Mapping Cu(II) Binding Sites in Prion Proteins by Diethyl Pyrocarbonate Modification and Matrix-assisted Laser Desorption Ionization-Time of Flight (MALDI-TOF) Mass Spectrometric Footprinting*

Although Cu(II) ions bind to the prion protein (PrP), there have been conflicting findings concerning the number and location of binding sites. We have combined diethyl pyrocarbonate (DEPC)-mediated carbethoxylation, protease digestion, and mass spectrometric analysis of apo-PrP and copper-coordinated mouse PrP23–231 to “footprint” histidine-dependent Cu(II) coordination sites within this molecule. At pH 7.4 Cu(II) protected five histidine residues from DEPC modification. No protection was afforded by Ca(II), Mn(II), or Mg(II) ions, and only one or two residues were protected by Zn(II) or Ni(II) ions. Post-source decay mapping of DEPC-modi-fied histidines pinpointed residues 60, 68, 76, and 84 within the four PHGGG/SWGQ octarepeat units and residue 95 within the related sequence GGGTHNQ. Besides defining a copper site within the protease-resistant core of PrP, our findings suggest application of DEPC footprinting methodologies to probe copper occupancy and pathogenesis-associated conformational changes in PrP purified from tissue samples. Prion diseases are associated with a pathogenic conformational transition in binding sites in native mouse PrP. Our results are discussed in the context of the physiology and pathobiology of PrP. 1:20) was used to digest the intact protein, carbethoxylated protein, and carbethoxy- lated copper-protein complex fo r 2 h at 37°C (trypsin) or at room temperature (chymotrypsin). The samples were analyzed by MALDI-MS without further purification. Mass Spectrometry and Structural Modeling— MALDI-TOF-MS analyses were carried out by using a Perseptive Biosystem Voyager-DE STR mass spectrometer as described previously (27). Post-source decay (PSD ) spectra were acquired at DE-reflectron mode. The accelerating voltage was set at 20 kV, grid voltage at 75%, guide wire voltage at 0.024%, and delay time at 100 ns. The timed ion selector was pre-set to the [M (cid:3) H] (cid:3) mass of the peptide. The spectra were acquired in 10–13 segments with mirror ratios 1.0–0.13 and then assembled by the instrument software. Three-dimensional images of PrP121–231 were gen- erated using the program RasMac v2.5 and the NMR co-ordinates from Protein Data Bank file IAG2 (2).

Prion diseases are associated with a pathogenic conformational transition in the cellular prion protein (PrP C ) 1 to an infectivity-associated form commonly denoted PrP Sc . Although it is clear that PrP gene-ablated (Prnp 0/0 ) mice have no overt phenotype in a laboratory environment, consistent with functional redundancy with other neuronal proteins, it is plausible that the physiological function of PrP C involves binding of Cu(II) ions. Subsequent to the removal of N-and C-terminal signal peptides, the mature form of mammalian PrP C contains residues 23-231 of a 253-amino acid sequence encoded by a single copy gene, Prnp (1). The C-terminal region of PrP C forms a globular structure comprising three ␣-helices and two short ␤-strands (2)(3)(4) and, during the process of prion infection, can re-fold into protease-resistant, ␤-sheet-enriched aggregates (5)(6)(7)(8)(9). In contrast, the N-terminal domain of PrP is highly flexible and typically includes five tandem copies of a conserved octapeptide repeat motif (3,4,6). The N-terminal domain of human PrP C contains four identical "octarepeat" sequences (PHGGGWGQ: within residues 60 -91) and one homologous sequence but lacking a histidine residue (PQGGGWGQ: residues 51-59) (10). Residues 51-91 of mouse PrP C consists of two octarepeat sequences (PHGGGWGQ: residues 59 -66 and 83-90), two octarepeat sequences (PHGGSWGQ: residues 67-81), and one homologous sequence lacking a histidine residue (PQGGTWGQ: residues 51-59) (11).
