|Home | About | Journals | Submit | Contact Us | Français|
Prion diseases are caused by conversion of a normal cell-surface glycoprotein (PrPC) into a conformationally altered isoform (PrPSc) that is infectious in the absence of nucleic acid. Although a great deal has been learned about PrPSc and its role in prion propagation, much less is known about the physiological function of PrPC. In this review, we will summarize some of the major proposed functions for PrPC, including protection against apoptotic and oxidative stress, cellular uptake or binding of copper ions, transmembrane signaling, formation and maintenance of synapses, and adhesion to the extracellular matrix. We will also outline how loss or subversion of the cytoprotective or neuronal survival activities of PrPC might contribute to the pathogenesis of prion diseases, and how similar mechanisms are probably operative in other neurodegenerative disorders.
Prion diseases result from conversion of a normal, cell-surface glycoprotein (PrPC) into a conformationally altered isoform (PrPSc) that is infectious in the absence of nucleic acids. Although we know a great deal about the role of PrPSc in the disease process, the normal, physiological function of PrPC has remained enigmatic. Attempts to deduce the function of PrPC from the phenotypes of PrP-null mice have been uninformative, since lines of these mice in which the adjacent Doppel (Dpl) gene is not artifactually up-regulated display no major anatomical or developmental deficits [1, 2]. Even mouse lines in which the PrP gene is deleted postnatally using a conditional Cre-Lox system are phenotypically relatively normal, arguing against the existence of other proteins that compensate for essential PrP functions in the adult . Parts of the PrPC sequence have been highly conserved in evolution , suggesting important biological roles for the protein.
Investigating the biological activity of PrPC is likely to be crucial for understanding the pathogenesis of prion diseases, since alteration of this function could play a role in the disease process. In this review, we will summarize some of the major proposed functions for PrPC, with a particular emphasis on a role for the protein in protection from cellular stress. We will also outline how gain, loss, or subversion of PrPC function can play a role in the pathogenesis of prion diseases, and how similar mechanisms may be operative in other neurodegenerative disorders.
PrPC is expressed beginning early in embryogenesis, and in the adult it is present at highest levels in neurons of the brain and spinal cord [5, 6]. PrPC is also found at lower levels in glial cells of the CNS as well as in a number of peripheral cell types [7, 8]. Most PrPC molecules are normally localized on the cell surface, where they are attached to the lipid bilayer via a C-terminal, glycosyl-phosphatidylinositol (GPI) anchor . The biosynthetic pathway followed by PrPC is similar to that of other membrane and secreted proteins, involving synthesis on ER-attached ribosomes, transit to the Golgi, followed by delivery to the cell surface [reviewed in 10]. PrPC is a glycoprotein, with two N-linked oligosaccharide chains of the complex type. Although most cell-surface PrPC is found in lipid rafts, some of the protein is transferred to clathrin-coated pits where is subject to constitutive endocytosis and recycling [11-14].
A powerful strategy for elucidating the physiological function of PrPC would be to identify other cellular proteins with which PrPC interacts. At least some of these interactors are likely to be components of the physiological pathways in which PrP plays a role. Over the years, a number of candidates have been identified as potential PrP-binding partners using conventional yeast two-hybrid screens, co-immunoprecipitation, cross-linking and other methods (Table 1) [reviewed in 15]. Several of the candidates listed in Table 1 (STI-1, N-CAM, Bcl-2, caveolin) will be discussed below, in conjunction with the cellular function of PrP in which they are purported to play a role. In almost every case, however, the physiological relevance of the proposed interactions remains uncertain. Some of the putative interactors are primarily or exclusively cytoplasmic proteins, and so would be unlikely to associate directly with PrP, which is localized to the outer face of the plasma membrane and to the lumen of organelles in the secretory pathway. One example of such a candidate is the laminin receptor precursor (LRP), which was originally identified in a yeast two-hybrid screen and has been claimed to serve as an endocytic receptor for cellular uptake of both PrPC and PrPSc [16, 17]. However, LRP lacks a signal peptide or transmembrane anchor that could target it to the secretory pathway, and the majority of the molecules are likely to be cytoplasmic.
Since PrP is a GPI-anchored protein, the entire polypeptide chain is located on the extracytoplasmic face of the lipid bilayer. Thus, direct PrP binding partners are most likely to be transmembrane and secreted proteins. Although PrP potentially exists as two transmembrane variants (NtmPrP and CtmPrP) that could theoretically interact with cytoplasmic partners [18, 19], these forms are normally present in minute amounts in the absence of predisposing mutations in the PrP molecule . PrP molecules that lie entirely in the cytoplasm have also been described in cultured cells , but our own data indicate that these forms are likely to be artifacts of over-expression or treatment with proteasome inhibitors . Of course, PrP could associate with cytoplasmic proteins indirectly, via involvement of intermediary proteins that serve as transmembrane linkers.
Several intriguing lines of evidence have emerged recently suggesting that PrPC may exert a cytoprotective activity, particularly against internal or environmental stresses that initiate an apoptotic program [reviewed in 22, 23]. This activity has been demonstrated in a variety of experimental systems, including cultured mammalian cells, yeast, and mice.
