Broadly speaking, three general possibilities have been considered, although the fundamental cause of neuronal dysfunction and loss remains uncertain. First, there is the possibility of a loss of normal PrP function, due to depletion of PrPC; second, the possibility of a directly neurotoxic effect of abnormal PrP (either relatively protease-resistant aggregated PrPSc or another abnormal form); and third, a toxic effect of some interaction of abnormal PrP with PrPC.
Depletion of PrPC
The possibility that depletion of PrPC is the key pathogenic factor brings into consideration the normal functions of PrPC, which is expressed not only in neurones but also in other cell types. The mature cellular PrPC is a GPI-anchored protein constructed of 208–209 amino acids. The overall composition of this membrane bound protein includes a structured, globular C-terminal and a flexible, unstructured N-terminal. Recent studies have demonstrated that PrPC is responsible for a variety of physiological cellular functions. These include neuronal growth (neurite outgrowth, neuronal survival and neuronal differentiation), regulation of ion channels and neuronal excitability, cell adhesion, lymphocyte activation and maintenance of myelination, as well as regulating intracellular signaling cascades, including those that control cell viability and death.[8–10]
There is evidence to indicate that PrPC is subject to proteolytic processing (α-cleavage, β-cleavage and shedding), which, in turn, appear to regulate these cellular functions under physiological conditions and produce 'biologically active fragments' that can influence neurodegenerative disease processes, such as prion and Alzheimer's disease.[9,11]
α-cleavage and shedding are considered to be the most relevant stages of proteolytic processing in relation to neurodegenerative disease processes and will, therefore, be discussed in more detail. The products of α-cleavage include a soluble N1 fragment and membrane anchored C1 fragment. Altmeppen et al. have reviewed the proteases that have been proposed as responsible for this cleavage process but the exact identity of the protease remains uncertain. The most compelling evidence supports the involvement of a disintegrin and metalloproteinase (ADAM) group of proteases. Most recently, inhibitor studies have suggested that ADAM 10 is responsible for 'constitutive cleavage', whereas ADAM 17 is 'stimulus dependant'. ADAM 9 was thought to be involved indirectly by regulating ADAM 10.[11,13,14] However, these studies have been challenged. For instance, transgenic overexpression of ADAM 10 in mice did not result in elevated proportions of N1 and C1 fragments. In addition, N1 and C1 fragments were not found to be reduced in brain homogenates and primary neurons of mice devoid of ADAM 10 in neural precursor cells.[11,12,15]
Although uncertainty remains surrounding the precise identity of the protease involved with α-cleavage, what is becoming increasingly evident is the physiological and neuroprotective importance of this proteolytic process.[16,17] α-cleavage occurs at site 106–126 of PrP and, thereby, interferes with the neurotoxic domain. The neurotoxic domain has been demonstrated to be a structural prerequisite for the conformational conversion of PrPC to PrPSc. It has been suggested that liberating the N-terminus and cleaving PrPC within the neurotoxic domain, is directly neuroprotective by preventing misfolding of the residual C1 and helping to reduce further propagation. Soluble N1 has also been reported to have roles in neuroprotective signaling.
With regard to the membrane anchored C1 fragment (i.e., before shedding has occurred – see below), the physiological function is less clear. However, similarly to the N1 fragment, there also appears to be a neuroprotective role. This has been demonstrated in relation to the upkeep of myelin in the peripheral nervous system. It has been suggested that axonal C1 expression is associated with a signaling cascade in trans that triggers neighboring Schwann cells to initiate protective signaling to help maintain and protect the myelin sheath. This is supported by the fact that PRNP0/0 mice and transgenic mice with a deleted α-cleavage site develop a demyelinating polyneuropathy.[20,21] Therefore, it was suggested that a potential therapeutic target against myelin-related toxicity would be the upregulation or stimulation of α-processing. By contrast, there is evidence to support an association between C1 fragments and toxicity via the initiation of a p53-dependent apoptotic process producing elevated levels of caspase-3 activation.
