Differentiation and Semiquantitative Analysis of an Isoaspartic Acid in

Anal. Chem. 2010, 82, 6384–6394

Differentiation and Semiquantitative Analysis of an Isoaspartic Acid in Human r-Crystallin by Postsource Decay in a Curved Field Reflectron Yuzo Yamazaki,*,† Norihiko Fujii,‡ Yutaka Sadakane,§ and Noriko Fujii‡ Applications Development Center, Analytical Applications Department, Shimadzu Corporation, 1, Nishinokyo-Kuwabaracho, Nakagyo-ku, Kyoto 604-8511, Japan, Research Reactor, Institute, Kyoto University, Kumatori, Sennan, Osaka 590-0494, Japan, and Kyushu University of Health and Welfare, 1714-1 Yoshino Nobeoka, Miyazaki 882-8508, Japan r-Crystallin, which forms a huge multimeric complex that is essential for maintaining eye lens transparency, is one of the major proteins in the lens. The protein, which exists as isoforms rA and rB, functions as a molecular chaperone to restore the original conformations of distorted constituent proteins in the lens. This function is important to maintain the transparency of the lens, because there is no protein turnover in the lens. Abnormal aggregation of constituent proteins in the lens has been reported in cataract patients, and deamidation of Asn as well as racemization and isomerization of Asp have been found in the r-Crystallin of these patients. While the establishment of a quick and facile analytical method is eagerly anticipated to investigate the relevance of the isomerization to pathological states such as cataracts, differentiating the isomerization states is still not performed routinely. Here, we report the differentiation and semiquantitative analysis of an isoaspartic acid (βAsp) in human r-Crystallin using postsource decay on a MALDITOF mass spectrometer incorporating a curved field reflectron. Our reproducible results of analyzing synthetic and tryptic peptides containing βAsp corroborated results obtained using a previously reported diagnostic ion, yl-n+1 46, for the differentiation of βAsp. The relative content of βAsp in the peptide was successfully estimated from a unique ratio, yl-n:yl-n+1, corresponding to cleavages at the C- and N-termini, respectively, of the isomeric residues. The βAsp content was consistent with measurements obtained independently by reversed phase HPLC analysis. Experiments in which neighboring amino acids adjacent to βAsp/ Asp were substituted revealed that the ratio between yl-n and yl-n+1 reflected the isomerization status, while the diagnostic ion was observed only in the peptides that included an arginine residue at their C-terminus. Postsource decay experiments utilizing both the diagnostic ion and the characteristic fragment pattern could be applied to various kinds of peptides containing βAsp. * Corresponding author. E-mail: [email protected]. Phone: +81-75823-1359. Fax: +81-75-841-9326. † Shimadzu Corporation. ‡ Kyoto University. § Kyushu University of Health and Welfare.

6384

Analytical Chemistry, Vol. 82, No. 15, August 1, 2010

Isomerization of aspartic acid (Asp) is frequently observed in proteins after long-term storage or exposure to radio/UV irradiation.1-3 Previous studies have revealed two processes of isomerization.4,5 One is a direct transformation of Asp to isoaspartic acid (βAsp), and another is transformation of asparagine to βAsp via deamidation. In both processes, the transformations occur through a cyclic succinimide as an intermediate form. A scheme of the transformation from Asp to βAsp is illustrated in Figure 1.6 A β-carboxyl group of the Asp is converted from a side chain to a component of the polypeptide chain, and concurrently, an R-carboxyl group is converted from the polypeptide chain to a carboxylic side chain. The conversion lengthens the polypeptide chain due to the insertion of an atypical methylene group. Since an Asp residue is frequently located at flexible regions in the 3D structure of proteins, such an alteration of the backbone length can disrupt protein structures, and this in turn can affect the quaternary structure by causing protein aggregation.7-9 Thus, isomerization in proteins is thought to be associated with degenerative diseases in human eyes,10,11 skin aging,2 and amyloidosis in the brain.12 Crystallin, which consists of three types proteins, R-, β-, and γ-crystallin, is one of the major proteins in the lens.13 The three constituent proteins form a huge multiple-protein complex that is essential for maintaining the transparency of the lens.14,15 Proteins in the lens can be damaged by various stimuli, and their structural (1) Fujii, N.; Momose, Y.; Ishibashi, Y.; Uemura, T.; Takita, M.; Takehana, M. Exp. Eye Res. 1997, 65, 99–104. (2) Fujii, N.; Hiroki, K.; Matsumoto, S.; Masuda, K.; Inoue, M.; Tanaka, Y.; Awakura, M.; Akaboshi, M. Photochem. Photobiol. 2001, 74, 477–482. (3) Hambly, D. M.; Banks, D. D.; Scavezze, J. L.; Siska, C. C.; Gadgil, H. S. Anal. Chem. 2009, 81, 7454–7459. (4) Fujii, N. Origins Life Evol. Biospheres 2002, 32, 103–127. (5) Robinson, N. E.; Robinson, A. B. Proc. Natl. Acad. Sci. U S A. 2001, 98, 944–949. (6) Fujii, N.; Tajima, S.; Tanaka, N.; Fujimoto, N.; Takata, T.; Shimo-Oka, T. Biochem. Biophys. Res. Commun. 2002, 294, 1047–1051. (7) Ecroyd, H.; Carver, J. A. Cell. Mol. Life Sci. 2009, 66, 62–81. (8) Fujii, N.; Takeuchi, N.; Fujii, N.; Tezuka, T.; Kuge, K.; Takata, T.; Kamei, A.; Saito, T. Amino Acids 2004, 26, 147–152. (9) Shimizu, T.; Matsuoka, Y.; Shirasawa, T. Biol. Pharm. Bull. 2005, 28, 1590– 1596. (10) Fujii, N.; Ishibashi, Y.; Satoh, K.; Fujino, M.; Harada, K. Biochim. Biophys. Acta 1994, 1204, 157–163. (11) Masters, P. M.; Bada, J. L.; Zigler, J. S., Jr. Nature 1977, 268, 71–73. (12) Shimizu, T.; Watanabe, A.; Ogawara, M.; Mori, H.; Shirasawa, T. Arch. Biochem. Biophys. 2000, 381, 225–234. (13) Graw, J. Exp. Eye Res. 2009, 88, 173–189. (14) Takemoto, L.; Sorensen, C. M. Exp. Eye Res. 2008, 87, 496–501. 10.1021/ac100310x  2010 American Chemical Society Published on Web 07/13/2010