Although the octarepeat motifs within PrP have no sequence homology to classical copper-binding proteins (12), there is now a broad agreement that these motifs bind to copper with a remarkable degree of selectivity; as such they may comprise the prototype of a new class of copper binding motif. Copper binding to the N-terminal octarepeat domain of PrP C was reported in recombinant human PrP23-98 using equilibrium dialysis (13) and in synthesized peptides using mass spectrometry (14,15), fluorescence spectroscopy (14), Raman spectroscopy (16), circular dichroism, proton nuclear magnetic resonance (NMR) spectroscopy (17), electron paramagnetic resonance (EPR), and electron spin-Echo envelope modulation spectroscopy (18). Copper binding has also been reported for nearly full-length forms of Syrian hamster (residues 29 -231) (19) and human PrP (residues 91-231) (20) and full-length mouse PrP (residues 23-231) (21). Affinity chromatography using immobilized copper ions has been used to purify mature, glycosylated PrP C isolated from hamster brain (22), while human PrP Sc in brain is inferred to be bound to transition metals (including copper) in situ, as assessed by the effects of chelators upon the size of protease-resistant fragments (23). With regard to stoichiometry, the number of Cu(II) binding sites in the N-terminal region of PrP C is reported between 2 and 5.6 being pH-dependent (13,15,(17)(18)(19)24). At neutral pH, copper binding to the N-terminal domain occurs in the micromolar range with positive cooperativity, with a remarkably close determinations of Hill coefficients by different laboratories: 3.4 (PrP23-98) (13), 3.3 (PrP58 -91) (17) and 3.6 (PrP23-98) (21). However, more recently, these stoichiometries and binding constants have been challenged, with the suggestion of two independent high affinity copper binding sites of 10 Ϫ14 and 4 ϫ 10 Ϫ14 M, deduced from the analysis of PrP58 -98 and PrP91-231, respectively (20). In this paper, we use chemical modification to map histidine-dependent copper binding sites in native mouse PrP. Our results are discussed in the context of the physiology and pathobiology of PrP.

MATERIALS AND METHODS
PrP Proteins-Two recombinant proteins were used for these studies. Purified human PrP23-98 (HuPrP23-98) contains 80 amino acids including four identical octarepeat motifs of the form PHGGGWGQ (13). Full-length mouse PrP (MoPrP23-231) was expressed and purified as described (25,26). MoPrP23-231 consists of 210 amino acids including four octarepeats in the N-terminal domain (two of the form of PHGGG-WGQ and two of the form PHGGSWGQ). All MoPrP23-231 samples were used within six months of preparation to circumvent altered aggregation properties seen in aged samples (26). Fig. 1, B and C show the amino acid sequences and predicted tryptic and chymotryptic peptides of HuPrP23-98 and MoPrP23-231.
Coordination and Modification of MoPrP23-231-5-fold molar excess DEPC was reacted with MoPrP23-231 (90 M) in NEMO-KCl buffer, pH 7.4, for 30 min at room temperature. In some cases, prior to the DEPC reaction, the protein was incubated with 10-fold molar excess CuCl 2 for 30 min. Sequencing grade modified trypsin (Promega) or chymotrypsin (Roche) (the ratio of enzyme:protein was 1:20) was used to digest the intact protein, carbethoxylated protein, and carbethoxylated copper-protein complex for 2 h at 37°C (trypsin) or at room temperature (chymotrypsin). The samples were analyzed by MALDI-MS without further purification.
Mass Spectrometry and Structural Modeling-MALDI-TOF-MS analyses were carried out by using a Perseptive Biosystem Voyager-DE STR mass spectrometer as described previously (27). Post-source decay (PSD) spectra were acquired at DE-reflectron mode. The accelerating voltage was set at 20 kV, grid voltage at 75%, guide wire voltage at 0.024%, and delay time at 100 ns. The timed ion selector was pre-set to the [MϩH] ϩ mass of the peptide. The spectra were acquired in 10 -13 segments with mirror ratios 1.0 -0.13 and then assembled by the instrument software. Three-dimensional images of PrP121-231 were generated using the program RasMac v2.5 and the NMR co-ordinates from Protein Data Bank file IAG2 (2).  1B), previously characterized in an equilibrium dialysis study. In earlier analyses with this recombinant polypeptide fragment, Cu(II) was used in the form of an Cu(II)-glycine 2 complex. These studies yielded a stoichiometry of 5.6 Ϯ 0.4 Cu(II) ions per peptide at saturation and binding curves indicating that between three and four Cu(II) ions are added to (or discharged from) PrP in a highly cooperative fashion (positive cooperativity, Hill coefficient ϭ 3.4) (13). Since the studies presented here involve mass determinations (rather than competition between glycine and peptide molecules for Cu(II) ions), Cu(II) was presented without the intervention of an amino acid chelator. Metal-protein complexes were measured using saturated sinapinic acid in 20% acetonitrile as matrix.  (Fig. 2B). These data are in close accord with other MALDI-MS data (14) and are inferred to represent a stable association with four copper binding sites in the octarepeat repeat region. In independent electrospray ionization MS analyses (ESI-MS), where protein/metal solutions mixtures are introduced into the instrument still in the aqueous phase it is possible to detect two additional, weaker binding sites (i.e. fifth and sixth binding sites) tentatively positioned at the free amino group of the N terminus and involving histidine 95, respectively (15,21).

