Address correspondence and reprint requests to Jochen W. Herms, Institut für Neuropathologie, Ludwig-Maximilians-Universität, Marchioninistraße 17, D-81377 München, Germany. E-mail: Jochen.Herms@inp.med.uni-muenchen.de
The fundamental physiological function of native cellular prion (PrPC) remains unknown. Herein, the most salient observations as regards prion physiology are critically evaluated. These include: (i) the role of PrPC in copper homeostasis, particularly at the pre-synaptic membrane; (ii) involvement of PrPC in neuronal calcium disturbances; and (iii) the neuroprotective properties of PrPC in response to copper and oxidative stress. Ultimately, a tentative hypothesis of basic prion function is derived, namely that PrPC acts as a sensor for copper and/or free radical stimuli, thereby triggering intracellular calcium signals that finally translate into modulation of synaptic transmission and maintenance of neuronal integrity.
prion protein peptide corresponding to residues 106–126 of the human prion protein
reactive oxygen species
stress-inducible protein I
voltage-gated calcium channel
Prion diseases, also known as transmissible spongiform encephalopathies, comprise a range of neurodegenerative disorders characterized pathologically by the triad of spongiform change, neuronal loss and reactive gliosis (Kretzschmar et al. 1996; Budka 1997). They include Creutzfeldt–Jakob disease (CJD), bovine spongiform encephalopathy (BSE) and scrapie. The aetiology of prion diseases can be infectious, sporadic, or genetic (for review, see Prusiner 1998). The agent responsible for both transmission (in infectious cases) and for causing neurodegeneration is now thought to be the prion protein itself, or rather a pathogenic isoform of the native prion PrPC, known as PrPSc (Pan et al. 1993). The latter is characterized by a rich β-sheet structure, a tendency for aggregation, resistance to protease-digestion and detergent insolubility (McKinley et al. 1983). Importantly, PrPSc has the ability to convert normal prion into its own aberrant structure by a template-directed refolding mechanism, consequently catalysing its own formation (Prusiner 1991). In the course of prion disease, PrPSc accumulates in the CNS. Because PrPSc is non-functional, the brain may be effectively depleted of the physiological function of prion protein. This in itself may be an important contributing factor to the pathophysiology of neurodegeneration.
The function of PrPC is still unknown. In the past decade or so, several significant inroads have been made in this regard, with a number of hypotheses being proposed. In this review, the most salient observations as regards prion physiology are critically evaluated. Ultimately, a tentative hypothesis of basic prion function is derived.
Cellular location of PrPC
The first step in identifying the physiological role of PrPC is to ascertain the precise cellular location where the protein carries out its function(s). Thus, prion protein is a sialoglycoprotein that is attached to the outer leaflet of the plasma membrane via a C-terminal glycophosphatidylinositol (GPI) anchor (Stahl et al. 1990). Although it is most abundant in the CNS, it is also expressed in many non-neuronal tissues, including blood lymphocytes, gastroepithelial cells, heart, kidney and muscle (Horiuchi et al. 1995; Fournier et al. 1998). This is significant, because it directly implies that prion protein has a widespread biological role, albeit most relevant to neuronal cells. Further, in the brain, PrPC is particularly localized to synaptic membranes, most likely the pre-synaptic domain. This is evidenced by immunohistochemical and immunoelectron microscopic studies (Fournier et al. 2000), as well as by finding prion enriched in synaptic plasma membrane fractions (Herms et al. 1999). The predominant pre-synaptic localization suggests a function that involves prion in synaptic transmission and neuronal excitability.
