Using Mass Spectrometry To Study Copper ... - ACS Publications

May 15, 2004 - Human β-2-microglobulin (β2m) is a 12-kDa subunit of the class I major histocompatability complex (MHC).1 Upon turnover of MHC, β2m ...
1 downloads 0 Views 118KB Size
Download PDF
Anal. Chem. 2004, 76, 3498-3504

Using Mass Spectrometry To Study Copper-Protein Binding under Native and Non-Native Conditions: β-2-Microglobulin Jihyeon Lim and Richard W. Vachet*

Department of Chemistry, University of MassachusettssAmherst, Amherst, Massachusetts 01003

A method based on metal-catalyzed oxidation (MCO) reactions and mass spectrometry (MS) has been used to determine the Cu(II) binding sites in both native and unfolded conformations of β-2-microglobulin (β2m). Recent studies have shown that β2m is destabilized and can form amyloid fibers in the presence of Cu(II). An increased affinity for Cu in unfolded states compared to that of the native state is suspected to facilitate overall protein destabilization. Cu-binding site information for native β2m is difficult to obtain using traditional techniques because of its propensity to form amyloid fibers at relatively high protein concentrations in the presence of Cu and because of the nonspecific paramagnetic peak broadening observed in NMR analyses. In addition, Cu-binding information of unfolded β2m is complicated by the high concentrations of denaturants (e.g., 8 M urea) needed to ensure protein unfolding. The MCO/MS approach has been successfully employed in this work to overcome these difficulties. The sensitivity of MS allowed the Cu-binding site of the native protein to be determined at the low concentrations of β2m necessary to avoid amyloid fiber formation. Results indicate that the N-terminus of the protein and His31 are responsible for Cu(II) coordination in the native state. The MCO/MS method was also successful at determining the Cu-binding site in the presence of 8 M urea with the N-terminus, His31, His51, and His81 found to be Cu-bound in the unfolded state. This result supports the existence of a well-defined but different coordination structure in the unfolded state, which leads to the greater affinity for Cu(II) observed in the unfolded state of the protein. In general, it appears that the MCO/MS method is capable of providing Cubinding site information for proteins that are difficult to study by traditional means. Human β-2-microglobulin (β2m) is a 12-kDa subunit of the class I major histocompatability complex (MHC).1 Upon turnover of MHC, β2m is usually transported to the kidneys and degraded, but in patients with renal failure, the concentration of β2m * Corresponding author. E-mail: [email protected]. Fax: 413-5454490. (1) Trinh, C. H.; Smith, D. P.; Kalverda, A. P.; Phillips, S. E. V.; Radford, S. E. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 9771-9776.