One of the clearest examples of a cytoprotective activity of PrPC is the protein's ability to protect human fetal neurons in culture against apoptosis induced by Bax. Bax is a pro-apoptotic member of the Bcl-2 family that plays a major role in postmitotic neurons of the central nervous system [24, 25]. When human fetal neurons in culture were microinjected with a plasmid encoding Bax, ~90% of the neurons underwent apoptosis; but when the neurons were co-injected with both Bax- and PrP-encoding plasmids, the percentage of apoptotic cells was reduced to ~10% [26, 27]. This cytoprotective effect of PrP appeared to be specific for Bax, since PrP did not prevent neuronal apoptosis induced by Bak, t-Bid, staurosporine, or thapsigargin .
It has also been reported that expression of PrPC rescued immortalized hippocampal neurons (HpL3-4 cells) derived from Prn-p0/0 mice from apoptosis induced by serum deprivation [29, 30]. This effect appeared to involve a Bax-mediated apoptotic pathway.
Another system involves MCF-7 cells, which are derived from a human mammary adenocarcinoma and undergo apoptosis in response to treatment with the cytokine, tumor necrosis factor (TNF-α). Diarra-Mehrpour et al.  isolated a variant sub-clone of MCF-7 cells that was resistant to TNF-α-induced apoptosis, and found that this clone displayed a dramatic up-regulation of endogenous PrP gene expression. Moreover, over-expression of PrPC in the parent MCF-7 line was sufficient to render it resistant to TNF-α-induced cell death. At least part of the protective effect of PrPC involved suppression of a mitochondria-dependent death pathway.
PrP has also been found to rescue cultured cerebellar granule neurons  and N2a neuroblastoma cells  from apoptosis induced by Dpl. Dpl is a PrP paralog which causes a neurodegenerative phenotype when ectopically expressed in the CNS of Prn-p0/0 transgenic mice (see Section 4.3). In N2a cells, it was claimed that the protective effect involved a physical interaction between PrP and Dpl on the cell membrane.
In a recently published study, it was reported that PrPC promoted self-renewal of hematopoietic stem cells during serial transplantation . This phenomenon could also relate to the cytoprotective activities of PrPC, since serial transplantation is likely to subject cells to apoptotic stress. In what could be a related effect, it was recently reported that PrPC positively regulates proliferation and differentiation of neural precursor cells in vitro and in vivo .
While these diverse model systems suggest a common mechanism for PrP cytoprotection, the specific cellular and molecular pathways involved remain largely unknown. Since Bax-dependent apoptosis is a common theme in several of these systems, one possibility is that PrP acts by inhibiting Bax-mediated cell death. Fig. 1 illustrates several possible mechanisms by which this might occur, based on known pathways for Bax action [24, 36]. First, GPI-anchored, PrP on the cell surface might interact with a putative transmembrane receptor, thereby initiating a signal transduction cascade. This cascade might then alter Bax activity directly, for example by inhibiting of its mitochondrial translocation, conformational change, or oligomerization (Fig. 1A). These same effects might also result from a physical interaction between Bax and cytoplasmic forms of PrP (Fig. 1B). Alternatively, a PrP-dependent signal might act upstream of Bax, causing either inhibition of pro-apoptotic, BH3-only activators of Bax (Fig. 1C), or enhanced association of Bax with anti-apoptotic regulators such as Bcl-2 or Bcl-XL (Fig. 1D). It is also possible that PrP suppresses downstream effects of Bax, such as release of cytochrome c, or activation of Apaf-1 and caspases (Fig. 1E). Finally, since Bax is known to have effects on calcium release and the unfolded protein response (UPR) in the ER, [37, 38], PrP traversing the secretory pathway could also affect Bax activity associated with this organelle, possibly via a transmembrane receptor (Fig. 1F).
At this point, there is a dearth of experimental data to distinguish among the different mechanisms outlined in Fig. 1. A recent study of the Bax suppressive effects of PrPC utilizing several different cell types (fetal neurons, HpL3-4 cells, MCF-7 cells) concluded that PrP inhibits the conformational change that occurs when Bax is activated and targeted to the outer mitochondrial membrane . This result would be consistent with mechanisms A and B in Fig. 1. Mechanism B is also suggested by a study showing that an engineered form of cytoplasmic PrP protects fetal neurons against Bax . However, other studies have shown that cytoplasmic PrP produces a toxic rather than protective effect by binding to and sequestering Bcl-2 . In any event, the physiological significance of cytoplasmic PrP is uncertain at this point, since very little of this form appears to be generated in vivo from the wild-type molecule  Clearly, further progress in elucidating the mechanism underlying PrP cytoprotection will require identification of other proteins that interact physically and functionally with PrP and connect it to cell death and survival pathways.
In order to investigate the anti-apoptotic effect of PrP in a genetically tractable experimental system, we have recently turned to baker's yeast, Saccharomyces cerevisiae. Extensive work by other laboratories has shown that heterologous expression of mammalian Bax in yeast is lethal . Although S. cerevisiae does not contain endogenous Bcl-2 family members or caspases, the initial events underlying Bax activity in yeast and mammalian cells are similar, including translocation of the protein to mitochondria, release of cytochrome c, and alterations in mitochondrial function [41, 42].
