (PK)-resistant protein that is devoid of detectable informational nucleic acid, herein referred to as PrPTSE (also known as PrPSc, PrPres or PrPd) (Refs Reference Prusiner1, Reference Prusiner2, Reference Prusiner40). The exact mechanism(s) of PrPTSE generation are not fully understood. According to the current state of knowledge, the main event is the conformational change of PrPC into PrPTSE. This transition occurs under unknown circumstances and gives rise to multiple conformers exhibiting a range of strain-specific phenotypes in afflicted hosts (Ref. Reference Weissmann41). Temporal and spatial deposition of PrPTSE coincides with a series of pathological events in the brain, resulting in spongiform degeneration, neuronal loss, and gliosis, which constitute the hallmarks of TSEs (Ref. Reference Unterberger, Voigtlander and Budka42). Prions utilise several routes of infection, which determine the length of the silent incubation period in an infected host. Usually, peripheral extra-neural exposures result in incubation phases that are longer than direct intracerebral routes. Factors and mechanisms underlying prion intra- and inter-species transmission leading to neurodegeneration are still under study, but significant progress has been made in the three decades following their discovery.
PrPC function
PrPC is a sialoglycoprotein of 253 amino acids (human PrPC) encoded by a single gene. Post-translational processing in the endoplasmic reticulum (ER), results in the removal of an amino (N)-terminal signal sequence peptide (residues 1 to 22), and a carboxi (C)-terminal sequence for the attachment of a glycosyl phosphatidyl inositol (GPI) anchor to Ser-231 (Ref. Reference Stahl43). The N-terminal domain of the protein contains a repetitive sequence (residues 52–91) of eight amino acids, the so-called octapeptide repeats (PHGGGWGQ) that appear five times in most mammalian species, a neurotoxic domain or central region (CR) (residues 106–126), and a hydrophobic domain (residues 112–135). Additionally, PrP harbours a disulphide bridge linking residues Cys-179 and Cys-214, and two glycosylation sites at residues Asn-181 and Asn-197 (human PrPC numbering) (Refs Reference Donne44, Reference Altmeppen45). Nuclear magnetic resonance spectroscopy analyses of full-length recombinant murine and hamster PrPC indicate that the secondary structure of the protein consists of a globular domain (residues 126–226) containing three α-helices, two β-strands, and a short helix-like segment comprising residues 222–226, a flexible random-coiled like N-terminal tail spanning residues 23–125, and a disordered C-terminal region (residues 227–231) (Refs Reference Donne44, Reference Riek46, Reference Riek47).
The physiological role of PrPC is still under debate, and defining its cellular role is complicated by the lack of major anatomical or developmental defects observed in early studies with PrPC-null (PrP−/−) mice generated after germline genetic ablation of PrPC expression (Refs Reference Bueler48, Reference Manson49). Likewise, PrP−/− cattle produced by sequential gene-targeting showed no physiological, immunological and reproductive abnormalities (Ref. Reference Richt50). Additionally, genetically engineered Prnp +/− and Prnp −/− goats (Refs Reference Yu51, Reference Yu52), and even those that are devoid of PrPC as a result of a naturally occurring nonsense mutation (Ref. Reference Benestad53), presented normal development and behaviour. Moreover, transgenic mice generated by cell-specific targeted cre-mediated post-natal ablation of PrPC in neurons, showed no evidence of neurodegeneration or other histopathological changes for up to 15 months post-ablation (Ref. Reference Mallucci54). However, certain alterations in physiological functions were reported in some Prnp −/− models. These included sleep disturbances, distorted circadian rhythm (Ref. Reference Tobler55), and abnormalities in synaptic transmission (specifically in cognition, olfactory physiology, and behaviour) [reviewed in (Refs Reference Westergard, Christensen and Harris56, Reference Wadsworth, Asante and Collinge57)]. Furthermore, one laboratory reported age-related defects in motor coordination and balance in PrP−/− mice; importantly, impaired mice displayed spongiform changes and reactive astrocytic gliosis in the brain, which usually accompany TSE pathology (Ref. Reference Nazor, Seward and Telling58). These findings suggested a plausible role for PrPC in neuroprotection during ageing. Electrophysiological studies pointed to a role for PrPC in modulating neuronal excitability. In these studies, PrP−/− mice exhibited long-term potentiation (LTP) impairment and reduced after-hyperpolarisation currents (Refs Reference Collinge59, Reference Manson60). These findings were later confirmed in experiments involving post-natal removal of neuronal PrPC expression (Ref. Reference Mallucci54).
The existing controversy in findings describing alterations in PrP−/− mice remains highly discussed, since variations in the genetic background of PrP−/− mouse models and PrP-flanking genes, rather than PrPC absence, may account for some of the observed phenotypes (Refs Reference Striebel, Race and Chesebro61, Reference Steele, Lindquist and Aguzzi62). Overall, what appears to be an irrefutable phenotype in PrP−/− mice is resistance to infection with prions (Refs Reference Bueler63, Reference Sailer64, Reference Manson65).
Many of the putative functions of PrPC are related to its cellular localisation. PrPC is attached to the outer leaflet of the plasma membrane through the GPI anchor (Ref. Reference Stahl43). In mammals, PrPC is expressed in various cell types throughout the body, with the highest levels reported in neurons (Refs Reference Kretzschmar66, Reference Moser67, Reference Brown68, Reference Tanji69, Reference McLennan70, Reference Bendheim71, Reference Ford72). While the exact PrPC localisation in the cell is still debated, three major sites have been identified: plasma membrane, Golgi apparatus and early and late endosomes (Ref. Reference Laine73). PrPC is mainly localised in cholesterol-rich microdomains, or lipid-rafts, at the plasma membrane (Refs Reference Laine73, Reference Gorodinsky and Harris74, Reference Naslavsky75). Since lipid rafts are platforms for signal transduction processes, it has been suggested that PrPC may trigger signalling pathways inside the cell (Ref. Reference Jacobson and Dietrich76), with the resulting modulation of neuronal survival (Refs Reference Mouillet-Richard77, Reference Chen78) or neuritic outgrowth (Refs Reference Chen78, Reference Graner79). Several groups localised PrPC at the membrane of synaptic specialisations, including pre- and post-synaptic membranes, and on synaptic vesicles (Refs Reference Mironov80, Reference Fournier81, Reference Herms82, Reference Haeberle83, Reference Moya84), further supporting its proposed role in neuronal synaptic transmission regulation (Refs Reference Collinge59, Reference Manson60). Additionally, PrPC was shown to interact with several proteins involved in synaptic release (Refs Reference Spielhaupter and Schatzl85, Reference Magalhaes86) and with various ion channels (Ref. Reference Senatore87). In line with these observations, PrP expression in Drosophila resulted in synaptic vesicle optimisation and higher vesicle release efficiency, supporting a functional role for PrPC in protein signalling and synaptic plasticity (Ref. Reference Robinson88).
In addition to its subcellular localisation, PrPC has been detected in the cytosol in certain subpopulations of neurons in the hippocampus, neocortex and thalamus, but not in the cerebellum (Ref. Reference Mironov80). These neurons may play a significant role in the pathogenesis of prion diseases (Refs Reference Mironov80, Reference Ma, Wollmann and Lindquist89). Differences in cytosolic PrPC distribution have never been addressed, but it has been suggested that cytosolic PrPC may have altered propensity for aggregation. Additional studies revealed that mammalian cells contain all co-factors required for cytosolic prion propagation and dissemination (Ref. Reference Hofmann and Vorberg90). Intriguing is the fact that, unlike mammals, the expression of PrPC in the CNS of adult chickens was observed in dendrites and axons of neurons associated with certain sensory systems, but not in neuronal bodies or glial cells (Ref. Reference Atoji and Ishiguro91).
