Cleavage of the Amino Terminus of the Prion Protein by Reactive Oxygen Species*

Relatively limited information is available on the processing and function of the normal cellular prion protein, PrP C . Here it is reported for the first time that PrP C undergoes a site-specific cleavage of the octapeptide repeat region of the amino terminus on exposure to reactive oxygen species. This cleavage was both copper-and pH-dependent and was retarded by the presence of other divalent metal ions. The oxidative state of the cell also decreased detection of full-length PrP C and increased detection of amino-terminally fragmented PrP C within cells. Such a post-translational modification has vast implications for PrP C , in its processing, because such cleavage could alter further proteolysis, and in the formation of the scrapie isoform of the prion protein, PrP Sc , because abnormal cleavage of PrP Sc occurs into the octapeptide repeat region. Prion diseases or transmissible spongiform encephalopathies are a of neurodegenerative affecting include scrapie

Prion diseases or transmissible spongiform encephalopathies (TSEs) 1 are a group of neurodegenerative diseases affecting both animals and humans (1). Such disorders include scrapie in sheep, bovine spongiform encephalopathy in cattle, and Creutzfeldt-Jakob disease (CJD), Gerstmann-Strä ussler-Scheinker syndrome, and fatal familial insomnia in humans (2,3). TSEs may be of sporadic, genetic, or infectious origin, and the agent believed to be responsible for these disorders is a protein molecule termed PrP Sc , which is a conformational variant of the normal cellular prion protein (PrP C ) (4). PrP C is a 33-35-kDa glycosyl-phosphatidylinositol-anchored protein that is expressed by most tissues and in particular at high levels by neuronal cells. Unlike PrP C , PrP Sc possesses a high ␤-sheet content and is partially protease-resistant (5). It still remains unclear how PrP C is converted to PrP Sc and whether the infec-tious agent consists solely of PrP Sc , because the infectivity of in vitro converted PrP Sc remains to be established (6). However, it is becoming increasingly evident that PrP C is essential for disease development on infection (7), although the function of the protein still remains obscure.
The amino terminus of PrP C contains a series of octapeptide repeats possessing the following consensus sequence: PHGGG-WGQ. This region, which is among the most conserved regions of mammalian PrP (8), has been implicated in the binding of divalent metal ions, and in particular copper (9). Whether this binding is of structural or functional significance is not known. However, NMR studies have shown that the recombinant protein possesses a flexible amino terminus lacking any given structure (10), and it is reported that copper binding not only adds structure to this region (11) but may also lend stability to the carboxyl terminus (12). Because of the capacity of PrP C to bind copper, this protein has been implicated in copper transport and metabolism (13) and in the defense mechanism of the cell against oxidative insult, possibly through a regulation of the Cu,Zn superoxide dismutase activity (14). However, a recent study may lead to a re-evaluation of the link between PrP and copper metabolism (15). In addition, although it would appear that the octapeptide repeat region may not be essential for disease development (16), the insertion of additional octapeptide repeats, which could increase metal binding, results in a pathogenic mutation (17). Thus the role of copper binding by this highly conserved region of PrP C still remains unclear.
Copper is an essential redox transition element and a crucial component of various enzyme systems (18), and yet under conditions of oxidative stress copper ion-mediated damage to proteins through reactive oxygen species (ROS) is very significant. Links between oxidative stress, ROS, and neurological disorders have been reported in Parkinson's and Alzheimer's diseases, amylotrophic lateral sclerosis (19), and aging itself. ROS, such as the superoxide radical (O 2 . ) and hydrogen peroxide (H 2 O 2 ), which although itself not an ROS is an important mediator of oxidative stress in neurons (20), can oxidatively modify biomolecules (21). The generation of such ROS is promoted by the presence of copper and iron through the Fenton reaction (22). In addition, when the metal ion involved is protein bound, the oxidative-reductive reaction can locally generate oxygen species that may react at specific sites in the protein itself, impairing activity or resulting in cleavage (23). Because oxidative stress may play a role in TSEs (24), it is possible that metal-catalyzed oxidation of PrP C , at its octapeptide repeat region, may be an important factor in TSEs. In fact in Alzheimer's disease, the ␤-amyloid precursor protein, which is another copper-binding protein, reduces Cu 2ϩ on binding, and it is believed that its processing occurs under conditions of oxidative stress (25). Considering this, and that PrP C may also * This work was supported by grants from the Cellule de Coordination Interorganismes sur les Prions and the CNRS. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ possess the capacity to reduce Cu 2ϩ (26), we examined the effect of ROS on the processing of PrP C .
