A Function for the Prion Protein?

Introduction

Protein function is often observed directly following protein isolation, or is deduced by loss of function following gene knockout or by analogy with proteins of known function and similar amino acid sequence. None of these is true in the case of prion proteins because aside from the association with the pathogenesis of the spongiform encaphalopathies, no single obvious function has been described for these molecules until recently. The first two topics in this volume (see refs. 1-3), concentrated on the characterization of the infectious agent, and led to the introduction of the term "prion" in 1982 (4). But it was not until the positive association of the infectious agent, PrPSc, with a normal host gene locus, prnp, that real opportunities to consider protein function in relation to the disease phenotype arose (5). The identification of the prion gene on chromosome 2 of the mouse (chromosome 20 in the human) (6), and the determination of its sequence (7), led to the translation of the encoded protein and speculation concerning its function.

Prion Sequences

Translation of the original DNA sequences of the mouse and hamster prion genes described the structural features of the protein. Prion protein is 254 amino acids long (253 in the human) and has an unusual structure, which was unique when first described Figure 1. The coding region for the mature protein is preceded by a 22 amino acid signal peptide and followed by a sequence of 23 amino acids that is removed following expression and replaced by a glycosyl phosphatidylinositol (GPI) tail at serine 231. These features mark the protein for expression on the cell surface and make it conceivable that PrPc may be grouped with other GPI anchored proteins that are commonly associated with cell to cell signalling, adhesion or cellular defense (8).


Schematic representation of the prion protein amino acid sequence. Details are taken from the mouse sequence, but the overall features are shared by all prion proteins. Numbers along the scheme indicate the amino acid positions of each feature and a correlation with the known three-dimensional structures is given below. The position of the single disulphide bond and the glycosylation sites are indicated.

Fig. 1. Schematic representation of the prion protein amino acid sequence. Details are taken from the mouse sequence, but the overall features are shared by all prion proteins. Numbers along the scheme indicate the amino acid positions of each feature and a correlation with the known three-dimensional structures is given below. The position of the single disulphide bond and the glycosylation sites are indicated.

Secondary structure predictions, even prior to the derivation of a three-dimensional structure, suggested that the carboxy terminal domain was likely to adopt a distinct fold mostly made up of a-helices, but that the N-terminal domain lacked clear structural features. The carboxyl domain also contains a single disulphide bond between residues 179 and 214 and two potential N-linked glycosylation sites at residues 181 and 197. Although lacking in unambiguous secondary structure predication, the amino terminal domain of PrPc has a unique sequence of eight amino acids, rich in glycine and proline, that is repeated 4x between residues 51 and 91, and is separated from the carboxyl half by a stretch of hydrophobic residues centered on residue 120. The PrPc sequence is unique, making impossible the prediction of a function based on linear or secondary structure comparisons.

The sequence of PrPc from a number of other species has since been determined. They show a remarkable degree of conservation at the amino acid level (Fig 2), and the same set of unique structural features. The sequence of avian PrP shows a greater number of repeats, with fewer amino acids in each (9), a feature shared with the recently determined reptilian (turtle) PrP (10). All other features of the protein have confirmed, however, an unprecedented homology among so diverse a set of organisms. It seems likely from these data that prion proteins with essentially the same structure are present in all vertebrates, a fact that may signify an important role for the encoded protein and a tight conservation of sequence throughout evolution.

Alignment of the sheep, cattle, human, mouse, turtle and chicken prion amino acid sequences showing the overall conservation of sequence and split between the mammalian and reptilian/avian sequences.

Fig. 2. Alignment of the sheep, cattle, human, mouse, turtle and chicken prion amino acid sequences showing the overall conservation of sequence and split between the mammalian and reptilian/avian sequences.

The original sequence of PrPc suggested that the molecule had features similar to some proteins with a tendency to aggregate (7), and that this may explain the predisposition to amyloid formation. More recently however, many proteins have been shown to be capable of forming amyloid, under the correct conditions, so that the requirement for particular sequence motifs is uncertain (11).