Recent studies highlight zebrafish as a useful model for studying proteins implicated in neurodegenerative diseases, including PrPC (Ref. Reference Malaga-Trillo and Sempou92). Zebrafish express high levels of duplicated PrP homologue proteins, namely PrP-1 and PrP-2, in the developing and adult brain (Refs Reference Rivera-Milla, Stuermer and Malaga-Trillo93, Reference Rivera-Milla94). Both proteins are functionally related and share similarities to mammalian PrPC in terms of domain composition, the presence of two N-glycosylation sites, and their binding to the plasma membrane via a GPI anchor (Refs Reference Malaga-Trillo and Sempou92, Reference Rivera-Milla94, Reference Miesbauer95, Reference Malaga-Trillo96). However, unlike findings in mammals, genetic silencing of PrP-1 or PrP-2 caused profound morphological defects in zebrafish embryonal development, with the knocking down of each protein affecting different stages of embryogenesis. Further studies revealed a role for PrP-1 in cell-to-cell adhesion, not only through homophilic interactions, but also by modulating the E-cadherin signalling cascade and regulating cell-to-cell communication in vivo (Refs Reference Malaga-Trillo and Sempou92, Reference Malaga-Trillo96). Moreover, interaction between Drosophila Schneider 2 cells separately expressing mouse and fish PrP resulted in cell aggregation and activation of an intracellular signalling cascade leading to the modulation of E-cadherin, suggesting that PrP trans-interactions are highly conserved and can take place across a wide range of species (Ref. Reference Malaga-Trillo and Sempou92).
Numerous putative neuroprotective functions attributed to PrPC, including cell surface signalling, antioxidant and anti-apoptotic effects have been proposed [reviewed in (Ref. Reference Roucou, Gains and LeBlanc97)]. Compelling evidence has been obtained for its role in myelination, autophagy regulation and trafficking of metal ions (Refs Reference Pauly and Harris98, Reference Mange99, Reference Bremer100, Reference Nishida101, Reference Oh102). Protein interaction analysis identified a subset of the ZIP (Zrt- Irt-like Protein) family of Zinc transporters as PrPC interaction partners. Additional sequence analysis across a wide range of species within the chordate lineage, suggested that the prion gene family is phylogenetically derived from a ZIP-like ancestral molecule, providing an explanation to the functional role of PrPC in the transmembrane transport of divalent cations (Ref. Reference Schmitt-Ulms103). PrPC is a copper (Cu2+)-binding protein that may play a role in Cu2+ homoeostasis by mediating Cu2+ transport or sequestration (Refs Reference Brown104, Reference Brown105). It has been debated whether binding of Cu2+ during PrP folding provides superoxide dismutase activity (SOD) on the protein (Refs Reference Brown104, Reference Jones106, Reference Hutter, Heppner and Aguzzi107) or if PrPC acts as an antioxidant by binding potentially harmful Cu2+ ions, quenching free radicals generated as a result of Cu2+redox cycling (Ref. Reference Davies108). A recent study performed with PrPC-deficient neuronal cells, suggested PrPC participates in anti-apoptotic and anti-oxidative processes by interacting with the stress inducible protein 1 (STI-1) to regulate SOD activation, reconciling previously discordant findings (Ref. Reference Sakudo109).
The PrPC octapeptide repeat region has limited structural similarity to the B-cell lymphoma 2 (Bcl-2) homology domain 2 (BH2) of the family of apoptosis regulating Bcl-2 proteins (Ref. Reference Roucou, Gains and LeBlanc97). Binding studies performed with the yeast two-hybrid system demonstrated a direct interaction between PrPC and the C-terminus of anti-apoptotic Bcl-2, but not of pro-apoptotic Bcl-2-associated X protein (Bax) (Refs Reference Kurschner and Morgan110, Reference Kurschner and Morgan111). In light of these findings, it was proposed that PrPC might assume the neuroprotective function of Bcl-2 proteins (Ref. Reference Roucou, Gains and LeBlanc97). Indeed, hippocampal neurons isolated from PrP−/− mice were more susceptible to serum deprivation-induced apoptosis than their wild-type counterparts, and overexpression of either PrPC or Bcl-2, rescued this phenotype (Ref. Reference Kuwahara112). Experiments with human primary neurons (Ref. Reference Bounhar113) and yeast models expressing mammalian prions (Ref. Reference Li and Harris114) provided further support to the neuroprotective effect of PrPC in pro-apoptotic Bax-mediated cell death. Expression in primary neurons of mutant PrPC harbouring a four octapeptide repeat deletion or the pathogenic substitutions T183A or D178N implicated in familial forms of TSE (Ref. Reference Bounhar113) fully or partially abolished PrPC neuroprotective function. Furthermore, Bax-mediated neuronal loss was reported in Tg(PG14) mice expressing, in analogy to a human mutation, PrPC with a nine extra octapeptide repeat insertion that perhaps lost its physiological function because of misfolding and/or protein aggregation (Ref. Reference Chiesa115). The finding that GPI attachment is not required for PrPC cyto-protective function remains unexplained (Refs Reference Bounhar113, Reference Li and Harris114). The role of PrPC in apoptosis will be further discussed.
The finding that PrP−/− mice show subtle abnormalities in immune system function, and that PrPC expression on bone marrow long-term hematopoietic stem cells seems to be important for self-renewal (Ref. Reference Zhang116), highlighted a physiological function for PrPC outside the CNS. PrPC was also implicated in T-cell cytokine response, where Prnp mRNA up-regulation and increased PrPC expression followed T-cell activation (Refs Reference Zomosa-Signoret117, Reference Ingram118), while PrPC expression ablation led to reduced induction of helper T-cell cytokine production (Refs Reference Ingram118, Reference Kubosaki119). Additional studies, suggested a role for PrPC at the immunological synapse by showing that absence of PrPC on antigen presenting dendritic cells resulted in a significant reduction of proliferative potential of responding T-cells (Ref. Reference Ballerini120).
PrPC interaction with proteins implicated in AD
The membrane localisation of PrPC places this protein in close proximity with the amyloid precursor protein (APP) and other proteins involved in amyloid-β (Aβ) processing. A putative role for PrPC in AD has been recently reviewed (Ref. Reference Gunther and Strittmatter121).
In contrast to the earlier work presented by Parkin (Ref. Reference Parkin122) indicating a protective role for PrPC, Lauren et al suggested that PrPC mediates oligomeric Aβ (oAβ) neurotoxicity. One manifestation of this role is that PrPC is required for oAβ-induced LTP impairment (Ref. Reference Lauren123), a finding later confirmed by others (Ref. Reference Barry124). Further studies provided evidence for the pathophysiological role of PrPC in mediating several other toxic effects of oAβ, including synaptic plasticity disruption (Refs Reference Barry124, Reference Freir125), axonal degeneration of serotonergic neurons, synapse loss and deficits in spatial learning and memory (Ref. Reference Gimbel126), neuronal cell death (Ref. Reference Kudo127) and synapse damage (Ref. Reference Bate and Williams128). Alternative studies assert that PrPC is not required for Aβ-dependent synaptic depression, LTP impairment (Refs Reference Kessels129, Reference Calella130, Reference Balducci131) or premature mortality, and abnormal neural network activity (Ref. Reference Cisse132). Others showed that, overexpression of PrPC prevents Aβ1–40-induced spatial learning and memory deficits by modulating programmed cell death pathways (Ref. Reference Rial133).
PrPC may play a role in neurotoxic signalling pathways by sensitising cells to toxic effects of β-sheet-rich conformers of different origins (Ref. Reference Resenberger134). Other mechanisms have been described involving additional interacting molecules. One study showed that Aβ neurotoxicity depends on interactions between Cu2+, PrPC and N-methyl-D-aspartate receptor (NMDA) (Ref. Reference You135). Others revealed that neuronal impairment occurs via Fyn activation through oAβ binding to synaptic PrPC (Ref. Reference Um136).