The cleavage of PrP C in the conserved middle region of the molecule, as well as the cleavage of PrP Sc close to the octapeptide region, has been well documented both in cultured cells and in the brain (27)(28)(29). However, the regulation and the physiological and pathological significance of these cleavages remain largely unknown. Here we demonstrate, in the presence of ROS, that the octapeptide repeat region of PrP C undergoes a copper-dependent cleavage. This cleavage can be inhibited by the presence of additional alternative divalent metal ions. It is possible that such ROS cleavage may be involved in PrP C processing, a post-translational modification that could also result in the activation of the protein. Additionally, because abnormal cleavage of PrP Sc occurs into the repeats, such cleavage of PrP C may be important in TSE disease development.
Chinese Hamster Ovarian (CHO) Cell Cultures-Construction of cDNA encoding the wild-type PrP that is derived from the Prn-p a allele and the generation and cultures of stably transfected lines of CHO cells expressing wild-type MoPrP have been described previously (30).
Preparation of Conditioned Medium-Subconfluent MoPrP CHO cells were rinsed twice with phosphate-buffered saline and were then overlaid with opti-MEM and incubated for 24 h at 37°C in a 5% CO 2 humidified atmosphere. Medium was then removed and centrifuged for 5 min at 10,000 rpm and 4°C; the supernatant was then recovered (conditioned medium) and used immediately, after the addition of the protease inhibitors (0.5 mM phenylmethylsulfonyl fluoride, 1 g/ml pepstatin, and 1 g/ml leupeptin), for assay purposes.
Effect of Reagents on PrP C Detection-Conditioned medium was incubated with the reagents tested under the conditions indicated under "Results". Proteins were then methanol-precipitated, resuspended in sodium dodecyl sulfate (SDS)-sample buffer and boiled for 5 min before 12% SDS-polyacrylamide gel electrophoresis (PAGE). The latter was then followed by electroblotting onto Immobilon membranes, and PrP C was detected by using the antibodies (see above) indicated under "Results" and a peroxidase-conjugated goat anti-mouse, or as appropriate anti-rabbit, secondary antibody. The blots were developed using enhanced chemiluminescence. Films were analyzed using Images Analysis software. The metal ions when used were CuSO 4, MnSO 4 , CoCl 2 , ZnSO 4 , and CaCl 2 .
To analyze the direct effect of H 2 O 2 on cells, confluent MoPrP CHO cells were rinsed twice with phosphate-buffered saline and overlaid with opti-MEM containing H 2 O 2 at the concentrations indicated under "Results". After 24 h of incubation at 37°C, the medium was removed, and the cells were lysed in lysis buffer (150 mM NaCl, 0.5% Triton X-100, 0.5% sodium deoxycholate, 50 mM Tris-HCl, 1 g/ml pepstatin, 1 g/ml leupeptin, 0.5 mM phenylmethylsulfonyl fluoride, and 2 mM EDTA) for 20 min at 4°C. The lysates were then spun at 14,000 rpm for 4 min, and the protein level in the supernatant was determined with the BCA protein assay kit (Pierce). After the addition of 2ϫ SDS-sample buffer, samples were analyzed by SDS-PAGE followed by immunoblotting.