3498 Analytical Chemistry, Vol. 76, No. 13, July 1, 2004

circulating in serum can increase up to 60-fold from its normal level of ∼0.1 µM.2,3 Furthermore, in dialysis patients, β2m forms amyloid deposits in joints, and these deposits are presumed to be the main pathogenic process underlying dialysis-related amyloidosis.4,5 It remains uncertain how amyloid fibrils are formed from β2m, but raised concentrations alone are not sufficient for fibril formation. Recent studies by Miranker and co-workers have shown that Cu(II) binds specifically to β2m at concentrations of Cu(II) that are accessible during dialysis therapy.6,7 These studies also demonstrated that β2m binds Cu(II) more strongly in its unfolded states than its native state, and the conformational stability of the protein is lower in the presence of Cu(II) according to both urea titrations and thermal denaturation studies. Because other divalent metals such as Ca(II) and Ni(II) do not exert the same effect and protein destabilization precedes amyloid fiber formation, Cu(II) likely plays an important role in the amyloidosis of this protein in vivo. A clearer picture of the differences between the Cu(II)binding sites in native and non-native conformations of β2m may provide insight into the mechanism of Cu(II)-induced fiber formation by indicating local protein substructures that lead to amyloid assembly. The metal-binding sites of both native and non-native conformations of β2m, however, are difficult to definitively determine using typical methods such as NMR or X-ray crystallography. At the relatively high protein concentrations needed for these techniques, stoichiometric concentrations of Cu(II) can trigger fiber formation, making binding site analysis difficult. In addition, high concentrations of Cu(II) can lead to nonspecific paramagnetic peak broadening in NMR analyses. The solution conditions used in previous studies to ensure protein denaturation (i.e., ∼8 M urea) also further complicate metal-binding site analysis for the nonnative states of β2m. Given the difficulty with binding-site analysis by traditional techniques, we decided to apply a method based (2) Floege, J.; Ketteler, M. Kidney Int. 2001, 59, 164-171. (3) Floege, J.; Ehlerding, G. Nephron 1996, 72, 9-26. (4) Gejyo F.; Yamada, T.; Odani, S.; Nakagawa, Y.; Arakawa, M.; Kunitomo, T.; Kataoka, H.; Suzuki, M.; Hirasawa, Y.; Shirahama, T.; Cohen, A. S.; Schmid, K. Biochem. Biophys. Res. Commun. 1985, 129, 701-706. (5) Chiti, F.; De Lorenzi, E.; Grossi, S.; Mangione, P.; Giorgetti, S.; Caccialanza, G.; Dobson, C. M.; Merlini, G.; Ramponi, G.; Bellotti, V. J. Biol. Chem. 2001, 276, 46714-46721. (6) Morgan, C. J.; Gelfand, M.; Atreya, C.; Miranker, A. D. J. Mol. Biol. 2001, 309, 339-345. (7) Eakin, C. M.; Knight, J. D.; Morgan, C. J.; Gelfand, M. A.; Miranker, A. D. Biochemistry 2002, 41, 10646-10656. 10.1021/ac049716t CCC: $27.50

© 2004 American Chemical Society Published on Web 05/15/2004

on metal-catalyzed oxidation (MCO) reactions and mass spectrometry (MS) to determine the Cu(II) binding sites in both native and unfolded conformations of β2m. The general approach of combining MCO reactions and MS involves first selectively oxidizing the protein under the appropriate reaction conditions. It has been well documented that biomolecules are susceptible to oxidative modification by reactive oxygen species (ROS) such as O2•- or OH• that are generated by metal reduction and subsequent reoxidation in the presence of O2, H2O2, or both.8-13 These ROS are generated at a redox-active metal center and can quickly react with the amino acid residues that are part of the metal-binding site. The oxidatively modified amino acids then can be identified using MS. With the appropriate MCO reaction conditions chosen, our group and others have shown that this approach is useful for selectively identifying the metal-binding sites of several Cu metalloproteins.14-18 Important to the selectivity of this approach appears to be the presence of relatively high concentrations of ascorbate or other radical scavengers,18,19 which are probably successful at both preventing diffusion of ROS far from the metal and preventing chain oxidation reactions. An ascorbate concentration that exceeds the protein concentration by at least 50-fold is usually sufficient to avoid the oxidation of nearby amino acids that are not part of the binding site. When several nonbinding Met residues are present in the protein, though, ascorbate/protein concentration ratios that exceed 1000 appear to be necessary to avoid the oxidation of this readily oxidizable amino acid.18,19 In this work, we demonstrate that the MCO/MS approach is capable of elucidating the Cu-binding sites of β2m under both native and non-native conditions. The sensitivity of MS allows low concentrations of β2m to be studied in the presence of Cu, thereby avoiding amyloid fiber formation. In addition, the MCO reaction chemistry seems to be unaffected by high concentrations of denaturants such as urea, so that the Cu-binding site of the unfolded protein can be determined. EXPERIMENTAL SECTION Materials. Human β-2-microglobulin (β2m) was obtained from Research Diagnostics, Inc. (Flanders, NJ). [Gln11]-amyloid-βprotein fragment 1-16 (Aβ1-16), dithiothreitol (DTT), L-ascorbate, Cu(CH3COO)2‚H2O, CuSO4, 3-morpholinopropanesulfonic acid (MOPS), and potassium acetate were purchased from Sigma (St. Louis, MO). Ammonium acetate was obtained from Fischer Scientific (Fair Lawn, NJ), and urea was purchased from Mallinck(8) Stadtman, E. R. Annu. Rev. Biochem. 1993, 62, 797-821. (9) Luo, S.; Ishida, H.; Makino, A.; Mae, T. J. Biol. Chem. 2002, 277, 1238212387. (10) Zhang, Z.; Barlow, J. N.; Baldwin, J. E.; Schofield, C. J. Biochemistry 1997, 36, 15999-16007. (11) Cao, W.; Barany F. J. Biol. Chem. 1998, 273, 33002-33010. (12) Hlavaty, J. J.; Benner, J. S.; Hornstra, L. J.; Schildkraut, I. Biochemistry 2000, 39, 3097-3105. (13) Platis, I. E.; Ermacora, M. R.; Fox, R. O. Biochemistry 1993, 32, 1276112767. (14) Kurahashi, T.; Miyazaki, A.; Suwan, S.; Isobe, M. J. Am. Chem. Soc. 2001, 123, 9268-9278. (15) Hovorka, S. W.; Williams, T. D.; Scho ¨neich, C. Anal. Biochem. 2002, 300, 206-211. (16) Scho ¨neich, C.; Williams, T. D. Chem. Res. Toxicol. 2002, 15, 717-722. (17) Scho ¨neich, C.; Williams, T. D. Cell. Mol. Biol. 2003, 49, 753-761. (18) Lim, J.; Vachet, R. W. Anal. Chem. 2003, 75, 1164-1172. (19) Bridgewater, J.; Vachet, R. W. In preparation.