We have found that expression of mammalian PrP in yeast efficiently suppresses cell death induced by Bax . Our assay system utilizes yeast expressing mammalian Bax from a galactose-inducible promoter (GAL10). The yeast also constitutively express a form of PrP containing a modified signal peptide that we have shown to be targeted to the secretory pathway . The assay involves comparing growth of the yeast on glucose vs. galactose medium. The basic effect is shown in Fig. 2A, B, which demonstrates that Bax-expressing yeast transformed with the PrP plasmid grow well on galactose medium, while Bax-expressing yeast transformed with the vector plasmid fail to grow on galactose. In contrast, both stains of yeast grow well on glucose medium, which does not induce synthesis of Bax. Fig. 2C demonstrates that PrP does not alter expression levels of Bax, ruling out a trivial explanation for PrP rescue activity. We have carried out a detailed structure-function analysis to define which domains of the PrP molecule are necessary for the Bax protective effect (Fig. 2D). We are currently using this yeast system to perform screens aimed at identifying genes that are essential for PrP rescue activity, and to characterize the cellular mechanisms involved.
The ability of PrP to rescue yeast from Bax-induced death has been confirmed by Bounhar et al.  using dye exclusion and colony-forming ability as measures of cell viability. In contrast to us, however, these authors found that non-secreted (cytoplasmic) forms of PrP possessed rescue activity; neither of the two PrP constructs they analyzed entered the secretory pathway to any significant extent. They also reported that PrP failed to prevent growth inhibition caused by Bax, although PrP expression did have effects on the yeast cell cycle as monitored by flow cytometry.
Two kinds of mouse models dramatically exemplify a cytoprotective activity of PrPC in vivo. In one model, Shmerling et al.  expressed a series of N-terminally deleted forms of PrP in transgenic mice. Deletions encompassing residues 32 through 106 did not cause any abnormalities, but two larger deletions (Δ32-121 or Δ32-134, collectively referred to PrPΔN) caused a progressive neurodegenerative illness in mice lacking both copies of the endogenous PrP gene (Prn-p). Strikingly, a single Prn-p allele was sufficient to completely abrogate the phenotype of these animals. A similar phenomenon was described in certain Prn-p0/0 mouse lines in which the prion-like protein Dpl was ectopically expressed in the brain as a result of intergenic splicing events between the adjacent PrP and Dpl genes [47-50]. Dpl is structurally similar to PrPΔN, since it lacks regions homologous to the flexible, N-terminal tail of PrP. Reintroduction of wild-type PrP also rescued Dpl-expressing mice from neurodegenerative illness [49, 51-53].
We have recently documented an analogous rescue effect in Tg(Δ105-125) mice that express PrP harboring a deletion of residues 105-125 . These mice spontaneously develop a severe neurodegenerative illness that is lethal within one week of birth in the absence of endogenous PrP. This phenotype is reversed in a dose-dependent fashion by co-expression of wild-type PrP, with 5-fold over-expression delaying death beyond 1 year. Thus, the phenotype of Tg(PrPΔ105-125) mice is reminiscent of, but much more severe than, those described in mice that express PrPΔN and Dpl. The fact that PrP suppresses the phenotypes induced by PrPΔN, PrPΔ105-125 and Dpl suggests that a common mechanism facilitates the protective effect of PrP against each of these toxic proteins. We have speculated that the greatly enhanced toxicity of PrPΔ105-125, coupled with the requirement for supraphysiological levels of wild-type PrP to rescue the neurodegenerative phenotype, is due to increased affinity of PrPΔ105-125 for a hypothetical receptor that transduces the toxic signal .
It is instructive to compare the sequence domains of PrP that are essential for its cytoprotective activity in yeast, cultured mammalian cells, and mice. Such comparisons suggest a number of similarities between these three systems, as well as a few differences (Table 2). Thus, this cytoprotective function of PrPC appears to be partially but not completely conserved in evolution.
The first region of importance (residues 23-31; KKRPKPGGW in mouse) lies at the extreme N-terminus of PrP and includes four positively charged residues. This segment is known to play a role in endocytic trafficking of PrP [14, 55] and in localization of the protein to lipid rafts . This region has been found to be essential for PrP protection against Bax in yeast  and Dpl in cerebellar granule neurons . In transgenic mice, deletions beginning at residue 32 (PrPΔ32-80 and PrPΔ32-93) do not affect the ability of PrP to suppress neurodegeneration induced by PrPΔN  (Li et al., manuscript in preparation). In contrast, deletion of residues 23-88 obliterates the ability of PrP to rescue mice from Dpl-induced toxicity . The latter two results together implicate residues 23-31 in protection against PrPΔN and Dpl, although generation of Tg(PrPΔ23-31) mice will be necessary to confirm this inference. The requirement for residues 23-31 suggests that PrP cytoprotective activity may depend on localization to the cell surface, lipid rafts, or endosomal compartments.
Another domain that has been tested is the C-terminal, GPI addition signal. In both yeast  and human neurons , deletion of this region does not impair the ability of PrP to suppress Bax-mediated apoptosis, implying that PrP cytoprotective activity does not require tethering to the plasma membrane. Similarly, PrP rescue of granule neurons from Dpl-induced apoptosis does not require the GPI anchor . To determine whether these observations in yeast and cultured neurons hold true in vivo, transgenic mice expressing GPI-anchorless PrP  could be crossed with mice expressing PrPΔN or Doppel. The lack of requirement for the GPI anchor suggests that the cytoprotective activity of PrP does not involve signal transduction through the PrP molecule itself, but perhaps through a transmembrane protein to which PrP binds on the extracellular surface.