Copper Binding Properties of Human
Chymotrypsin digestion was used to confirm the location of the four copper binding sites detected by MALDI-MS.  Histidine residues are shown in bold type. In this panel a sequence alignment of tryptic peptide T5 from MoPrP23-231 (see "Results") is presented below the sequence of PrP23-98. Dots indicate sequence identity, dashes represent deleted residues, and non-conserved residues are represented by the single letter amino acid code. Note that tryptic peptide T5 extends beyond the C-terminal residue of PrP23-98 to position 105. C, amino acid sequence and tryptic and chymotryptic map of MoPrP23-231. A disulfide bond is indicated. In B and C residues added for expression in Escherichia coli are indicated in italics (13,25).
A distinct mass spectrum was obtained for carbethoxylated HuPrP23-98 formed subsequent to pre-incubation with Cu(II).
Here the major species were determined as the starting peptide HuPrP23-98, 1 Cu(II) (7876. 5 3C). Since the maximum number of mono-carbethoxylations observed for HuPrP23-98 is nine (Fig. 3B) while the maximum number of mono-carbethoxylation of pre-formed copper-HuPrP23-98 complexes is four (Fig. 3C), these data imply that five histidine residues within HuPrP23-98 are typically protected by Cu(II) from DEPC modification. Of interest, the largest peak heights in Cu(II)-treated samples corresponded to unmodified peptide or metal ion-peptide complexes with 1-3 copper ions. These results suggest that the 3-4 histidine residues that bind to copper with the greatest avidity are the residues most susceptible to DEPC modification in the naked peptide. Presumably a fraction of the metal-peptide complexes formed during the preincubation with Cu(II) are stripped of the metal ions during the MALDI-MS procedure, yielding a peak of free peptide. An alternative (but perhaps less likely) explanation is that PrP23-98-copper complexes with multiple metal ions undergo a conformational change such that all DEPC-reactive residues are shielded from or are less reactive with the chemical modification reagent. Again, dissociation of some metal-peptide complexes during the MALDI-MS procedure could yield the observed peak of free PrP23-98 peptide.
To determine whether any other cations are involved in  Fig. 3B), except that maximum peak heights were observed at 3-and 4-mono-carboxylations, rather than at 4-and 5-mono-carboxylations. Thus, these three metals provide no significant protection of the HuPrP23-98 peptide. For Zn(II) and Ni(II) the largest peak amplitudes corresponded to yet lesser degrees of mono-carboxylation: 2-and 3-mono with Zn (II), or 1-and 2-mono with Ni(II) (Fig. 3, G and C). Signals deriving from intact peptide (7811.9 Da in Fig. 3G, 7812.8 in Fig. 3H) can also be observed in these analyses. For HuPrP28 -98 incubated with Ni(II) or Zn(II), the maximum number of mono-carbethoxylations was seven or eight, respectively. The results suggest that Zn(II) or Ni(II) might partially protect one or two histidine residues from DEPC modification, with a fraction of the metal-peptide complexes dissociating during the course of the MALDI-MS analysis to yield free peptide. While attempts were also made to extend these studies to Fe(II) ions, a surfeit of background noise in the corresponding MS spectra precluded meaningful analysis (not presented; also see Fig. 2H versus Fig. 2 (Fig. 4A). Since attempts to detect copper adducts by direct MALDI-MS analysis of MoPrP23-231 were not successful, we were prompted to explore the DEPC footprinting technique. In addition to the terminal NH 2 group, there are 9 histidine, 13 tyrosine, 12 serine, and 11 arginine residues in this protein. After incubation with DEPC the mass of carbethoxylated MoPrP23-231 increased, and the peak of carbethoxylated MoPrP23-231 became wider. The largest peak amplitude was observed m/z of 24,321.6 Da (Fig. 4B) indicating that the most abundant form of the modified MoPrP23-231 had ϳ17 mono-carbethoxylations (calc. 24,330.4 Da). The greatest peak amplitude of the carbethoxylated copper-MoPrP23-231 complex corresponded to m/z 23,972.9 Da (Fig. 4C) indicating that the most abundant form of the modified copper-MoPrP23-231 complex had ϳ12 mono-carbethoxylations (calc. 23,972.1 Da). Based on the difference between DEPC modification in the presence and absence of copper we can infer that ϳ5 histidine residues (i.e. 17 minus 12) are protected by copper coordination. This conclusion was supported by analyses from two additional experiments. The averaged mass difference of 344.9 Ϯ 7.7 Da derived from all three experiments yielded a figure of 4.79 Ϯ 0.11 sites protected by copper per MoPrP23-231 molecule. This stoichiometry is close to that obtained using the N-terminal PrP23-88 fragment.