A third physiological proteolytic process occurs close to the GPI anchor and has been termed 'shedding'. This is also a cleavage process that liberates the C1 fragment from the membrane. Like α-cleavage, there is evidence to suggest that a protease is involved in the release of the C1 fragment from its membrane. Cell culture experiments have suggested ADAM 10 as the responsible protease. Similar to α-cleavage, ADAM 9 was also thought to play an indirect role by regulating ADAM 10. However, ADAM 17 is not thought to have a role in proteolytic shedding.[11,12] However, the precise roles played by the anchorless C1 fragment and the process of shedding are uncertain, with evidence that they may contribute to neurotoxicity (as discussed in the 'Possible toxic protein species' section).
PrPC has been shown to interact with many other proteins that are potentially relevant to some of its proposed functions. Studies exploring some of these interactions have shown potentially relevant effects, for example, a study using pharamacological interference with the interaction of PrPC and STIP1 in rats, reported demonstrable memory deficits. Two reviews contain comprehensive detail of these aspects of PrPC.[23,24] Some protein interactions of PrPC may be relevant in neurodegenerative diseases other than prion disease. For example, in Alzheimer's disease, the key neurotoxic species, the Aβ oligomer, is thought to bind to the N-terminus of PrPC, which subsequently induces its toxic signaling. α-cleavage prevents this interaction and additionally produces free N1, which binds the oligomers and ultimately blocks the toxicity pathway. In support of this concept, it has been reported that transgenic mice expressing N-terminal truncation or deletion and where α-cleavage does not occur, display signs of neurological dysfunction. It has been suggested that this blocking and neuroprotective function of N1 may not be unique to Aβ oligomers, but could be extended to include other neurodegenerative diseases that involve β-sheet toxic species in the disease pathogenesis. It has, therefore, been proposed that attempts to stimulate the α-cleavage process, could potentially offer a therapeutic option against protein diseases, such as prion and Alzheimer's disease.
The loss-of-function possibility has been explored experimentally, in particular, with the creation of PrP-null mice. The interpretation of these rodent studies has been a little difficult, partly because of the employment of different techniques to produce the mice. Two initial reports described null mice whereby the genetic modifications were restricted to the Prnp open reading frame, which showed normal development and reproduction.[26,27] Some models, using a different genetic production method, reported the development of an ataxic syndrome with loss of Purkinje cells in later life. However, this was shown to result from changes in a related gene (the PRND gene).[28–30] This led to considerations of the possible role of the doppel protein and its relationship to PrPC in prion disease pathogenesis. Other studies have reported a variety of abnormalities in PrP-null mice, including the following. Prestori et al. reported cerebellar abnormalities with impaired motor control in juvenile mice distinct from the later ataxic syndrome associated doppel. Collinge et al. described abnormal synaptic function in hippocampal slices from PrP-null mice. Abnormalities of sleep regulation and circadian rhythms have been described.[33,34] Criado et al. reported deficits in hippocampal-dependent spatial learning (although not nonspatial learning) in PrP-null mice; the deficits being rescued by PrPC and using methods making it unlikely that the deficits arose through other mechanisms (such as doppel). A variety of studies have shown PrP-null mice to be more sensitive to certain stresses, such as ischemic brain injury and convulsant-induced seizures.[36–38] A mouse model depleting PrPC at 9 weeks of age, via a Cre-loxP mechanism, showed no serious effects, with no neurodegenerative changes or clinical illness, although some neurophysiological effects were seen.
Given the many potential functions of PrPC and its complex interactions with other proteins, both in development and in adults, it is perhaps not surprising that a definitive conclusion cannot easily be drawn from these data. The overall trend in the literature has been away from loss of PrPC function explanations; however, this is not universal.
PrPC is subject to several important post-translational modifications including N-linked glycosylation, the attachment of a GPI anchor, the formation of a single disulfide bond and these would be expected to have significant functional consequences. Interestingly and importantly, it has been demonstrated that PrPC is in fact essential for the development of prion disease.[41,42] These aspects are discussed in the 'PrPC & PrPSc interaction' section.
Direct Neurotoxicity of Abnormal PrP
Although PrPSc has a crucial role, not least in being a diagnostic pathological hallmark of prion disease, it is not clear if it is the critical neurotoxic factor and, if it is, which form of PrPSc is relevant and by what precise mechanism. The concept that neuronal cell death is a consequence of direct toxicity by PrPSc has been increasingly questioned, although, of course, any such discussion needs to first define what particular molecular species of PrPSc is being considered and, therefore, the precise meaning of 'PrPSc' being employed.