A related issue concerns the necessity for expression of PrP in the secretory pathway. In our experiments, deletion of the N- and C-terminal signal peptides (PrP23-231) completely abolishes the ability of PrP to rescue yeast from Bax-induced cell death . This result indicates that PrP must be targeted to the secretory pathway to be active in yeast, and also makes it unlikely that PrP interacts directly with cytosolic Bax (mechanism B in Fig. 1). In contrast, PrP23-231 retains full rescue activity against Bax in cultured human neurons . Whether this discrepancy implies a fundamental difference in the cytoprotective pathways operative in yeast and mammalian cells remains to be determined. The protective activity of cytoplasmic PrP in human neurons is surprising, since in other cultured cell types and in transgenic mice cytoplasmic PrP is toxic [39, 59].
The role of the PrP octapeptide repeats has also been analyzed in several systems. The repeats are known to bind copper ions, which have been postulated to play a role in the function of PrPC (see Section 6). Deletion of the octapeptide repeats abolishes the ability of PrP to protect cortical neurons from Bax , granule neurons from Doppel , and immortalized hippocampal neurons from serum deprivation . In contrast, removal of the repeats does not affect the ability of PrP to protect yeast from Bax  or transgenic mice from PrPΔN . These observations are difficult to reconcile, perhaps because the octapeptide repeats perform distinct functions in different cell types or in the context of different toxic insults.
Another interesting point of comparison concerns the effect of disease-causing mutations on the rescuing activity of PrP. D178N and E199K mutants retain the ability to rescue yeast from Bax-mediated cell death . In contrast, PG14, a mutant harboring an insertion of nine additional octapeptide repeats, lacks rescue activity . Biochemical analysis of these mutant proteins suggests that the loss of protective activity of PG14 may be due to its aggregated state, whereas D178N and E199K are completely soluble in yeast. This observation is consistent with studies in transgenic mice, since PG14 PrP is aggregated when expressed in the brain  and is also partially deficient in its ability to suppress the neurotoxicity of PrPΔ32-134 (Li et al., manuscript in preparation). PrP carrying the E199K mutation, which has been shown to be completely soluble in mouse models , maintains the ability to suppress Doppel-induced neurodegeneration . Although soluble in yeast, D178N PrP is partially detergent resistant in cultured neurons , which may explain the inability of this mutant to rescue neurons from Bax . Taken together, these results indicate that the cytoprotective activity of PrP requires a soluble form of the protein, and that mutations inducing aggregation impair activity.
A frequently discussed hypothesis to explain the pathogenesis of several neurodegenerative disorders involves chronic oxidative stress. Dysfunction of any of several interconnected cellular pathways is sufficient to cause oxidative stress in the brain, including impaired mitochondrial function, increased oxidative damage, defects in the ubiquitin–proteasome system, the presence of aggregated proteins, changes in iron metabolism, excitotoxicity, and inflammation [reviewed in 64].
Several lines of evidence suggest that PrPC may play a role in protecting cells from oxidative stress [reviewed in 65]. Perhaps the most compelling observation is that neurons (cerebellar granular and neocortical) cultured from Prn-p0/0 mice are more susceptible than neurons from wild-type mice to treatments with agents that induce oxidative stress, including hydrogen peroxide, xanthine oxidase and copper ions [66, 67]. Consistent with these cell culture results, brain tissue from Prn-p0/0 mice exhibits biochemical changes indicative of oxidative stress, such as increased levels of protein carbonyls and lipid peroxidation products . In addition, brain lesions induced by hypoxia and ischemia are larger in Prn-p0/0 compared to Prn-p+/+ mice [69-71]. Since hypoxia and ischemia probably cause neuronal death via oxidative damage, these observations also tie PrPC to protection from oxidative stress.
How might PrPC protect cells from oxidative stress? One possibility is that PrPC itself acts directly to detoxify reactive oxygen species (ROS). Consistent with this idea, it has been claimed that PrPC (both recombinant, and immunoprecipitated from brain tissue or cultured cells) displays a copper-dependent superoxide dismutase (SOD) activity [72, 73]. However, the biological significance of these results is questionable for several reasons. First, the SOD activity measured for recombinant PrP depended on refolding the protein from a denatured state in the presence of supra-physiological concentrations of copper. Second, even small organic molecules like amino acids can bind copper and exhibit weak dismutase activity. Finally, copper binds much more weakly to PrPC than to known cuproenzymes like Cu-Zn SOD , arguing against the possibility that the copper plays a specific catalytic role in PrPC. In a recent report, Jones et al.  failed to detect SOD activity above background levels in recombinant PrP refolded from a denatured state in the presence of excess copper (as in the original studies of Brown and colleagues), or in PrP loaded with copper after folding.
A second hypothesis is that PrPC acts indirectly to protect cells from oxidative stress by up-regulating the activities of other proteins, such as Cu-Zn SOD, that detoxify ROS. It has been reported that the enzymatic activity and the 64Cu loading of Cu-Zn SOD from the brains of Prn-p0/0 mice is 10-50% of normal [66, 76, 77]. Conversely, it has been claimed that the activity and copper loading of Cu-Zn SOD are increased in PrP-over-expressing mice. However, we  and others  have been unable to replicate these results. The activities of other anti-oxidant enzymes such as catalase and glutathione reductase have been reported to be decreased in Prn-p0/0 mice [77, 80], but whether PrPC plays a direct role in regulating these molecules remains to be determined.
It is also possible that PrPC acts either upstream or downstream of ROS to protect cells from oxidative stress. In some situations, for example, oxidative stress may activate apoptotic pathways . In such cases, the anti-apoptotic effects of PrPC described in Section 4 may account for the protein's ability to protect cells against oxidative stress.