Five Cu(II)-protected Residues Are Located within an N-terminal Tryptic Fragment of Mouse PrP23-231-To extend the above studies, enzymatic digestion of MoPrP23-231 was used to produce peptide fragments and peptide fragment adducts amenable to more accurate sizing. Accordingly, MoPrP23-231 (90 M) in NEMO-KCl buffer, pH 7.4, was digested with trypsin. In some cases MoPrP23-231 at the aforementioned concentration was pre-incubated with or without 10-fold molar excess CuCl 2 then reacted with 5-fold excess DEPC prior to tryptic digestion (a tryptic peptide map of MoPrP23-231 is presented in Fig. 1C). In the resulting analyses, with the exception of tryptic peptide T5, no copper-dependent changes were apparent in mass spectra of the tryptic peptides. By way of example, a peak at m/z 4987. Whereas peaks indicative of intact metal-peptide complexes (and free peptide) were noted in the copper-protection experiment involving PrP23-98 (Fig. 3C), peaks of unmodified metalpeptide complexes were not detected in the analysis of the octarepeat containing peptide T5 derived from copper-protected MoPrP23-231 (Fig. 5C). From this observation we infer that additional sequences present in PrP23-98 versus peptide T5 (i.e. residues 23-48), or minor changes within the homologous regions of peptide T5 and PrP23-98 (see Fig. 1, B and C) stabilize copper interactions, such that metal-peptide complexes remain intact for MALDI-MS analyses of HuPrP23-98. In this regard a tendency for copper-binding constants to increase in affinity in parallel with increasing PrP fragment length has been noted previously (21). Alternatively, extra residues present at the C terminus of peptide T5 versus PrP23-98 (i.e. WNKPPSKPK; Fig. 1B, lower line) may destabilize copper-peptide interactions or desorption such that intact metal-peptide complexes are not detected in the MALDI-MS analysis of peptide T5.
Subsequent to a Cu(II) pre-incubation, a peak at m/z 795.40 Da was observed and indicated that His-60 and His-84 in chymotryptic peptides C3 and C6, respectively, coordinated to Cu(II) and thereby protected against DEPC modification (Fig.  6C). Similar results were observed for peptides C4 and C5, where a peak at m/z 825.36 Da indicated that copper protection of His-68 and His-76 (Fig. 6C). In comparisons with Fig. 6B, a peak at m/z 1113.35 Da (1-mono-carbethoxylated C7) disappeared, whereas the peak at m/z 1041.50 Da (unmodified C7) exhibited an amplitude similar to that of the starting material in Fig. 6A. These data indicate that His-95 in full-length PrP is capable of coordinating to Cu(II). In contrast, little change could be observed in the peaks of C1 and carbethoxylated C1 (Fig. 6, B and C), indicating that Cu(II) did not coordinate to the C1 peptide. It is also of note that mass spectra presented in Fig. 6 include two peaks corresponding to partial digest products: C18ϩC19ϩC20 (YRENMY, calc.875.35 Da) and C21ϩC22 (RYPNQVYY, calc. 1102.53 Da). These peptides do not contain histidine residues, and their corresponding mass signatures remained invariant throughout the manipulations presented in Fig. 6, B and C. It is likely that the tyrosines located within these peptides do not react with DEPC as in all but one case these aromatic residues are disposed toward the hydrophobic interior of the native PrP molecule (2).
