Address correspondence and reprint requests to D. R. Brown, Department of Biochemistry, Tennis Court Road, University of Cambridge, Cambridge CB2 1QW, UK. E-mail: firstname.lastname@example.org
We have developed a method to affinity-purify mouse prion protein (PrPc) from mouse brain and cultured cells. PrPc from mouse brain bound three copper atoms; PrPc from cultured cells bound between one and four copper atoms depending on the availability of copper in the culture medium. Purified PrPc exhibited antioxidant activity, as determined by spectrophotometric assay. Incubation of PrPc with the neurotoxic peptide, PrP106-126, inactivated the superoxide dismutase-like activity. Culture experiments showed that PrPc protects cells against oxidative stress relative to the amount of copper it binds. These results suggest that PrPc is a copper-binding protein which can incorporate varying amounts of copper and exhibit protective antioxidant activity.
The prion protein, a glycoprotein expressed by neurones (Brown 1999a), glia (Moser et al. 1995; Brown et al. 1998a) and other cells (Brown et al. 1998b) is believed to bind copper (Brown et al. 1997a). Mouse prion protein (PrPc) is highly conserved throughout evolution, yet apart from the association with prion diseases (Prusiner 1991) its cellular function remains undefined. An abnormal isoform (PrPSc) of the normal cellular prion protein (PrPc) is associated with the prion diseases, a group of fatal neurodegenerative diseases. Understanding the function of PrPc may have important implications for unravelling the nature of prion diseases (Prusiner 1991). Evidence suggests that the prion protein is important for synaptic activity (Collinge et al. 1994; Colling et al. 1996) but its expression in many cells is indicative of a more cosmopolitan function. Studies with recombinant prion protein and peptides related to its sequence have suggested that PrPc is a copper binding protein (Hornshaw et al. 1995; Brown et al. 1997a; Stöckel et al. 1998). Furthermore, studies of PrPc deficient mice suggest that these mice have altered copper metabolism and deficiencies in copper content of synaptosomal fractions (Brown et al. 1997a).
We have recently demonstrated that recombinant PrPc binds and retains copper during controlled refolding with urea (Brown et al. 1999). The recombinant protein bound between four and five copper atoms. At least four of these copper atoms were specifically retained by the octameric repeat region and a deletion mutant protein lacking this region showed little binding of copper. Additionally, we also showed that once copper is bound to the octameric repeat region, PrPc exhibits an activity like that of superoxide dismutase (Brown et al. 1999). This activity was measured using several different assays and was found to occur at a rate within an order of magnitude of the rate of Cu/Zn superoxide dismutase. This activity was not due to non-specific binding to PrP, since PrPc with deletion of the octameric repeat region and with a hexa-histadine tag that bound six atoms of copper did not have any significant activity in any of the assays we used. Despite these findings it has not previously been shown that native PrP binds equivalent levels of copper or has superoxide dismutase activity.
In the present report we demonstrate that native prion protein binds three copper atoms and can bind up to four copper atoms in vivo. Purified protein exhibits similar activity to that of superoxide dismutase and expression of PrPc with copper bound enhances cellular resistance to oxidative stress.
Unless otherwise described, all reagents were from Sigma (Poole, UK). Animals were 129(Ola) mice from Harlan (Bicester, UK). The antibody to mouse Cu/Zn superoxide dismutase was from Europa Research Products (Cambridge, UK).
Peptides PrP106-126 and βA25-35 were prepared as previously described (Brown 1999a).
Affinity purification of PrPC
The IgG fraction of a rabbit polyclonal antiserum raised against mouse PrP was isolated with protein A sepharose (Sigma). The antiserum was raised against a peptide corresponding to amino residues (89–103) of the mouse prion protein (Cymbus Biotechnology, Chandlers Ford, UK). Its specificity for the prion protein was tested by ELISA (against the peptide antigen) and by comparing its specificity for the whole protein to known specific antibodies (data not shown). The IgG fraction was coupled to cyanogen bromide activated beads (Amersham, Little Chalfont, UK). A column with a bed volume of 3 mL was prepared from this. An extract from 3-month-old mouse brain prepared by homogenization of 10 mouse brains in 10 mL of phosphate-buffered saline (PBS) containing 1% NP-40 and 0.5% Tween 20 at room temperature for 20 min followed by vortexing. Solid material was spun out at 14 k r.p.m. The volume of the extract was increased tenfold with dilution in PBS. This extract was passed over the bead-immobilized antibody column. The column was washed with wash solution (50 mm NaCl, 50 mm NaH2PO4 at pH 8.0). The specifically bound material was eluted in 250 µL steps with elution buffer (50 mm glycine, 50 mm NaCl pH.3) and then neutralized by addition of one-tenth volume of 100 mm Tris-HCl pH.8. The initial fractions containing PrP were measured to be pH 5–6. The purity was tested by PAGE electrophoresis in the presence of sodium dodecyl sulfate (SDS) and silver staining, and by Western blotting for prion protein using a monoclonal antibody which recognizes amino residues 142–160 of mouse PrP. A similar method was used to prepare PrPc from 8-day-old cultures of mouse cerebellar cells. Samples of the purified protein were taken for analysis using transmission electron microscopy following phosphotungstinic acid staining on electron microscope grids. No aggregated material was detected.
Total X-ray reflection fluorescence spectroscopy
Metal ion content of protein samples was determined using total reflection X-ray fluorescence spectroscopy (TXRF) as previously described (Prange et al. 1989; Dogan et al. 1993). Elemental determinations were carried out using an EXTRA II Energy Dispersive X-ray Fluorescence Spectrometer (Rich, Seifert and Co., Ahrensburg, Germany) fitted with multiple total reflection optics. Determinations were made by operating the X-ray tube at 50 keV and 10 mA with a count time of 300 s. Background readings for the quartz reflectors were recorded (count time 50 s-ISIS Processor Unit, Rich Seifert and Co.); acceptable reflectors were identified as those reflectors which when scanned gave less than 1.2 counts per second (cps) for the elements of interest. All sample handling was performed in a positive pressure filtered air cabinet using Gilson microman (5–25 mL) positive displacement pipettes. A 10-µL aliquot of each protein sample together with a 10-µL volume of internal standard (10 p.p.m. of cobalt) was placed directly onto the surface of a quartz reflector and air dried by placing the reflector in a positive pressure filtered air cabinet.
Superoxide dismutase assays
The superoxide dismutase assay used in these experiments employed superoxide production from xanthine oxidase; detection of this product was by spectrophotometric quantification of a coloured formazan product formed from nitro blue tetrazolium, as previously described (Oberley and Spitz 1984). The second assay was also based on superoxide production by xanthine oxidase but used cytochrome C (Sigma) to detect changes in superoxide consumption, as previously described (Flohé and Ötting 1984). This assay was performed with 50 µm EDTA added.
Cerebellar cells from wild-type mice were prepared as previously described (Brown et al. 1996). Prion proteins were added directly from stock solutions at the concentrations indicated. Exposure to xanthine oxidase (Sigma) and 500 µm xanthine (Sigma) was for 24 h one day after plating. The 3,[4,5dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT)-based survival assay was as previously described (Brown et al. 1996). Cells were exposed to 200 µg/mL MTT in medium for 1 h. The cells were lysed and the formazan product dissolved in dimethylsulfoxide (DMSO). Absorbance was measured at 570 nm in a spectophotometer. For cultures grown under different culture conditions, cerebellar cells were either grown in serum free medium consisting of Dulbecco's modified Eagle medium (DMEM; Gibco, Paisley, UK) and TCM culture supplement (ICN) or in serum containing medium (DMEM plus 10% fetal calf serum). Increased copper concentration was achieved by adding copper sulfate in its histidine-chelated form.
In order to confirm that native prion protein binds copper, we used an immuno-affinity procedure to purify mouse PrPc from brain tissue. The resulting material was confirmed to be pure PrPc following Western blot and detection with a PrPc-specific antibody and silver staining of a PAGE gel (Fig. 1). The purified material was examined by TXRF. The purified protein retained approximately three atoms of copper (Table 1) and no other divalent cation. Further experiments were carried out to determine if this native PrPc could be stripped of its copper content or whether it would bind more copper. Native PrPc (1 µg/mL solution) was unfolded in 8 m urea and refolded into water containing either no copper or 5 mm copper solution. It has previously been shown that recombinat PrP treated this way can be reconstituted with copper binding specifically at the octameric repeat sequence (Brown et al. 1999). Following TXRF analysis, it was found that native material refolded in water bound no detectable copper (n = 6 determinations) while PrPc refolded in the presence of copper bound five atoms of copper (n = 6). When copper sulfate solution (5 µm) was added to native PrPc without refolding no further addition of copper to the molecule was noted (n = 6). The molar ratio remained at three atoms/molecule. Incubation of copper complexed with histidine at a 2000 : 1 ratio (His : Cu) with PrPc resulted in further binding of copper to PrPc (five atoms : one molecule, n = 4). These results suggest that PrPc can bind up to five atoms of copper.
Table 1. Divalent cation content of isolated prion proteins
TXRF values are μg metal ion/2 mg of protein and were determined by TXRF analysis.
Cu values are approximate molar values from TXRF analysis of medium samples. Additionally, for all samples tested, other divalent cations, including Ni and Fe, were detected at less than 0.01 μg/mg protein. A minimum of six measurements for each value were made.
The molar ratio indicates the approximate number of copper atoms per molecule pf PrP (2 mg PrP = 80 nmol).
+ Cu, copper sulfate was added to the medium to increase copper concentration; cells, PrP was extracted from cerebellar cell cultures; serum or serum-free, the conditions under which the cerebellar cells were grown.
Experiments with cerebellar cell cultures derived from mice were carried out to determine whether PrPc can bind differing amounts of copper in a cellular context. Cells were grown under varying conditions with respect to copper concentration within the medium. After 7 days in these conditions the cells were placed in media with low copper concentration and left for 12 h. The half-life of PrPc is in the order of an hour or less (Shyng et al. 1993), therefore the majority of PrPc expressed by such cells should be newly synthesized. The cells were harvested, PrP affinity purified and the copper content assessed using TXRF analysis. Table 1 shows the concentration of copper in the medium and the amount of copper incorporated into PrPc. The higher the extracellular copper content, the higher the level of copper incorporated into PrPc. However, in these experiments, PrPc bound no more that four atoms of copper per molecule. Furthermore, none of the conditions resulted in PrPc without bound copper.
Previous studies have shown that PrPc-deficient cerebellar cells are more sensitive to oxidative stress than wild-type cells (Brown et al. 1996, 1997b). In addition, PrP-deficient mice show decreased levels of Cu/Zn superoxide dismutase activity (Brown et al. 1997b; Brown and Besinger 1998). Although these changes in oxidative metabolism might be dismissed as secondary effects of PrPc depletion, the availability of a source of pure native PrPc that binds copper allowed us to consider the possibility that PrPc may, itself, act as an antioxidant. Accordingly, native and recombinant PrPC preparations were tested for superoxide dismutase (SOD) activity using a formazan formation assay (Oberley and Spitz 1984). Native PrPc, affinity-purified from mouse brain significantly inhibited the rate of formazan production (anova, n = 4, p < 0.05 for all concentrations from 0.1 µg/mL upwards) in a dose-dependent fashion (Fig. 2a). This was confirmed using a second assay for SOD activity which used cytochrome C for detection (Fig. 2b) (Flohé and Ötting 1984). This assay, which was peformed in the presence of EDTA and native PrP, showed a similar activity in this assay. The purified PrPc was free of contamination with a known cellular Cu/Zn or Mn superoxide dismutase as none was detected by Western blotting using a specific antibody against Cu/Zn superoxide dismutase (Fig. 1); TXRF analysis indicated no presence of Zn or Mn in the preparations (Table 1), which would be necessary for this activity. We also detected a similar activity associated with recombinant PrPc reconstituted with copper (Brown et al. 1999). Addition of copper during the assay procedure did not increase or inhibit the activity measured for native PrPc (Fig. 2a) suggesting the activity does not result simply from the presence of copper.
Native PrPc prepared from cell cultures, as described in Table 1, allows PrPc to incorporate on average 1, 2, 3 or 4 atoms of copper, each of which should be tested for SOD-like activity (Fig. 2c). SOD-like activity was observed only once PrPc had bound two atoms of copper. PrPc with three or four atoms of copper showed significantly higher activity (Student's t-test, n = 5, p < 0.05).
Other studies have shown that PrP106-126, the neurotoxic peptide based on the sequence of PrPSc, reduces cellular resistance to oxidative stress and decreases measurable SOD activity in cerebellar cell cultures (Brown et al. 1996, 1997b). This decrease in resistance to oxidative stress is possibly part of the neurotoxic effect of the peptide. Affinity-purified PrPc from mouse brain was incubated with increasing concentrations of PrP106-126. As a control, a peptide based on the β-amyloid sequence (βA-25–35) was also tested. Only PrP106-126 inhibited the SOD-1 like activity of PrPc significantly (Student's t-test, p < 0.05) (Fig. 2d). These results suggest that PrP106-126 may have its effect on neuronal resistance to oxidative stress by inhibiting the activity of PrPc itself.
These results indicate that PrPc may act as an antioxidant, protecting cells from oxygen radical damage. To test this, cell cultures were prepared by culturing cerebellar cells for 7 days in differing concentrations of copper to produce cells expressing PrPc with different amounts of copper incorporation. After 7 days of treatment, the culture medium was replaced with a medium consisting of serum-free, copper-deficient medium and xanthine oxidase, an oxygen-radical producer added at various concentrations. Survival after 24 h was determined using an MTT-based survival assay. Cells grown in increasing concentrations of copper showed significantly increased resistance to oxidative stress (Student't t-test, p < 0.05, for values of 100 µU of xanthine oxidase or above) (Fig. 3a). One possible explanation for the increased resistance of the cells to oxidative stress is that PrPc binding higher levels of copper is more effective as an antioxidant. Phosphatidylinositol-specific phospholipase C (PIPLC) is an enzyme that strips PrPc from the surface of cells (Stahl et al. 1990). Similar cultures prepared in different concentrations of copper were treated with 0.2 U/mL PIPLC for 2 h before exposure to xanthine oxidase. There was no significant difference in the resistance of the cultures to xanthine oxidase treatment after PIPLC digestion (Fig. 3b) (Student's t-test, n = 4, p > 0.05). One possible explanation for this is that PrPc protecting the cells (as in Fig. 3a) was stripped away by PIPLC. To test this further, native PrPc affinity purified from cultured cells grown under different conditions was added back to cerebellar cell cultures grown in serum-free conditions and exposed to xanthine oxidase. In this situation, PrPc reduced the toxicity of xanthine oxidase dependent on level of copper incorporation into the molecule (Fig. 3c). This protection was not due to an effect from the copper added with PrPc. Addition of 1 µg/mL PrPc loaded with four copper atoms is approximately the same as addition of 0.2 µm copper. The addition of this amount of copper, in the form of CuSO4, to the same cells as in Fig. 3c did not significantly alter the toxicity of xanthine oxidase (0.1 ± 0.1% reduction in toxicity, n = 4, p > 0.05).
The work presented in this paper was carried out in order to determine if results demonstrated for recombinant protein hold true for native PrPc isolated from or present in mouse brain and cultured cells. Previous results suggested that full-length recombinant prion protein can bind copper and, on doing so, gains a function like that of superoxide dismutase (Brown et al. 1997a, 1999). The results of experiments with native PrP confirm these findings and highlight some novel findings of major significance.
PrPc isolated from mouse brain retains three copper atoms. It is possible that some of this copper could have been acquired from the protein during the purification procedure. However, this seems unlikely as there is very little or no free copper in the brain. For PrPc to bind copper not already associated with it in situ the copper would have to have been captured from molecules with a much lower affinity for copper. As the affinity of PrPc for copper (Brown et al. 1997a; Stöckel et al. 1998) is not far greater than other known copper-binding molecules, it seems unlikely that a substantial proportion of the copper bound to isolated PrPc would arise from non-specific chelation during the extraction procedure.
Additionally, we have shown that the amount of copper associated with PrPc depends on the level of copper in the environment of the cell. Reducing the copper concentration in the medium of cells reduces the amount of copper bound to PrPc to one atom per molecule. This suggests that PrPc may be synthesized with at least one atom of copper bound. Increasing extracellular copper concentrations increases the number of copper atoms bound to PrPc to as much as four atoms per molecule. The binding of four copper atoms agrees with previous studies that have analysed the ability of the octameric repeat region of the molecule to co-ordinate copper atoms between the four histidine residues present in the repeat region (Viles et al. 1999). Previous studies have also indicated that increasing copper concentrations in the medium of cultured cells increases the turnover rate of PrPc (Pauly and Harris 1998). Changes in PrPc expression levels also alter the uptake of copper into cells (Brown 1999b). Cells expressing high levels of PrPc show the highest level of copper uptake. Copper taken up in association with PrPc can be incorporated into molecules such as Cu/Zn superoxide dismutase (Brown and Besinger 1998), or released at the synapse (Brown 1999b).
Previous studies with peptides containing the octameric repeat sequence of the protein have shown that all four copper atoms are bound in the Cu2+ state (Whittal et al. 2000). It is likely that the full-length protein binds copper in the same way. However, we have shown previously that a peptide based only on the octameric repeats of PrPc has no SOD activity (Brown et al. 1999). Dismutation of superoxide involves a change in the copper's redox state to Cu+, which is an unstable state. We have also shown that recombinant full-length protein with bound copper contains large amounts of oxidized methionine (Wong et al. 1999), implying that the copper can readily transfer the extra electron elsewhere in the protein or possibly out of the protein altogether via the formation of hydrogen peroxide. This implies that the copper bound to PrP can undergo redox cycling. However, the details of this process have yet to be deduced.
Experiments with PrPc from mouse brain indicate that PrPc can be induced to bind a fifth copper atom. A similar result was obtained with recombinant PrPc using equilibrium dialysis (Brown et al. 1997a). Although it is possible that this reflects binding to another position on the protein, either to the histidine at position 111 or to another amino residue in the N-terminal region (Whittal et al. 2000), it could also be an artefact, especially as this was not observed with experiments on PrPc from cultured cells.
The results present here confirm that PrPc could act as an antioxidant such as a superoxide dismutase, but show for the first time that its activity is tightly regulated by the level of copper available. The binding of two atoms of copper per molecule endows the protein with antioxidant activity. Doubling the amount of copper binding to PrPc increases this activity. This increased activity could possibly result from PrPc adopting a conformation that facilitates higher activity, rather than resulting directly from the presence of more copper. Changes in the secondary structure of PrP peptides have been noted following copper binding (Miura et al. 1996).
The level of copper bound to PrPc influences the survival of cells when they are exposed to oxidative stress. Growing cells in higher copper concentrations could lead to increased activity of other proteins which might protect against oxidative stress. However, these proteins would also have to be glycosylphosphatidylinositol (GPI) anchored, as stripping GPI-anchored proteins from the cells with PIPLC abolishes the effect of increased copper concentration on increasing cell survival in the presence of oxidative stress generated by xanthine oxidase. Increased numbers of copper atoms being bound to PrPc appears to have a protective effect. This is probably because PrPc can then exert antioxidant activity via superoxide dismutation.
PrPc binding only one atom of copper has no detectable SOD-like activity. Nevertheless these experiments show that when PrPc binds more copper it gains an antioxidant activity similar to that of a superoxide dismutase. This activity can be inhibited by the neurotoxic PrP106-126 peptide which is known to inhibit neuronal resistance to oxidative stress in culture (Brown et al. 1996, 1997b). Since it has been found that PrP-deficient mice remain healthy and do not die from prion disease, it has been agreed that the loss of prion protein function does not cause prion disease (Büeler et al. 1992). Oxidative stress resulting from activated microglia has already been implicated from in vivo observations of prion disease (Giese et al. 1998) and in vitro models of neurodegeneration. However, in the in vitro models decreased neuronal resistance to oxidative stress was also shown to be necessary for neurodegeneration (Brown et al. 1996, 1997b). Therefore, loss of resistance to oxidative stress due to inactivation of PrPc function, either due to conversion or through interaction with PrPSc, may be important to the neurodegeneration seen in prion disease.
A suggested association between copper and prion disease dates back many years to the work of Pattison and Jebbett (1971) who suggested that pathology induced by the copper chelator, cuprizone, was similar to the pathology of scrapie, especially as regards the formation of spongiform changes. Also, a small number of biochemical markers were found to be shared by cuprizone treatment and scrapie in rodents (Kimberlin et al. 1974). These findings predate the discovery of the prion protein. Nevertheless, it is a possibility that changes in the ability of the prion protein to retain or utilize copper might have some consequence for the spongiform degeneration seen in prion disease.
This work was supported by the BBSRC of the United Kingdom.