The PrP-like Protein Doppel Binds Copper*

Doppel (Dpl) is a glycosylphosphatidylinositol-an-chored protein expressed in the testis. It exhibits 26% sequence identity with the prion protein (PrP) but lacks the octarepeat region implicated as the major copper-binding domain. Contrary to expectations, Cu(II) induced a 26% reduction in the intrinsic fluorescence of Dpl(27–154) and a calculated K d for a single-site model of 0.16 (cid:1) 0.08 (cid:2) M . Other metals had minimal effects on fluorescence quenching. Matrix-assisted laser desorption ionization mass spectrometry of a Dpl peptide re-vealed binding of copper (but not other metals) to the helical (cid:3) B/B (cid:1) -loop- (cid:3) C subregion of Dpl. Fluorescence quenching and equilibrium dialysis analyses of this Dpl(101–145) peptide were compatible with a binding site of K Reactive Carbonyl Groups— carbonyl groups into oxidative OxyBlot oxida- kit (cid:3) secondary


Doppel (Dpl) is a glycosylphosphatidylinositol-an-
our data indicate that the ␣-helical region of mouse Dpl possesses a selective copper-binding site with a submicromolar K d and perhaps one or more lower affinity sites. Although metallated forms of Dpl might exist in vivo, analyses of Tg(Dpl)10329 mice were inconsistent with reports that Dpl expression is associated with increased carbonylation and nitrosylation of brain proteins. Thus, rather than comprising an important source of free radical damage, copper binding may serve to modulate the activity, stability, or localization of the Dpl protein.
The cellular prion protein (PrP C ) 1 and Doppel (Dpl) are glycosylphosphatidylinositol-anchored proteins with ␣-helical C-terminal domains. Whereas PrP C is expressed in the central nervous system and many peripheral tissues, the most notable site of Dpl expression is the testis (2,3), although transient expression in endothelial cells of the developing central nervous system has also been described (4). Both proteins can engender neurodegenerative diseases. PrP C serves as a precursor to a malfolded isoform denoted PrP Sc in prion infections such as scrapie and is mutated in familial prion diseases such as familial Creutzfeldt-Jakob disease and Gerstmann-Sträussler-Scheinker syndrome (5). In the case of Dpl, neuronal expression results in the death of cerebellar cells by apoptosis (6,7).
Although PrP C was first described in 1985 (8), discerning its physiological function has proven to be challenging (9). The absence of PrP in homozygous null mice was once thought to cause a profound cerebellar ataxia (10), but it is now clear that this syndrome is an artifact of the genetic engineering used to create the deleted PrP gene (Prnp) alleles, resulting in ectopic expression of Dpl (2,6,7,11). Recent analyses have instead focused on a potential role for PrP C in signal transduction (12), interactions with other cell surface proteins (13), and perhaps elements of the extracellular matrix (14 -16). Several reports have established that under conditions of oxidative or cellular stress, PrP C can serve a protective role (17)(18)(19).
One particularly intriguing connection concerns the ability of PrP C to bind copper ions. Many studies have documented the selectivity, multivalence, and pH dependence of Cu(II) binding, as well as binding constants potentially compatible with Cu(II) occupancy in vivo (20 -23). There is a broad consensus that the N-terminal region of PrP is the major contributor to copper binding ability. This glycine-rich domain, absent from Dpl, contains five "octarepeat" motifs. Although the first repeat is of the general form PGGGGWGQ, the four following identical repeats PHGGGWGQ bear histidine residues and bind copper in a cooperative fashion (Hill coefficient, ϳ3.5) (21,23,24). One plausible scenario for physiological activity, based upon stimulation of endocytosis by extracellular copper (25,26), is that PrP C plays a role in copper uptake or scavenging, perhaps germane to the synaptic location of this molecule. An alternative viewpoint is that PrP C possesses a superoxide dismutaselike enzymatic activity (27). Irrespective of these two contrasting views of the physiological role of PrP C , it is to be emphasized that genetic evidence exists for the importance of the main copper-binding domain, insofar as alterations in the number of octarepeats cause familial prion diseases (28 -31).