rodt Chemicals (Phillipsburg, NJ). Formic acid, HPLC-grade acetonitrile, tris(hydroxymethyl)aminomethane (Tris), and Tris hydrochloride (Tris-HCl) were obtained from EM Science (Gibbstone, NJ). Trypsin was from Roche Diagnostics (Indianapolis, IN). Chelex 100 resin (200-400 mesh, sodium form) was purchased from Bio-Rad Laboratories (Hercules, CA). Distilled, deionized water was prepared from a Millipore (Burlington, MA) Simplicity 185 water purification system. Metal-Catalyzed Oxidation Reactions. The MCO reactions of Aβ1-16 were performed at room temperature with 100 µM peptide, 100 µM CuSO4, 1 mM ascorbate, 1 mM H2O2, 6 M urea, and 50 mM Tris-HCl/Tris, buffered to a pH of 7.4. The MCO reactions of native β2m were performed at room temperature in aqueous solutions containing 2.5 µM β2m, 0-2.5 µM Cu(CH3COO)2‚H2O, 0-100 µM ascorbate, 50 mM MOPS, and 150 mM potassium acetate, buffered to a pH of 7.4. The MCO reactions of denatured β2m were performed under the same conditions but with 8 M urea added to ensure the unfolding of the protein in the solution. In all cases, the protein solutions were added to buffers containing the copper salts and then diluted to the stated concentrations with buffers containing ascorbate to initiate the MCO reactions. Reaction times ranged from 15 min to 1 h. Ureacontaining buffers were treated prior to the experiment with 1% (w/v) Chelex 100 to remove excess trace metals. Proteolytic Digestions. Before digestion, urea and buffer salts were removed from the β2m solution using a Hi-Trap desalting column (Amersham Pharmacia Biotech, Piscataway, NJ). The protein was then lyophilized, dissolved in 200 µL of a 10 mM ammonium acetate buffer (pH 7.4), and digested overnight with 5 µg of trypsin and 100 µM DTT at 36.5 °C. The trypsin was inactivated by adjusting the pH to 2, and the samples were immediately frozen until ready for further analysis. Instrumentation. The MCO reactions were monitored by removing an aliquot of the reaction mixture at selected time intervals and analyzing the products by HPLC-ESI-MS. An HP1100 (Agilent, Wilmington, DE) HPLC system with a C18 column (4.6 mm × 50 mm, Metachem, Lake Forest, CA) was used. Proteins and peptides were eluted using a linear gradient of acetonitrile with 0.1% formic acid over 10 min (10%/min) at a flow rate of 0.5 mL/min. The LC effluent was split at a ratio of 1:4 using a T-splitter (Upchurch Scientific, Oak Harbor, WA), and the smaller fraction of split flow was fed into the ESI source of a Bruker Esquire-LC quadrupole ion trap mass spectrometer. The ESI source was operated at a spray voltage of 3 kV, and the capillary temperature was set at 250 °C. Typically, 30-40 V was applied to skimmer 1, and the capillary exit offset voltage was set between 20 and 60 V. Tryptic fragments were analyzed by direct injection using similar source parameters. RESULTS AND DISCUSSION MCO Reaction of Aβ1-16 in the Presence of Urea. Aβ1-16 was first examined to test the viability of the MCO reactions in the presence of urea. In previous work, we demonstrated that MCO reactions and MS are able to determine the Cu-binding site of Aβ1-16 as His6, His13, and His14.18 As evidenced by Figure 1, Cu-bound Aβ1-16 is readily oxidized after an MCO reaction in the presence of 6 M urea. The main reaction product is the addition of one oxygen atom (see inset in Figure 1), but a small peak corresponding a second oxygen addition can be seen too. The Analytical Chemistry, Vol. 76, No. 13, July 1, 2004