Fine Mapping DEPC Modification Sites by Post-source Decay Analysis-Further analysis of the chymotryptic peptides C3/ C6, C4/C5, and C7 was attempted with the PSD technique of tandem analysis to verify the inference that the copper-sensitive DEPC modifications resided on histidine residues. In postsource decay experiments, fragmentation of the peptides orig-inates from collisional activation within the field-free region, with the collision-induced peptide fragment ions mostly derived from bond breakage along the peptide backbone (28,29) and often with transfer of one or two hydrogens to create stable ion structures. In the nomenclature system, the a-, b-, and c-ions all contain the N terminus of the peptide, while the x-, y-, and z-ions all contain the C terminus. The major N terminuscontaining ion series is the b-ion series, and the major C terminus-containing ion series is the y-ion series (30). Fig. 7A shows the scheme of nomenclature for peptide PSD fragment ions of C3 or C6 chymotryptic fragments of intact MoPrP23-231.
Post-source decay spectra of peptides of C3 or C6 of intact MoPrP23-231 ([MϩH] ϩ 795.32 Da) and their mono-carbethoxylated counterparts ([MϩH] ϩ 867.35 Da) are presented in Fig. 7, B and C, respectively. A complete listing of PSD fragment ions is presented in Table I. The largest peak amplitude was obtained for the y 6 ion PHGGGW in the unmodified sample, whereas this peak was considerably reduced in the DEPCmodified sample, to be replaced by a species of m/z 682.22 corresponding to y 6 plus 72 mass units. Of the other characteristic fragment ions, b 3 (sequence GQP) was detected in both intact and carbethoxylated C3 or C6, b 4 (sequence GQPH, and thus encompassing a histidine) was seen only in intact C3 or C6, while b 4 plus a 72.06-Da adduct was only observed in FIG. 7. Fine mapping of DEPC modification sites by post-source decay analysis. A, scheme of nomenclature for post-source decay peptide fragment ions. N-terminal fragment ions (those containing the N terminus of the peptide) are labeled with the letters of a, b, whereas C-terminal fragment ions (which contain the C terminus of the peptide) are labeled with the letter y. The internal fragment ions are those ions containing neither the N terminus nor the C terminus of the peptide, and their mass usually corresponds to a structure with a y-type cleavage at one end and a-or b-type at the other. They are labeled as y i b j or y i a j , where i or j represents the number of peptide backbone. B, tandem post-source decay mass spectra of chymotryptic peptide C3 or C6 of MoPrP23-231. C, tandem post-source decay mass spectra of carbethoxylated C3 or C6 of MoPrP23-231. m ϭ mono-carbethoxylation. See also Table I. carbethoxylated C3 or C6. These data indicate that histidine residues at positions 60 and 84 were modified by DEPC. The 1 mono-carbethoxylated forms found in all the fragment ions containing histidine residues, but absent from the fragment ions without histidine residues underscore a similar behavior regarding the other variety of octapeptide repeat-derived chymotryptic peptides, namely C4 and C5 (Table I, Fig. 1C). Here the unmodified peptide has a mass of (825.32 Da, Table I Fig. 6B), while the signal of b 7 (GQGGGTH) deriving from the C7 peptide was replaced by a signal of b 7 ϩ 72 Da (Table I). In sum, these PSD data confirm DEPC modification on His-60, His-68, His-76, His-84, and His-95 of the fulllength MoPrP23-231.

Metal-Protein Binding and Neurodegenerative Disease-
Many proteins associated with neurodegenerative diseases have metal binding properties and/or metal-responsive expression (31). Besides PrP, the ectodomain of the Alzheimer precursor protein APP (32), A␤ peptide (33), and SOD-1 (34) exist in copper-metallated forms, while relationships to Fe metabolism are suggested for frataxin, APP, and huntingtin (35)(36)(37). In the documented examples metal binding relates to pathogenesis via an impact on aggregation or production of oxidative damage. Thus defining binding sites and the molecular details of complex formation may provide important and practical insights into pathogenic processes and neuronal biology.