A recent investigation (12) has identified the first prion protein-like gene (prnd) in the mouse. Prnd expresses a short protein with homology to the C-terminal domain of PrPc. The protein, termed Doppel, is not normally expressed in adult mice, but is expressed in two strains of Prnpo/o mice. These mice differ from others, in that they develop late-onset neuropathological changes including Purkinje cell degeneration (13). Speculation concerning the mechanism of Doppel expression has centered on the possibility that the genetic manipulation used to create the two strains of Prnpo/o mice caused a deletion of specific inhibitor sequences between the prnp and prnd genes resulting in the prnd gene being transcribed by the prnp promoter. Despite the similarities between the two proteins, Doppel lacks most of the more highly conserved regions, including the hydrophic core region and the octameric repeats. It is unlikely, therefore, that Doppel and PrPc have a common function although interference of either by the other remains a possibility.

Prion Structure

It is not unusual for proteins to exhibit similar folds in the absence of significant sequence homology (e.g., those shared by HIV matrix antigen and interferon y [14]). Thus, although direct alignment of prion sequences with those in the databases failed to identify matches, homology based on the tertiary structure of the protein could be instructive. The solution of the structure for the carboxyl domains of mouse and hamster PrPc, obtained by nuclear magnetic resonance spectroscopy (15,16) has, however, failed to suggest a role for the molecule. The C-domain is well ordered, containing three a-helices and a short section of antiparallel P-sheet (Fig 3). The GPI anchor occurs at the end of the final helix and suggests an orientation of the molecule with respect to the cell membrane (Fig 3). The two longest helices are held together by the single di-sulphide bond. Recent evidence suggests the carboxyl domain folds very rapidly, and is mostly unaffected by variation in pH or temperature (17). The N-terminal domain, by contrast, which includes the octameric repeats, has no defined structure. The lack of distinct secondary and tertiary structure to the N-terminal domain of PrPc coupled with the unusual structure of the octarepeats suggest that ligand binding in this region may be necessary for the adoption of a stable tertiary structure.

Using peptides representing only the octarepeats Hornshaw et al. showed that copper, as Cu++, was bound by both the mouse and chicken sequences (18). Equilibrium dialysis experiments, first using a recombinant fragment of PrPc equivalent to the N-terminal region to amino acid 98, and later with full length recombinant or wild-type PrPc, has since confirmed that mouse PrPc binds several atoms of copper (19-21). Copper binding was not observed however following expression and purification of full length chicken prion protein (22). Moreover, further studies on peptides representing the octarepeat region have not yet allowed an unambiguous mechanism of copper binding to be formulated. Viles et al., using a variety of spectroscopic techniques, concluded that copper was co-ordinated by the histidine residues of each octarepeat in a tetrad planer arrangement reminiscent of Cu/Zn superoxide dismutase (23).

The structure of the Golden hamster prion protein taken from the published work of James et al., (16). The 142-amino acid protein extends from position Gly90 to Ser23i and contains three y-helices and two short sections of p-sheet. A structure is unavailable for the region prior to amino acid 90. The addition of the GPI tail to Ser231 suggests the orientation to the cell membrane as shown, but this is not proven.

Fig. 3. The structure of the Golden hamster prion protein taken from the published work of James et al., (16). The 142-amino acid protein extends from position Gly90 to Ser23i and contains three y-helices and two short sections of p-sheet. A structure is unavailable for the region prior to amino acid 90. The addition of the GPI tail to Ser231 suggests the orientation to the cell membrane as shown, but this is not proven.

By contrast, Miura et al., concluded that the form of contact with copper depended heavily on the pH of the interaction, and suggested that, at neutral and basic pH, each copper ion was bound in an intrachain configuration by two adjacent glycine and one histidine residues. Under weakly acid conditions, however, copper binding changed to an interchain configuration, with implications for the formation of prion aggregates (24). Similar uncertainty surrounds the redox state of the bound copper. Ruiz et al., using a copper chelator, bathocuproine disulfonate (BC), suggested that the bound copper was present in reduced form, and that the tryptophan residues present in each repeat sequence were the likely redox acceptor (25). Shiraishi et al., also used BC to measure the redox state of bound copper but concluded that the metal was present in the nonreduced form (26). They suggested that sequestering of the metal as the divalent cation prevented copper induced generation of reactive oxygen implicating PrPc as a protective metal chelator molecule. Wong et al. reported extensive methionine oxidation in preparations of recombinant mouse and chicken prion proteins following refolding in the presence of copper suggesting that the copper ion is redox active when present within the full length protein (27). Other copper binding proteins involved in the transport of copper across mammalian cell membranes are thought to bind copper in the Cu+, rather than the Cu++ form (28).