3499

Table 1. Control Conditions Used in This Work

β2m copper acetate ascorbate MOPS potassium acetate reaction time

control 1

control 2a

control 3a

2.5 µM

2.5 µM 2.5 µM

2.5 µM

25 mM 150 mM 30 min

25 mM 150 mM 30 min

100 µM 25 mM 150 mM 30 min

a Control 2 and control 3 were also performed in the presence of 8 M urea.

Figure 1. Mass spectrum of unfolded Aβ1-16 after a 1-h MCO reaction, and an expanded view around the triply charged species (inset).

specificity of the oxidation can be determined by tandem MS, and the results indicate that collections of isomeric peptides exist in which His6, His13, or His14 is modified. The product ion spectrum of [Aβ1-16 +2H + O]2+ (data not shown) contains an almost complete series of b-type product ions from b5 to b15 and y-type product ions (y3-y5, y8-y15), which facilitates identification of the oxidized residues. While the nature of Cu binding in Aβ1-16 under non-native conditions is not known, it probably does not differ significantly from metal binding under native conditions. Structural studies show that the first 9 residues of the 40-residue amyloid-β-protein, of which Aβ1-16 is a part, have a disordered structure in water.20 Also, molecular dynamics simulations suggest significant flexibility in the first 15 residues,21 so very likely, Aβ1-16 has little higher order structure in water, and this structure likely remains disordered in the presence of urea. Furthermore, because this peptide binds Cu strongly, metal binding probably enforces any peptide structure under both native conditions in water and nonnative conditions in 6 M urea. Thus, these experimental results suggest that the MCO reactions are still specific in the presence of high concentrations of urea. Consequently, we felt the MCO reactions could be effectively applied to the Cu binding of β2m in the presence of urea. β2m Control Experiments. Before the MCO reactions of β2m were performed, three different control experiments were done under native (0 M urea) conditions and two were done under denatured (8 M urea) conditions. The control experiments are important because they allow us to confirm that oxidative modifications occur only as a result of the MCO reactions. The three control experiments done under native conditions were (1) no reaction of β2m, (2) a 30-min reaction of Cu-bound β2m in the absence of ascorbate, and (3) a 30-min reaction of β2m with ascorbate only (see Table 1). Control experiments 2 and 3 were (20) Zhang, S.; Iwata, K.; Lachenman, M. J.; Peng, J. W.; Li, S.; Stimson, E. R.; Lu, Y. A.; Felix, A. M.; Maggio, J. E.; Lee, J. P. J. Struct. Biol. 2000, 130, 130-141. (21) Straub, J. E.; Guevara, J.; Huo, S.; Lee, J. P. Acc. Chem. Res. 2002, 35, 473-481.