In addition to the octarepeat Cu(II) binding site of PrP C , there has been interest as to the existence of additional Cu(II)binding sites C-terminal to these motifs. In the course of our studies to inventory Cu(II)-binding sites in recombinant PrP, we reasoned that Dpl, a PrP-like protein lacking octarepeats, would comprise a useful control. Unexpectedly, as described below, Dpl was found to exhibit selective Cu(II) binding with an affinity comparable with that of PrP. ) refolded by a copper catalysis method (3) and mouse ) refolded in the presence of glutathione (32) were prepared as described previously. Intrinsic copper content was determined by atomic absorption spectrophotometry. A MoDpl helix-loop-helix peptide was synthesized by standard Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry. This peptide corresponds to residues 101-145 of the mouse doppel and includes the interrupted ␣-helix B (␣B/BЈ), the ␣-helix C (␣C), and the loop between these two helical structures (32). The disulfide bond was introduced by treating the peptide overnight in an aerated solution of ammonium bicarbonate. Protein concentrations were calculated by amino acid hydrolysis (33). Far-UV CD measurements, endoproteolysis, and S-carbamidomethylation of cysteines to determine the presence of integrity of disulfide bonds were carried out as described previously (3,33).

Dpl Peptide and Proteins-Mouse
Fluorescence Spectroscopy and Titrations-Steady-state fluorescence spectra were recorded on a Photon Technology International QM-1 spectrophotometer. Emission spectra were collected from 310 to 400 nm ( ex ϭ 280 nm, 1 s/nm, band pass ϭ 1 nm for excitation and emission Biochemicals) was used to digest the intact peptide and copper-peptide complex for 2 h at room temperature (the ratio of enzyme:protein was 1:20). In some cases, before chymotrypsin digestion, the samples were incubated with 5-fold molar excess diethlpyrocarbonate (DEPC) at room temperature for 30 min. The samples were then analyzed by using MALDI-MS as described previously (1).
Equilibrium Dialysis-MoDpl(101-145) and control protein (Hu-PrP(23-98): (21)) were prepared in NEMO-KCl buffer, pH 7.4. Chelate solution (Gly:Cu:Gly) was prepared by mixing CuSO 4 solution with a 2-fold molar excess of glycine and then adjusting the pH to 7.4. The equilibrium dialysis experiments were carried out in multiple 5-cell equilibrium dialyzers (Spectrum Industries). One ml of protein plus chelated Cu(II) mix (retentate) and 1 ml of buffer (dialysate) were placed on opposite sides separated by a 1-kDa cut-off cellulose ester membrane (Spectrum Industries) in a dialysis chamber. For a given experiment, the concentration of protein was held constant, and the concentration of chelated Cu(II) was varied over a Cu(II):protein ratio from 0.5:1 to 40:1. The system was rotated to come to equilibrium for 108 h at 4°C (verified by analysis of "blank reactions" not containing protein), and the concentration of all forms of copper in each halfchamber was measured by atomic absorption spectroscopy. 2.5 M protein was used in all of the equilibrium dialysis experiments.
Copper Determinations-Copper was determined by means of graphite furnace atomic adsorption spectrophotometry. Protein samples for analysis were diluted with 0.5% nitric acid (nitric acid Ultrex II; J. T. Baker) to concentrations within the linear range of graphite furnace copper measurements. The dilution range varied from 10 to 520 times. The diluted samples were transferred to Eppendorf tubes and placed within the autosampler of the atomic adsorption spectrophotometer (Varian Graphite Furnace GTA 100/Varian Spectra 880), and absorbance was measured using furnace conditions developed in the trace element laboratory. Calibration was against working standard solutions prepared by dilution with 0.5% nitric acid from a 1000 mg/liter copper standard (J. T. Baker). Concentrations of stock solutions used for fluorescence quenching studies were verified by atomic absorption spectroscopy.
Mice-Tg(Dpl) and non-Tg littermate mice (Tg(Dpl) line 10329) have been described previously (7). Mice were between 15 and 20 weeks of age at the time of analysis. Antibodies-The anti-Dpl rabbit polyclonal antibody E6977 raised against recombinant mouse doppel (7) was used as described below. The anti-nitrotyrosine antibody was from Cayman Chemical (a gift from N. Cashman), whereas antibodies against 2,4-dinitrophenylhydrazine were obtained from Intergen.