PrPC binds copper(II) ions: copper buffering and uptake at the pre-synaptic membrane
One of the most documented, and perhaps one of the most salient, features of prion protein is its ability to bind copper(II) ions. Electronspray ionization mass spectroscopy and tryptophan fluorescence spectroscopy studies have demonstrated that, among divalent metal ions, PrPC selectively binds copper(II) (Stockel et al. 1998). The major copper(II)-binding site has been identified as the N-terminal region (encompassing residues 60–91 in human PrPC), specifically an octarepeat domain consisting of four sequential repeats of the sequence PHGGGWGQ (Hornshaw et al. 1995; Brown et al. 1997a). The relevance of the copper(II) binding to the physiological function of PrPC is reflected in the high degree of conservation of the octarepeat region among prion proteins from mammalian and avian species. Most studies have concluded that copper(II) to octarepeat binding stoichiometry is 1 : 1, i.e. the octarepeat region humPrP60-91 binds four copper(II) ions (Hornshaw et al. 1995; Brown et al. 1997a; Miura et al. 1999; Viles et al. 1999; Kramer et al. 2001; Garnett and Viles 2003). Another significant aspect of copper binding to PrPC is its high pH dependence: binding is most favoured at a physiological pH of ∼7.4, and falls sharply under mildly acidic conditions (pH < 6.0; Miura et al. 1999; Viles et al. 1999; Kramer et al. 2001). Affinity data from full-length PrPC showed that the protein is saturated at about 5 µm copper(II) (Kramer et al. 2001). Similar dissociation constant (Kd) values were reported for shorter fragments spanning residues 23–98 (5.9 µm; Miura et al. 1999) and residues 60–91 (6.7 µm; Hornshaw et al. 1995). These values are compatible with the physiological copper(II) concentration released within the synaptic cleft during synaptic vesicle release (15 µm); higher concentrations of 100–300 µm are achieved during neuronal depolarization (Hartter and Barnea 1988). It has been proposed that the N-terminal octarepeat region contains a tight binding site for a single copper(II) ion with a Kd of 10−14 M. A second tight copper site (Kd = 10−13 M) was detected outside of the octarepeat region, around histidines 96 and 111 of human PrPC (Jackson et al. 2001). However, a recent study contradicts these findings, arguing against fentomolar affinity of PrPC for copper(II) on the primary basis that glycine is able to compete strongly for copper (Garnett and Viles 2003). Therefore, given its µm affinity, PrPC would not be expected to take up copper(II) ions from inside the cell. Importantly, exposure of PrPC-expressing N2a mouse neuroblastoma cells to high copper(II) concentrations (i.e. in the millimolar range) stimulated endocytosis of the prion protein (Pauly and Harris 1998). In addition, deletion of the four octarepeats or mutation of the histidine residues (His68 and His76) in the central two repeats abolished endocytosis of PrPC expressed in human neuroblastoma SH-SY5Y cells (Sumudhu et al. 2001). This indicates that copper(II)-mediated endocytosis of PrPC is unlikely to be mediated by a high-affinity site around histidines 96 and 111. PrPC-mediated endocytosis of copper has also been demonstrated by expression of GFP- PrPC constructs in SN56 cells (Lee et al. 2001).
Having recognized that PrPC binds copper(II) ions at the surface membrane in a co-operative and pH-dependent manner, causing structural folding of the N-terminal region, the next question to consider is whether this process has physiological relevance in vivo. In support of this, copper concentration in synaptosomal preparations of Prnp°/° mice (which do not produce any PrP protein) was decreased by 50% with respect to wild-type (Herms et al. 1999). This 50% reduction is much too great to be solely attributed to loss of the copper bound to PrPC. A straightforward explanation would be that PrPC has a direct role in brain copper metabolism, with the protein transporting copper(II) ions from the extracellular milieu to acidified cellular vacuoles (Pauly and Harris 1998). However, no differences were observed in the uptake of radioactive copper between Prnp°/° and wild-type synaptosomes (Bucholz M. et al., unpublished results). Another recent study also concluded that PrPC does not participate in the uptake of extracellular copper(II), at physiological concentrations of copper (Rachidi et al. 2003). Moreover, total brain copper content in Prnp°/° and Tga20 (which express ∼10 times the normal level of PrP) mice was not found to be significantly different from that in wild-type mice (Kretzschmar et al. 2000; Waggoner et al. 2000). Similarly, copper, zinc superoxide dismutase (Cu, Zn-SOD) and cytochrome c oxidase activities (both cuproenzymes) in brain extracts from Prnp°/°, wild-type and Tga20 mice did not reveal any significant differences (Waggoner et al. 2000). Determination of the levels of various divalent cations in whole brain homogenates of control and scrapie-infected mice did show lower levels of copper(II) in the latter; however, the levels of zinc, magnesium and calcium were also decreased, so the effect was not specific to copper (Wong et al. 2001a). Given all these results, it appears unlikely that PrPC is primarily responsible for uptake of copper into neurones, or that its chief role is in specialized trafficking pathways for delivery of the metal to Cu, Zn-SOD or other cuproenzymes.