If copper is a natural ligand of PrPc, then it is a reasonable premise that prion protein function depends on its acquisition, or that imbalances in copper level, and the consequences of it, would go hand in hand with the presence or absence of PrPc.

Resistance to Oxidative Stress

The experimental production of prion protein knockout mice allowed an investigation of the role of PrPc in prion disease. Prnpo/o mice do not express the product of prnp, yet remain healthy, suggesting that loss of PrP expression does not directly result in disease. A role for the protein is clear, however, from the observation that Prnpo/o mice cannot be infected with the scrapie agent. Although susceptibility to infection is an identifiable phenotype for PrPc, it seems nonsensical to suppose that this is its only cellular role, given the evolutionary conservation apparent in PrP sequences.

Analysis of the neurotoxicity of both PrPSc and the neurotoxic peptide, PrP106-126, in cell culture experiments using cells derived from Prnpo/o mice indicated that PrPc expression is also necessary for the observed toxicity (29,30). This was later confirmed in the mouse model of scrapie. Following transplantation of PrPc expressing brain tissue into Prnpo/o mouse brains, and infection of the transplanted mice with the scrapie agent, PrPSc accumulated in PrPc-expressing tissue. The surrounding tissue was free of neurodegeneration, indicating that PrPSc was not toxic to PrF-deficient neurons (31).

Further analysis of PrP106-126 toxicity indicated that the peptide kills neurones in culture by causing toxic radical release from microglia and by inducing a reduced resistance to those radicals in neurones. PrP106-126 could only reduce the resistance to oxidative stress in neuron cultures that express PrPc (32,33). The reactive oxygen species produced by wild-type microglia in the presence of PrP106-126 were insufficient to kill neurones that did not express PrPc (32), or neurones not exposed to the peptide (D. R. Brown, unpublished data).

Although the brains of Prnpo/o mice (lacking Doppel expression) are normal (34), neuronal cultures produced from neonatal Prnpo/o mice show greater sensitivity to culture conditions, and died at a faster rate than wild-type neurons (33). A similar observation has been made for cell lines generated from other Prnpo/o mice (35). The decreased viability of Prnpo/o neurons in culture was found to result from increased sensitivity to oxidative stress (33). Superoxide, generated enzymatically in the cultures, also killed more Prnpo/o neurons than wild-type neurons (33). Using PC12 cells, increased resistance to oxidative stress was found to correlate with increased PrPc expression (36). This also correlated with increased sensitivity to the toxic effects of PrP106-126. There is, therefore, a strong parallel between the in vitro phenotype of PrPc-deficient cells and the phenotype induced by PrP106-126, at least in terms of resistance to oxidative stress.

Studies on the activity of the antioxidant, cytosolic enzyme, Cu/Zn superoxide dismutase SOD-1 also support the idea that PrP106-126 may induce a PrPc-deficient phenotype in neurons. PrP106-126 treatment induced decreased activity of SOD-1 in cultured cerebellar cells (33). Studies of the brains of two strains of Prnpo/o mice indicated that these mice have reduced SOD-1 activity in vivo without treatment (19,33). The reduction in SOD-1 activity was not caused by decreased expression of protein or transcription of messenger RNA, but was likely to have resulted from decreased incorporation of copper, necessary for activity, into the SOD-1 molecule (37). Furthermore, mice expressing higher levels of PrPc than wild-type mice had correspondingly higher SOD-1 activity in their brains.