Association of PrPSc with toxicity Experimental studies on cultured neurones have shown toxic effects of either full-length PrPSc or shorter fragments of it.[43,44] However, the general association of PrPSc and neuronal pathology (at least in a simple, direct manner) is brought into question by a variety of data. A number of transmission studies have demonstrated neuropathological change (with or without clinical disease) in the presence of minimal amounts of detectable PrPSc. Lasmezas et al. studied bovine spongiform encephalopathy transmission to mice and reported clinical disease and associated neuropathological abnormality in the recipients, but with many mice showing no PrPres accumulation; interestingly, on serial passage in mice, with evidence of species adaptation, PrPres accumulation was increasingly found. In another experiment, Manson et al. reported that successful transmission of disease (using human GSS brain homogenate) to transgenic mice was characterized by extremely low levels of abnormal PrP in the recipient mice. In a transgenic overexpression mouse model related to GSS, Hsiao et al. reported spontaneous neurodegenerative disease with characteristic neuropathological features, but very low levels of protease-resistant PrP. Collinge et al. reported successful transmission of fatal familial insomnia to mice with fatal results associated with neuropathological disease, but no detectable PrPSc in the recipient mice. Other experimental studies have demonstrated the presence of PrPSc without clinical dysfunction, including transmission studies of variant CJD into transgenic mice.[49,50] Mallucci et al. reported continued PrPSc tissue deposition despite reversal of early neurodegenerative change, prevention of neuronal loss and absence of clinical disease after successful experimental neuronal PrPC depletion in their scrapie Cre-lox mouse model. In an interesting graft experiment, Brandner et al. showed that PrPSc was generated in a PrPC-expressing graft into PrPC-deficient mice, but that the graft-generated PrPSc did not cause neuropathological change in the surrounding brain into which it migrated.
Abnormal PrP in neurons or other cells Another consideration relating to PrPSc toxicity is its tissue location. Cells other than neurons express PrPC and Diedrich et al. reported that, in their mouse scrapie model, abnormal PrP accumulated in astrocytes prior to the development of the cardinal neuropathological changes, although Race et al. reported an experiment indicating that astrocytic expression of PrP was not an essential prerequisite for the development of prion disease. Two experimental studies have produced results suggesting that extraneuronal accumulation of PrPSc is not directly toxic and that neurotoxicity relates to the generation of intraneuronal abnormal PrP molecular species. Chesebro et al. studied transgenic mice that expressed anchorless PrP (PrPC without the GPI anchor). This anchorless PrP is significantly underglycosylated. After inoculation with scrapie, these mice did not develop typical scrapie, but demonstrated minimal clinical illness with limited neuropathological changes unlike those seen in scrapie despite accumulation of PrPres in amyloid plaques. Mallucci et al. studied a mouse model in which neuronal PrPC was deleted during the course of infection but glial PrPSc production was unaffected; although PrPSc accumulation therefore continued, there was no neuronal loss nor any clinical illness. However, it is notable that Jeffrey et al. demonstrated the development of prion disease in transgenic mice that expressed PrPC only in astrocytes. In their mouse model, following infection with scrapie, there was typical neuronal pathology even though the neurones lacked PrP expression. There was abnormal PrP around astrocytes (which demonstrated only reactive changes themselves) in the extracellular spaces of the neuropil, indicating that disease pathogenesis did not require direct intraneuronal production of abnormal PrP.
Possible toxic protein species The accumulated evidence of uncoupling of tissue aggregated PrPSc and neurotoxicity has led to two (not necessarily mutually exclusive) considerations: prion neurotoxicity may be due to oligomeric intermediates between PrPC and PrPSc and that there are important interactions between PrPC and some molecular forms of PrPSc (this second possibility being discussed in the 'PrPC & PrPSc interaction' section).[24,54] Collinge and colleagues have proposed that a hypothetical neurotoxic form of PrP named PrPL may be generated during the PrPC to PrPSc conversion (as an intermediate form or a by-product).[5,54,57] On this suggestion, PrPL may be rapidly 'cleared' or recruited into large PrPSc aggregates, with the opposing possibility that preventing the conversion of PrPC to PrPSc could limit the amount of PrPL or could ultimately lead to increasing levels of PrPL. If the rate of neurodegeneration was dependent on the 'level of steady state' of PrPL then this could help to explain a dissociation of PrPSc and neurodegeneration.[5,54,57] Aguzzi and Falsig suggested that, as PrPL has not been physically defined, it may not exist and argue that it is at least equally possible that neurotoxicity results from an interaction of PrPSc with PrPC. However, Zhou et al. have recently reported a highly toxic form of PrP that is potentially significant for prion neuropathogenesis. They subjected recominbant PrP to denaturation and dilution refolding and then separated out different PrP forms, testing their toxicity in cell cultures, brain slices and in laboratory mice via intracerebral inoculation. The most toxic PrP was found to be monomeric and highly α-helical. They called this toxic PrP, but it has not yet been isolated from prion disease brains.