While discrepancies in the contribution of PrPC to AD pathogenesis may have arisen from differences in the animal models and Aβ preparations employed, all these studies suggest that the prion protein binds oAβ. Further investigations will clarify the interrelationship of both proteins and neurodegeneration.
Neurodegeneration pathways in prion disease
Owing to the temporal and spatial coincidence of PrPTSE accumulation and the appearance of the first neurodegenerative changes in the brain, it has been suggested that either the loss of a critical biological function of PrPC or the acquisition of toxic properties upon conversion into PrPTSE triggers neurodegeneration and disease.
Loss of PrPC function
PrPC has been implicated in several mechanisms leading to neuronal protection from oxidative stress or other types of pro-apoptotic insults (Refs Reference Roucou, Gains and LeBlanc97, Reference Kovacs and Budka137). Therefore, it seems logical that conversion of PrPC into PrPTSE may render the former unable to perform its normal biological function, leading to neurodegeneration. This theory is strongly challenged, however, by compelling data showing normal embryonic development and absence of major anatomical or functional phenotypes in mammals where PrPC expression has been permanently or conditionally knocked out (Refs Reference Bueler48, Reference Manson49, Reference Richt50, Reference Yu51, Reference Yu52, Reference Benestad53, Reference Mallucci54).
Gain of toxic function
PrPTSE is considered the surrogate marker of prion diseases. It accumulates in various regions of the brain, either in a diffused pattern or in the form of aggregates of different types depending on the strain of the agent and the host species (Refs Reference Collinge and Clarke34, Reference Sikorska138). The co-occurrence of PrPTSE accumulation and spongiform changes in the brains of the majority of TSE patients, and in the brains of naturally and experimentally infected animals, prompted investigators to attribute a toxic function to PrPTSE. This assumption was supported by in vitro studies reporting neurotoxicity of micromolar concentrations of short PrPC peptides, encompassing residues 106–126 of human PrPC (Ref. Reference Forloni139), and nanomolar concentrations of purified PrPTSE (Ref. Reference Hetz140).
While these data suggest a direct involvement of PrPTSE as the toxic agent in TSE pathogenesis, several lines of evidence do not support a causative relationship. Some forms of TSE, including FFI, have very restricted brain pathology with little or no spongiform change or detectable PrPTSE (Ref. Reference Budka141). PK-resistant PrPTSE has been identified in non-CJD brains suggesting that the accumulation of PrPTSE is probably not neurotoxic (Refs Reference Zou and Gambetti142, Reference Yuan143). Dissociation between PrPTSE and neurodegeneration has also been reported in several cases of natural and experimental human prion disease, where neuropathological changes occurred in the absence or with very limited PrPTSE accumulation (Refs Reference Hsiao144, Reference Tateishi145, Reference Hayward, Bell and Ironside146, Reference Collinge147). A number of studies have reported no correlation between PrPTSE deposition and neurodegeneration in the brains of infected mice expressing half the normal levels of PrPC (Refs Reference Sandberg148, Reference Chesebro149, Reference Bueler150, Reference Sandberg151). Bueler and colleagues showed that although these mice accumulated PrPTSE and infectivity titres similar to those found in clinically sick, wild-type animals, they had a significant delay in disease onset and progression indicative of relative resistance to the toxic effects of prions (Ref. Reference Bueler150). Likewise, it is well documented the existence of subclinical carriers of prion infection, with wild-type mice living a typical lifespan despite harboring PrPTSE titers similar to mice at the end-stage of the disease (Refs Reference Hill and Collinge152, Reference Race153, Reference Thackray154, Reference Thackray, Klein and Bujdoso155, Reference Asante156, Reference Hill157). Compelling experimental evidence against the direct neurotoxic effect of PrPTSE was first obtained in the seminal studies by Aguzzi and colleagues (Refs Reference Brandner158, Reference Aguzzi159) who showed that intra-cerebral injection of prions in PrPC-null mice with grafted neural tissue from a mouse overexpressing PrPC, resulted in severe spongiform changes, infectivity, and PK-resistant PrPTSE in the grafts. However, while in some instances PrPTSE was found in brain areas outside the graft, these regions were spared from neurodegenerative changes (Refs Reference Brandner158, Reference Aguzzi159). Work by Mallucci et al. (Ref. Reference Mallucci160) provided additional proof against a direct role of PrPTSE in neurotoxicity. By specifically knocking down PrPC expression in mouse neurons after established prion neuroinvasion, they showed reversion of early spongiform changes and prevention of disease progression, despite continuous accumulation of PrPTSE in the neuropil (the latter coming from the conversion of PrPC produced in nonneuronal cells) (Ref. Reference Mallucci160).
Multiple studies suggested that soluble oligomeric species, intermediates in the formation of protease-resistant PrPTSE, are the pathogenic entity rather than insoluble PrPTSE aggregates (Refs Reference Kazlauskaite161, Reference Masel, Genoud and Aguzzi162, Reference Haass and Selkoe163, Reference Weissmann164, Reference Simoneau165, Reference Novitskaya166). The idea has been further developed that PrPTSE replication and neurotoxicity occur in two separate phases, where the production of a neurotoxic species (PrPL) is catalysed by PrPTSE. According to this model, when PrPTSE amplification saturates, the autocatalytic production of infectivity (phase 1) switches into a toxic pathway (phase 2) leading to the formation of toxic PrP species. Importantly, the formation of PrPL is linearly dependent on PrPC concentration (Refs Reference Collinge and Clarke34, Reference Sandberg148, Reference Sandberg151).
Subversion of PrPC function
The experimental observation that PrPC expression at the neuronal membrane is required to transduce a toxic signal inside the cell (Refs Reference Mallucci54, Reference Brandner167) established the molecular basis of a new mechanism to explain neurotoxicity in prion diseases: subversion of PrPC function (Ref. Reference Harris and True168), whereby interaction between PrPC and PrPTSE, or intermediate species, subverts or modifies the normal function of PrPC, triggering a toxic signal inside the cell (Fig. 2). While the prevailing data argue for the necessity of neuronal PrPC expression for neurotoxicity, some challenging evidence still exists (Refs Reference Mallucci54, Reference Chesebro149, Reference Simoneau165, Reference Novitskaya166, Reference Brandner167, Reference Chesebro169, Reference Zhou170). Studies by Zhou et al. identified a monomeric, highly α-helical form of PrPC, the so-called toxic PrP (TPrP), as the most neurotoxic PrP species in vitro and in vivo. TPrP was generated in vitro by size fractionation following dilution refolding of full-length mouse recombinant PrPC, and it was shown to elicit autophagy, apoptosis and a molecular signature similar to that observed in the brains of prion-infected animals (Ref. Reference Zhou170). TPrP was toxic to PrP−/− mouse-derived hippocampal neurons in vitro, supporting the hypothesis that endogenous neuronal PrPC is not required to propagate a toxic signal inside the cell (Ref. Reference Zhou170). However, even if TPrP toxicity does not rely on membrane-bound PrPC expression, experimental evidence suggests that it should be generated within neurons, given that post-natal ablation of PrPC expression in these cells reverses early neurodegenerative changes and prevents disease progression in mice, even though glial replication and accumulation of PrPTSE continues (Ref. Reference Mallucci54). Indeed, growing evidence indicates that prion-induced pathology comprises cell-autonomous mechanisms, resulting in cellular dysfunction and neurodegeneration, and noncell-autonomous processes leading to prion spread (Ref. Reference Halliday, Radford and Mallucci171).