Further study of cultured cells provided other examples of diminished cellular resistance to oxidative stress in PrPc-deficient neurons. Although the activity of glutathione peroxidase and catalase appear to be unaltered in neuronal cultures from Prnpo/o mice, there is evidence for altered glutathione metabolism resulting from changes in the activity of glutathione-S-transferase (GST) (38). A similar result was obtained for cells treated with PrP106-126, which also diminished the activity of GST, as well as depleting cells of the reduced form of glutathione (39).

Despite a wealth of evidence from cell culture experiments which suggests that PrPc expression is linked to resistance to oxidative stress it is important to note that there is, as yet, no evidence that this is a significant role for PrPc in vivo.

Copper Metabolism

Experiments with PC12 cells indicated that those expressing high levels of PrPc were more resistant to oxidative stress. Additionally, the same cells were found to be more resistant to copper toxicity (40). Furthermore, a PC12 cell line selected for its resistance to copper toxicity, was also more resistant to oxidative stress and showed higher levels of PrPc-expression than standard PC12 cells (40).

Toxicity of copper and cobalt to cultures of 6-d-old cerebellar cells. The cells were treated for 2 d with CuCl2 or CoCl2. They were then assayed for relative cell survival, using a standard MTT assay. Control = 100% survival o-o wt; •-• Prnpo/o.

Fig. 4. Toxicity of copper and cobalt to cultures of 6-d-old cerebellar cells. The cells were treated for 2 d with CuCl2 or CoCl2. They were then assayed for relative cell survival, using a standard MTT assay. Control = 100% survival o-o wt; •-• Prnpo/o.

Cultures of primary neurons or astrocytes from Prnpo/o mice were also found to be more sensitive to the toxicity of copper (41). Cultures from mice overexpressing PrPc were more resistant to copper toxicity than wild-type cells. Other divalent cations were not more toxic to Prnpo/o cerebellar neurones than to wild-type neurons suggesting PrPc-expression selectively protects against copper toxicity (Fig 4).

Peptides based on the octarepeat sequence bind copper (see Subheading 3.) (42,43). When a 32-amino acid peptide encoding this sequence was added to cultures of cerebellar neurons, it protected against the toxicity of copper. This effect was strongest on Prnpo/o cerebellar neurons (41). Additionally, this pep-tide protected against superoxide toxicity. Copper can convert superoxide to other toxic substances, and the mechanism of peptide action could have been through copper binding and prevention of these reactions. Depletion of the cellular ability to bind copper has been shown to increase cellular sensitivity to copper toxicity (44). An antibody that binds near the octameric region was found to specifically enhance the toxicity of copper (41) possibly through prevention of copper binding to the PrPc expressed by wild-type neurons. Together, these results suggest that PrPc can act as a copper chelator.

PrPc immunoprecipitated from mouse brain contains large amounts of copper but no other divalent cation (21). Cultured cells from wild-type cells contain more copper than those from Prnpo/o mice (19). The difference between wild-type and Prnpo/o cerebellar cell membrane fractions in terms of copper content can be abolished by treatment with an enzyme that cleaves GPI-anchored proteins from cells, suggesting that the difference in copper content of wild-type cells result from binding of copper by one or more GPI anchored proteins, such as PrPc (45). The synaptosomal fraction of mouse brain also contains large amounts of copper, much higher than in similar preparations from Prnpo/o mice (Fig 5); (19). It is clear that PrPc is highly expressed at the synapse (46) , leading to the conclusion that PrPc binds copper both in vitro and in vivo.

The brain has high levels of copper, second only to the liver. The brain shows sensitivity to imbalances in copper levels. In Wilson’s disease and Menke’s disease, mutations in P type adenosine triphosphatases (ATPases) alter copper metabolism (for reviews, see refs. [47-49]). In Menke’s disease there is failure of copper transport from the intestine which leads to copper deficiency (50) and the inability of the brain to develop normally. Copper deficiency may lead to neurodegeneration (51). In Wilson’s disease, mutations in a P type ATPase (52) found mostly in the liver, lead to an increase in deposition of copper in the brain and kidneys (53). This is probably due to failure of the Wilson type ATPase to transport copper across the canalicular membrane of the liver into the bile (54). In addition, release of the main serum transporter of copper, ceruloplasmin, is impaired. This is probably caused by the failure of the Wilson-type ATPase to donate copper to the necessary proteins in the excretory pathway. The accumulation of copper in the brain subsequently leads to neurodegeneration. The sensitivity of the brain to copper suggests that control of copper uptake and detection of abnormal levels of extracellular copper are important. Copper is necessary to the brain in terms of the activity of molecules such as SOD, cytochrome C and tyrosinase, and also for synaptic transmission. However, the exact mechanism of copper uptake by the brain remains unclear.