The evidence from the transgenic Cre-lox mouse experiments (mentioned above) undertaken by Mallucci et al., whereby disease was rapidly reversed by PrPC depletion, before extensive PrPSc accumulation, also supports the possibility of a transient toxic intermediate.[5,51,54,57,59]
Two important reviews discuss the biosynthesis of PrPC and note that minor 'aberrent' pathways could be followed.[40,60] In particular, while biosynthesis mostly produces a fully translocated and glycosylated PrP (denoted SecPrP from 'secretory') that results in the typical cell-surface PrPC protein, minority transmembrane forms (<10%). These transmembrane forms have been designated NtmPrP and CtmPrP with 'Ntm' and 'Ctm' referring to transmembrane locations with either the N-terminus or C-terminus, respectively, being on the extracytoplasmic side).[40,60,61] Stewart et al. reported experiments concerning using a mutant form of PrP that is synthesized exclusively with the CtmPrP topology, showing that it is retained in the endoplasmic reticulum and is degraded by the proteasome and they discuss possible pathogenic implications of this. In addition, through 'aberrant' processing, PrP may be released into the endoplasmic reticulum cytosol and this has been termed cytosolic PrP (cyPrP). In considering the possible roles of these 'aberrant' PrP forms, most of the discussion has centered on CtmPrP and cyPrP and, as is discussed in the review by Chakrabarti et al., while these two are apparently produced at low levels in normal PrP biosynthesis, there is evidence that higher levels are detrimental. In murine experiments, wild-type mice produced CtmPrP at levels around 1% of PrP; however, mice with certain PrP gene mutations (in the region responsible a sequence of hydrophobic amino acids in the central region of the protein) produced significantly elevated levels and they developed a spontaneous neurodegenerative disease with features similar to those of some genetic prion diseases, in the absence of PrPSc.[61,63] A strong correlation between CtmPrP levels and disease severity has been found, which provides support for CtmPrP having a pathogenic role.[61,63,64] Interestingly, several of the recognized pathogenic PRNP mutations responsible for human genetic prion disease affect PrP properties in a manner that favors CtmPrP production and the expression of one of these (A117V) in mice causes a neurodegenerative illness.[60,61,63] However, mutations affecting other PRNP regions do not lead to increased CtmPrP levels indicating that this is not a straightforward mechanism in all genetic prion diseases.[40,65] As a result of these accumulated data, it is suggested that CtmPrP has a key role in prion disease pathogenesis and, while a significant amount of the evidence has related to genetic disease, Hegde et al. proposed that CtmPrP is part of a mechanism common to both genetic and transmissible prion diseases. There is evidence also that cyPrP has a pathogenic role. One series of experiments in cell lines (including exposure to proteasome inhibitors) and in transgenic mice, elevated levels of cyPrP were shown to be neurotoxic, with resulting disease in the mice. However, in some cell-culture experiments, cyPrP was found not to be toxic, or even to have a protective role.[67,68] The possible physiological function of the membrane anchored C1 fragment (i.e., before shedding has occurred), was discussed in the 'Depletion of PrPC' section. However, there is evidence to support an association between C1 fragments and toxicity via the initiation of a p53-dependent apoptotic process producing elevated levels of caspase-3 activation. The process of shedding (described in the 'Depletion of PrPC' section) has also been implicated in terms of possible harmful effects although there is a paucity of in vivo data detailing the exact relevance of shedding in neurodegenerative diseases.