Putative molecular mechanisms of prion-induced neurodegeneration. Schematic representation of different mechanisms by which PrPC misfolding and PrPTSE accumulation may result in cellular death and neuronal damage. PrPC synthesis: solid grey double lines. Subversion of function/Excitotoxic stress: dotted grey line. The mitochondrial pathway of apoptosis: solid black line. Endoplasmic reticulum stress-induced apoptosis: dotted black double lines. Endoplasmic reticulum stress: dashed black line. The preemptive quality control (pQC) pathway: dotted grey double lines. The unfolded protein response, IRE-1 arm: dotted black line. The unfolded protein response, PERK arm: solid black double lines. The unfolded protein response, ATF-6 arm: dash-dotted black line. The ubiquitin–proteasome system (UPS) and the aggresome: dash-dotted grey line. Normal PrPC degradation by the proteasome: dash-dotted black double lines. NAD+ starvation: solid grey line. Cyt. C: Cytochrome C, E1: Ubiquitin activating enzyme, E2: Ubiquitin conjugating enzyme, E3: Ubiquitin protein ligase. 
Activates route. 
Inhibits route.
Studies performed with Tg44 transgenic mice expressing PrPC that lacks the GPI membrane anchoring signal (GPI−-PrPC) provided additional supporting evidence to the dissociation between prion replication and neurotoxicity (Refs Reference Chesebro149, Reference Chesebro169, Reference Trifilo172). In anchorless-PrPC transgenic mice, about 90% of GPI−-PrPC is secreted, the rest appearing in the ER and Golgi complex, but not on the plasma membrane. Scrapie infection of Tg44 mice resulted in a substantially different disease than that observed in wild-type mice, with an incubation period, clinical signs and neuropathological abnormalities characteristic of cerebral amyloid angiopathy (CAA) (Refs Reference Chesebro149, Reference Chesebro169, Reference Klingeborn173). Spongiform grey matter degeneration was minimal or not present in the brains of diseased mice, and this feature was preserved through subsequent multiple passages in Tg44. However, the brains of these mice contained large deposits of amyloid PrPTSE and high levels of infectivity that caused classical grey matter spongiosis with PrPTSE accumulation in wild-type mice (Ref. Reference Chesebro169). Experimental studies involving peripheral scrapie infection of Tg44 mice revealed a crucial role for membrane-bound PrPC in neuroinvasion and neuronal PrPTSE spread (Ref. Reference Klingeborn173), and for the induction of a typical TSE pathogenic process (Ref. Reference Chesebro149). Interestingly, whereas the neuropathogenic processes found in Tg44 mice i.c. injected with prions were distinct from typical prion disease, they were reminiscent of changes found in familial forms of TSEs associated with STOP mutations at codons 145, 163 and 226 of PRNP (Refs Reference Ghetti174, Reference Revesz175, Reference Jansen176). These changes involved dense amyloid PrPTSE plaque deposits with CAA, but without grey matter spongiosis.
Because prion diseases are a group of diverse disorders characterised by different disease phenotypes and different pathological features (Ref. Reference Gambetti177), it is likely that neuropathogenesis in these disorders is triggered by different mechanisms, the majority of which depend on PrPC anchoring to the neuronal membrane. In typical prion disease, neuronal PrPC is required for neuroinvasion and for PrPTSE-mediated neurotoxic membrane interactions (Ref. Reference Chesebro149). Neurotoxicity independent of neuronal PrPC expression was observed in Tg44 mice after i.c. injection of scrapie, where neuropathogenesis was likely the result of tissue distortion by amyloid plaques, obstruction of interstitial fluid flow, and vascular occlusion triggered by the accumulation of PrPTSE amyloid within basement membranes and interstitial space between neurite and glial processes. Proteomic analysis of CAA and nonamyloid TSE disease phenotypes in mice revealed similarities and differences in the mechanism of pathogenesis (Ref. Reference Moore178). Following scrapie infection with the scrapie strain RML, the brains of wild-type and Tg44 mice showed evidence of a neuroinflammatory response and complement activation. However, ER-associated degradation (ERAD) and mitochondrial induced apoptosis pathways were implicated only in wild-type animals exhibiting nonamyloid disease phenotype, whereas metal binding and synaptic vesicle transport were more profoundly disrupted in Tg44 mice with PrPTSE-CAA accumulation (Ref. Reference Moore178). A similar unique mechanism may be responsible for the phenotypic differences observed in certain forms of human TSE (Ref. Reference Chesebro149).
Molecular and cellular mechanisms of neuronal death
With just few exceptions, TSEs are characterised at the neuropathological level by various degrees of spongiform vacuolation of the neuropil, accompanied by neuronal cell loss and gliosis, which together constitute the classic neuropathological triad of TSEs (Ref. Reference Unterberger, Voigtlander and Budka42). Ultrastructural studies revealed that typical ‘spongiform vacuoles’ develop within neuronal elements or within myelinated axons or myelin sheaths. The origin of these vacuoles is still debated, but the prevailing view attributes their occurrence to autophagy rather than to abnormalities in membrane permeability leading to increased water retention (Refs Reference Kovacs and Budka137, Reference Budka179).
Autophagy in prion diseases
Basal autophagy plays an important role in maintaining cell homeostasis and physiological function. It is characterised by the formation of cytoplasmic autophagic vacuoles that fuse with lysosomes to degrade and recycle the vesicular content. This is a tissue-specific, tightly regulated process mediated via the lysosomal degradation pathway. Alternatively, autophagy induced by various cellular insults is recognised as one of the three mechanisms of programmed cell death in eukaryotes (Ref. Reference Klionsky180).
Autophagic vacuoles have been identified in several neurodegenerative diseases, including various forms of TSEs (Refs Reference Heiseke, Aguib and Schatzl181, Reference Rubinsztein182, Reference Boellaard, Schlote and Tateishi183). Since, under normal conditions, autophagy takes place at low levels in the CNS an increase in the number of autophagosomes in prion-infected brains was interpreted by some as cause of neurodegeneration (Refs Reference Zhou170, Reference Liberski, Gajdusek and Brown184, Reference Liberski185). However, based on recent studies providing evidence that PrPC may play a role in autophagy regulation in neurons, and since the absence of PrPC expression resulted in autophagy up-regulation (Ref. Reference Oh102), it can be argued that loss or subversion of PrPC function during prion infection may trigger a similar response.
Although some authors consider autophagosomes to be part of the mechanism leading to neuronal death (Refs Reference Zhou170, Reference Liberski, Gajdusek and Brown184), a debate continues as to whether they play a neuroprotective role through degradation of intraneuronal deposits of PrPTSE. Early studies showed impaired PrPTSE aggregation and protection against oxidative damage following trehalose treatment of prion-infected cells (Ref. Reference Beranger186). Schatzl and colleagues advanced these observations and provided direct in vitro evidence of autophagy-induced PrPTSE degradation (Ref. Reference Heiseke, Aguib and Schatzl181). Pharmacological treatment of prion-infected neuronal and non-neuronal cells, with the autophagy-inducing agents imatinib, rapamycin, lithium and trehalose, increased cellular clearance of PrPTSE (Refs Reference Aguib187, Reference Heiseke188, Reference Ertmer189). Early administration of imatinib after peripheral inoculation of prions in mice delayed disease onset and accumulation of PrPTSE in the CNS (Ref. Reference Yun190). However, neither intraperitoneal nor intracerebroventricular delivery of the drug showed beneficial effect on PrPTSE clearance in the CNS (Ref. Reference Yun190). Likewise, the survival of mice i.p. infected with scrapie was not influenced by trehalose treatment despite the delayed appearance of PrPTSE in the spleen (Ref. Reference Aguib187).
Recent observations derived from the pharmacological manipulation of autophagy with either autophagy-enhancing or -inhibiting drugs showing no changes on the time course or amplitude of neuronal death in response to TPrP exposure, strongly suggested that the observed autophagy in protein misfolding diseases is a secondary mechanism in the neurodegenerative process (Ref. Reference Zhou191). Consistent with this idea are in vivo studies where Tg(PrP-A116V) mice, a model of GSS, were chronically treated with i.p. injections of rapamycin. Drug treatment led to a dose-dependent delay in disease onset, a reduction in symptom severity, and improved survival concomitant with increased levels of the autophagy-specific marker LC3-II (microtubule-associated protein 1A/1B-light chain 3-phosphatidylethanolamine conjugate), reduced levels of insoluble PrP-A116V, and a near to complete absence of PrP amyloid plaques in the brain. However, despite the reported reduction in amyloid burden, these mice eventually reached terminal levels of motor impairment and succumbed to the disease (Ref. Reference Cortes192).