Western Blotting-10% (w/v) homogenates of half brains were prepared in 0.32 M sucrose supplemented with one complete-mini protease inhibitor tablet (Roche Diagnostics) on ice. Homogenates were aliquoted and stored at Ϫ80°C until needed. The total protein concentration of each sample was determined using Bradford assay reagents (Bio-Rad) as per the manufacturer's instructions. For the immunoblot analysis of Doppel expression and nitrosylation, samples of 50 g of total brain protein were prepared and boiled in loading buffer for 9 min and then centrifuged at 20,800 ϫ g for 5 min prior to fractionation by SDSpolyacrylamide gel electrophoresis. The samples were loaded onto a 10 -20% Tricine gradient gel (Novex) and electrophoresed at 100 V for ϳ100 min. The samples were then transferred onto 0.2 m nitrocellulose (Schleicher & Schuell) for ϳ80 min at 25 V (Novex semi-dry apparatus). The blots were blocked with 5% (w/v) milk powder in Tris-buffered saline (pH 7.4), 0.02% Tween 20 (Buffer A) for 1 h with gentle agitation at room temperature and then probed with the appropriate primary antibody (anti-Dpl E6977 used 1:4000, anti-nitrotyrosine used 1:5000) in 0.5% milk powder in Tris-buffered saline (pH 7.4), 0.02% Tween 20 overnight at 4°C. The appropriate horseradish peroxidase-conjugated secondary antibodies were used. The blots were visualized using ECL detection reagents (Amersham Biosciences) and exposed on X-Omat AR film (Eastman Kodak Co.).
Detection of Nitrotyrosine Residues-A positive control for the nitrosylation blot was made just prior to SDS-PAGE. Peroxynitrite was added to 50 g of a wild type mouse brain homogenate in sodium phosphate buffer while vortexing to create a final concentration of 15 mM peroxynitrite. After 2 min, 2ϫ gel loading buffer was added to quench the reaction, and the sample was kept on ice. The positive control as well as the sample homogenates of 50 g of total brain protein were subjected to Western blot analysis with a monoclonal anti-nitrotyrosine primary antibody in Buffer A.
Detection of Reactive Carbonyl Groups-The amount of carbonyl groups introduced into the total protein of the brain homogenates via oxidative modification was detected using the OxyBlot protein oxidation detection kit (Intergen). Samples of 15 g of total brain proteins in 0.32 M sucrose and 2% 2-mercaptoethanol were denatured and then derivatized with 2,4-dinitrophenylhydrazine, electrophoresed, and electroblotted as per the manufacturer's instructions. The blots were blocked with 1% (w/v) bovine serum albumin in phosphate-buffered saline (pH 7.4), 0.05% Tween 20 for 1 h at room temperature and then probed with a rabbit anti-DNP primary antibody, used at a concentration of 1:150 in blocking solution overnight at 4°C. Following incubation with the goat anti-rabbit horseradish peroxidase-conjugated secondary antibody, blots were visualized using ECL or ECL-Plus detection reagents as described above or subjected to protein quantification using ImageQuant software (Molecular Dynamics, Inc.), respectively.