High copper content has been previously localized to the secretory apparatus of neuronal terminals (55,56) from where it is released upon depolarization (57). In this situation, there is a high local accumulation of copper that must be dealt with rapidly after the transmission event. The copper that is released at these sites is initially taken up by a high-affinity copper binding process (58).

Analysis of subcellular fractions of cells from 10-d-old wild-type (open bars) and PrP knockout mice (closed bars) for the levels of two divalent metal cations, copper and zinc. The mitochondrial data are unpublished work of DRB.

Fig. 5. Analysis of subcellular fractions of cells from 10-d-old wild-type (open bars) and PrP knockout mice (closed bars) for the levels of two divalent metal cations, copper and zinc. The mitochondrial data are unpublished work of DRB.

The presence of copper in the synaptic cleft is important for the modulation of many kinds of synapses. High copper reduces y-aminobutyric acid (GABA) and glutamate uptake (59) and reduces GABA-induced currents (60). However, copper enhances dopamine uptake (61). Copper also regulates the distribution of muscarinic cholinergic receptors, enhancing uptake at low concentration, and inhibiting at high concentrations (62,63). Copper-deficient mice show increased levels of GABA receptors and muscarinic cholinergic receptors (64). Copper can inhibit transmission at N-methyl-D-aspartate (NMDA) receptors, and shows a higher affinity for NMDA receptors to which the agonist has already bound (65). Collinge et al. (66) found defects in synaptic activity of PrPc-deficient mice, which included reduced long-term potentia-tion, which is dependent on NMDA receptor activity, and also a reduction in GABA-type inhibitor currents (66). Therefore, it is possible that PrPc may assist synaptic transmission by preventing the deleterious effects of copper release at the synapse.

A Molecular Function for PrPc

Copper is an abundant cation with an ability to capture electrons. Metabolic control of copper is essential both for prevention of harmful effects, such as generation of oxidative damage, by its ability to generate reactive oxygen species in the presence of water, and also because it can be utilized to control electron transfer such as occurs in respiration or dismutation of superoxide (67). Copper binding proteins are therefore essential for normal cellular metabolism. Because PrPc is a membrane-associated protein, it may function similarly to other membrane-associated copper binding proteins which eliminates a role for PrPc purely as a storage protein, such as the copper binding metallothionens. Similarly, although copper sequestration may be an advantage of PrPc expression, the abundance of extracellular copper transporting proteins, such as albumin and transferrin, eliminate this as a likely sole function.

Three classes of copper-binding proteins are membrane-associated (68). First, oxidases such as cytochrome C, are mostly associated with the mitochondria, and are involved in respiration, an unlikely function for PrPc. Second, extracellular SOD exists in a membrane-bound form, and, as previously stated, cells from Prnpo/o mice have reduced resistance to oxidative stress, and have reduced levels of SOD activity (33). The changes in phenotype were originally considered to result from decreased incorporation of copper into the cytosolic SOD. PrPc is a small molecule and it seems likely that any substrate molecule would also be small. Investigation of possible SOD activity of PrPc, therefore, has some validity. Third, some membrane-associated copper binding proteins are involved in copper translocation. Use of Cu67 has identified two possible copper-uptake mechanisms into neuronal tissue with high and low affinities (69). However, the nature of the proteins responsible for this has not been identified.