RESULTS
PrP C Undergoes a Site-specific Cleavage in the Presence of H 2 O 2 -PrP C and the epitopes that are recognized by the antibodies used in this study are diagrammatically presented in Fig. 1A. Subconfluent stably transfected MoPrP CHO cells, which were incubated for 24 h in Opti-MEM, released into the medium (conditioned medium) a large quantity of soluble fulllength glycosylated PrP C lacking its glycosyl-phosphatidylinositol anchor (data not shown). Western blots of conditioned medium showed a major band at 33 kDa that reacted with P45-66, SAF 32, 8G8, and SAF mix antibodies ( Fig. 1, B, C, D, and E, lane 1, respectively). In addition, a set of bands corresponding to amino-terminal fragments of PrP C around 6.5 kDa was detected using the antibodies P45-66, SAF 32, and 8G8 ( Fig. 1, B, C, and D, respectively). The size of the aminoterminal fragments and the fact that they were recognized by both P45-66 and 8G8 could indicate that a mixture of aminoterminal fragments are released into the medium (this was confirmed using 15% Tricine gel; data not shown). Such fragments could arise from multiple cleavage of the amino terminus of PrP C , from cleavage into the octapeptide repeat region and the conserved region. A band around 14 kDa, corresponding most likely to carboxyl-terminally truncated PrP, was also detected by using SAF mix (Fig. 1E).
To examine the effect of H 2 O 2 on PrP C , conditioned medium (in the presence of protease inhibitors) was incubated with 5 mM H 2 O 2 at pH 7.0 and 37°C in the presence of 10 M Cu 2ϩ over varying time periods (0 -2 h). Exposure to H 2 O 2 induced a slight increase in the molecular mass of PrP C (Fig. 1B, compare lanes 1 and 2) (this observation will be discussed later). Within 5 min a dramatic reduction in the full-length PrP C P45-66 signal was detected (Fig. 1B) concomitant with the appearance of a P45-66-negative but SAF 32-, 8G8-, and SAF mix-positive truncated PrP C at 28.5 kDa (Fig. 1, B, C, D, and E, respectively). This latter fragment remained relatively unaffected by further exposure, indicating a H 2 O 2 cleavage that was not random, but one that occurred in the repeat region. In addition, within 5 min of exposure the intensity of the signal for the amino-terminal band (as detected by P45-66) increased by 18% (Fig. 1B, compare lanes 1 and 2). This latter increase occurred in conjunction with the loss of full-length P45-66-positive PrP C and could consequently represent an increase in detection of a cleavage product. After a further 25 min this amino-terminal band was no longer detected (Fig. 1B, lane 4) Fig. 2A). To observe an effect with lower concentrations of H 2 O 2 , an incubation period of 24 h at 37°C was chosen. H 2 O 2 induced a concentration-dependent loss of the P45-66 signal, with a complete loss at a concentration of 2.5 mM (Fig. 2B).
To facilitate an elucidation of the mechanism and dependence of the reaction, the effect of chelators, Me 2 SO, and metal ions on the cleavage was investigated (Fig. 2, C and D). In the presence of 10 M Cu 2ϩ and 2.5 mM H 2 O 2 (the concentration of H 2 O 2 allowing for a complete disappearance of the P45-66 signal in 24 h at 37°C ( Fig. 2A)), the P45-66 signal was lost (Fig. 2C, lane 12). Incubation of PrP C with 10 M Cu 2ϩ alone at 37°C did not result in cleavage (Fig. 2C, lane 2), indicating that the cleavage with H 2 O 2 was most probably a non-enzymatic event. In addition, in the absence of added copper, H 2 O 2 induced a slight increase in PrP C molecular mass without inducing cleavage (Fig. 2C, lane 3). The chelator (at a 1 mM concentration) EDTA; a broad range metal chelator, bathocuproine disulfonate; a Cu 1ϩ chelator, diethylenetriaminepentaacetic acid (DTPA); a Cu 2ϩ chelator; and the reagent Me 2 SO (10% v/v), a hydroxyl radical ( ⅐ OH) scavenger, had no effect on their own on the migration pattern (Fig. 2C, lanes 4 -7) or on the shift-up observed with H 2 O 2 in the absence of added copper (Fig. 2C, lanes 8 -11). This would indicate that the increase in molecular mass induced by H 2 O 2 was metal ion-independent. It is possible that the increase may have been due to amino acid oxidation, a phenomenon observed with H 2 O 2 (21); this is currently under investigation. Recently it was reported that storage of PrP C under oxidative conditions induced amino acid modifications that resulted in a similar increase in the M r of PrP C (32); this is consistent with the idea that the increased M r observed on H 2 O 2 exposure may also be due to amino acid modification. Importantly, both EDTA and DTPA protected PrP C from cleavage induced by H 2 O 2 (Fig. 2C, lanes 13 and 15); in contrast bathocuproine disulfonate and Me 2 SO were without effect (Fig. 2B, lanes 14 and 16).