3500 Analytical Chemistry, Vol. 76, No. 13, July 1, 2004

Figure 2. Mass spectrum of β2m after control experiment 3 without urea. The deconvoluted spectrum is shown in the inset.

also separately performed in the presence of 8 M urea. After each control experiment, the intact protein was analyzed by MS to note any oxidation. Then, the protein was digested, and any oxidized residues were identified by tandem MS (MS/MS). Figure 2 shows a representative mass spectrum after control experiment 3 without urea. In each control experiment, a significant degree of oxidation (∼20%) was observed. Proteolytic digestion of β2m after each control experiment and examination of the resulting tryptic fragments indicated that the oxidative modification in each case occurred solely on Met99. Furthermore, the digestion process added an additional degree of oxidation (∼10%), so that 30 ( 6% of the protein was oxidized at Met99. Analysis of the protein directly from the manufacturer revealed that ∼20% of β2m was oxidized, indicating that no additional oxidation occurs during the control experiments. In total, these results show that in subsequent experiments oxidations occurring on residues other than Met99 will be a result of a specific oxidation, and thus, these amino acids will be identified as Cu-binding residues. Cu-Binding Site of Native β2m. The MCO reactions of native β2m were performed as described in the Experimental Section. The reactions were initiated by adding ascorbate and quenched by separating ascorbate from the solution via a desalting column. Figure 3 shows a mass spectrum obtained after a 30-min MCO reaction. At least two pieces of information are evident after deconvoluting this mass spectrum (inset, Figure 3). First, β2m has been modified by the addition of more than one oxygen atom, and second, peaks corresponding to mass losses are evident. Careful examination of the mass spectrum before deconvolution leads to the identification of only one definitive mass loss, which

Figure 3. Mass spectrum of native β2m after a 30-min MCO reaction with 2.5 µM β2m, 2.5 µM Cu(CH3COO)2‚H2O, 100 µM ascorbate, 50 mM MOPS, and 150 mM potassium acetate, buffered to a pH of 7.4. The inset shows the deconvoluted spectrum. Table 2. Observed m/z Ratios for the Proteolytic Fragments of Oxidized Native β2m m/zobs [1-3-1]+

[1-3]+ [1-3]2+ [4-6]+ [7-12]+ [7-12]2+ [13-19]+ [13-19]2+ [20-41]2+ [20-41]3+ [20-41+16]2+ [20-41+16]3+ [42-45]+ [42-45]2+ [46-48]+ [46-48]2+ [46-58]+ [46-58]2+ [46-58]3+ [49-58]+

415.3 416.3 208.6 345.3 765.7 383.3 752.4 376.8 1249.1 833.3 1257.1 838.4 475.3 238.1 389.3 195.1 1518.9 760.0 507.0 1148.6

m/zobs [49-58]2+ [59-81]2+ [59-81]3+ [82-91]+ [82-91]2+ [82-91]3+ [92-94]+ [92-94]2+ [92-97]+ [92-97]2+ [92-97]3+ [92-99]+ [92-99]2+ [92-99+16]+ [92-99+16]2+ [95-99]+ [95-99]2+ [95-99+16]+ [95-99+16]2+ [98-99]+

574.8 1471.0 981.2 1122.7 561.9 375.0 359.3 180.1 816.7 408.8 272.8 1062.7 531.8 1078.6 539.8 722.6 361.7 738.5 369.7 265.1

corresponds to ∼87 Da. The relatively low mass accuracy of the ion trap makes a confident assignment of the actual mass loss from these multiply charged ions difficult. Nonetheless, such cleavages after MCO reactions are not unprecedented, and they usually occur at sites adjacent to where the metal is bound.8,10-13 This mass decrease possibly corresponds to the loss of H2NCH2C(CH3)CH2CH3, which could arise from Ile on the Nterminus. This result would suggest that the N-terminus is bound to Cu. Because this mass loss assignment is tentative, though, the conclusion that the N-terminus is bound to Cu is also tentative. Fortunately, other data also implicate the N-terminus as part of the Cu-binding site (vide infra). To pinpoint the oxidation sites, β2m was subjected to proteolysis, and the oxidized peptide fragments were sequenced by MS/ MS. The peptide fragments observed after digestion of oxidized β2m are listed in Table 2. Shifts were observed in the m/z ratios of four different proteolytic fragments: Ile1-Arg3, Ser20-Lys41, Ile92-Met99, and Trp95-Met99. All of these fragments had 16-

Figure 4. Product ion spectrum of (Ile1-Arg3 + H - 1)+. The inset shows peak assignments for the product ions of (Ile1-Arg3 + H - 1)+.