Copper is an essential cofactor for a number of enzymes that catalyze redox reactions. Because ionic copper is highly reactive, however, cells have evolved specialized mechanisms for its uptake and transport . Defects in copper metabolism have been linked to a number of human diseases, including several neurodegenerative disorders . An extensive body of evidence has accumulated suggesting a connection between PrPC and copper ions [reviewed in 83, 84].
The most widely agreed upon observation is that PrPC is a copper-binding protein [85-88]. The histidine-containing octapeptide repeats specifically bind up to four Cu2+ ions copper in a pH-dependent and negatively cooperative manner, with an affinity that may be as high as 0.1 nanomolar (depending on binding site occupancy) . Binding involves coordination with nitrogen atoms in the imidazole side chains of histidine residues, as well as with nitrogen and oxygen atoms in main-chain amide linkages involving glycine residues. Two additional copper binding sites exist at residues 96 and 111 . Copper binding has been shown to cause conformational changes in the flexible, N-terminal tail of PrP [91, 92].
Copper ions also alter the biochemical and cell biological properties of PrPC. Copper causes PrPC in brain homogenates to assume an aggregated and protease-resistant form that is distinct from PrPSc . In addition, micromolar concentrations of copper stimulate endocytosis of cell-surface PrP via clathrin-coated pits [94-96]. This effect requires binding of copper to the octapeptide repeats, and exit of PrP from lipid raft domains . Based on the effects of copper on PrP trafficking, we have hypothesized that the protein serves as a receptor for cellular uptake or efflux of copper ions . It is also possible that PrPC acts as a sink for binding of copper ions at the cell surface without actually undergoing endocytosis.
One piece of evidence that would strongly support a role for PrPC in copper uptake or efflux would be a correlation between PrPC expression levels and the copper content of cells or tissues. An initial report indicated that the content of copper, but not of several other transition metals, is only 10% of normal in crude membranes, synaptosomes, and endosomes derived from the brains of Prn-p0/0 mice . A subsequent study from the same authors  reported that synaptosomes from Prn-p0/0 mice had a copper content that was 50% of the wild-type level, a considerably smaller difference than in the original report. Based on these results, the authors proposed that PrPC may play a role in regulating copper release at the synapse .
We have re-examined this subject by using mass spectrometry to measure the concentrations of several transition metals in brain tissue from wild-type and Prn-p0/0 mice, as well as in Tga20 mice that over-express PrP by 10-fold. We were unable to find any differences in metal content in either whole brain or of several subcellular fractions among mice of these three genotypes , and we believe that the results of Brown and colleagues are likely to be in error.
We have also utilized yeast as an experimental system to test the role of PrPC in copper metabolism . Much of what we know about cellular utilization of copper ions in eukaryotes is derived from studies of S. cerevisiae . We therefore tested the effect of PrP expression on the growth phenotypes of yeast strains harboring deletions of genes that encode key components of copper utilization pathways, including transporters, chaperones, pumps, reductases, and cuproenzymes. We failed to find an effect of PrP expression on the growth deficiency of any of the yeast strains tested.
Taken together, these studies in yeast and mammalian systems suggest that PrPC is unlikely to be part of a major pathway for copper uptake or efflux by cells and tissues. Of course, the results do not rule out involvement of PrPC in more specialized processes for metabolism or trafficking of this metal ion. Additionally, copper binding may play a regulatory role by modulating the physiological activity of PrPC.
Since most PrPC is localized to the cell surface, it is reasonable to hypothesize that the protein could participate in transmembrane signaling processes. Like other GPI-anchored proteins, PrPC resides in lipid raft domains on the plasma membrane, which are known to serve as molecular scaffolds for signal transduction [99, 100]. Since its polypeptide chain is entirely extracellular, PrPC would presumably need to interact with transmembrane adaptor proteins in order to transmit signals into the cytoplasm. Yeast two-hybrid screens have identified several signaling molecules that bind to PrP, including Grb2 and synapsin 1b . However, because these molecules are cytoplasmic, they are unlikely to directly associate with membrane-bound PrPC in a cellular environment. Clearly, a crucial challenge in the field is to identify the molecular components of putative PrPC-mediated signal transduction pathways.
There are now a number of studies suggesting that PrPC can activate transmembrane signaling pathways involved in several different phenomena, including neuronal survival, neurite outgrowth, and neurotoxicity. In some of these cases, signal transduction is initiated by interaction of PrPC with specific protein or peptide ligands. In other cases, PrPC appears to act constitutively.
An artificial situation in which PrPC-mediated signaling has been observed involves antibody-induced cross-linking of cell-surface PrPC. Presumably, the antibodies in this type of experiment are mimicking the action of naturally occurring PrPC ligands. Antibody-induced cross-linking of GPI-anchored proteins on the cell surface is a technique commonly used to activate phosphorylation-dependent signaling cascades in lymphocytes . Similarly, antibody-mediated cross-linking of PrPC on a neuroectodermal cell line (1C11) was found to stimulate the activity of the non-receptor tyrosine kinase, fyn . This effect was reported to require interaction of PrPC with the raft protein caveolin. In a subsequent study, it was shown that antibody-induced fyn activation in 1C11 cells led to downstream stimulation of NADPH oxidase and extracellular-regulated kinases (ERKs), as well as production of reactive oxygen species . The activities of several G protein-coupled serotonin receptors found on the surface of these cells were also altered by PrP cross-linking . The signaling pathways engaged by anti-PrP antibodies in 1C11 cells are postulated to have pro-survival effects .