Some metal binding sites are defined by well understood structural criteria and hence amenable to bioinformatic identification. These motifs would include "zinc fingers" (38), and calcium "EF hand" binding sites (39). However proteins with other binding modalities will elude identification by this bioinformatic approach, with PrP being a case in point (6,24). In this paper we have used a mass spectroscopic-based footprinting technique to position His-dependent metal coordination sites in PrP. Our approach employs DEPC, a widely used protein modification reagent capable of reacting with histidine residues to produce an N-carbethoxy-histidyl derivative (mono-N-carbethoxyhistidine; Fig. 1A). DEPC can also react with other nucleophilic residues including sulfhydryl, arginyl, and tyrosyl residues, as well as with ␣and ⑀-amino groups (40). However, in contrast to arginine, tyrosine, and amino acids with sulfhydryl side chains, the imidazole ring of histidine is known as a copper binding site, and many studies have demonstrated that metal coordination to histidine residues of peptides or proteins protects these amino acids from DEPC modification (41)(42)(43)(44). Since DEPC-modified histidines increase in mass of 72.06 Da for each mono-carbethoxylation of the pyrrole type of nitrogen in the imidazole ring in exposed residues (di-substituted adducts are only formed at high DEPC concentrations; not presented), resulting adducts can be detected and enumerated by MALDI-MS analysis (45)(46)(47). If metal ions such as Cu(II), Zn(II), Ni(II), Mg(II), Ca(II), or Mn(II) are coordinated to histidine residues prior to DEPC exposure, the pyrrole nitrogens in these particular residues are protected from acquiring carboxyethyl adducts. Thus the total number of histidines involved in coordination with copper or other metal ions can be determined by measuring mass differences between apo-and metal-coordinated proteins or peptides. If necessary, copper binding histidine residues can then be further delimited by endoproteolytic digestion and post-source decay analysis of the resulting peptides (48). Several points emerging from our analysis of PrP bear emphasis. First, there are parallels between metal selectivity observed in the footprinting studies and selectivity observed by classical means of biochemical analysis (strong interactions with copper, weaker or undetectable interactions for Zn(II) and Ni(II), interactions for other metals undetectable (17,19,20). Second, there is close concordance between the number of histidines protected from DEPC modification by copper and the stoichiometry of copper binding sites per polypeptide determined by classical analytical techniques. Data drawn from seven other studies are compiled in Table II. With one excep-tion, a report describing a binding stoichiometry of one copper per polypeptide containing four octapeptide repeat units where the authors suggest that all four octarepeat histidines within the PrP52-98 peptide converge upon a single copper ion (20), the data indicate the same stoichiometry. Third, the locations of the histidine residues protected from DEPC modification confirm and extend assignments made by other methodologies. Our data define the four octarepeat histidines (residues 60, 68, 76, and 84 of MoPrP), compatible with other studies (13,15,20,21), and also a second type of site involving His-95 (Table II), perhaps in conjunction with His-110 (equivalent to His-111 in human PrP) (20,49). These sites, as present in PrP23-98, PrP90 -231, or PrP23-231, have binding affinities estimated in a range from 2.2 ϫ 10 Ϫ6 to 10 Ϫ14 M for the octarepeats and ϳ5 ϫ 10 Ϫ6 to 4 ϫ 10 Ϫ14 M for the site involving His-95 (Table  II). In addition to His-110, surface-exposed histidines exist at positions 139, 176, and 186 in the NMR structure of the Cterminal region of PrP (not shown), yet interactions with copper were not detected. There was no copper-dependent alteration for tryptic peptides T10ϩT14 containing His-176 (T11 and the hydrophobic peptide T7 were not detectable in direct analyses of proteolytic digests), nor was there any increase in overall copper binding stoichiometry when comparing PrP23-98 with MoPrP23-231 (Table II). The failure to detect additional copper sites when proceeding from analyses of PrP23-98 to full-length MoPrP23-231 has been noted previously (21) and argues against the presence of strong copper binding sites in the ␣-helical C-terminal domain of PrP. These data again speak to the issue of selectivity, making it unlikely that low-affinity sites are detected in the DEPC protection studies. The only apparently discordant determination concerns experiments using EPR analysis. Although precise stoichiometries were not emphasized, spectra for two coordination geometries not seen for N-terminal fragments were indicative of two varieties of Cu(II) binding sites specific within PrP121-231 (50). One type of site was compatible with an oxygen-dependent ligation, perhaps via aspartic or glutamic acid residues and was seen at pH values Ͻ7.0 (and therefore potentially compatible with our failure to detect His-dependent coppercoordination in the C-terminal domain of PrP when analyzed at pH 7.4). However, the other type of site was speculated to involve a nitrogen ligand. In practice, these are restricted to the aforementioned spatially dispersed histidines 139, 176, and 186. None of these residues are perfectly conserved in vertebrate PrP sequences (51), and our studies failed to detect involvement of His-176. We conclude that further studies will be required to define Cu(II) binding constants of sites putatively involving His-139 and His-186 and address this seeming discrepancy.