Role of apoptosis in TSEs
Not all histopathological changes identified in TSE-affected brains can be attributed to the activation of cellular autophagy. Indeed, several studies conducted in natural disease and experimental models of TSEs observed DNA fragmentation and the activation of several caspases, indicating that neuronal death may occur via apoptotic pathways (Refs Reference Giese193, Reference Lucassen194, Reference Jesionek-Kupnicka195).
The mitochondrial pathway
This was identified in experimental models where aggregated PrP peptides, like PrP106-126, or recombinant mutant PrP were used (Refs Reference Pillot196, Reference O'Donovan, Tobin and Cotter197, Reference Lin198). These toxic forms of PrP caused alterations in the mitochondrial membrane leading to mitochondrial stress, cytochrome c release, caspase activation, and ultimately, neuronal death (Fig. 2, solid black line). Early findings suggested that cerebellar granule neurons (CGNs) in Tg(PG14) mice died via a Bax-dependent process (Refs Reference Chiesa115, Reference Chiesa199). However, crossing these animals with Bax−/− mice had no effect on disease onset and duration. Additionally, despite Bax inactivation significantly inhibited apoptotic death of CGNs, it did not rescue synaptic degeneration and did not prevent neurological disease. Instead, these findings supported synaptic degeneration, and not apoptotic neuronal loss, as the primary pathologic event contributing to the clinical signs observed in this model (Ref. Reference Chiesa115). Experiments with Bax deficient mice provided additional evidence against the primary role of proapoptotic Bax in neuronal cell death. Inoculation of Bax−/− and wild-type control mice with mouse-adapted BSE prions showed no differences in terms of PrPTSE accumulation, neurodegeneration, disease onset and clinical signs, providing compelling evidence that Bax-mediated cell death was not involved in the pathological mechanism induced by BSE (Ref. Reference Coulpier200). Nevertheless, cleaved caspase-3 and -9 were found in the brains of Bax−/− mice, suggesting that apoptosis may occur through an alternative mechanism in TSEs of infectious origin. These observations were consistent with previous findings of apoptotic features in the brains of wild-type mice infected with RML, in the absence of Bax upregulation (Ref. Reference Siso201). More recently, proteomic analyses of wild-type mice injected with the same agent revealed upregulation of proteins involved in cell death and survival; in particular, in the levels of proteins associated with the mitochondrial inner membrane, proteins associated with the ubiquitin/proteasome pathway, and proteins involved in the ERAD (Ref. Reference Moore178), but not in the brains of RML-infected Tg44 mice. This evidence further supported the notion that distinct pathways may be activated in TSEs of different aetiology.
The endoplasmic reticulum pathway
ER stress has been recently discovered as a novel apoptotic-regulatory pathway, with implications in β-amyloid cytotoxicity (Ref. Reference Nakagawa202). Stress of the ER results from changes in Ca2+ homoeostasis or accumulation of aggregated proteins. Either situation will induce Ca2+ release from the ER and activation of the ER-membrane-resident caspase-12. Following activation, caspase-12 is released into the cytoplasm where it activates downstream caspases of the apoptotic response (Fig. 2, dotted black double lines). Additionally, Ca2+ regulates calcineurin (CaN), a type 2 phosphatase involved in synaptic function, memory and cell death (Ref. Reference Mansuy203).
Dysregulation of intracellular Ca2+ balance has been described in a number of neurodegenerative proteinopathies, including TSEs (Ref. Reference Hetz140). Early studies identified ER stress and caspase-12 activation in murine neuroblastoma (N2a) cells infected with highly purified prion preparations. Similar changes were observed in prion-infected mice as well as in brains of patients affected with vCJD and sCJD (Ref. Reference Hetz140). But, even if these findings initially suggested the involvement of a caspase-12-dependent apoptotic pathway in naturally occurring prion diseases, infectivity studies with caspase-12 knockout and wild-type mice revealed identical disease onset and progression, irrespective of caspase-12 expression, arguing against its role in neurotoxicity (Ref. Reference Steele204).
Further investigations on the role of ER stress and Ca2+ homoeostasis in neurodegeneration suggested that PrPTSE accumulation caused synaptic dysfunction and neuronal death via CaN activity dysregulation (Ref. Reference Soto and Satani205). This hypothesis stemmed from in vitro studies with N2a cells that were either treated with brain-derived PrPTSE, or engineered to overexpress mutant PrP forms associated with familial TSEs. Thus, pharmacological manipulation of Ca2+ homoeostasis in scrapie-infected N2a cells led to ER stress (Ref. Reference Torres206). Moreover, treatment of cells with brain-derived PrPTSE induced upregulation of the unfolded protein response (UPR)-responsive chaperones glucose regulated protein (Grp) 58, Grp78, and Grp94, which was indicative of ER stress. Additionally, overexpression of SERCA (ER-Ca2+ ATPase) in cells made them highly susceptible to PrPTSE-induced cell death. The study of N2a cells overexpressing murine PrP carrying mutations associated with FFI (PrPD177N/Met128) (murine PrP numbering) or GSS (PrPPG14), or neurotoxic transmembrane forms, showed decreased ER Ca2+ content upon treatment with Ca2+ agonists (Ref. Reference Torres206). According to this model, hyperactivation of CaN, because of increased cytosolic levels of Ca2+, results in the dysregulation of the pro-apoptotic molecule Bcl-2-associated death promoter (Bad), and the transcription factor cAMP response element-binding (CREB), among other targets. Upon dephosphorylation, Bad interacts with Bax causing mitochondrial stress and apoptosis. Dephosphorylated CREB cannot translocate into the nucleus, where it regulates the transcription of proteins involved in synaptic plasticity, resulting in synaptic degeneration (Ref. Reference Mukherjee207) (Fig. 2, dashed black line). Consistently, disruption in the expression of proteins involved in Ca2+ homoeostasis and synaptic vesicle transport were also found in Tg44 mice with CAA (Ref. Reference Moore178).
Other mechanisms of neurodegeneration
The studies described above suggest that autophagy and apoptosis are rather a secondary consequence of prion-induced neurodegeneration. To better understand the mechanism behind this process many laboratories have hypothesised alternative pathways triggering neurodegeneration based on other pathophysiological changes observed in the cell.
Excitotoxic stress
Important information about PrPC function and prion neurotoxicity has been gathered from the analysis of phenotypic variations produced by several PrPC deletion mutants in transgenic mice. It has been repeatedly shown that deletions within the N-terminal half of PrPC cause massive neuronal death in transgenic mice (Refs Reference Shmerling208, Reference Li209, Reference Baumann210). This phenotype was rescued in a dose-dependent manner by co-expression of full-length PrPC. Therefore, it has been suggested that wild-type and truncated PrPC bind to a common molecular target eliciting two different responses in the cell: while truncated-PrPC binding triggers a toxic signal, binding of wild-type PrPC restores a nontoxic physiological function (Ref. Reference Solomon, Schepker and Harris211) (Fig. 2, dotted grey line).