In addition, because it has been reported that PrP C may interact with metal ions other than copper (33), Cu 2ϩ was replaced by 10 M Co 2ϩ , Zn 2ϩ , Mn 2ϩ , or Ca 2ϩ (Fig. 2D, lanes  3-6). Cleavage by H 2 O 2 was only detectable with Cu 2ϩ (Fig.   2D, lane 2); interestingly, when the other ions were added in combination with Cu 2ϩ , all but Co 2ϩ had a significant protective effect on H 2 O 2 cleavage (Fig. 2D, lanes 7-10).
Effect of pH on the Cleavage of PrP C by H 2 O 2 -It has been shown that binding of copper by PrP C is highly pH-dependent (34). Therefore, because cleavage of the amino terminus by H 2 O 2 appeared to be dependent on Cu 2ϩ , the effect of pH on PrP C cleavage by H 2 O 2 , in the presence of 10 M Cu 2ϩ , was determined by following the P45-66 signal of full-length PrP C (Fig. 3A). Although H 2 O 2 can induce significant cleavage of PrP C at low concentrations ( Fig. 2A), a concentration of 5 mM H 2 O 2 was selected for this experiment to increase the rate of cleavage. Loss of signal by H 2 O 2 was highly pH-dependent, with the rate of cleavage decreasing with decreasing pH (Fig. 3B).  (Fig. 4). After 1 and 24 h of incubation at 37°C, in the presence and absence of Cu 2ϩ , the P45-66 signal was lost by 90, 40, 42, and 26%, respectively (Fig.  4, A and B), whereas the SAF 32 signal was lost by 99, 65, 90, and 40%, respectively (Fig. 4, C and D). Similar to H 2 O 2 , O 2 . induced a loss of the P45-66 signal, and this loss of signal was augmented by the presence of Cu 2ϩ . In addition, the loss of the P45-66 signal was more rapid than that of SAF 32, which is similar to the cleavage of PrP C by H 2 O 2 in Fig. 1.

Effect of H 2 O 2 on PrP C in Cells-
To confirm the physiological relevance of the finding that PrP C can be cleaved by H 2 O 2 into the repeat region, confluent MoPrP CHO cells were exposed for 24 h to a range of H 2 O 2 concentrations (0, 10, and 100 M, which is within the range reported to be reached in rat brain under oxidative stress (25-160 M) (36)) (Fig. 5). Because the concentration of H 2 O 2 used on the cells was lower than that used in previous experiments, it was possible to observe a concentration-dependent increase in the detection of an amino-terminal fragment within cells (Fig. 5, A and B). For the latter observation a long film exposure was employed, but with a reduced exposure a decrease in full-length PrP C was also observed (Fig. 5C, compare lane 1 with lanes 2 and 3). This would indicate an increased cleavage of full-length PrP C under conditions of oxidative stress within the cell. In addition, because the protein level of another protein, actin, was unaffected by this treatment (Fig. 5D), it can be inferred that exposure to H 2 O 2 did not result in an increase of all protein cleavage. DISCUSSION The cell surface glycosylated protein, PrP C , has been recognized as a metal-binding protein since 1992 (5). Although the role of copper binding by PrP C remains obscure, it may be structural, because copper binding appears to induce structure to the otherwise flexible amino terminus of the protein (11). The present study suggests that copper binding by PrP C allows for the processing of the protein to occur under conditions of oxidative stress. In the presence of H 2 O 2 and copper, PrP C was observed to be cleaved into the octapeptide repeat region, yielding a protein of 28.5 kDa. H 2 O 2 cleavage was most efficient in inducing a loss of the full-length P45-66 signal; exposure also affected the SAF 32 signal (albeit at a lower rate) and resulted in further cleavage of the amino-terminal P45-66-positive fragment that was formed during the reaction. This would indicate that oxidative processing of PrP C most likely occurred within each of the metal-binding repeats, but that highest affinity was toward the repeat region after amino acid residue 66.