Another kind of pro-survival signaling pathway that has been characterized involves an interaction between PrPC and stress-inducible protein 1 (STI-1). STI-1 was originally described as a co-chaperone found in macromolecular complexes with heat shock proteins of the Hsp70 and Hsp90 families . Although STI-1 lacks a signal sequence and is primarily localized to the cytoplasm and nucleus, some molecules have been reported to reside on the plasma membrane and to co-immunoprecipitate with PrPC . Interaction with cell surface STI-1 has been proposed to mediate PrPC-dependent protection of retinal explants from anisomycin-mediated cell death . This effect was found to depend on activation of a cAMP/protein kinase A pathway . A recent study also demonstrated that incubation of cultured hippocampal neurons with recombinant STI-1 stimulated neurite outgrowth in a PrP-dependent manner, an effect requiring signaling through a mitogen-activated protein kinase (MAPK) pathway .
Two recent studies have suggested a potential role for the PI3 kinase/Akt signaling pathway in the neuroprotective effects of PrPC. In one study , infarct volumes were measured in Prn-p+/+ and Prn-p0/0 mice that had been subjected to focal cerebral ischemia. It was found that Prn-p0/0 mice displayed significantly larger infarct volumes, demonstrating a protective role for PrPC in response to brain injury. Additionally, the level of phosphorylated Akt was diminished in Prn-p0/0 mice, indicating that PrPC may enhance Akt-dependent cell survival pathways to prevent damage inflicted by ischemic brain injury. In a second study , Akt activity was found to be diminished in neurons and brain tissue from Prn-p0/0 mice compared to Prn-p+/+ mice. Moreover, it was reported that pharmacological inhibition of Akt reduced the ability of PrPC to protect cells against oxidative damage.
Another context in which PrPC-mediated signal transduction events have been observed involves neurite outgrowth. Cell surface PrPC facilitates axonal outgrowth via cis and trans interactions with N-CAM, a process that involves recruitment of N-CAM to lipid rafts and activation of fyn kinase . Treatment of cultured neurons with recombinant PrP also enhances neurite outgrowth and neuronal survival, concomitant with activation of several kinases, including fyn, PKC, PKA, PI-3 kinase/Akt, and ERK [113, 114].
The signaling processes described thus far involve positive effects of PrPC on neuronal survival or differentiation. However, there is also evidence that PrPC can mediate neurotoxic effects via activation of specific signaling cascades. One example involves the synthetic peptide PrP106-126, which displays certain biochemical properties of PrPSc and has been used to mimic the effects of PrPSc on cultured cells . PrP106-126 is toxic to cultured neurons and neuronal cell lines, but only those that express PrPC, suggesting that the toxic action is mediated by a PrPC-dependent signaling pathway [116, 117]. Consistent with this idea, PrP106-126 has been reported to stimulate a number of intracellular kinase cascades, including those involving p38, ERK1/2, and JNK1/2 [118-120]. In a second example of PrPC-mediated neurotoxicity, PrPC expression has been reported to sensitize neurons and neuronal cell lines to the apoptotic action of the kinase inhibitor, staurosporine [121-123]. This effect required endocytosis of PrPC, and was mediated by increased activity of the key transcriptional regulator p53.
Several experimental observations suggest that PrPC could play a role in synaptic structure, function or maintenance. This hypothesis is consistent with the fact that synaptic pathology is often a prominent feature of prion diseases . Light and electron microscopic immunocytochemical studies, as well as localization of a PrP-EGFP fusion protein, indicate that PrPC is preferentially concentrated along axons and in pre-synaptic terminals [125-130]. In addition, PrPC is subject to anterograde and retrograde axonal transport [131, 132], and PrP-EGFP fusion proteins can be visualized in what appear to be axonally transported synaptic vesicles (Medrano and Harris, unpublished observations). Incubation of cultured hippocampal neurons with recombinant PrP induces rapid elaboration of axons and dendrites, and increases the number of synaptic contacts . This result suggests that PrPC could play a regulatory role in synapse formation. There is evidence that PrPC could serve a function at peripheral as well as central synapses. It was reported that PrPC is concentrated at the neuromuscular junction where it is localized in the sub-synaptic sarcoplasm, possibly associated with endosomal structures . In addition, nanomolar concentrations of recombinant PrP have been found to potentiate acetylcholine release at the neuromuscular junction .
Electrophysiological recordings from brain slices of Prn-p0/0 mice also support a functional role for PrP in synaptic transmission. In hippocampal slices from Prn-p0/0 mice, it was initially reported that long term potentiation was impaired and receptor-mediated fast inhibition involving GABA-A receptors was decreased [135, 136]. However, this result was subsequently disputed . More recent studies have demonstrated a positive correlation between the expression level of PrPC and the overall strength of glutamatergic transmission in the hippocampus, with PrP-over-expressing mice exhibiting supra-physiological responses . At least part of this effect seemed to result from more efficient recruitment of pre-synaptic fibers as the level of PrPC increased. Correlating with reduced after-hyperpolarization seen in hippocampal CA1 neurons of Prn-p0/0 mice , cerebellar Purkinje cells from these animals have been reported to show decreased Ca2+-activated K+ currents . Reduced after-hyperpolarization was observed in pyramidal neurons regardless of whether the PrP gene was deleted pre- or postnatally .
Prn-p0/0 mice have been reported to display several other neurobiological abnormalities that may also relate to the participation of PrPC in synapse formation and function. These include alterations in nerve fiber organization , circadian rhythm , and spatial learning .