Copper Uptake and Release

Little or no free copper exists in the brain. Copper is mostly present on transport proteins or as chelates with other compounds, such as peptides or amino-acids. Accordingly, uptake of copper into neurons is greatly enhanced when the copper is provided in the form of a chelate. The fate of added copper was followed using radioactive copper with Cu67 because its half life is greater than Cu64 and because Cu64 often contains a higher percentage of contaminants. Three strains of mice were used for these studies: mice overexpressing PrPc, wild-type mice, and PrPc-deficient mice. Uptake of nonchelated Cu67 was identical for all three strains of mice. However, Cu67 provided as a histidine chelate was taken up at a rate that could be related to the level of expression of PrPc (70). Membrane fractions from cerebellar cells overexpressing PrPc showed the highest binding of Cu67 whilst the lowest binding was by PrPc-deficient cells. The rate of entry of Cu67 into the cell was assessed in greater detail and kinetic parameters determined. Values for Vmax increased with increased expression of PrPc. On the other hand values of Km (in the nM range) were not greatly different, the only significant difference being between overexpressing and PrPc-deficient cells (70). These differences in Cu67 uptake are consistent with the idea that there is an increase in the number of Cu67- binding sites between the three strains, which may be related to the level of PrPc expression.

Immunoprecipitation of the cytosolic enzyme, SOD-1, from cells loaded with Cu67 indicated that Cu67, could be incorporated into SOD-1 in proportion to the level of PrPc expressed by the cells (37). This suggests that Cu67 can be incorporated into cellular proteins when taken up in association with PrPc.

The highest level of PrPc expression is at the synapse (above), and, because copper is released during quantal release, it is possible that PrPc may have an important role in regulating copper concentrations there. Cells loaded with Cu67 for 30 min were allowed to release copper spontaneously over a period of 1 h, and reached a plateau after 30 min, when the amount of copper was stable (70). Cells prepared in this way were then treated with veratridine, which is a depolarizing agent that induces release similar to synaptic release. Veratridine induced release of copper from cells at a level that could be related to the expression of PrPc. Cells expressing no PrPc released almost no copper when induced by veratridine; PrPc-overexpressing cells released more copper than wild-type cells (70).

Electrophysiological experiments indicate that copper applied to cerebellar slices inhibited the amplitude and frequence of inhibitory currents measured on Purkinje cells of PrPc-deficient cells but not on wild-type cells (19). This suggests that some form of protection against copper is missing at PrPc-defi-cient synapses, and that such protection could be mediated by PrPc. These results suggest that PrPc expression alters a number of aspects of copper metabolism including copper uptake, copper utilisation, synaptic release and may aid in protective sequestration of copper.

A Synaptic SOD

Sources of recombinant PrP protein (rPrPc) from bacterial expression systems have allowed direct assessment of the ability of PrP to bind and use copper (21). After purification of rPrPc and subsequent refolding of the protein from urea, rPrPc was found to be both proteinase sensitive and to possess a secondary structure similar to that found by other researchers (71).

Refolding of the protein in the presence of copper led to copper binding to the octameric repeat region as shown by the failure of a specific deletion mutant lacking residues 59-91 to bind the same level of copper (21). Direct measurement of the copper bound suggested that four copper ions bound per complete octarepeat region confirmed the measurements made in earlier work (23). Copper bound to the Histidine repeat region at the C-terminus of the protein, added to facilitate purification, could be eliminated by removal of the tail by partial trypsin digestion (21). The solubility of the protein was increased when the protein was refolded in the presence of copper, suggesting that it adopted a more compact structure in keeping with the increase in secondary structure measured for octarepeat peptides following copper binding (43).

The rPrPc protein with and without specifically bound copper was also quantitatively assayed for SOD activity. These assays indicated that rPrPc (from both chicken and mouse) could catalyze superoxide dismutation at a rate equivalent to one tenth that of SOD-1 (based on protein concentration), one of the most potent enzymes known which catalyzes superoxide dismutation at around 100,000x the spontaneous rate (72). Thus, rPrPc has SOD-like activity in vitro, a finding that was confirmed for native protein immunoprecipitated from mouse brain (21).