This H 2 O 2 cleavage of PrP C was Cu 2ϩ -dependent, because EDTA (a broad range metal ion chelator) and DTPA (a Cu 2ϩ chelator) inhibited cleavage, whereas bathocuproine disulfon- ate (a Cu 1ϩ chelator) was without effect. Under oxidative stress ⅐ OH is one of the most destructive biological molecules, and it is primarily generated through the breakdown of H 2 O 2 in the presence of copper or iron in the Fenton reaction (22). Although copper ions are invariably protein bound in vivo, binding does not prevent copper from participating in oxygen radical reactions but limits radical formation to the site of copper binding (23). Importantly, Me 2 SO, a known trapping agent for ⅐ OH, did not inhibit cleavage. Therefore, it can be assumed that in the case of PrP C ⅐ OH radical formation does not occur in solution; rather oxidation of PrP-copper by H 2 O 2 occurs at the site of copper binding. Additionally, because binding of Cu 2ϩ by PrP C is highly pH-dependent (34,37), the pH dependence of the H 2 O 2 cleavage of PrP C (Fig. 3) would further emphasize the participation of protein-bound copper in the reaction. Thus, similar to Cu,Zn superoxide dismutase in amylotrophic lateral sclerosis (23) and ␤-amyloid precursor protein in Alzheimer's disease (25), the results presented here would suggest that PrP C is also subject to a copper-dependent cleavage in the presence of H 2 O 2 with the following proposed reactions.
Because cleavage of PrP C , in the presence of Cu 2ϩ , was antagonized by additional divalent metals (Zn 2ϩ , Ca 2ϩ , and Mn 2ϩ , but not Co 2ϩ ), it is possible that PrP C can interact with more than one metal ion at a time or that the presence of additional metal ions could interfere with Cu 2ϩ binding. Either way the other metal ions are not known to participate in the Fenton reaction. In fact Mn 2ϩ has the capacity to inhibit it (38); therefore, the binding of these other metal ions could inhibit ROS cleavage by preventing the site-specific generation of ⅐ OH radicals.  Fig. 2A), and exposure of cells to H 2 O 2 at a concentration as low as 10 M induced a significant increase in PrP C cleavage. Consequently, it is most likely that PrP C could be susceptible in vivo to ROS-mediated cleavage of its amino-terminal octapeptide repeat region. In the brain normal PrP C cleavage is believed to occur within the conserved region, yielding a fragment called C1, whereas PrP Sc is thought to be cleaved outside this region, yielding a 27-30-kDa species (27). Jimenez-Huete et al. (28), on studying PrP C proteolytic cleavage using deglycosylated PrP C , were able to detect a 21-22-kDa fragment. Although amino-terminal sequencing is not available for this fragment, its size would indicate that cleavage of PrP C can occur physiologically into the repeats without cleavage of the conserved region (such cleaved PrP has also been observed in our cell models (data not shown)). However, it was not known if the 21-22-kDa species could be an obligatory intermediate for C1 generation. One possibility is that the 21-22 kDa fragment may have been generated through oxidative cleavage. Such cleavage could then initiate or facilitate further proteolysis of the molecule. In fact, it has been reported that such modified proteins are more susceptible to proteolysis than their native counterparts (21). Jimenez-Huete et al. (28) also demonstrated that PrP cleavage producing amino-truncated peptides could be stopped by EDTA and EGTA. Interestingly, a return of cleavage could only be achieved by adding the metal ions Cu 2ϩ and Fe 3ϩ but not by adding Mg 2ϩ , Zn 2ϩ , or Ca 2ϩ . Hence because both Cu 1ϩ and Fe 2ϩ can participate in the Fenton reaction, it is possible that the chelators may not have inhibited an enzymatic event but rather a reactive oxygen species event.