Da increases except for Ile1-Arg3, which had a 1-Da decrease. Unfortunately, we had difficulty finding a proteolytic fragment that contained the 87-Da loss noted above. The modification to Ile1Arg3, however, provides more concrete evidence that Cu binds to the N-terminus of the protein. Figure 4 shows the product ion spectrum of (Ile1-Arg3 - 1 + H)+. Unmodified y1 and (y2 2NH3)+ product ions and a b2 - 1 product ion suggest that Ile1 has been modified in the Ile1-Arg3 proteolytic fragment. The 1-Da loss is perplexing, but it might result from a reaction involving the loss of NH2 and H from Ile followed by oxygen addition, a process that is likely initiated by •OH-mediated hydrogen abstraction. Experiments are currently underway to determine the exact nature of the modification. Oxidation of Ser20-Lys41 allows another binding site of β2m to be identified. Figure 5 shows the product ion spectrum of (Ser20-Lys41 + 16 + 3H)3+. This spectrum shows a series of b ions (b3-b6, b8, b10, b11) that have m/z ratios identical to the same product ions from the unoxidized peptide (data not shown) and a b12 product ion with a m/z ratio 16 Da higher. Also, the y4, y6, and y10 product ions appear with no m/z shifts, while a y-series of product ions from y11 through y20 appear 16 Da higher than in the product ion spectrum of the unoxidized peptide. As a whole, these data indicate that His31 is the only amino acid modified in this peptide fragment, which elucidates it as one of the Cu-binding sites. The product ion spectra of the other two oxidized fragments, Ile92-Met99 and Trp95-Met99, were also collected, and in both cases, Met99 was found to be oxidized (data not shown). Met99 was found to be oxidized in 30 ( 6% of the protein molecules during the control experiments, but after the MCO reactions, Met99 was found to be oxidized in about 40 ( 10% of the protein molecules. This value was determined by comparing the abundances of the unoxidized and oxidized ions of Ile92-Met99 and Trp95-Met99. Despite the slight increase in Met99 oxidation as compared to the control experiments, we conclude that Met99 is not part of the binding site for two reasons. First, the additional oxidation of Met99 after the MCO reactions is very small as compared to the 30 ( 10% oxidation of His31. Met is known to be Analytical Chemistry, Vol. 76, No. 13, July 1, 2004

3501

Figure 6. Mass spectrum of unfolded β2m after a 30-min MCO reaction with 2.5 µM β2m, 2.5 µM Cu(CH3COO)2‚H2O, 100 µM ascorbate, 8 M urea, 50 mM MOPS, and 150 mM potassium acetate, buffered to a pH of 7.4. The inset shows the deconvoluted spectrum. Table 3. Observed m/z Ratios for the Proteolytic Fragments of Oxidized Unfolded β2m Figure 5. (a) Product ion spectrum of (Ser20-Lys41 + 16 + 3H)3+. (b) Peak assignments for the product ions of (Ser20-Lys41 + 16 + 3H)3+. Product ions with 16-Da shifts relative to the unoxidized peptide are underlined.

much more susceptible to oxidation than His,22,23 and in our previous studies in which Met was Cu-bound, Met was oxidized to a much greater extent than all other Cu-bound amino acids.18 One would then expect Met99 to be oxidized to a much greater extent if bound to Cu. Second, circular dichroism (CD) data of native β2m indicates no change in the folded conformation of the protein upon addition of Cu.6 Because His31 is quite distant from Met99 in the apoprotein structure,24,25 simultaneous Cu binding by both residues would require a significant conformational change in the protein, but this does not occur. Met99 oxidation, then, is more likely to be explained by one of two possibilities. First, the small degree of oxidation of this residue may be due to a nonspecific oxidation reaction, which we and others have had difficulty completely avoiding during MCO reactions, especially for Met residues.14,18 A second possibility is that a second weaker Cu-binding site exists. Miranker and coworkers have previously shown that, in the presence of a significant excess of Cu, β2m can bind more than one Cu.7 Met99 might be part of this second binding site, especially given the presence of two adjacent Asp residues according to the crystal structure.24 Currently, acidic residues are transparent to our method. Our data suggest that Cu binding to the native structure of β2m involves His31 and the N-terminus. This conclusion is consistent with both CD and NMR data taken previously.6,7,25 The crystal structure of apo β2m indicates the proximity of the (22) Maleknia, S. D.; Ralston, C. Y.; Brenowitz, M. D.; Downard, K. M.; Chance, M. R. Anal. Biochem. 2001, 289, 103-115. (23) Edwards, A. M.; Ruiz, M.; Silva, E.; Lissi, E. Free Radical Res. 2002, 36, 277-284. (24) Collins E. J.; Garboczi, D. N.; Wiley, D. C. Nature 1994, 371, 626-629. (25) Verdone, G.; Corazza, A.; Viglino, P.; Pettirossi, F.; Giorgetti, S.; Mangione, P.; Andreola, A.; Stoppini, M.; Bellotti, V.; Esposito, G. Protein Sci. 2002, 11, 487-499.