Transgenic mice expressing PrPC harbouring a deletion within the highly conserved CR of the protein, residues 105–125 (PrP∆CR) (murine PrP numbering), have been generated in an attempt to define the PrPC sequence determining neurotoxicity (Ref. Reference Li209). When PrP∆CR was expressed on the Prnp −/− background, mice died within the first week of life, whereas co-expression of wild-type PrPC alleviated the phenotype in a dose-dependent fashion. Neuropathological examination of mice co-expressing PrP∆CR and one Prnp allele revealed dramatic degeneration of CGNs and vacuolation of white matter regions, features observed in TSEs (Ref. Reference Christensen212). Biochemical and morphological analyses could not attribute CGN death to apoptosis or autophagy, since no activation of caspases 3 or 8, or increased levels of the autophagy marker LC3-II were detected in these animals. Instead, degenerating CGNs displayed a cytoplasmic morphology reminiscent of certain forms of excitotoxic neuronal death characterised by heterogeneous condensation of the nuclear matrix without formation of discrete chromatin masses (Ref. Reference Christensen212). The same morphology was present in the neurons of transgenic mice expressing a different deletion mutant, PrP∆32–134. These findings suggested common neurotoxic mechanisms for PrPC proteins missing the CR (Ref. Reference Christensen212) (Fig. 2).
The preemptive quality control (pQC) system
To further explain the necessity of a continuous supply of PrPC for neurotoxicity, Rane and colleagues, advancing an original idea by Orsi et al. (Ref. Reference Orsi213), proposed that the alteration of the cellular metabolism resulted in the synthesis of a neurotoxic form of PrPC (Ref. Reference Rane214). Given the implication of this pathway in PrPTSE pathogenesis (Ref. Reference Hetz140), the authors highlighted chronic ER stress as an example of altered cellular metabolism, and suggested the existence and involvement of the pQC pathway (Refs Reference Rane214, Reference Kang215).
There are three mechanisms by which proteins may be degraded in the cytosol: failed targeting, retrotranslocation and pQC. However, only the pQC pathway is activated under ER stress. There is no evidence that PrPC is synthesised in the cytosol or is retrotranslocated from the ER during normal and stress conditions (Ref. Reference Drisaldi216). In light of these data, it has been suggested that PrPTSE accumulation triggers ER stress. Consequently, PrPC is no longer translocated into the ER but rapidly degraded by the pQC pathway. Persistent, accelerated routing of PrPC through the pQC results in neuronal damage and corresponding clinical symptoms by an unknown mechanism (Ref. Reference Rane214) (Fig. 2, dotted grey double lines).
Since the model did not fully recapitulate all clinical and histopathological changes found in prion diseases, it was proposed that several pathways lead to neurodegeneration in these disorders, including increased production of transmembrane forms of PrP (CtmPrP), decreased proteasome activity, or generation of a toxic intermediate during prion conversion (Ref. Reference Rane214).
The UPR
UPR activation is a cellular response to reestablish homeostasis by synthesis of properly folded proteins. It affects the expression of chaperones, enhances degradation of unfolded and mutated proteins, and inhibits protein synthesis (Ref. Reference Walter and Ron217). When the concentration of unfolded proteins in the lumen reaches a threshold, a set of intracellular signal transduction pathways involving three transmembrane ER-resident signalling components, namely inositol requiring enzyme 1 (IRE1), activating transcription factor 6 (ATF6) and double-stranded RNA-activated protein kinase (PKR)-like ER kinase (PERK), is activated. Prolonged activity of the UPR is indicative of chronic ER stress and results in cellular apoptosis (Refs Reference Walter and Ron217, Reference Lindholm, Wootz and Korhonen218).
IRE1, a bifunctional transmembrane kinase-endoribonuclease protein, initiates the nonconventional splicing of mRNA to propagate the UPR signal (Refs Reference Walter and Ron217, Reference Hetz219). Activated IRE1 facilitates generation of an active form of the X-box binding protein 1 (XBP-1), a UPR-specific transcription factor that enhances the transcription of UPR genes involved in protein quality control, and activates ERAD among other targets (Ref. Reference Teske220) (Fig. 2 dotted black line). While early in vitro studies in N2a cells indicated a protective role for UPR activation against PrPC misfolding under ER stress mediated by XBP-1 (Ref. Reference Hetz219), infection of conditional XBP-1 knockout mice with prions demonstrated that ablation of this gene had no effect on neuronal function and prion pathogenesis in vivo, providing compelling evidence against the hypothesised neuroprotective role of the XBP-1 branch of the UPR (Ref. Reference Hetz, Castilla and Soto221).
Later studies in PrPC overexpressing Tg37 MloxP transgenic mice explored the involvement of the arm of the UPR responsible for translational control. This mechanism is directed by PERK phosphorylation of the ubiquitous eukaryotic translation initiation factor 2 (eIF2α) (Fig. 2, solid black double lines) (Ref. Reference Moreno222). Upon ER stress, PERK aggregates and phosphorylates itself and eIF2α (PERK-P and eIF2α-P). Phosphorylation inactivates eIF2α, and inhibits mRNA translation, therefore reducing the protein load in the ER to alleviate ER-stress (Ref. Reference Walter and Ron217). However, sustained elevation of eIF2α-P leads to a decline in global translation rates and loss of synaptic proteins, contributing to neuronal death. Importantly, increased levels of PERK-P and eIF2α-P were detected throughout the course of the disease in RML-infected Tg37 mice (Ref. Reference Moreno222). Reduction of eIF2α-P levels by lentiviral-induced overexpression of GADD34, an eIF2α-P-specific phosphatase, or inhibition of PrPC expression by anti-PrP shRNA, rescued the pathologic phenotype and significantly increased mouse survival (Ref. Reference Moreno222). In contrast, pharmacological inhibition of eIF2α-P dephosphorylation further enhanced neurotoxicity and significantly accelerated the disease (Ref. Reference Moreno222). These data strongly supported the conclusion that chronic UPR stress, with persistent expression of eIF2α-P and continuing inhibition of protein synthesis, leads to synaptic failure, spongiosis and neuronal loss in TSEs (Ref. Reference Moreno222).
The ubiquitin–proteasome system (UPS) and the aggresome
Ubiquitination is an efficient system for targeting cellular proteins for degradation. However, impairment of the proteasomal machinery can result in atypical ubiquitination of proteins and accumulation of ubiquitinated proteasomal substrates. Aberrations in the regulation of the UPS have been associated with a wide range of neurodegenerative diseases (Refs Reference Ciechanover and Brundin223, Reference Ma and Lindquist224). The 26S proteasomal complex comprises a 20S proteolytic core complexed at one or both ends with a 19S regulatory complex. The 20S proteasome is composed of 14 α and 14 β subunits arranged in four stacked rings, the two outer rings consisting of 7 α units and the two inner rings of 7 β subunits each. This organisation gives the 20S proteasome the appearance of a hollow barrel. Under physiological conditions, a small fraction of PrPC undergoes ubiquitination and proteasomal degradation (Ref. Reference Yedidia225). Positive ubiquitin staining of proteins was found in brains of CJD patients and experimental scrapie-infected mice (Refs Reference Lowe226, Reference Ironside227). In humans afflicted with CJD, extensive aggregated deposits of PrPTSE frequently co-localise with ubiquitin, while fine granular deposits do not appear to be ubiquitinated (Ref. Reference Kovacs228). The notion that in scrapie-infected mice protein ubiquitination increased during infection while proteasomal activity declined, led to the suggestion that ubiquitination occurs after the formation of protease-resistant PrPTSE (Ref. Reference Kang229) (Fig. 2, dash-dotted grey line). Studies using scrapie-infected N2a cells showed that PrPTSE co-localised with the cytosolic marker heat shock cognate protein 70 (Hsc70) (Refs Reference Kristiansen230, Reference Kristiansen231) and that mild proteasome inhibition led to PrPTSE accumulation in the form of large cytoplasmic perinuclear aggresomes (Ref. Reference Kristiansen230). In the aggresome, PrPTSE co-localised with 20S proteasome and ubiquitin, and PrPTSE aggresome formation was associated with caspase-8 and -3 activation and apoptosis (Ref. Reference Kristiansen230), hence confirming previous evidence suggesting caspase-3 activation in affected brain areas of scrapie-infected mice (Refs Reference Siso201, Reference Jamieson232). Later studies provided strong evidence that accumulation of PrPTSE in the cytosol inhibits the UPS, and partially explained the molecular mechanisms behind this inhibition. Thus, in N2a cells stably expressing the fluorogenic proteasome reporter substrate UbG76V-GFP, scrapie infection led to increased levels of the reporter protein, while curing cells of infection with the anti-PrP monoclonal antibody ICMS18 normalised substrate degradation (Ref. Reference Kristiansen231). Aggregated β-sheet rich PrP (β-PrP) and small PrPTSE oligomers specifically inhibited the proteolytic activities, namely chymotrypsin-like and caspase-like activities, of the 26S proteasome beta subunits as a result of decreased gate opening of the 20S particle, but not via dissociation of the 26S proteasome (Ref. Reference Kristiansen231). Supporting evidence was obtained in vivo with transgenic mice expressing the UbG76V-GFP reporter protein. These mice displayed diminished capacity to degrade the reporter protein, which accumulated as aggregated deposits in the cytosol of neurons in the most affected brain regions during scrapie infection (Ref. Reference Kristiansen231). Importantly, the recent identification of changes in the relative levels of heat shock protein HSPA5 and chaperonin TCP1 in mice infected with RML, is consistent with the involvement of the UPS in prion neurodegeneration (Ref. Reference Moore178).