Recombinant Dpl Proteins and a Synthetic Dpl Peptide-To
exclude trivial or idiosyncratic effects pertaining to expression and purification of Dpl proteins from Escherichia coli, we used two recombinant polypeptides prepared by different methods (3,32) (in the absence of the N-terminal sequences of mature Dpl isolated from mammalian cells, the lengths of these two proteins, Dpl  versus Dpl(26 -157), merely reflect ambiguities in assigning N-and C-terminal signal peptides by different algorithms). For Dpl  prepared by oxidative refolding in catalytic concentrations of Cu(II), we used atomic adsorption spectroscopy to quantify residual metal content (Ͻ3 ϫ 10 Ϫ4 mol of copper/mol of dialyzed protein). Extrapolating from studies attributing special significance to the helix B/BЈ-loop-helix C region of Dpl (34), we hypothesized that this area of Dpl may comprise a site with biological activity and synthesized a corresponding peptide, Dpl(101-145) (Fig. 1A). This peptide encompasses one of the two disulfide bonds present in full-length Dpl, the "inner" Cys 109 -Cys 143 linkage located at an equivalent position to the single disulfide bond present in PrP. Measured in NEMO-KCl buffer at pH 7.4, Dpl(101-145) exhibited a CD spectrum with minima at 208 and 222 nm, indicative of a high ␣-helical content (Fig. 1B). MALDI-MS analysis of Dpl(101-145) peptide showed a single charged peptide signal at m/z 5162.9 ( Fig. 2A), in excellent agreement with the calculated molecular mass of 5162.9 Da as [MϩH] ϩ . S-Carbamidomethylation (3,33) was used to verify the presence of appropriate disulfide linkages in the recombinant polypeptides (not shown) and Dpl(101-145) peptide (see  Fig. 3, no signals corresponding to metal-peptide complexes were detected in these other analyses (Fig. 2, C-H).
Copper-Protein Interactions Assessed by Fluorescence Quenching-Fluorescence quenching analysis for recombinant mouse Doppel (recDpl) was undertaken exploiting the intrinsic properties of four tryptophan residues at positions 35, 85, 136, and 151, and five tyrosine residues at positions 78, 79, 84, 91, and 92 (present within both Dpl(27-154) and Dpl (26 -157)). To confirm that these residues undergo a change in environment, we measured the change of fluorescence intensity of Trp and Tyr residues upon Cu(II) addition. The diminutions in intrinsic fluorescence (integrated from 310 to 380 nm) induced by Cu(II) addition were 26% for Dpl  and 30% for Dpl (26 -157). However, the change in the intensity was not accompanied by a significant shift in the wavelength position of the maximum ( max ), suggesting that the solvent accessibility of Trp residues are not affected significantly by copper binding. Also, CD spectra were not notably altered by copper addition (not shown), suggesting that copper binding is not accompanied by a significant change in the secondary structure of recDpl. MoDpl(101-145) contains a single tryptophan residue at position 136 (Fig.  1A), facilitating similar fluorescence spectroscopic studies to those performed for the aforementioned full-length recDpl molecules.
Binding affinities and stoichiometries for copper binding to the various proteins and peptides were quantified by fitting of the conventional model expressed in Equations 1 and 2 to the fluorescence data in Fig. 3, where F is the measured fluorescence, normalized by taking the difference in the first and last data point as 1, F 0 is the fitted fluorescence before addition of copper, and ⌬F is the fitted change in fluorescence.  concentration. In accord with data presented in Figs. 2 and 5, binding was considered in terms of "one-site" and "two-site" models. The protein concentration terms in Equations 1 and 2 were first set to the measured value of 0.7 M (i.e. one copperbinding site/molecule), and the best fits of the model to the data are shown in the lower panels of Fig. 3 (B, D, and F). The affinities of Dpl  and Dpl  for copper were essentially identical. For Dpl , a unique binding site would have a K d value of 0.16 Ϯ 0.08 M (Fig. 3B), and for Dpl-(26 -157), the analogous value is K d ϭ 0.20 Ϯ 0.05 M (Fig. 3D). Remarkably, copper was found to interact with the 44-mer, Dpl(101-145) with nearly the same characteristics as seen for the larger fragments; for a unique site, the K d for a copperpeptide complex is estimated to be 0.36 Ϯ 0.04 M (Fig. 3F).