[11,69] Altmeppen et al. speculated that ADAM 10 has a conflicting role with regard to prion diseases with the suggestion that it acts in a protective and inimical manner. It is well recognised that PrPC acts as the substrate for the conversion to PrPSc and neurotoxicity in prion diseases. It has been suggested that ADAM 10 exerts its protective effect by reducing cell surface levels of full length PrPC and, therefore, reducing the load of PrPSc. However, in instances where the normal PrP has already converted to PrPSc and has been shed and there has been misfolding of shed PrPC, thus producing anchorless PrPSc, it has been argued that this could facilitate spread of disease throughout the brain. There is evidence to support that shedding also occurs in cerebrospinal fluid and blood, which could potentially enhance transmissibility of the disease.[11,70–73] Altmeppen et al. went on to suggest that these contradictory roles could potentially explain the concept that prion propagation is not directly associated with neurotoxicity. In support of this, prion infection in transgenic mice expressing anchorless, secreted PrPC, resulted in a prolongation of the time to onset of clinical signs despite extensive PrPSc load.[12,55,74]
Explorations of toxic protein species via genetic disease In the case of pathogenic PRNP mutations, disease appears to develop spontaneously and, therefore, it is considered that the resulting mutant PrP is liable to spontaneously convert to PrPSc and, possibly, to be directly neurotoxic. Given this, the study of genetic prion disease may well provide insights into toxic PrP species and prion pathogenesis; indeed some of the experiments discussed in the section above illustrate this. Harris et al. reviewed several studies that used cell lines expressing murine homologs of mutant PrPs associated with the three types of human genetic prion disease (genetic CJD, fatal familial insomnia, GSS). In a set of studies, the mutant PrPC molecules formed in these cells were found to have biochemical properties similar to those of PrPSc including detergent insolubility and protease resistance. In further studies, three intermediate biochemical steps were identified in the development of PrPSc-like properties by the mutant PrPC and that pathogenic mutations may impair the delivery of PrP molecules to the plasma membrane with resultant partial accumulation in the endoplasmic reticulum.[40,75,76] It has been suggested that PrPSc accumulation may directly or indirectly lead to accumulation of these PrP byproducts and subsequently exert neurotoxic effects that cause cellular dysfunction.[60,63] However, these mechanisms have not been definitively implicated in prion disease. In studies involving transgenic, Tg(PG14), mice expressing a 9 octapeptide insertional PrP mutant (associated with human genetic prion disease), the mice developed a progressive neurodegenerative disease with the accumulation of mutant PrP.[77–79] This mutant PrP, termed 'PG14Spon', showed several biochemical properties similar to those of PrPSc. The development of disease reflected the level of expression of the mutant transgene, although disease did occur even at levels of expression equivalent to that of endogenous PrP (i.e., it did not depend on significant overexpression). In further analysis, it was shown that the mutant PrP accumulates throughout the life of the animal, but that neuropathological abnormality and clinical disease did not develop until a critical threshold is reached. The experimental results indicated that the PG14 PrP accumulated in the brains of the mice due to a slower turn over or less efficient clearance, rather than due to an increased rate of synthesis. Interestingly, further experiments using this mouse model provided support for the distinction between neurotoxic and infectious aspects of prion disease with brain homogenates from the Tg(PG14) mice failing to transmit disease to recipient mice via cerebral inoculation. However, inoculation of the Tg(PG14) mice with RML scrapie resulted in an accumulation of an abnormal protein they termed PG14RML and brain homogenates from these mice transmitted disease. The biochemical properties of both PG14 protein types were examined; although conformationally similar, it was found that PG14RML formed larger, less easily dissociated aggregates than PG14Spon, which tended to be found in smaller oligomeric forms. Soloman et al. commented on this and drew a parallel with the findings indicating that small Aβ oligomers, rather than large amyloid fibers, are the primary toxic species in Alzheimer's disease. However, it remains unclear how these oligomers induce neurotoxicity.