We think that another more likely hypothesis is that, in the unique setting of the synapse, PrPC acts to buffer copper(II) levels in the synaptic cleft following release of copper ions as a result of synaptic vesicle fusion. Thus, the micromolar Kd of PrPC, together with its localization at the pre-synaptic membrane, would allow prion to effectively buffer copper at the site from where it is being released. Further, PrPC ensures redistribution of copper back into the pre-synaptic cytosol, thereby maintaining synaptosomal copper concentrations. Such copper uptake can occur by transfer of copper ions from PrPC to specialized copper transport proteins in the membrane. These would have a higher affinity for copper than PrPC. The maintenance of copper levels in the pre-synaptic cytosol is physiologically important, because nerve endings release copper into the synaptic cleft upon depolarization (Kardos et al. 1989). Importantly, such a hypothesis would explain why the total copper content in brain lysates from Prnp°/° mice remains unchanged, because PrPC is essentially altering the distribution (but not the overall amount) of copper within the brain. This proposed function of prion protein would additionally serve to protect the synaptic membranes from copper toxicity, in view of the propensity of copper to participate in Fenton-type redox reactions that lead to generation of harmful reactive oxygen species (ROS; Halliwell and Gutteridge 1984).
Electrophysiological studies in Prnp°/° mice have shown abnormalities that can be explained on the basis of increased synaptic copper concentrations as a result of decreased buffering by PrPC (Herms et al. 1999; Kretzschmar et al. 2000). For instance, impaired long-term potentiation and weakened GABAA receptor-mediated currents (Collinge et al. 1994) and reduced late after-hyperpolarization currents (Colling et al. 1996) have been reported. After-hyperpolarization currents are mediated by Ca2+-activated K+ channels. In fact, patch-clamp studies on cerebellar Purkinje cells from Prnp°/° mice revealed impaired activation of Ca2+-dependent K+ channels (Herms et al. 2001). Indeed, Prnp°/° cerebellar granule neurones (CGN) were found to exhibit low resting [Ca2+]i levels and a decreased response of [Ca2+]i to high K+-induced depolarization (Herms et al. 2000). Significantly, disturbances in intracellular calcium homeostasis have also been observed in scrapie-infected GT1-1 neuronal cells, and in CGN treated with the prion fragment PrP106-126 (which has most of the pathogenic characteristics of PrPSc; Thellung et al. 2000). Several lines of evidence point to a selective modulation of L-type voltage-gated calcium channels (VGCCs) as the basis of the changes in calcium homeostasis caused by a lack of PrPC (Herms et al. 2000, 2001; Thellung et al. 2000). Because copper is known to reduce influx of calcium through VGCCs at concentrations similar to those found in the synaptic cleft (Nam and Hockberger 1992), it can be argued that PrPC acts to protect the VGCCs from inhibitory copper concentrations, thereby maintaining intracellular calcium homeostasis.
Taken together, the experimental data considered so far seems to be fairly supportive of a hypothesis whereby the location of PrPC in the synaptic membrane allows the protein to maintain copper content in the pre-synaptic cytosol, and at the same time buffer against toxic levels of copper(II) ions in the synaptic cleft, e.g. during high synaptic activity. These two complementary activities assist in preserving neuronal electrophysiology, particularly via maintenance of intracellular calcium homeostasis.