Residual plots and chi-squared values for the best fits to the data of Equation 1 were also ascertained for when [P] was set at 1.4 M, which would hold if two copper ions bound to two similar sites in one protein or peptide molecule. Smaller deviations between model and data for the 1:1 model than for the 2:1 model were apparent in the residual plots (Fig. 3, B, D, and  F, lower panels). Chi-squared values, which increase with differences between the model and data, were smaller in all cases for the 1:1 binding model than for the 2:1 model (Fig. 3, legend). Modeling the stoichiometry at levels of 3:1 and beyond by increasing [P] during fitting led to progressively poorer agreement with the data (not shown). Taken together, these results do not support the hypothesis of two similar copper sites/ polypeptide, in agreement with the absence of palindromic or repeated amino acid sequences in Dpl. These data are compatible with an interpretation that each mole of protein binds 1 mole of copper at a single predominant site under the particu-lar conditions of the assay. However, the presence of a weaker binding site or of a second binding site associated with a metalpeptide complex with different fluorescent properties cannot be excluded based solely upon these fluorescence titrations.
To evaluate metal binding selectivity of the Dpl protein, we calculated the fluorescence intensity change of 0.7 M Dpl-    (Fig. 3G), Cu(II) induced a 25.85 Ϯ 0.39% quench of fluorescence intensity (Fig. 3G, first column). Although we cannot exclude that Dpl complexes (which do not result in fluorescent changes) with metals other than copper exist, we note that similar selectivity was found in MALDI-MS analyses (Fig. 2). Also, these data are strikingly reminiscent of prior studies of selective interactions between copper and hamster PrP (22). binding site/monomer and that fluorescence provides a reliable signal for the rapid measurement of copper binding to Dpl. (27-154), we exploited the ability of copper to protect certain reactive amino acid residues (predominantly histidine) from chemical modification with DEPC (1). Upon incubation of native protein (15 M) (Fig. 5A) with DEPC, up to nine DEPC adducts were detected per mole of Dpl(27-154) (Fig. 5B), assuming an increment of 72.06 Da/mono-carbethoxylated adduct. In the case of protein preincubated with a molar excess of copper, there was a shift of n Ϫ 1 in the spectrum of adducts (Fig. 5C), indicating that one DEPC-reactive residue was protected from modification by the presence of copper.

DEPC Footprinting of Dpl(27-154)-In a further technique to interrogate copper binding by Dpl
Copper Binding in the Region of Dpl(122-136)-Chymotrypsin digestion was used to map the location of Cu(II)-binding site(s) detected by the prior MALDI-MS analysis of Dpl(101-145). In the absence of Cu(II) (Fig. 6A) bated with 5-fold excess DEPC and then digested with chymotrypsin. Compared with the signal in Fig. 6A, the signal at m/z 1910.09 corresponding to unmodified residues 122-136 (C3 ϩ C4 ϩ C5 ϩ C6, calculated 1910.03 Da) was reduced in amplitude (Fig. 6C). Two new peaks were found at m/z 1982.09 and 2055.08 and indicated one mono-and two mono-carbethoxylations of fragment 122-136 (calculated 1982.09 and 2054.14 Da, respectively). A different mass spectrum was obtained when Dpl(101-145) peptide was preincubated with 10-fold excess Cu(II) prior to reaction with DEPC (Fig. 6D). Here the major peaks were at m/z 1910. 34, 1973.28, 2036.21, 1982.38, and 2046.31. Diminution of the mono-carbethoxylated peak at 1982.09 and an increase in the peak corresponding to unmodified 122-136 C1 ϩ C2 ϩ C3 peptide (compare with Fig. 6D) indicated that a site of potential DEPC modification was blocked by copper binding. As per Fig. 6B, peaks at m/z 1972.53 and 2036.03 are indicative of the 122-136 peptide with one or two bound copper ions. Additionally, a peak at 2046.31 is likely equivalent to one copper plus one mono-carbethoxylation (calculated 2045.59 Da) (Fig. 6D). These data can be reconciled with hypotheses that (i) a second copper-binding site within Dpl(101-145) is independent of the His 131 residue or (ii) binding of a second copper ion involves the second nitrogen in the imidazole ring, with this nitrogen being unmodified in the mono-carbethoxylated adduct. Also, in contrast to Fig. 6B, in-stead of a peak at m/z 3286.84 Da corresponding to fragment (101-121)-S-S-(137-145), a new peak at m/z 3358.53 Da corresponding to one mono-carbethoxylation (calculated 3358.71) indicates one DEPC modification within this disulfide-bridged fragment (Fig. 6C). No change was observed in this peak (3358.83 Da) in copper and DEPC-treated versus DEPC-treated samples (Fig. 6D), indicating that Cu(II) did not interact with this fragment.