3502 Analytical Chemistry, Vol. 76, No. 13, July 1, 2004

m/zobs [1-3-1]+ [1-3]+ [1-3+16]+ [1-3]2+ [4-6]+ [7-12]+ [7-12]2+ [13-19]+ [13-19]2+ [20-41]2+ [20-41]3+ [20-41+16]2+ [20-41+16]3+ [42-45]+ [46-48]+ [46-58]2+ [46-58]3+ [46-58+16]2+ [46-58+16]3+

415.3 416.3 432.2 208.6 345.3 765.4 383.3 752.5 376.7 1249.6 833.3 1257.4 838.4 475.2 389.3 760.0 506.9 768.0 512.3

m/zobs [49-58]+ [49-58]2+ [59-81]3+ [82-91]+ [82-91]2+ [82-91]3+ [82-91+16]+ [82-91+16]2+ [92-94]+ [92-97]2+ [92-99]2+ [92-99+16]2+ [95-97]+ [95-97]2+ [95-99]+ [95-99]2+ [95-99+16]+ [95-99+16]2+ [98-99]+

1148.4 574.8 981.2 1122.5 561.8 375.2 1138.5 569.8 359.3 408.7 531.8 539.8 476.2 238.0 722.3 361.7 738.4 369.3 265.0

N-terminus to His31, and the absence of a conformational change in β2m with Cu present, as determined by CD,6 make this the likely binding site. Furthermore, NMR data on β2m in the presence of substoichiometric concentrations of Cu suggested His31 as a likely binding site.7,25 MCO Reactions of Non-Native β2m. The Cu-binding site of non-native β2m was also determined by the MCO/MS method. The mass spectrum acquired after a 30-min MCO reaction of nonnative β2m is shown in Figure 6. The deconvoluted mass spectrum (Figure 6, inset) indicates that up to two and possibly three oxygen atoms are incorporated into β2m. Unlike the mass spectrum of the native protein after the MCO reaction, however, no cleavage is observed in this case. Upon digestion of β2m, six peptide fragments are found to have oxidative modifications, and the results are summarized in Table 3. Figures 7 and 8 show the product ion spectra of two of these fragments, Ser82-Lys91 and Ile46-Lys58, as examples of how the oxidized residues can be identified. In each spectrum, an almost complete series of b and y product ions are observed. In Figure 7, a series of oxidized b ions from b3 to b8, a series of unoxidized y ions from y2 to y7, an

Figure 7. Product ion spectrum of (Ser82-Lys91 + 16 + 2H)2+. The inset shows peak assignments for the product ions of (Ser82Lys91 + 16 + 2H)2+. Product ions with 16-Da shifts relative to the unoxidized peptide are underlined.

Figure 8. Product ion spectrum of (Ile46-Lys58 + 16 + 2H)2+. The inset shows peak assignments for the product ions of (Ile46Lys58 + 16 + 2H)2+. Product ions with 16-Da shifts relative to the unoxidized peptide are underlined.