In sharp contrast are findings in mouse models of inherited prion disorders (Ref. Reference Quaglio233). Double-Tg mice, co-expressing PrP with a nine-octapeptide-repeat insertional mutation (PG14) and the UbG76V-GFP reporter, displayed no evidence of UPS impairment despite showing cytosolic PrP accumulation. Pharmacological induction of cytosolic PrP accumulation in hippocampal neurons from UbG76V-GFP and wild type mice was not accompanied by changes in proteasome activity. No difference in disease onset and progression was reported as compared to Tg(PG14) mice that do not express UbG76V-GFP. In addition, proteasome activity was not altered in primary CGN isolated from Tg mice expressing the D177N/128V mutation (murine PrP numbering). A similar mutation in humans is associated with fCJD (Ref. Reference Quaglio233).
A genetic basis for the role of the UPS in TSEs was found in a recent study showing association between the haplotypes of the HECTD2 gene, encoding an E3 ubiquitin ligase, and the susceptibility to sCJD, vCJD and kuru in humans, and the length of the incubation time in mice (Ref. Reference Lloyd234). Genotype-associated differential expression of Hectd2 mRNA was also reported in mouse brains and human lymphocytes, with a significant up-regulation of Hectd2 expression in mice during the course of prion disease (Ref. Reference Lloyd234).
While different mechanisms of neurodegeneration may occur in inherited TSEs and prion forms acquired by infection (Refs Reference Jeffrey235, Reference Chiesa236), further research is needed to reconcile the existing discrepancies. In this line, of particular relevance are the findings derived from the proteomic analysis of RML-infected wild-type and Tg44 mice described above (Ref. Reference Moore178).
Nicotine adenine dinucleotide (NAD+) starvation
Following up on the empirical observation that changes in the composition of culture medium delayed neuroblastoma cell death after exposure to TPrP, Zhou and colleagues identified NAD+ starvation as a novel mechanism of neuronal death in protein misfolding neurodegenerative diseases (Fig. 2, solid grey line) (Ref. Reference Zhou191). NAD+ levels were significantly reduced in neuroblastoma cells three days following TPrP treatment, but not after exposure to nontoxic monomeric PrPC. Such an effect was normalised by the addition of NAD+ or the NAD+ precursor nicotinamide. Since NAD+ is a co-enzyme critical for energy production, redox homeostasis, Ca2+ signalling, and post-translational modifications, the authors underlined NAD+ starvation and related metabolic failure as the most likely primary cause of neuronal death. NAD+ levels were elevated in astrocytes, a finding consistent with the previously observed resistance to TPrP toxicity (Ref. Reference Zhou170). Because reactive astroglyosis is a well defined characteristic of prion pathology with astrocytes replacing dying neurons (Ref. Reference Lasmezas237), it was suggested that elevated astrocytic NAD+ levels compensated the reduced neuronal NAD+quantities, therefore explaining the unchanged levels of total NAD+ observed in prion afflicted mouse brains (Ref. Reference Zhou191). Further investigations demonstrated that NAD+ depletion triggered autophagy activation, probably as a result of limited ATP supply, suggesting that autophagy, similarly to apoptosis, is a secondary mechanisms of neurodegeneration in prion diseases (Ref. Reference Zhou191).
Therapeutic approaches
At present, no efficient therapies are available to treat prion diseases (Ref. Reference Trevitt and Collinge238). Therapies can be directed to various targets: PrPC, PrPTSE, PrPC to PrPTSE conversion, and specific neurodegenerative pathways. To date, most approaches are based on inhibiting PrPTSE accumulation. Although targeting PrPTSE is a potentially powerful approach, it has some drawbacks. Since no pre-symptomatic tests are currently available to diagnose prion diseases, these treatments are administered during the clinical phase, after PrPTSE accumulation has long been established. However, it is well known now that loss of synapses and dendrites followed by synaptic dysfunction are early events in prion neurodegeneration preceding PrPTSE accumulation. Therefore, therapies aiming at removing PrPTSE aggregates may provide little benefit to clinically sick patients. Moreover, in light of the prevalent view that oligomeric rather than fibrillar accumulations of PrPTSE are the neurotoxic species, breaking up PrPTSE aggregates may exacerbate or even prolong the disease.
Alternatively, targeting PrPC offers a number of advantages. PrP-null mice, goats, and cattle are refractory to prion infection and live normal lives (Refs Reference Bueler48, Reference Richt50, Reference Yu51, Reference Yu52, Reference Benestad53, Reference Bueler63). Likewise, knocking out neuronal PrPC half-way through the incubation period prevents the development of prion disease (Ref. Reference Mallucci160). Unfortunately, transgenic strategies to eliminate PrPC expression are far from being applicable to human therapy at the present time. Instead, it is possible to screen for, or design, pharmaceutical inhibitors of PrPC biogenesis. Notwithstanding, uncertainty arises from zebrafish studies highlighting the importance of PrPC in cell-to-cell adhesion, and until the observed phenotypes in zebrafish and mammalian knock-out models are further explained, approaches targeting PrPC expression should be considered with caution. In addition, PrPC has been shown to negatively regulate BACE-1 (β-site APP cleaving enzyme 1), and increased levels of Aβ1–40 and Aβ1–42 have been found in PrP null mice, in mice harboring mutations in the Prnp gene associated with some forms of fCJD and GSS, as well as in mice infected with various strains of PrPTSE (Ref. Reference Parkin122). Thus, removal of PrPC or interfering with its physiological function may contribute to the development of AD.
In light of recent observations (Ref. Reference Moreno222), targeting the UPR appears as a very attractive approach. Reduction of eIF2α-P levels by lentiviral-induced overexpression of an eIF2α-P specific phosphatase, rescued the pathologic phenotype associated with sustained translational repression during ER stress, and significantly increased mouse survival in scrapie infected Tg37 transgenic mice (Ref. Reference Moreno222). Moreover, pharmacological modification of the UPR by selectively inhibiting PERK phosphorylation and activity upstream of eIF2α, prevented translational repression with subsequent neuroprotection. The concomitant pancreatic toxicity observed in treated mice, which led to significant body weight loss and a mild increase of glucose levels in blood, questioned the clinical application of UPR inhibitors and stimulated the search for new compounds with reduced toxicity (Ref. Reference Moreno239). Additional studies from the same group identified the small molecule ISRIB (integrated stress response inhibitor), which prevents translation inhibition downstream of eIF2α-P, as a good drug candidate. Daily intraperitoneal administration of ISRIB to prion infected Tg37+/− mice from 7 weeks post infection, showed that partial restoration of global translation rates sufficed to confer neuroprotection in the absence of pancreatic toxicity. Notably, despite significantly increasing mouse survival as compared to untreated controls, ISRIB administration resulted in significant body weight loss similarly to PERK inhibition (Ref. Reference Halliday240). Understanding whether the observed weight loss results from UPR inhibition, it is a side effect of ISRIB treatment, or else it arises as a consequence of persistent prion infection will have important implications in the clinical implementation of this and similar drugs (Ref. Reference Halliday240).