PrPC has been found to interact with several proteins involved in cell adhesion. For example, cross-linking experiments have identified a binding interaction between PrP and neural cell adhesion molecule (N-CAM) . In cultured hippocampal neurons, this interaction results in redistribution of N-CAM to lipid rafts, activation of fyn kinase, and enhancement of neurite outgrowth . Laminin, a major structural component of basement membranes, has also been shown to be a binding partner of PrPC. In neurons, laminin plays a significant role in cell proliferation, supports neurite outgrowth, and aids in cellular migration. Graner et al.  have shown that PrP binding to laminin promotes neurite outgrowth in PC12 cells and hippocampal neurons. Laser ablation of cell-surface PrP caused retraction of neurites . Finally, expression of PrP has been found to enhance aggregation of neuroblastoma cells, although the molecular mechanisms underlying this phenomenon are uncertain .
A great deal of effort in the prion field has been devoted to understanding of the chemical nature of the infectious agent, and testing the validity of the protein-only mechanism of infectivity. In contrast, much less attention has been paid to the question of how PrPSc or other abnormal forms of PrP cause CNS pathology. It is now becoming clear, however, that prions may kill nerve cells by virtue of their ability to perturb the normal, physiological activities of PrPC. In this section, we will outline three different hypotheses explaining how changes in PrPC function can cause neurodegeneration [reviewed in 148, 149].
The most widely discussed hypothesis is that prion pathology is attributable to a toxic gain of function mechanism (Fig. 3A). In this view, PrPSc possesses novel toxic properties that are not related to the normal, physiological function of PrPC. For example, aggregates of PrPSc may block axonal transport, interfere with synaptic function, or trigger apoptotic pathways.
An alternative hypothesis for prion toxicity postulates that PrPC possesses a biological activity that is lost upon conversion to or contact with PrPSc (Fig. 3B). Loss of this putative PrPC function would then cause neurodegeneration. In principle, loss of any of several of the putative functions of PrPC discussed above could produce pathogenic consequences. However, the anti-apoptotic activity of PrPC (Section 4) is most easily accommodated in such a mechanism, since loss of this activity could lead directly to neuronal death.
Two considerations would seem to argue against a loss of function mechanism in prion diseases. First, genetic ablation of PrP expression, either prenatally [1, 2] or postnatally , has relatively little phenotypic effect, and does not produce any features of a prion disease. Thus, loss of PrPC function cannot, by itself, account for prion-induced neurodegeneration. However, it is possible that a loss of function mechanism exacerbates pathology caused by toxic gain of function or other mechanisms. For example, a cytoprotective activity of PrPC that is dispensable under normal conditions may become essential in the disease state due to cellular or organismal stress. Loss of PrPC function as a pathogenic mechanism also appears to be incompatible with the dominant mode of inheritance of familial prion diseases. In this regard, however, PrPSc or mutant PrP may sequester wild-type PrPC into aggregates that lack functional activity, thereby producing a dominant-negative effect.
A consideration which argues for a loss rather than a gain of function in some familial prion diseases relates to the effect of PrP mutations on the thermodynamic stability and biochemical properties of PrP. Although some mutations markedly destabilize the protein, favoring misfolding and formation of PrPSc-like aggregates, others do not [150, 151]. It is difficult to explain the pathogenicity of the latter mutants on the basis of the accumulation of toxic protein aggregates. Rather, it seems more likely that these proteins are deficient in some functional property normally displayed by wild-type PrP. Consistent with this proposal, we have shown that an octapeptide insertional mutation in PrP partially impairs the cytoprotective activity of the protein in both a yeast model  and a transgenic mouse model (Li et al., manuscript in preparation).
A third possible hypothesis for prion pathogenesis involves a subversion of the normal neuroprotective function of PrPC (Fig. 3C). In this mechanism, interaction with PrPSc converts PrPC from a transducer of neuroprotective signals into a transducer of neurotoxic signals. Consistent with this model, there are now a number of experimental situations in which expression of PrPC in neurons, rather than being neuroprotective, appears to be essential for conferring sensitivity to PrP-related neurotoxic insults. First, Prn-p0/0 neurons are resistant to the toxic effects of PrPSc supplied from grafted brain tissue  or from nearby astrocytes . Second, scrapie-inoculated mice expressing a GPI-negative version of PrP develop minimal brain pathology and neurological dysfunction despite the accumulation of numerous PrPSc-containing amyloid plaques . This result implies that PrP must be membrane-anchored to efficiently transduce a toxic signal. Third, Prn-p0/0 neurons in culture have been found to be resistant to apoptosis induced by exposure to the synthetic peptide PrP106-126, which has been used as a model for PrPSc .
How might the neuroprotective activity of PrPC be subverted to produce a neurotoxic effect? Presumably, a physical interaction of PrPSc with PrPC is required. One possibility is that PrPSc induces aggregation of cell surface PrPC, thereby generating a neurotoxic rather than a neuroprotective signal. Consistent with this model, cross-linking of PrPC with anti-PrP antibodies induces apoptosis of CNS neurons in vivo . Alternatively, PrPSc may bind to and block specific regions of PrPC, thereby altering the signaling properties of the latter.