unoxidized b2 ion, and oxidized y8 and y9 product ions indicate that His84 is oxidized in the Ser82-Lys91 proteolytic fragment. Similarly, in Figure 8, His51 is identified as the oxidized residue in the Ile46-Lys58 fragment by a series of oxidized b ions from b6 to b12, a series of unoxidized y ions from y2 to y7, a series of unoxidized b ions from b3 to b5, and a series of oxidized y ions from y8 to y11. When a similar analysis of the product ion spectra is performed for the oxidized ions of Ile1-Arg3, Ser20-Lys41, Ile92-Met99, and Trp95-Met99, the residues Ile1, His31, and Met99, respectively, are found to be oxidized. Met99 is found to be oxidized in 40 ( 8% of the protein molecules. Because this percentage is only slightly greater than in the control experiments, its identity as a Cu-binding residue is again ambiguous. As in the case of native β2m, the low degree of additional oxidation on Met99 allows it to be ruled out as part of the primary binding site. When the data from the MCO reaction of non-native β2m are considered in total, the Cu-binding site appears to include the N-terminus, His31, His51, and His84.

Cu binding by these residues is somewhat consistent with data from previous work by Miranker and co-workers.7 Fluorescence measurements indicated that the non-native state of β2m has a greater affinity for Cu than its native state. Certainly, Cu binding to three His residues in the non-native state as opposed to one in the native state would explain the greater affinity in the former case. In the same work by Miranker and co-workers,7 His13, His51, and His84 were suggested as the Cu-binding site in nonnative β2m. These suggestions were based upon site-directed mutagenesis studies and the importance of these residues in the Cu-mediated destabilization of β2m. The high degree of cooperativity usually involved in the protein-metal binding, however, makes site-directed mutagenesis studies inherently unreliable for providing binding site information. Furthermore, the binding of His13, His51, and His84 was inferred only because their removal reduced the ability of β2m to be destabilized by Cu. The MCO/ MS measurements presented here should be more accurate because they are made while the entire coordination sphere is intact, and thus, they fully take into account any cooperativity in binding. A comparison of the binding site information gathered in this work and the results obtained by Miranker and co-workers7 provide some interesting insight into the role of Cu and several His residues (i.e., His13, His31, His51, and His84) in the destabilization of β2m. His13, His51, and His84 were all found to be important for the stabilization of the unfolded structure in the presence of Cu.7 For His51 and His84, this stabilization most likely arises from direct coordination to Cu. In contrast, the role of His13 in stabilizing the unfolded structure is not due to direct binding to Cu. Instead, perhaps His13 is involved in destabilization of the native state by interacting with amino acids that are bound to Cu, so that it is essentially an outer-sphere ligand. Finally, while we find that His31 is bound to Cu in the unfolded state, its coordination is only marginally important for the Cu-mediated destabilization of β2m. CONCLUSION A method based on MCO reactions and MS to determine Cubinding sites of proteins has been further utilized in this work. In particular, this technique has demonstrated the ability to determine the amino acids bound to Cu in β2m, which readily precipitates in the presence of this metal. Also, Cu-protein binding was determined in very complex solution conditions, namely, 8 M urea. Measurements by traditional techniques under these conditions would be difficult or impossible. In this work, Cu was found to bind native β2m through the N-terminus and His31, while Cu binding to non-native β2m involved the Nterminus, His31, His51, and His84. These results support the existence of a well-defined but different coordination structure in the non-native state of β2m, which leads to a greater affinity for Cu(II) than in the native state. In understanding the mechanism of protein destabilization and possibly amyloid fiber formation, these data might be useful in two ways. First, because Cu binding to the native protein occurs at the N-terminus, destabilization of this part of the protein structure may be the first step in the pathway that leads to protein aggregation. Second, the protein structure enforced by Cu under non-native conditions might also be an important step along the pathway. Clearly, more biophysical studies are necessary to better understand the amyloidosis of this Analytical Chemistry, Vol. 76, No. 13, July 1, 2004

3503

protein, but together with molecular modeling studies, the picture of Cu(II)-protein binding given by our MCO/MS method will provide some additional insight.

2002 American Society for Mass Spectrometry Research Awards (to R.W.V.). The authors also acknowledge Prof. Andrew D. Miranker from Yale University for the helpful discussions.

ACKNOWLEDGMENT The authors acknowledge the generous financial support of Thermo Finnigan Corp. through the sponsorship of one of the

Received for review February 19, 2004. Accepted April 13, 2004.

3504

Analytical Chemistry, Vol. 76, No. 13, July 1, 2004

AC049716T