One of the most promising therapies for treating human prion diseases is passive immunisation with anti-PrP antibodies (Refs Reference Wisniewski and Goni241, Reference Roettger242). This strategy has been extended to several neurodegenerative diseases, including TSEs, encouraged by the observed reduction in amyloid plaque burden in a transgenic mouse model of AD treated with Aβ-directed antibodies (Ref. Reference Schenk243). Additionally, several in vitro studies demonstrate the feasibility of this approach to cure or prevent cellular infection with several strains of prions (Refs Reference Enari, Flechsig and Weissmann244, Reference Peretz245). Importantly, a recent in vivo study demonstrated PrPTSE clearance, and increased survival, after passive immunisation with anti-PrP monoclonal antibodies administered before disease onset (Ref. Reference White246). These findings indicate that therapeutic treatments aiming at blocking the conversion mechanism leading to PrPTSE generation are good approaches to treating prion diseases. Nonetheless, the practical application of these therapies depends upon the development of reliable diagnostic tests to allow early antibody administration.
Active immunisation has been evaluated in a number of studies with highly encouraging outcomes [reviewed in (Refs Reference Wisniewski and Goni241, Reference Roettger242)]. Initial studies with bacterially-expressed recombinant full length PrPC and various PrP peptides served to highlight the difficulty in breaking tolerance to a self-protein. Additionally, they stressed the importance of breaking tolerance to raise antibodies interfering with PrPTSE replication and propagation, while at the same time, minimising an autoimmune response (Ref. Reference Wisniewski and Goni241). Along these lines, mucosal immunisation using bacterial vectors has been suggested as an ideal means to achieve immunomodulation by inducing a secretory IgA response with limited systemic IgG levels, and minimal autoimmune inflammatory effects. This approach was validated in vivo when wild-type mice previously immunised with a mouse PrPC-expressing attenuated Salmonella strain, showed resistance to orally administered prions. These mice were characterised by significant anti-PrP mucosal IgA and systemic anti-PrP IgG response (Ref. Reference Goni247). Mucosal immunisation to prevent CWD infection has also been attempted. White-tailed deer immunisation with attenuated Salmonella expressing deer-PrPC resulted in increased titers of anti-PrP IgG and IgM in the plasma, and anti-PrP IgA in saliva and faeces as compared with control deer, showing for the first time the generation of humoral responses against self-PrP in the biological fluids of large cervid animals (Refs Reference Wisniewski and Goni241, Reference Goni248). Vaccinated deer had a significantly prolonged incubation period than control animals following oral infection with a 100% lethal dose of CWD prions. One deer in the vaccinated group remained free of CWD symptoms and PrPTSE for 3 years and 7 months, as determined by immunohistochemical evaluation of tonsil and rectoanal mucosa-associated lymphoid tissue biopsies. Importantly, this animal showed the highest levels of mucosal and systemic immune response (Ref. Reference Goni248).
This study revealed the feasibility of utilising bacterial vectors to induce mucosal immunisation and prevent prion disease infection, albeit being at an early stage of development. Mucosal immunisation represents an important approach to prevent acquired forms of TSE in humans (vCJD), and interrupt transmission of BSE, scrapie, and CWD in animals in which the gut is the major route of entry. Additionally, monoclonal antibodies produced in vaccinated animals could be isolated and used for passive immunisation, extending the potential clinical application of these approaches to other forms of TSE of different aetiology (Ref. Reference Wisniewski and Goni241).
The recent discovery of NAD+ starvation as potential cause of neuronal death, led to the therapeutical investigation of NAD+ replenishment in in vivo models of mouse scrapie. Stereotaxic injections of TPrP in the presence or absence of NAD+ resulted in TPrP toxicity abrogation in a dose-dependent manner (Ref. Reference Zhou191). When NAD+ was intra-nasally administered at disease onset, mice infected with RML, showed slower disease progression and weight loss than phosphate-buffered saline (PBS)-treated control mice. A similar result was obtained when RML-infected mice were treated during the clinical phase of the disease. Although no differences in survival were observed between treated and control groups, NAD+ administration significantly increased motor function in mice (Ref. Reference Zhou191).
Concluding remarks
Prion diseases are devastating disorders of the CNS. At the present time there is no efficient treatment or cure for these diseases, and therefore, affected individuals die within months after the first clinical symptoms appear. Identification of the neurotoxic molecule(s) and the cellular pathways leading to neurodegeneration would facilitate the development of new therapies. Likewise, development of early, noninvasive diagnostic tests is extremely important; epidemiologically, to prevent secondary transmissions, and therapeutically, to allow the administration of effective therapies before extensive brain damage takes place. This is of particular relevance in the case of hereditary/genetic forms of TSE.
Understanding the biology underlying prion-mediated neurotoxicity is challenging given the unprecedented nature of the infectious agent. PrPTSE replicates by conferring its aberrant conformation onto a cellular protein, PrPC. Studies conducted with PrP−/− mice indicated that loss of PrPC function cannot account for prion neurotoxicity, as these mice show neither a deleterious phenotype nor altered lifespan. Experimental evidence has been accumulated demonstrating the dissociation between PrPC to PrPTSE conversion with accumulation of the latter and neurotoxicity. While PrPC expression is prerequisite for prion neuroinvasion and propagation, its requirement at the neuronal membrane for neurotoxicity is still matter of controversy. Intriguing is the fact that astrocytes are refractory to TPrP toxicity, similarly to findings derived from the post-natal removal of PrPC expression highlighted above. These observations suggest that the neurotoxic entity, namely TPrP or PrPL, should be generated within neurons to cause toxicity to these cells. Importantly, the confirmation of these hypotheses relies on the successful isolation of these molecules from naturally or experimentally induced TSE; task that remains highly elusive.
Similarly controversial is the nature of the entity responsible for the neurotoxic cascade observed in TSEs. Whether prion-related neurodegeneration is triggered by the gain of toxic function of low molecular weight PrPTSE oligomers, the generation of toxic intermediates, or the subversion of PrPC biological function remains to be elucidated. But in light of the heterogeneity in clinical and pathological presentation of this diverse group of disorders, it is very likely that distinct mechanisms are involved in TSE forms of different aetiology.
Altogether, the findings described in this review indicate that significant progress has been made in recent years towards the elucidation of prion-induced mechanisms of neurodegeneration and, more importantly, potential treatments to halt these devastating disorders. Further confirmation of these mechanisms can be accomplished by refining the identification of the molecules involved in the cellular pathways activated during neurodegeneration. Genetic and biochemical methods can be utilised to identify more precisely cellular partners interacting with variants of PrP molecules, and may reveal signalling cascades activated in TSEs and other neurodegenerative diseases that culminate in neuronal death. The involvement of identified partners will subsequently need to be validated in in vitro (cell cultures, organotypic slices, mini brains) and in vivo models of prion disease. This can be accomplished, for example, by deleting the expression of the identified molecules in cells or in mice and by creating double-knockout mice and evaluating the pathologic process under these conditions. Eventually, based on findings derived from these studies, appropriate early diagnostics and targeted treatments can be developed and implemented in real life to treat humans and animals.
Acknowledgements
We thank Donna Sobieski and Dr Erin Moritz for editorial assistance. Work in the laboratory of DAH is supported by grants from the National Institutes of Health (R01 NS065244 and R01 NS040975).