Since MALDI-MS and ESI-MS have been used previously to analyze interactions between metal ions and proteins (15,24,52) questions arise as to the attributes and/or advantages of the DEPC "footprinting" methodology. For conventional MALDI-MS (53), direct analysis of native metal-peptide complexes is complicated by the required excess of a matrix that is typically strongly acidic and thus favors histidine protonation and protein denaturation. Consequently, MALDI can have a reduced ability to detect low affinity sites when compared with other techniques, although novel sample preparation matrices have now been developed to offset the effect (52). Another approach is to use large excesses of metal ions, although this has a potential to reveal low affinity sites, leading to the problem of discerning of nonspecific interaction metal/peptide interactions. Electrospray MS has different attributes (reviewed by Loo in Ref. 54)). These include the sensitive detection of metal-peptide complexes at low metal concentrations, as samples are not desorbed from a matrix but introduced into the instrument while still in an aqueous buffer (although this too is often used at an acidic pH). In the case of the DEPC technique, copper binding sites are revealed indirectly by protection from carboxyethlation in a chemical reaction performed prior to introduction into the instrument; hence metalprotein complexes need not remain intact for MS analysis, avoiding some technical confounds listed above. These general considerations can be illustrated for the specific case of copper binding to PrP. MALDI-MS analyses failed to reveal binding to full-length protein (not shown) and failed to reveal binding to a "fifth" copper site either in intact PrP23-98 (Fig. 2B) or chymotryptic digests of MoPrP23-231 (not shown). In contrast, this site was detected by DEPC protection (Figs. 3 and 5), by ESI-MS (15,21), and by other techniques (Table II).
Chemical Modification and Prion Biology-In pioneering experiments to demonstrate the proteinaceous nature of prions (55), DEPC was found to reduce the infectious titer of prion preparation by two to three log units (55). Since these experiments used PrP Sc preparations enriched by proteolysis with proteinase K, they indicate that DEPC-reactive amino acid residues modulating prion infectivity must reside in the Cterminal protease-resistant "core" of PrP Sc (i.e. PrP27-30) (6). These results, taken together with the footprinting experiments described here, have implications for the nature of studies that might be possible in the future. The first is that DEPC modification need not be restricted to recombinant prion proteins. Instead, DEPC can be reacted with purified mammalian proteins (55) or perhaps even tissue samples. In the latter instance, the modification status of protein samples might be investigated by high-resolution electrophoresis and immunoblotting or by using affinity chromatography to purify samples for MS analysis. In short, because footprinting uniquely records an "imprint" of metal binding, rather than requiring the metal to remain intact during purification or denaturing procedures, future iterations of this technique might provide us with the chemical signatures necessary to identify metallo-or apo-forms of PrP C in tissue samples. A second implication is that mapping DEPC modification sites associated with the drop in infectious titer of PrP27-30 could be of great interest, irrespective of whether metal-binding is involved in modulating the infectious properties of prion proteins (histidines 60, 68, 76, and 84 are absent from PrP27-30 and His-95, putatively involved in Cu(II) binding with very high affinity (20,49), might be predicted to be coordinated in vivo and hence protected from modification). There is still a paucity of precise structural information about the conversion of PrP C to PrP Sc , in major part because PrP Sc has poor solubility and is difficult to study by high-resolution biophysical methods. Also, though spectroscopic comparisons between PrP C and PrP Sc indicate a profound change in structure to a more ␤-sheet-rich conformation (8,9), obtaining antibodies specific to PrP Sc neo-epitopes has proven difficult, albeit with one possible exception (56). Rather, in the case of PrP Sc , published single chain and monoclonal antibodies define an occlusion (rather than exposure) of epitopes associated with the conversion of PrP C to PrP Sc (57)(58)(59). We therefore suggest that, since PrP27-30 is clearly reactive with DEPC, comparative mapping of DEPC adducts in PrP C and PrP27-30 by site-specific endoproteolysis and MS may comprise a powerful strategy to define infectivity-associated surface determinants in PrP Sc .