Fine Mapping DEPC Modification of His 131 within Dpl(122-136)-The post-source decay technique of tandem analysis was used to verify DEPC modification and copper protection of His 131 . Partial post-source decay spectra of fragment 122-136 and its counterpart in DEPC-and copper-DEPC-treated samples are presented in Fig. 7. Fig. 7A shows the scheme of nomenclature for peptide post-source decay fragment ions of chymotryptic peptide 122-136 derived from intact Dpl(101-145). In unmodified samples, a peak at m/z 109.9 corresponds to an immonium histidine ion (His 131 ) (calculated 110.0 Da) (Fig. 7B). After DEPC treatment, a new peak at m/z 181.7 indicated one mono-carbethoxylation of histidine (calculated 182.0 Da) (Fig. 7C). Incubation of the sample with Cu(II) prior to DEPC treatment resulted in the disappearance of one monocarbethoxylation of histidine and indicates that Cu(II) coordination to this histidine protected the imidazole ring from DEPC modification (Fig. 7D). It is also of note that peaks at m/z 128.7, corresponding to an immonium arginine ion (Arg 123 or Arg 133 ) (calculated 129.1 Da), and at m/z 226.9, corresponding to a b 2 -NH 3 ion (SR-NH 3 , calculated 227.1 Da) were always present (Fig. 7, B-D), indicating that neither fragment ion reacted with DEPC or Cu(II). Another peak for the y 9 SKLHQRVLW ion at m/z 1166.7 (calculated 1166.7 Da) in the unmodified sample (Fig. 7B) was replaced by a signal at m/z 1239.6, corresponding to one mono-carbethoxylation of the peptide ion (calculated 1238.8 Da) in the DEPC-treated sample (Fig. 7C). Incubation of the sample with Cu(II) before DEPC treatment led to the disappearance of the mono-carbethoxylation signal and replacement by a signal at 1230.9 (calculated 1230.2 Da), indicating the binding of one copper ion. In sum, the results demonstrate that Cu(II) coordination to a y 9 ion protected the peptide ion from DEPC modification and, in turn, the involvement of His 131 in coordinating copper ions in the context of the peptide. Given the similarities between the peptide and protein in terms of their sequences, copper affinities, and fluorescence changes in response to copper binding, these results suggest that His 131 will coordinate copper in the context of the fulllength protein as well.
Dpl Expression and Oxidative Damage-The proposal that Dpl expressed in the central nervous system is a source of oxidative damage (35) has a potential parallel in the notion that Dpl is metallated in vivo and thus may be a source of free radicals. Prior analyses in this area examined the Dpl-expressing Rcm0 line of Prnp 0/0 mice (35). To address Dpl as a source of oxidative damage in vivo, we exploited Tg(Dpl) mice created with the hamster PrP gene promoter, exhibiting potent central nervous system expression of Dpl and ataxia at about 1 year of age (7). Brain homogenates of non-Tg and Tg(Dpl)10329 mice were subjected to polyacrylamide gel electrophoresis and probed with antibodies to known markers of oxidative stress. As seen in Fig. 8A, Tg(Dpl) mice express Dpl in the central nervous system, as demonstrated by the strong heterodisperse signals at ϳ34 kDa, when compared with non-Tg controls. When samples of the same homogenates were analyzed for nitrotyrosine modification of proteins using an anti-nitrotyrosine antibody, low signal intensities were obtained, and no obvious differences in the levels of nitrotyrosine formation between Tg(Dpl) and non-Tg mice (Fig. 8B) were found. Furthermore, modification of carbonyl groups, another direct indication of oxidative stress, also appeared to be unrelated to increased doppel expression (Fig. 8C). Quantification of this blot for reactive carbonyl groups is presented in Fig. 8D, revealing no significant differences between the Tg(Dpl) and non-Tg sample groups (p ϭ 0.97).