A neuroprotective function for PrPC in response to copper and oxidative stress: more than a synaptic SOD?
There are several other experimental findings related to prion function which cannot be explained on the basis of copper buffering and uptake alone. In particular, a considerable amount of data has been accumulated, implicating PrPC as a neuroprotective molecule, especially in response to pro-oxidative insults. The suggestion that copper-bound PrPC is involved in free radical pathways should not be surprising. The considerable overlap between systems controlling homeostasis of redox-active metals (such as copper, iron, manganese) and oxygen radical metabolism has been extensively documented (Avery 2001). The physiologically intimate connection between transition metals and oxidative stress is exemplified by the activities of metal-detoxifying proteins on the one hand, and antioxidant enzymes on the other. Thus, yeast and mammalian metallothioneins harbor potent antioxidant activities. Conversely, an antioxidant metalloenzyme like Cu, Zn-SOD also participates in intracellular copper and zinc homeostasis (Wei et al. 2001). Returning to PrPC, therefore an additional role for this copper(II)-chelating protein in relation to oxidative stress should be seriously considered.
To start with, lack of PrPC results in neuronal phenotypes sensitive to oxidative stress induced by superoxide (O2.–) and hydrogen peroxide (H2O2). Thus, prion-deficient cerebellar neurones showed increased sensitivity to O2.– generated by xanthine/xanthine oxidase (Brown et al. 1997b). Increased sensitivity of Prnp°/° CGN to H2O2-induced cell death was also clearly evident, when compared to wild-type (White et al. 1999). Importantly, Prnp°/° CGN were found to have lower glutathione reductase (GR) activity (White et al. 1999). Because GR functions in the regeneration of cellular GSH (glutathione), lower GR activity would decrease the breakdown of H2O2 by glutathione peroxidase, and hence be related to the increased sensitivity of Prnp°/° CGN neurones to H2O2. Interestingly, rabbit kidney epithelial cells (RK13) expressing murine PrPC had significantly higher levels of GR activity and GSH (Rachidi et al. 2003). Thus, it appears that regulation of the GSH antioxidant system is a central feature of PrPC expression. It should be noted, however, that high levels of PrPC expression in the RK13 cells actually made them more susceptible to H2O2 toxicity, i.e. despite raised GR activity and intracellular GSH. This phenomenon is most likely due to excessive generation of Fenton-derived ROS on application of H2O2 concentrations greater than 200–300 µm.
Additional neuroprotective effects of prion were evidenced when immortalized cell lines from Prnp°/° hippocampal cells were found to be more sensitive to serum withdrawal than control wild-type neurones. Transfection of a PrPC or Bcl-2 expressing construct into the Prnp°/° hippocampal neurones protected them from apoptosis induced by serum deprivation, thus abolishing the sensitive phenotype (Kuwahara et al. 1999). Experiments using PrPC constructs have demonstrated that prion protein also potently inhibits Bax-induced apoptosis in primary human neurones. Significantly, deletion of the octapeptide repeat domain abolished this novel neuroprotective function of PrPC, whilst elimination of the GPI-anchoring sequences had no effect (Bounhar et al. 2001). Both removal of serum from neuronal cultures and Bax expression are known to induce intracellular oxidative stress (Atabay et al. 1996; Kirkland et al. 2002). However, ROS levels were not measured in the above-mentioned experiments on serum-withdrawal and Bax-induced apoptosis, and therefore the antiapoptotic effect of PrPC cannot be directly correlated with protection against oxidative stress.