Support for the latter mechanism comes from Tg(PrPΔ105-125) mice we have made that express PrP containing a deletion of a highly conserved block of 21 amino acids in the unstructured, N-terminal tail of the protein  (see Section 4.3). The neurotoxicity of PrPΔ105-125 observed in these animals suggests a model in which residues 105-125 constitute a critical binding site for interaction between PrPC and a hypothetical, cell-surface receptor (designated Tr) that can transduce either neuroprotective or neurotoxic signals. PrPC binding to Tr normally elicits a neuroprotective signal, but deletion of the 105-125 domain of PrPC subverts this interaction in such a way that a neurotoxic signal is generated. We speculate that exogenous PrPSc perturbs binding interactions between Tr and the 105-125 domain of PrPC, thereby producing a neurotoxic effect equivalent to deletion of residues 105-125 . A similar phenomenon might occur with the synthetic peptide PrP106-126, which is toxic to cultured neurons in PrPC-dependent fashion . In this scenario, the neurotoxic pathways activated by PrPSc and PrP106-126 would be the same as those activated by PrPΔ105-125, PrPΔN and Dpl. However, wild-type PrPC would have opposite effects in the two situations: it would suppress the toxicity of PrPΔ105-125, PrPΔN and Dpl, but it would be required for the toxicity of PrPSc and PrP106-126.
Toxic gain of function is commonly invoked to explain the phenotypes of other dominantly inherited neurodegenerative disorders including Alzheimer's disease, Huntington's and other polyglutamine expansion diseases, Parkinson's disease, frontotemporal dementia, and amyotrophic lateral sclerosis. In these disorders, it is postulated that intracellular or extracellular aggregates of the relevant misfolded protein (Aβ, huntingtin, α-synuclein, tau, or SOD1) possess a neurotoxic activity that is not directly related to the normal, physiological function of the parent protein . However, in at least some of these disorders, a loss of function mechanism similar to what we have proposed for prion diseases (Section 10.2) may also be operative. Huntington's and Parkinson's diseases provide especially clear examples of this idea.
Huntington (Htt), like PrP, has been found to possess physiological activities that promote neuronal survival, and loss of these activities may contribute to disease pathogenesis . For example mice lacking Htt show extensive cell death in the embryonic ectoderm . Moreover, it has been reported that wild-type Htt up-regulates transcription of BDNF, and that mutant Htt lacks this activity, leading to compromised neuronal differentiation and survival . Wild-type Htt may also facilitate vesicular transport of BDNF along axons . Consistent with a neuroprotective effect of Htt, over-expression of wild type Htt in transgenic mice significantly reduced the cellular toxicity elicited by various Htt mutants [160, 161].
Similarly, there is evidence that α-synuclein, which comprises the Lewy bodies of Parkinson's disease, may have anti-apoptotic activities. Expression of wild type α-synuclein in primary neurons and neuronal cell lines protects the cells from death induced by serum deprivation, H2O2-induced oxidative stress, and glutamate-induced toxicity [162-164]. This protective effect was not observed with the two disease-associated mutants of α-synuclein (A30P and A53T) .
These results suggest that some of the pathogenic mechanisms operative in prion diseases may be closely related to those underlying other, more common neurodegenerative disorders. This idea is consistent with evidence that prion neurotoxicity is attributable to small, non-infectious PrP oligomers that are distinct from PrPSc, and that are similar to the pathogenic aggregates seen in other protein misfolding disorders [148, 165].
For many years the PrPSc isoform of the prion protein has consumed the attention of investigators because of its relationship to infectivity, and its specific accumulation during the disease process. PrPC, the cellular isoform, was of interest primarily because it served as a necessary precursor to PrPSc, but its intrinsic physiological function seemed irrelevant to understanding the disease process. It is now clear, however, that alterations in the normal function of PrPC may play a crucial role in causing or contributing to the disease phenotype. For this reason, elucidating the physiological activity of PrPC has become a major priority in the prion field. In this review, we have discussed each of the major functions that have been proposed for PrPC. Although the evidence is not conclusive for any them, we have emphasized the anti-apoptotic activity of PrPC as being of particular interest, because it has been demonstrated in a variety of experimental systems, from yeast to mice. In addition, there is evidence that loss or subversion of this activity may be related to neuronal death in prion diseases.
Understanding the normal function of PrPC has important implications for the therapy of prion disorders. At present, most therapeutic strategies are directed at inhibiting the formation of PrPSc . If alterations in PrPC function play an important role in prion-induced pathology, then an alternative approach is to target the cellular pathways mediating the biological actions of PrPC. In this regard, it might be possible to use the physiological activity of PrPC to develop in vitro assays to screen for drugs that have therapeutic potential. In addition, if the toxicity of PrPSc is partly attributable to a loss of PrP function, then over-expression of wild-type PrP may represent a strategy for suppressing the disease phenotype. Conversely, reduction of PrP expression, a strategy that has been proposed for preventing or treating prion diseases [153, 167], may have detrimental consequences due to loss of the neuroprotective activity of PrPC.
Future progress in understanding the normal function of PrPC will require identification of physiologically relevant PrP-interacting partners, and elucidation of the cellular pathways in which they participate. Accomplishing these objectives is likely to be facilitated by the use of model systems such as yeast  and Drosophila  that are amenable to genetic analysis, as well as by the development of additional cultured cell and transgenic mouse models. Given the rapid pace of work in the prion field, new discoveries will certainly be forthcoming.
Work in the Harris laboratory is supported by grants from the NIH (NS052526 and NS040975) and the Hope Center for Neurological Disorders at Washington University. H.M.C. was supported by a pre-doctoral fellowship (NS04691003) from the NIH.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.