DISCUSSION
Metal Binding to Dpl in Vitro-Many studies have documented binding of one copper ion/histidine containing octarepeat within PrP, when measured at neutral pH (Ref. 36 and references therein). In addition to these four sites, there is growing agreement upon the existence of a fifth copper site within PrP located slightly C-terminal to the octarepeats and likely involving histidines 95 and 111 (mouse PrP numbering scheme) (1,(37)(38)(39). Although sequence alignments reveal that none of these residues have an equivalent in Dpl, our in vitro studies have revealed, nonetheless, that Dpl also possesses copper binding properties.
It is unlikely that the copper binding detected in our studies has a trivial origin. Although contaminants from E. coli capable of binding copper can be hypothesized, they would have to be present in substantial quantities to account for the observed stoichiometry of Ն1 mol of copper/mol of polypeptide. We also note that similar fluorescence quenching properties were produced by two forms of recDpl prepared by different chromato-graphic procedures. Furthermore, a Dpl(101-145) peptide synthesized in vitro also had the ability to bind copper (Fig. 2), as did a Dpl(122-136) proteolytic derivative of this peptide (Fig.  6). Another notable feature of the interaction with copper is the degree of specificity. Assessed by fluorescence quenching, the Dpl substrates exhibited no discernable affinity for other divalent cations, a pattern of selectivity reminiscent of PrP itself (1,20,22). Although it is possible that peptide complexes formed by metals other than copper have fluorescent properties similar to that of Dpl apo-protein and thus appear not to interact in this assay, the similar pattern of metal specificity derived from MALDI-MS analysis (Fig. 2) tends to argue against this interpretation and in favor of an intrinsic metal selectivity.
Our experiments demonstrate that Cu(II) binding to recDpl can be largely attributed the ␣B/BЈ-loop-␣C domain; using fluorescence quenching, K d values for full-length protein and Dpl(101-145) differ only by a factor of two. Binding assessed directly by equilibrium dialysis and indirectly by fluorescence  1 and 2) and Tg(Dpl) (lanes 3 and 4) mouse brain homogenates immunoblotted with the anti-Dpl E6977 antibody demonstrate the relative levels of Dpl expression in the samples used. Cross-reactive bands between 78 and 210 kDa present in all samples served to demonstrate uniform sample loading. B, analysis of nitrotyrosinylation. Lane 1, peroxynitratetreated wild type positive control (ϩCtl) brain homogenate. Non-Tg (lanes 2 and 3) and Tg(Dpl) (lanes 4 and 5) brain homogenates were probed with an anti-nitrotyrosine antibody. These analyses demonstrated similar, very low levels of immunoreactivity (indicative of nitrotyrosine modification) in non-Tg and Tg(Dpl) mice. C, analysis of protein carbonyl groups. 15-g samples of non-Tg (lanes 1 and 2) and Tg(Dpl) (lanes 3 and 4) total brain protein was derivatized with 2,4dinitrophenyhadrazine and immunoblotted to detect relative levels of reactive carbonyl groups. Underivatized control samples yielded no signal (not shown). D, quantitative representation of carbonyl immunoreactivity via phosphorimaging analysis. A paired t test yielded a p value of 0.97. There is no significant difference between the non-Tg and Tg samples. quenching analysis can be accounted for by a single-site model, with the additional site detected by MALDI-MS analysis, perhaps reflecting contribution of a low affinity site only filled by high concentrations of Cu(II). In the NMR structure for Dpl (32), the ␣B/BЈ-loop-␣C domain lies at the opposite end of the molecule from the free C terminus (the site of the glycosylphosphatidylinositol anchor addition) and may therefore be displayed toward the extracellular environment in vivo. The notion of copper-binding sites within the ␣-helical domain of a cellular prion protein has a precedent from studies of PrP (40). Of potentially greater importance, this region of the Dpl molecule encompassing a kinked helix B (␣B/BЈ) contributes to a triangular hydrophobic pocket with no exact equivalent in PrP (32). It will be of interest to determine the details of how amino acid residues in this vicinity contribute to copper binding. The observation that a cluster of fCJD mutations in the helix Bloop-helix C region of PrP recapitulate conserved residues in Dpl provides further impetus for deciphering the biological properties of this region of the protein (34).