Under normal conditions, amino-terminally truncated PrP C , resulting from ROS cleavage, could be rapidly degraded within the cell or on the cell surface. However, because PrP Sc proteinase K digestion results in cleavage into the octapeptide repeats and because additional repeats in PrP C lead to abnormal properties in cells (41) and CJD in familial TSEs, it is possible that oxidative cleavage of the octapeptide repeat region may be a key factor in TSEs. This theory is further emphasized because ROS, oxidatively modified proteins, and CJD phenotypic expression increase with age. Additionally, amino-terminal sequence data available for PrP Sc , without proteinase K digestion, would indicate that a quantitative cleavage in the repeat region of PrP Sc does occur in vivo (29,42). The cleavage pat- terns that have been reported for size variants of mouse scrapie-associated fibrils are far from random, occurring three to four amino acid residues away from the copper-binding histidine in each of the octapeptide repeat regions. Because increasing evidence indicates that oxidative stress is involved in TSEs (43,44) and because H 2 O 2 at a concentration as low as 10 M altered PrP C processing, it is likely that PrP Sc is susceptible to ROS cleavage, and it is possible that such cleavage may account for in vivo amino-terminal truncation. In addition, because a number of reports have now separated PrP Sc proteinase K resistance from infectivity and neurodegeneration (45,46), it is possible that ROS modification of the molecule could modulate its toxicity and/or its effect on PrP-mediated signal transduction, for example (47,48).
Shmerling et al. (48) observed that expression of aminoterminally truncated PrP in Zurich Prnp o/o mice led to abnormalities that were reversed by the introduction of a single PrP allele. They proposed that the carboxyl terminus of PrP C may interact with a ligand and that the full-length amino terminus, through an attachment to a secondary site on the same ligand, may induce a cell survival signal. It was also proposed that in Zurich Prnp o/o mice an additional PrP-like molecule named could carry out the same function but that the presence of truncated PrP could also interact with the ligand, at a higher affinity than , preventing cell survival signaling. Additionally, it was recently suggested (49) that expression of Doppel, a PrP-like molecule, except that it resembles truncated PrP C lacking the repeat region (50), could also result in disease through a mechanism similar to that proposed for truncated PrP C . From the data presented here, we proposed that an increase in ROS-truncated PrP C may be pathogenic and lead to an alteration in cellular signaling. Additional repeats within PrP C , as is associated with CJD in familial TSEs, could increase copper binding and as a result susceptibility to ROS. Consequently, in CJD cases disease may present with age through a mechanism similar to that proposed by Shmerling et al. (48).
Whether ROS cleavage is fundamental to the function of PrP C or to the production of PrP Sc remains to be elucidated. However, it may be proposed that such ROS processing, at a rate allowing for the formation of a low level of amino-terminally cleaved PrP C , may be important for the function of the protein, because a number of proteins require ROS post-translational modification for activation. It has been suggested that PrP C may be involved in synaptic transmission, and it was reported recently that synaptic activity can be stimulated by H 2 O 2 in a PrP C -dependent manner (51). The H 2 O 2 effect was believed to be due to the higher synaptosomal copper content of PrP C mice over PrP Ϫ/Ϫ mice. In light of the results reported here, it is possible that the activation may have been due to a direct interaction between PrP C and H 2 O 2 , leading to PrP C cleavage and possibly increased copper release. The results presented here show for the first time that PrP C processing occurs under conditions of oxidative stress. Because cleavage of PrP C by ROS occurs into a region similar to that in PrP Sc , it is possible that such cleavage may be relevant in TSEs. Consequently, the findings presented here may lead to further approaches in the prevention and treatment of such diseases.