The 'end point' tissue deposition of some forms of aggregated protein may conceivably be part of a protective mechanism or, at least, represent the least toxic forms of abnormal PrP with the implication that therapeutic strategies designed to block or disrupt PrPres aggregation may be harmful.[3,81] A related possible concern was expressed by Radford and Mallucci that preventing PrPC to PrPSc conversion, might lead to an increase in an intermediate toxic form. However, in animal experimental models, depletion of PrPC (via a Cre-lox mechanism or lentivirus introduction of short hairpin RNAs) has proved beneficial in prion disease.[51,82]
PrPC & PrPSc Interaction
As indicated above, the accumulated evidence of uncoupling of tissue aggregated PrPSc and neurotoxicity has led not only to considerations of directly toxic non-PrPSc protein species, but also to considerations of possible toxicity arising from important interactions between PrPC and some molecular forms of abnormal PrP. Soloman et al. described this as a 'subversion-of-function' as opposed to a loss-of-function or gain-of-function, suggesting that PrPSc (or other abnormal forms of PrP) could interact with PrPC resulting in a neurotoxic signal. This idea has been discussed by other authors; for example, Harris and True outlined the idea that, in subversion of PrPC function, the normal, neuroprotective activity of PrPC is subverted by binding to PrPSc (or some toxic abnormal PrP form) leading to a change in its signaling properties such that a neurotoxic rather than a neuroprotective signal is delivered. They comment that, in the absence of the GPI anchor attaching PrPC to the membrane, no signal would be delivered and disease would not occur, in keeping with the results of the already discussed anchorless PrP experimental results. There are various lines of evidence that lend support to these notions. The Brandner et al. experiment described above, where overexpressing PrPC neural tissue was grafted into the brain of PrPC deficient mice, showed that the surrounding PrPC deficient tissues failed to demonstrate neuropathological change even though PrPSc migrated into it from the graft. The authors concluded that "brain tissue devoid of PrPC is not damaged by exogenous PrPSc." In the Cre-lox mouse model described above, early degenerative neuropathological change (including PrPSc deposition) and associated clinical illnesses was reversed by PrPC depletion a few weeks after innoculation after prion infection with the control mice rapidly dying from typical prion disease. This recovery occurred despite continued accumulation of extraneuronal PrPSc. In the study of scrapie brain inoculation in transgenic mice expressing mostly unglycosylated GPI-anchorless PrP, the anchorless mice did not develop clinical disease, but showed brain PrPres amyloid plaques with some limited neuropathological change, implying that GPI-membrane anchoring of PrPC plays an important role in prion disease. Flechig et al. reported an experiment showing that mice devoid of PrPC were indeed not susceptible to scrapie, that susceptibility was restored by the introduction of transgenes expressing PrP, but that the introduction of truncated PrP (devoid of the N-terminal octarepeat copper-binding site) led to significant modification of the resultant disease. The full-PrP mice developed typical scrapie in the manner expected for this murine model, but the truncated-PrP mice developed a fatal illness with a longer incubation period, significantly lower levels of PrPSc and scrapie histopathological features limited to the spinal cord (being absent from the brain). Their conclusion was that the octapeptide region of PrPC was not essential for prion pathogenesis, but that it modulates certain aspects of the disease. Experiments involving deletions in the PrP gene that result in murine spontaneous fatal illness have been reported and discussed in the context of the role of PrPC in either neuroprotective or neurotoxic signalling. Schmerling et al. reported on Δ32–121 and Δ32–134 PrP deletions resulting in a spontaneous fatal murine illness that was totally abolished by introducing one copy of wild-type PrP. Their experimental results indicated that the resulting disease was not likely to be owing a specific direct toxicity of the truncated PrP (particularly as the pathological changes were anatomically circumscribed, while the truncated protein was found throughout the brain). They also felt that a direct interference between the truncated PrP and normal PrPC was unlikely as the pathological changes developed in mice devoid of PrPC. Instead, they proposed a competitive interaction between the truncated PrP and another protein for a hypothetical ligand for which PrPC also has an affinity. In more recent deletion murine experiments, concerning Δ105–125, a spontaneous neurological illness was found, again reversed by the expression of wild-type PrP. The authors comment on some similarities between their illness model and those of the Schmerling model and also those reported related to doppel (as discussed above). They speculate that the 105–125 PrP region is a crucial determinant of the neuroprotective and neurotoxic roles of PrPC and that, therefore, it plays an important role in generating neurotoxicity/neuropathology in prion disease. In discussing this, they make suggestions as to how PrPC function may be subverted in prion disease pathogenesis. In considering these data, the overall implication is that, whatever its normal function, PrPC contributes to neurotoxicity in prion disease, but the exact mechanism of this remains unclear.
Future Neurology. 2014;9(2):135-147. © 2014 Future Medicine Ltd.