The role of PrPC in modulating neuronal antioxidant homeostasis is further strengthened by evidence of oxidative stress in models of prion disease. GT1–7 cell lines infected with prion had significantly raised levels of lipid peroxidation, increased sensitivity to glutathione depletion, and lowered Cu, Zn-SOD, manganese SOD (Mn-SOD), GR and glutathione peroxidase (GPX) activities (Milhavet et al. 2000). In essence, this study on prion-infected cells again confirmed that depletion of normal cellular PrPC is associated with an impaired response, and increased susceptibility, to oxidative stress. Markedly increased levels of lipid and protein oxidation were also found in brain lysates from sporadic CJD disease subjects (Wong et al. 2001b). Similarly, studies of oxidative stress markers have been carried out on brain tissues from Prnp°/° mice. Once again, higher levels of lipid and protein oxidation were measured (Wong et al. 2001c), as well as decreased SOD activities (Klamt et al. 2001).
Despite such strong evidence favouring a neuroprotective role of cellular prion, particularly against oxidative stress, the actual biological activity of the protein that mediates the neuroprotection is as yet unknown. One has to presume that it would involve copper(II) binding, given the fundamental importance of the N-terminal octarepeat region in defining the biochemical, molecular and cellular activities of PrPC. It is in this context that the concept of PrPC as a synaptic SOD was proposed. In vitro experiments using recombinant mouse and chicken PrPC refolded to incorporate copper(II), revealed that PrPC has SOD activity (Brown et al. 1999). The rate constant for the dismutation reaction was estimated to be 4 × 108 M/s, i.e. a magnitude lower than that of Cu, Zn-SOD (6.4 × 109 M/s) but similar to Mn-SOD (6 × 108 M/s). Prion protein immunoprecipitated from wild-type mouse brains was found to contribute 10–15% of total SOD activity in vivo (Wong et al. 2000). In addition, SOD activity of PrPC was dependent on the level of copper(II) incorporated into the molecule; thus, incorporation of three or four copper atoms resulted in higher activity than two, whilst no activity at all was detected with only one copper ion bound to PrPC (Brown et al. 2001). The actual mechanism of the dismutation reaction is still unknown. Presumably, it involves reduction of copper(II) to copper(I), with the formation of H2O2. Both neuronal and non-neuronal cells expressing copper-loaded PrPC were significantly more resistant to superoxide toxicity (Brown et al. 2001; Rachidi et al. 2003).
Notwithstanding its attractiveness as a hypothesis, we believe there are several arguments that do not favour the proposal that the primary function of PrPC is to act as a synaptic SOD. Firstly, it is uncharacteristic of a dismutating enzyme to exhibit pH-dependent binding to its catalytic metal, as does PrPC. Secondly, affinity of the apoprotein to its metal is normally much higher than the 5–7 µm estimated for copper(II)-binding to PrPC; Cu, Zn-SOD, for example, has an affinity of about 10−14M for copper(II) (Rae et al. 1999). The Kd of 10−14M for copper(II)-binding to the N-terminal octarepeat segment of PrPC that has been measured (Jackson et al. 2001), is now strongly contested (Garnett and Viles 2003). Further, the fentomolar Kd value represented binding of the four octarepeats to a single copper(II) ion, i.e. one tight site per polypeptide. As mentioned above, PrPC incorporating a single copper ion did not manifest SOD activity. Thirdly, PrPC was found to contribute only 10–15% of total SOD activity in brain extracts – this is in concordance with the modest increased susceptibility of Prnp°/° cells to O2•–. Indeed, PrPC offered greater protection against H2O2 and Bax-induced apoptosis. Lastly, and perhaps most importantly, the SOD activity of PrPCper se explains neither the neuroprotective functions that have been described for the protein, nor the gross disturbance in antioxidant defenses found in infected brains or in brains lacking native prion.
Concluding hypothesis: redox signalling by PrPC modulates intracellular calcium homeostasis
At this point, then, it can be confidently concluded that PrPC maintains synaptic integrity and neurophysiological function by going beyond mere copper buffering and/or dismutase activity. The question remains, however: how can PrPC influence resistance to a wide range of insults? If copper(II)-bound PrPC were to activate an intracellular signalling pathway, then such an effect would not be unexpected. The most likely candidate for the intracellular messenger would be calcium. Evidence has already been presented above suggesting that loss of normal prion protein is associated with a disturbance in neuronal calcium homeostasis. Although reduced Ca2+ influx through VGCCs has been indicated as a likely cause, other data suggest that prion also affects the mechanism of Ca2+ release from internal endoplasmic reticulum stores; for example, scrapie-infected mouse neuroblastoma cells have a 90% reduction in the IP3 second messenger response to bradykinin stimulation (Kristensson et al. 1993; Wong et al. 1996). Therefore, a disturbance of calcium homeostasis is conceivably the basic physiological alteration in prion-deficient neuronal cells. Intracellular calcium is a critical component of signalling pathways that lead to structural and functional changes in neurones. The calcium signal itself is usually in the form of an elevation of [Ca2+]i as a consequence of synaptic activity, for instance depolarization. Hence, PrPC could influence both neuronal survival and synaptic physiology by regulating calcium homeostasis.
It has been proposed that a neuroprotective signalling pathway, involving cAMP, is activated by interaction of PrPC with stress-inducible protein I (STI 1) at the cell surface (Chiarini et al. 2002; Zanata et al. 2002). PrPC completely lacking the copper-binding domain still interacted with STI 1 and activated the signalling pathway. Given that, as discussed in earlier sections, the affinity of PrPC for copper has important implications for its function, the physiological relevance of PrPC-STI 1 binding has yet to be demonstrated. Indeed, several molecules have been reported to associate with PrPC, such as Bcl-2, the laminin-receptor, heparin and dystrophin (Martins et al. 2001). However, support for relevance in the in vivo situation is generally lacking.
How then does the physiological copper(II)-binding/release activity of PrPC link with regulation of calcium homeostasis? One potential scenario is that PrPC acts as a modulator of calcium flux in response to copper(II) and/or ROS, with the role of copper being that of enabling redox signalling and triggering metal-dependent biochemical and cellular responses. Thus, exposure to raised copper levels (e.g. during high synaptic activity) would cause several copper(II) ions to bind prion and initiation of Ca2+-activated signalling cascades. Moreover, PrPC attached to copper(II) allows the protein to participate in redox reactions. As we have seen, PrPC incorporating at least two copper atoms is able to dismutate superoxide into H2O2. The latter could presumably function as a redox signalling molecule, activating an as yet unidentified membrane protein kinase. PrPC can, in addition, react with H2O2. In the presence of physiological concentrations (as low as 10 µm) of H2O2 and copper(II), PrPC was cleaved into the octapeptide repeat region (McMahon et al. 2001). Interestingly, O2.– also induced a site-specific cleavage of the prion amino terminus (McMahon et al. 2001). In other words, extracellular H2O2 and/or possibly the H2O2 generated by PrPC itself (through dismutation), can react with the copper(II) ions bound to the octarepeat region of PrPC As a consequence of Fenton-type reactions, PrPC is cleaved and copper released. We further propose that such physiological ROS processing may also directly deliver a redox signal that induces protective responses and/or influences synaptic activity. In support of this, application of 0.01% H2O2 enhanced inhibitory synaptic activity in wild-type, but not in Prnp°/° mouse Purkinje cells (Herms et al. 1999).
To conclude, we endorse a hypothesis (Fig. 1) in which PrPC, due to its localization at the plasma membrane, acts as a sensor for strong copper and/or ROS stimuli, with the role of copper being that of generating a signal through redox chemistry, thereby switching on Ca2+-dependent signalling pathways leading to downstream effects that modulate synaptic transmission and maintain neuronal integrity. Identifying the signalling pathways in which PrPC is involved is thus of prime importance for further elucidating the enigmatic biological function of prion protein.
We thank Joe V. Bannister for his help in preparation of the manuscript. This work was supported by the Deutsche Forschungsgemeinschaft, BMBF and the Bayerische Forschungsverbund For Prion.