Stringent controls were carried out to ensure that the activity measured was a form of enzymatic catalysis, and was not caused by simple Fenton chemistry arising from the ligating of copper to the protein. Ethylenediamine tetracetic acid does not inhibit the SOD-like activity of rPrPc. Deletion of the specific octameric repeat region of the protein abolishes the activity, even when copper is bound to the uncleaved His-tag (21). A peptide based on the octameric repeat region, and with copper bound to it, has no SOD-like activity. Similarly, if rPrPc is refolded without copper and copper is added afterwards, the copper/rPrPc mixture does not show the SOD-like activity of rPrPc refolded with copper (21). Therefore, the observed catalytic activity is not the result of presence of copper alone, suggesting that a catalytic site is required for the activity. Deletions of regions in the C-terminus of the protein abolish the activity, but do not prevent copper binding to the octarepeat region (D. R. Brown, unpublished observations). Further work is required to delineate the active site, but it is clear that regions outside the octameric repeat region are required for the SODlike activity. There was a high degree of methionine oxidation associated with copper-refolded rPrPc which, as most methionines are in the C terminal domain, indicated the involvement of the whole molecule in a redox-based reaction (27). This is characteristic of anti-oxidant copper binding enzymes, such as SOD-1 (73).

Currently, there are three accepted mammalian SODs (74). The first of these, Cu/Zn superoxide dismutase (SOD-i), is found only in the cytosol. It consists of a dimer of two identical subunits, one of which binds copper and the other zinc. Manganese superoxide dismutase (MnSOD, or SOD-2) is found in the mitochondria and binds manganese. These two enzymes are found in all cells at varying concentrations, and often show increased expression or activity in the presence of oxidative stress. Similarly, PrPc expression increases when PC12 cells are grown in the presence of oxidative stress (36). The third mammalian SOD is known as extracellular superoxide dismutase (EC-SOD or SOD-3) (75). It exists in three different isoforms, and, like PrPc, binds four copper atoms, and is either released into the extracellular environment or is bound to the cell surface. However, the expression of SOD-3 is very low in the brain (76). PrPc, on the other hand, has its highest expression in the brain, and is also highly expressed at neuromuscular junctions (77). In the brain, the expression of PrP is highest at synapses, and it is conceivable that it may act as a synaptic SOD that may even be released during transmission. Superoxide is known to inhibit synaptic transmission and the presence of SOD activity at the synapse may have protective effects.

The expression of PrPc is not restricted to neurons or even to the synapse. In neurons, mostly one glycoform of PrPc is transported from the soma to the axonal terminals (78). Astrocytes (79,80), microglia (81), leukocytes (82) and cells of other nonneuronal tissues (83,84) also express PrPc. Therefore, its activity as an SOD, although of less direct consequence, may not be limited to the synapse or the brain.

Clearly, the identification of SOD activity associated with purified PrPc requires further verification. If confirmed, however, a defined enzymatic function for PrPc has broad implications for understanding prion disease. The changes in secondary structure, following conversion of PrPc to PrP c, mean that it is unlikely that PrPSc would retain SOD activity. As PrPc levels are depleted during the accumulation of PrPSc, oxidative damage to the expressing cells would ensue. Alternatively, if PrPSc retained some capacity to bind copper, Fenton chemistry, unchecked by the action of PrPc could inducing oxida-tive damage to neurons in the vicinity of PrPSc deposits. Alternatively, deposition of chelated copper within PrPSc could cause local copper depletion and induce reduced resistance to oxidative stress, an effect that has already been observed in culture for the neurotoxic PrP peptide, PrP106-126.

The finding that PrPc can exhibit a specific antioxidant activity which requires copper links together the physical measurement of copper binding by PrPc with the biology of PrPc loss. Lack of PrPc expression results in reduced resistance to oxidative stress, and results in a depletion of copper found at the synapse. This observation is not just true for neurons. Astrocytes deficient in PrPc expression are also more sensitive to oxidative stress (85), and cell membrane fractions from such cells have lower copper content (D. R. Brown, unpublished observations).

Together, these results indicate a role for PrPc consistent with its distribution and known biochemical properties to date, and strongly suggest the possibility that PrPc functions primarily as a copper dependent antioxidant protein.

Next post:

Previous post: