Copper Binding to the Octarepeats of the Prion Protein

The prion protein (PrP) is a Cu2+ binding cell surface glycoprotein. There is increasing evidence that PrP functions as a copper transporter. In addition, strains of prion disease have been linked with copper binding. We present here CD spectroscopic studies of Cu2+ binding to various fragments of the octarepeat region of the prion protein. We show that glycine and l-histidine will successfully compete for all Cu2+ ions bound to the PrP octapeptide region, suggesting Cu2+ coordinates with a lower affinity for PrP than the fm dissociation constant reported previously. We show that each of the octarepeats do not form an isolated Cu2+ binding motif but fold up cooperatively within multiple repeats. In addition to the coordinating histidine side chain residues, we show that the glycine residues and the proline within each octarepeat are also necessary to maintain the coordination geometry. The highly conserved octarepeat region in mammals is a hexarepeat in birds that also binds copper but with different coordination geometry. Finally, in contrast to other reports, we show that Mn2+ does not bind to the octarepeat region of PrP.

The prion protein (PrP) is a Cu 2؉ binding cell surface glycoprotein. There is increasing evidence that PrP functions as a copper transporter. In addition, strains of prion disease have been linked with copper binding. We present here CD spectroscopic studies of Cu 2؉ binding to various fragments of the octarepeat region of the prion protein. We show that glycine and L-histidine will successfully compete for all Cu 2؉ ions bound to the PrP octapeptide region, suggesting Cu 2؉ coordinates with a lower affinity for PrP than the fM dissociation constant reported previously. We show that each of the octarepeats do not form an isolated Cu 2؉ binding motif but fold up cooperatively within multiple repeats. In addition to the coordinating histidine side chain residues, we show that the glycine residues and the proline within each octarepeat are also necessary to maintain the coordination geometry. The highly conserved octarepeat region in mammals is a hexarepeat in birds that also binds copper but with different coordination geometry. Finally, in contrast to other reports, we show that Mn 2؉ does not bind to the octarepeat region of PrP.
Prion diseases, which include Creutzfeldt-Jacob disease in humans, mad cow disease in cattle, and scrapie in sheep, involve the misfolding of the benign cellular prion protein (PrP C ) 1 to the infectious disease-causing scrapie isoform PrP Sc (1)(2)(3). The prion protein (PrP C ) is a copper-binding cell surface glycoprotein (4,5). The role of copper in the normal function of PrP, as well as in prion diseases, has been the subject of a number of excellent reviews (6 -8). The mature cellular form of PrP consists of residues 23 to 231 and is tethered to the cell surface via a glycosylphosphatidylinositol anchor at the C terminus. There are now a number of NMR solution structures of copperfree mammalian PrPs (9 -12). A crystal structure of PrP C has also been published; this structure is dimeric involving domain swapping of the monomeric form (13).
PrP C contains a C-terminal domain that is largely ␣-helical. In the absence of Cu 2ϩ the N-terminal half of the protein, residues 23-124, is unstructured (11), with a large degree of backbone flexibility (14). Residues 60 -91 consist of an octapeptide sequence, PHGGGWGQ, that is repeated four times. It is this unstructured region that binds four Cu 2ϩ ions in the fulllength protein and similarly in fragments from the octarepeat region (5). This octarepeat region binds four Cu 2ϩ ions cooperatively with identical coordination geometry (5). There are now a number of studies to suggest that copper binds to the octarepeat region of PrP C with affinities ranging between fM (15) and M (16). In addition, a fifth Cu 2ϩ binding site centered at residues His-96 and His-111 has also been observed (15,17).
The function of PrP is still the subject of debate. Elevated copper levels promote endocytosis of PrP suggesting that PrP transports copper into the cell (18,19). An enzymatic role for Cu-PrP is also proposed as it exhibits superoxide dismutase activity (20 -22). It is also suggested that PrP has a protective role binding Cu 2ϩ in a redox-inactive state (23)(24)(25). Mice deficient in cellular PrP show a reduction in copper concentration in the brain relative to wild type mice and a reduction in activity of copper/zinc superoxide dismutase (4), although this observation is contested (26).
Cu 2ϩ has also been linked with prion diseases. For example, the presence of copper can confer different strains of prion disease with different protease resistance properties (27,28). Elimination of octarepeats slows disease progression (29). A mutant form of PrP associated with familial prion disease contains nine additional octarepeats and fails to undergo copper-mediated endocytosis (19). Copper can enhance reversibility of scrapie inactivation (30). Metal binding to prion protein is altered in human prion disease (31). Copper can convert the cellular prion protein into a protease-resistant species (32). Metal imbalances are a feature of prion disease (33). Coppercatalyzed redox damage of PrP has been implicated in prion disease (34,35).
There is now an impressive list of techniques that have been directed at determining the structure and the copper binding properties of the octarepeat region of PrP C . These include the following: CD (5,36,37), electron paramagnetic resonance spectroscopy (EPR) (5,38,39), NMR spectroscopy (5,15), mass spectrometry (40,41), Raman spectroscopy (42), infrared spectroscopy (43), voltammetry (37), and fluorescence spectroscopy (36,44). Despite this, the precise structure of the copper-binding region is not fully established. Recently, however, the crystal structure of Cu-HGGGW has been published (45). The structure is in agreement with EPR data suggesting type II coordination geometry involving three nitrogen ligands and oxygen ligand coordination (5,38). The structure is pentacoordinate with four equatorial ligands in a square-planar arrangement with an additional axial water molecule. Coordination includes an imidazole nitrogen ligand, two deprotonated amides from the next two glycines with the second glycine also contributing a carbonyl oxygen. The axial water ligand is stabilized by hydrogen bonding to the tryptophan indole.
Despite numerous studies of copper binding to PrP there is still intense disagreement in the literature as to the affinity of Cu 2ϩ for PrP and therefore its functional role (see Refs. 15 and 16). The specificity of PrP for Cu 2ϩ ions is disputed (see Refs. 44 and 46), and the number of Cu 2ϩ ions binding to PrP is in dispute. Copper binding to the octarepeats is often described in terms of isolated single copper centers; this is inconsistent with the cooperative nature of copper binding found in multiple repeats. It is the aim of this paper to use direct spectroscopic methods to resolve some of these inconsistencies.

EXPERIMENTAL PROCEDURES
Peptide Synthesis and Purification-Peptides representing various lengths of the repeat region of PrP were synthesized employing solid phase Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry using the University of London service at Imperial College. After removing from the resin and deprotection the samples were purified using reverse phase high pressure liquid chromatography and characterized using mass spectrometry and 1 H NMR. To mimic the peptides within the full-length prion protein all peptides were blocked at the N terminus with N-acetyl and with ethyl ester at the C terminus. Peptides synthesized included those shown in Table I.
pH and Buffers-The pH was measured before and after each spectrum was recorded. Unless otherwise stated all spectra were recorded at pH 7.5. The effect of various buffers on Cu 2ϩ binding to PrP (58 -91) was investigated. We have found that Tris buffer will compete very successfully for Cu 2ϩ ions. HEPES, acetate, and N-ethylmorpholine buffer were found not to interfere with the Cu 2ϩ binding to the octarepeats as CD signals were unaffected by the addition of the buffers at constant pH. Typically, 50 mM N-ethylmorpholine buffer was used for CD studies in the visible region whereas in the UV region the pH was adjusted using small aliquots of 10 mM NaOH or HCl. The effect of ionic strength on the UV-CD data was investigated for a number of peptides with addition of NaCl; no effect was observed.
Titration of Metal Ions-The peptide concentrations were determined using extinction coefficients at 280 nm. The extinction coefficient of the peptides were calculated for the octarepeat peptides using 5690 M Ϫ1 cm Ϫ1 multiplied by the number of Trp residues or 1280 M Ϫ1 cm Ϫ1 for Tyr residues (47). Typically, the freeze-dried peptides contained 5 to 10% moisture by weight. The addition of metal ions to the octarepeat peptides was performed using small aliquots from stock aqueous solutions of CuCl 2 ⅐2H 2 O and MnCl 2 ⅐4H 2 O.
CD-CD spectra were recorded on an AVIV instrument at 25°C. Typically a cell with a 0.1-cm path length was used for spectra recorded between 185 and 260 nm with sampling points every 0.5 nm. A 1-cm cell path length was used for data between 260 and 800 nm with a 2-nm sampling interval. A minimum of three scans were recorded, and baseline spectra were subtracted from each spectrum. It was not necessary to apply smoothing methods to any of the data presented. Data were processed using KaleidaGraph spread sheet/graph package. The direct CD measurements (, in millidegrees) were converted to molar ellipticity, ⌬⑀ (M Ϫ1 cm Ϫ1 ) using the relationship ⌬⑀ ϭ /33,000 ϫ c ϫ l ϭ []/3,300, where [] ϭ /cl, c is the concentration, and l is the path length. The molar ellipticity [] is in units deg cm 2 Absorption Spectroscopy (UV-visible)-UV-visible electronic absorption spectra were obtained with a Hitachi U-3010 double beam spectrophotometer, using a 1-cm path length.

RESULTS AND DISCUSSION
Stoichiometries: Each Histidine-containing Octarepeat Will Bind a Single Cu 2ϩ Ion-To ascertain the minimal binding motif of the Cu-PrP octarepeat complex, CD spectra of a series of peptides representing various lengths of the PrP octarepeat region were obtained. On the addition of increasing amounts of Cu 2ϩ changes in the visible and UV region of the CD spectra were observed. Ellipticities, at specific wavelengths for each peptide, have been plotted against copper addition to assess the copper saturation point, as shown in Fig. 1 and Table II. The binding curves exhibit a sharp transition at the saturation point indicating tight binding at peptide concentrations of typically 0.05 mM (K d Ͻ 50 M). The Cu 2ϩ stoichiometries for the various peptides are summarized in Table II.
The number of Cu 2ϩ ions binding to PrP has been the subject of disagreement. Stockel et al. (44) reported two Cu 2ϩ ions binding to the four octarepeat peptide whereas others (5) have reported four Cu 2ϩ ions to four octarepeats but only one Cu 2ϩ ion bound to two octarepeats. Disagreements as to the stoichiometries of Cu 2ϩ binding to PrP are probably due to binding Cu 2ϩ below physiological pH. In this study we confirm that each histidine-containing octarepeat will bind a single Cu 2ϩ ion (see Table II). In particular, the 15-mer, GGGWGQPHGGG-WGQP, which represents the longest sequence from the octarepeat region that contains one histidine residue, binds only a single Cu 2ϩ ion. The peptide HGGG represents the shortest element that binds a single Cu 2ϩ ion. Surprisingly the crystal structure of Cu-HGGGW (45) suggests that that HGG might represent the shortest peptide to bind a single Cu 2ϩ ion, as direct coordination comes from these three residues. Presumably the shorter less bulky tripeptide facilitates a second HGG peptide to bind a single Cu 2ϩ ion, giving the 2:1 stoichiometry.
Affinity of Cu 2ϩ for PrP(4octa): Gly Competes Strongly for Cu 2ϩ Ruling Out 10 15 Affinity-There is much disagreement as to the affinity of the Cu 2ϩ binding to PrP in the octarepeat region. A recent study (15) using tryptophan fluorescence quenching and glycine competition suggests that the affinity is as much as eight orders of magnitude higher than the M dissociation constants reported previously (16). With this in mind we have used CD spectroscopy to study the binding of Cu 2ϩ under the competitive effects of the free amino acid glycine. Fig. 2 shows the CD spectrum of Cu-PrP(4octa). Before the addition of glycine the CD spectrum gives characteristic CD bands because of d-d electronic transitions at 580 and 680 nm. Addition of increasing amounts of glycine gradually decreases the intensity of all the CD bands in the visible region. The intensity of the band at 580 nm is plotted versus Gly addition, as shown in Fig. 2B. After only 8 mol equivalents of glycine relative to the PrP(4octa) peptide, there is almost no visible CD signal. Two glycine residues bind to a single Cu 2ϩ ion; therefore 8 mol equivalents of glycine gives the same 2:1 ratio with Cu 2ϩ ions. Note that the d-d absorption bands for the Cu(Gly) 2 complex are CD silent. To confirm that there is strong competition for Cu 2ϩ from glycine, even for the first equivalent of copper bound to PrP(4octa), spectra of PrP(4octa) at 0.9, equivalent to Cu 2ϩ , were obtained, as shown in Fig. 2C. Subsequent addition

GQAHGGGW
of glycine caused almost complete loss of the CD signal, also shown in Fig. 2C.
Glycine is able to coordinate to Cu 2ϩ via its amino and carboxylate group. The affinity of Cu 2ϩ for glycine at pH 7.4 is reported to be K a ϳ1.2 * 10 8 (48). The binding curve shown in Fig. 2 shows a rapid loss of Cu 2ϩ bound to PrP(4octa) with glycine addition. At Cu:Gly ratios of 2:1, little copper remains bound to PrP(4octa). We can conclude that Gly competes strongly with Cu 2ϩ indicating that the affinity of Cu 2ϩ for PrP(4octa) is less than that of Cu 2ϩ for glycine. The tight binding observed in the CD experiments at 10 M concentrations therefore puts the K d in the 10-M to 10-nM range.
The affinity we observed is in sharp contrast the study that reported fM dissociation constants for the four octarepeats. The study describes a biphasic binding of Cu 2ϩ to the octarepeats, with a single Cu 2ϩ ion binding with a K d of 8 fM followed by weaker Cu 2ϩ binding of 10 M (15). There is, however, no evidence for two modes of binding to the octarepeats. As we can see from Fig. 2C, the shape of the CD bands for 0.9 mol equivalents of Cu 2ϩ is identical to that observed for subsequent additions of Cu 2ϩ to PrP(4octa) as seen in Fig. 3A and Ref. 5. The EPR spectra for 1 equivalent Cu 2ϩ added to PrP (58 -91) is also not significantly different from EPR spectra of subsequent additions of copper (5). The addition of Gly as shown in Fig. 2C should not be sufficient to have any effect on the 8 fM site (if one was present), but it is clear from Fig. 2C that Gly is a very effective competitor. To reconcile the observations made by Jackson et al. (15), we suggest that the copper-bound Gly complex, (Gly) 2 Cu, interferes with apo-PrP(4octa) tryptophan fluorescence signal, causing it to partially quench.
A Single Repeat Is Found in a Different Environment than Multiple Repeat Complexes-From the stoichiometries alone (see Table II) one might conclude that HGGG (or HGGGW) represents the minimal binding motif for Cu 2ϩ . Indeed, the EPR spectra for HGGGW are indistinguishable from larger multiple repeats (38). However, it is clear from a comparison of the CD spectra in the visible region (Fig. 3) that for various peptides, including, HGGG, HGGGW, GGGWGQPHGGG-WGQP, PrP(2octa), PrP(3octa), and PrP(4octa), there is a distinct difference between those peptides binding a single Cu 2ϩ ion and multiple repeat peptides. Fig. 3A compares the spectra for PrP(2octa), PrP(3octa), and PrP(4octa) peptides containing 2, 3, and 4 mol equivalent of Cu 2ϩ added, respectively (i.e. the peptides are fully loaded with copper). The CD spectra have been plotted using ⌬⑀ (divided by 4, 3, and 2, reflecting the Cu 2ϩ ion concentration added) for a direct comparison. As observed previously (5) for the fouroctarepeat peptide, the visible region of the CD spectrum shows two bands due to d-d electronic transitions. These are centered at 625 nm (visible absorption E 625 nm ϭ 40 cm Ϫ1 M Ϫ1 ) with a positive ellipticity band at 580 nm and negative band at 680 nm. In addition, there is a positive band at 330 nm of comparable intensity. It is clear that the multiple octarepeats give strikingly similar visible CD spectra. In contrast, if we compare these multiple octarepeats with octapeptides containing a single histidine residue, the single octarepeats spectra show some significant differences (Fig. 3A). The position of the maximum shifts to longer wavelength from 580 nm (Ϯ3 nm) in the multiple octarepeats to 610 nm (Ϯ3 nm) for various versions of the single octapeptide. The crossover point from positive to negative also shifts by ϳ40 nm to a longer wavelength. The intensity of the CD bands at 330, 580, and 680 nm is very similar for 4, 3, and 2 octarepeats (divided by 4, 3, and 2 respectively, reflecting the number of copper ions bound); however, this is quite different in the single repeat peptides, as shown in Fig. 3. The pH dependence of the CD spectrum in the visible region was obtained to confirm that the differences in the spectra were not because of slight differences in pH. The maximum positions of the CD bands and crossover points did not change by more than 2 nm between pH 7.4 and 7.9. The pH dependence of binding has been described previously that shows complete release of Cu 2ϩ below pH 6 (5).
The visible absorption bands and accompanying CD bands are very sensitive to ligand coordination geometry around the Cu 2ϩ ion. For example, increasing the number of peptide nitrogens coordinating the Cu 2ϩ ion for the complex with acetyl-Gly-Gly-His results in a decrease in wavelength maximum from 765 nm for 1 nitrogen to 540 nm for 4 nitrogens (49). The hexadecant rule has been used in an attempt to interpret CD bands in terms of coordination geometry around the metal center (50). The sign of the CD band indicates the position of ligands around the Cu 2ϩ coordination plane. The single octarepeat has the same sign of CD bands as the multiple repeats, and we can conclude that the difference in ligand position is small, as it is not sufficient to move coordinating ligands into neighboring hexadecant sectors.
EPR data indicate that a very similar coordinating geometry exists in HGGGW, relative to multiple repeats, with the fundamental coordination ligands unchanged (38). It appears that, in this instance, the visible CD spectra are more sensitive to differences in the coordination geometry than the EPR spectra. The CD spectra are sensitive to the interactions between Cu 2ϩ centers, and this is reflected in the differences seen in the CD spectra between multiple and single repeats. The strong cooperativity of Cu 2ϩ binding in multiple repeats (see Table II) also indicates that the four Cu 2ϩ ions are not isolated from each other.
Octarepeats Are Not Isolated from Each Other, but the Main Chain Folds Together to Produce Cooperative Binding-Addition of Cu 2ϩ to various octarepeat peptides causes a profound structuring of the main chain from an unstructured conformation, as indicated by the CD spectrum in the UV region. Comparisons of the CD spectra in the secondary structure region (185-260 nm) have been made (see Fig. 4) in an attempt to establish whether the main chain conformation of the Cu 2ϩbound octarepeat region behaves differently in a single octarepeat compared with when present in multiple octarepeats. Fig.   FIG. 1. Direct Cu 2؉ binding curves for peptides from the octarepeat region. A, HGGGW; B, HGGG; C, HGG; D, PrP(2octa). Change in ⌬⑀ with Cu 2ϩ was measured at 220, 197, 202, and 590 nm, respectively. The stoichiometries of binding are summarized in Table II. Binding curves for PrP3octa and PrP(4octa) have been published previously (5).
4B shows the resultant spectrum in the UV region when the CD spectrum for HGGGW (multiplied by four) is subtracted from the four-octarepeat peptide spectrum. EPR data suggest that HGGGW encompasses the residues directly binding a single Cu 2ϩ ion in multiple octarepeats (38,45). The difference between the spectra should therefore give an indication of the difference in the backbone conformation for an isolated octarepeat relative to multiple octarepeat peptides. Interestingly, the resultant difference between the spectra is not characteristic of an irregular/unstructured peptide. A resultant random coil spectrum would be expected if the Cu 2ϩ binding residues HGGGW were isolated from each other in successive octarepeats. The resultant CD spectrum shown in Fig. 4B gives a maximum at 205 nm and a negative band at 225 nm. These bands indicate profound structuring of the intervening (GQP) residues between successive HGGGW motifs.
For a control comparison, the difference between GGG-WGQPHGGGWGQP (the longest possible peptide containing a single Cu 2ϩ ion) and HGGGW is made as shown in Fig. 4, C and D. In this case, the resultant spectrum is similar to that found for unstructured/random coil peptides with a single negative CD band at 200 nm and with a ⌬⑀ per mean residue of Ϫ1.5 M Ϫ1 cm Ϫ1 . In this case, the residues on either side of HGGGW do not form a regular secondary structure. This is direct evidence to indicate that the Cu 2ϩ bound to HGGGW is not an isolated entity but interacts with neighboring Cu 2ϩ centers to cause the intervening residues to fold up in an ordered conformation.
The endocytosis of PrP is triggered by the presence of Cu 2ϩ (18,19). The folding together of the four copper centers, as demonstrated by our CD measurements, is critical to endocytosis. Recent work shows that mutations of just two histidine residues in the octarepeats is sufficient to restrict endocytosis (19). The cooperative fold of all four copper centers is essential to the recognition process that triggers endocytosis.
Cu 2ϩ Binds to Avian PrP Repeat Region but Forms a Different Complex than That of Mammalian PrP-The octarepeats are the most highly conserved region of mammalian prion proteins primary sequence. In chicken and other avian species, a hexameric repeat (PHNPGY) 7 is conserved (51). There has been disagreement as to whether avian PrP also binds Cu 2ϩ . Studies have suggested that synthetic peptides of the chicken hexarepeat will bind copper with M affinity (36), others (52) have indicated no Cu 2ϩ binding to full-length chicken PrP, and others (20,53) have observed binding of Cu 2ϩ to full-length chicken PrP.
There is little spectroscopic evidence of Cu 2ϩ binding to avian PrP. With this in mind a peptide representing the longest avian sequence containing just two histidine residues has been synthesized, a 17-mer, NPGYPHNPGYPHNPGYP. Fig. 5 shows CD spectra in the visible and UV region on the addition  of increasing amounts of Cu 2ϩ to this avian PrP two-hexarepeat peptide. It is clear from changes in the spectra with copper addition that Cu 2ϩ does bind to avian PrP.
The mode of binding is very different from that of the mammalian octarepeat sequence. The visible CD bands are quite different from those observed for the mammalian repeats. Most strikingly, there is a negative band observed at 330 nm rather than the positive band seen in the mammalian octarepeats. The binding curve for a CD band at 330 nm (Fig. 5B) indicates the stoichiometry is also very different from mammalian PrP with only one Cu 2ϩ ion binding to the two-hexarepeat peptide. The d-d bands have the same sense as the mammalian spectra but with a positive band at 550 nm, a crossover at 625 nm, and a negative band at 700 nm. The intensities of the bands are much weaker by an order of magnitude relative to mammalian PrP. Changes in the UV region of CD spectrum with Cu 2ϩ addition (Fig. 5A) are also very different from the mammalian PrP(octas) (see Fig. 4A for comparison). The copper-free form of avian PrP hexarepeat peptide is not characteristic of random coil. This is presumably because of the high proportion of proline in the sequence. Addition of excess copper causes a decrease in the negative CD band at 210 nm.
The crystal structure of Cu-HGGGW (45) indicates copper coordination by an amide nitrogen from the second glycine after the histidine residue. In the case of PHNPGY the second residue following the histidine residue is proline, which is not available for coordination to the copper in the avian hexarepeat. For this reason, a very different complex must form. The stoichiometry for the two-hexarepeat peptide strongly suggests that both histidine residues coordinate to a single copper ion. The preservation of copper binding in the repeat region in avian, as well as mammalian PrP (although with different coordination geometry), supports a functional role for this domain.
Proline and Glycine Are Also Critical Residues in the Copper-Octarepeat Complex-To investigate which residues are essential in the Cu-HGGGW complex, in addition to histidine and tryptophan, whose side chains are involved in coordination, two alternative sequences were synthesized. In one peptide the glycine residues were replaced with alanine to produce the peptide HAAAW. In the second peptide, the proline residue preceding the histidine is replaced with alanine to produce the peptide GQAHGGGW. Both changes in the sequence were found to have a dramatic effect on Cu 2ϩ binding. Fig. 6A shows the visible CD spectrum of HAAAW with Cu 2ϩ addition. Cu 2ϩ still binds to the histidine residues but with a very different coordination geometry. The spectra in both the visible and UV regions (see Fig. 6B) are very different from HGGGW, which is also shown in Fig. 6A for comparison. Indeed, the inset in Fig. 6A shows that the stoichiometry is Cu(HAAAW) 2 rather than 1:1 observed for HGGGW. It is clear that glycine residues are required for main chain coordination. The additional and space available with glycine residues is necessary for coordination, in the manner observed for the octarepeats of PrP.
The presence of proline is also necessary to form the complex observed in the octarepeats, although it is not directly involved in coordination. Fig. 6, C and D, shows the CD spectrum of GQAHGGGW with copper addition. Again both the visible and UV regions of the CD spectrum are profoundly different when compared with the octarepeat spectrum. Cu 2ϩ ions tend to coordinate main chain amides to the N terminus of His residues when available (49). The presence of proline proceeding the His residue in the PrP octarepeats forces an alternative coordination geometry.
These data show that the sequence of amino acids required to form the complex observed in the octarepeats is quite specific. Not only are the side chains of His and Trp required but so are the intervening Gly residues and also the proceeding proline. The sequence of amino acids in four successive repeats is also necessary for Cu 2ϩ to bind to PrP with the required geometry for recognition associated with Cu 2ϩ -induced endocytosis of PrP C . Mn 2ϩ Does Not Bind to the Octarepeat Region of PrP-It has been reported that Mn 2ϩ can bind to PrP and can substitute for Cu 2ϩ in the octarepeat region (46). Mn 2ϩ appears to alter PrP C to a protease-resistant conformation that forms fibrils. Furthermore, PrP C expression influences uptake of Mn 2ϩ into cells. It is speculated that Mn 2ϩ could have a role in the formation of the scrapie isoform of the PrP generated in sporadic prion diseases. The possibility that imbalances in environmental cations may induce conditions favoring the formation of protease-resistant PrP in sporadic Creutzfeldt-Jacob disease is controversial (54, 55) but has gained interest in the British popular press. Evidence for Mn 2ϩ binding to PrP is based on equilibrium dialysis studies. As yet there have been no spectroscopic studies of Mn 2ϩ binding to PrP. Fig. 7A shows a CD spectrum in the UV region of the PrP(4octa) peptide before and after the addition of Mn 2ϩ . The apopeptide gives a negative CD band at 195 nm, ⌬⑀ Ϫ1.5 M Ϫ1 cm Ϫ1 per mean residue, which is  6. A and B, mutated PrP peptides HAAAW and GQAHGGGW, respectively. The CD spectrum of HAAAW in the visible and UV regions is shown, and Cu-HGGGW is also shown for direct comparison. C and D, GQAHGGGW visible and UV regions. The insets are direct binding curves for HAAAW and GQAHGGGW showing change in ⌬⑀ monitored at 650 and 550 nm, respectively, with Cu 2ϩ addition.