Regarding the issue of binding affinity, some of our analyses use copper in the form of CuCl 2 . Such solutions contain oxy and hydroxy polymers of copper, which may be kinetically inert (41). Accordingly, some studies of PrP have instead used copper presented in the form of glycine-or histidine-chelated complexes. These experiments still yield binding at micromolar concentrations of copper, but back-calculations based upon the dissociation constants of the metal-amino acid complexes can be used to derive a concentration for free ionic metal and protein-metal binding constants estimated in the range of 10 Ϫ14 M (38). Our equilibrium dialysis binding analysis of Dpl(101-145) was performed with glycine-chelated copper and a K d value for a high affinity site of 0.4 Ϯ 0.2 M. Irrespective of the validity of back-calculating ionic copper concentrations from experiments using copper chelates (where ternary complexes of protein/copper/amino acid complexes might also have to be considered (42)), our numerical data (i) are nonetheless quite similar to estimates of binding constants derived from "uncorrected" binding data for PrP obtained under similar conditions (21) and (ii) illustrate that Dpl, like PrP, can extract copper from an amino acid chelate, a form that might better approximate presentation in biological fluids than in solutions of copper salts. "Exchangeable" copper in plasma bound to amino acids has been estimated at 3.6 M, whereas the total copper levels in plasma and seminal fluid are measured at 15.5 Ϯ 9 and 5.9 Ϯ 3.7 M, respectively (43,44) and are thus about 1 order of magnitude above the K d values presented herein. These data strongly suggest that metallated forms of Dpl could exist in vivo, which is of potential relevance to physiological function(s) in spermatogenesis and also to the pathological effect of central nervous system expression.
Central Nervous System Expression of Dpl: Oxidative Damage and Neurotoxicity-Based upon prior connections between copper-polypeptide complexes and the generation of reactive free radical species, we investigated indicators of oxidative damage in Tg(Dpl) mice (35). Ectopic expression of Dpl in the brain is known to be toxic, and genetically engineered mice with this property succumb to an ataxic syndrome characterized by apoptotic death of cerebellar cells (2,6,7). Our studies exploited the Tg(Dpl)10329 line of mice (7). In contrast to another study (35), we found no evidence of enhanced oxidative damage to brain proteins (Fig. 8). This discrepancy cannot be attributed to the penetrance of the disease phenotype in the Tg(Dpl)10329 mice studied here. Tg(Dpl)10329 mice develop an ataxic syndrome at 375 Ϯ 8 days, as compared with Rcm0 mice, which develop ataxia much later in life at 611 Ϯ 12 days of age. Furthermore, the Tg(Dpl)10329 mice at age of sacrifice in the analyses of oxidative markers described here were approximately twice as old as Rcm0 mice analyzed for markers of oxidative damage (35). It is possible that the discrepancy between our findings and those of Wong et al. (35) reflects a contribution of modifier loci deriving from different genetic backgrounds. Nonetheless, because the Tg(Dpl)10329 mice described here have levels of protein carbonyl and nitrotyrosine formation no different from those seen in non-Tg mice (yet are certainly prone to loss of cerebellar cells), it seems quite unlikely that the neurotoxic properties of Dpl depend upon causing oxidative damage to proteins.
Competition between Dpl and PrP C : a Role for Cu?-Despite our failure to implicate oxidative damage in the neurotoxic action of Dpl, it remains plausible that an ability to bind copper ions is necessary for neurotoxicity proceeding by a different, but as yet undefined, mechanism. The studies presented here were not designed to elucidate what this mechanism might be. Nonetheless, by revealing that the ability to bind copper is a property of both Dpl and PrP C , they may speak to "competition" between the PrP C and Dpl proteins in modulating this pathologic process (7). Specifically, because expression of PrP C can nullify the neurotoxicity of Dpl to cerebellar cells, the question now arises as to whether the competing actions of these two glycosylphosphatidylinositol-linked proteins somehow derive from a shared predilection for this particular transition metal. Experiments to recapitulate Dpl-mediated neurotoxicity in a tractable system may allow us to appraise this possibility.