characteristic of an unstructured peptide. Addition of 8 mol equivalent of MnCl 2 , pH 7.5, has no effect on the CD spectrum. In contrast, subsequent addition of Cu 2ϩ shows a substantial change in the CD spectrum. The changes in the CD spectrum in the presence of Cu 2ϩ indicate marked structuring of the peptide with Cu 2ϩ binding. Fig. 7B shows a CD spectrum for PrP(4octa) in the visible region in the presence of 4 mol equivalent of Cu 2ϩ . The CD spectra gives characteristic CD bands due to d-d electronic transitions at 580 and 680 nm. Addition of as much as 400 mol equivalent of Mn 2ϩ has no effect on the spectrum.
The complete lack of change in the CD spectrum on the addition of Mn 2ϩ suggests that even at high levels of Mn 2ϩ and peptide (0.4 and 0.05 mM, respectively), Mn 2ϩ does not bind to the octarepeat region of PrP. The CD spectrum indicates that upon subsequent addition of Cu 2ϩ , the copper binds to PrP(4octa) in an identical manner as PrP(4octa) with no Mn 2ϩ present. Clearly Mn 2ϩ does not inhibit Cu 2ϩ binding to PrP(4octa). In the reverse experiment, shown in Fig. 7B, the metal complex is pre-formed with Cu 2ϩ . In this situation Mn 2ϩ is unable to displace Cu 2ϩ from PrP(4octa) even when a 100fold excess of Mn 2ϩ is added.
It is clear that Mn 2ϩ does not bind to the octarepeat region under the conditions used in this study (pH 7.5 and no buffer present). However, equilibrium dialysis studies have indicated that Mn 2ϩ can compete with Cu 2ϩ ions bound to PrP and displace them (46). It may be that Mn 2ϩ binds elsewhere to PrP other than the octarepeat region, but it is clear from our studies that Mn 2ϩ will not directly displace Cu 2ϩ bound to the octarepeat region.
L-His Competes for Copper Binding to PrP but Does Not Form a Ternary Complex-The effects of L-histidine competition on Cu 2ϩ binding to PrP(4octa) were also investigated. As with the addition of glycine, L-His will strongly compete for Cu 2ϩ ions. The Cu 4 -PrP(4octa) signal at 580 and 320 nm is completely lost when between 4 and 5 mol equivalent of L-His is added (Cu-His 1:1 ratio). This is consistent with L-His having a higher affinity for Cu 2ϩ than PrP. We considered the possibility of a ternary complex of L-His, PrP, and Cu 2ϩ , which is thought to occur for the square-planar complex of Cu 2ϩ bound to serum albumin. However there is no evidence of a ternary complex with PrP. In Fig. 8A we see a simple diminution of the Cu 4 -PrP signals with L-His addition and the appearance of new CD signals identical to that observed for Cu-His complex only. Fig. 8B shows the CD spectra of Cu-His complex for comparison.

CONCLUSIONS
The affinity of Cu 2ϩ for PrP C has important implications for its function. The extracellular levels of Cu 2ϩ can vary between nM and M, reaching as high as 15 M in the neocortex (56). Cu 2ϩ affinity for PrP is in the same range. PrP C is therefore able to bind extracellular copper. The affinity of PrP C for Cu 2ϩ is in a range in which it could act as a sensor with binding triggered during increased levels of extracellular Cu 2ϩ . The copper binds cooperatively causing the unstructured N terminus to fold up in a specific manner; this then triggers increased endocytosis of Cu-PrP C . The metal ion binding appears to be specific for copper, and Mn 2ϩ binding to the octarepeat region is not observed. It is clear that the sequence of amino acids required for cooperative binding and endocytosis is highly specific. The four copper centers are not isolated, and the intervening residues between copper centers become structured in the presence of copper.