Figure 1. Scheme of isomerization and racemization of aspartic acid. Adapted with permission from ref 6. Copyright 2002 Elsevier B.V.

alterations may be responsible for some diseases. R-Crystallin, which exists as isoforms RA and RB, functions as a molecular chaperone to restore the original conformations of distorted constituent proteins in the lens.16-18 This function is important to maintain the transparency of the lens, because there is no protein turnover in the lens. Therefore, R-Crystallin is one of the most intensively studied proteins in relation to cataract development, which could be attributed to isomerization and abnormal aggregation of constituent proteins in the lens. To investigate the relevance between pathological states such as cataracts and the content of isomerized residues in proteins, a quick and a facile analytical method is necessary. To date, some analytical methods, including enzymatic methylation, NMR, Edman-based sequencing, HPLC, and mass spectrometry, have been applied to investigate βAsp in biological systems. NMR is a very informative technique to probe for the presence and position of βAsp in a peptide sequence.19 However, it typically requires milligram quantities of analyte. Although methylation with PIMT (protein isoaspartyl methyl transferase) can detect βAsp,20-22 it does not provide any information as to the site of the βAsp residue(s) nor the relative abundance. The Edman degradation reaction is blocked by βAsp residues, and this can be used to demonstrate the presence and site of the βAsp Delaye, M.; Tardieu, A. Nature 1983, 302, 415–417. Horwitz, J. Exp. Eye Res. 2009, 88, 190–194. Andley, U. P. Prog. Retinal Eye Res. 2007, 26, 78–98. Kumar, P. A.; Reddy, G. B. IUBMB Life 2009, 61, 485–495. Chazin, W. J.; Ko ¨rdel, J.; Thulin, E.; Hofmann, T.; Drakenberg, T.; Forsén, S. Biochemistry 1989, 28, 8646–8653. (20) Alfaro, J. F.; Gillies, L. A.; Sun, H. G.; Dai, S.; Zang, T.; Klaene, J. J.; Kim, B. J.; Lowenson, J. D.; Clarke, S. G.; Karger, B. L.; Zhou, Z. S. Anal. Chem. 2008, 80, 3882–3889. (21) Clarke, S. Annu. Rev. Biochem. 1985, 54, 479–506. (22) Johnson, B. A.; Aswad, D. W. Anal. Biochem. 1991, 192, 384–391. (15) (16) (17) (18) (19)

in conjunction with HPLC separation. However, like NMR, this method also consumes a significant amount of the sample. Although there have been some reports of successful quantification by protein digestion followed by reversed phase HPLC,23 separation of the isomeric peptide from a peptide mixture is still challenging, even when a shallow gradient is applied, because the retention time of a peptide containing βAsp frequently overlaps with not only the counterpart containing Asp but also the one containing racemic residues, due to very similar physicochemical properties of the peptides. Furthermore, these separations usually require a very long run time. Recently, mass spectrometry (MS) has been applied to analyze small amounts of analytes containing βAsp. One might think that the molecular weight obtained using MS would not be useful in differentiating between isomers, due to their identical mass. However, the MS/MS technique may be applied to the differentiation of isomers. For instance, unique side chain fragment ions produced by high energy collisions can be useful for differentiation between leucine and isoleucine,24 and the specific relative intensities of several fragment ions of carbohydrates obtained by low energy collision induced dissociation (CID) in an ion trap and by postsource decay (PSD) may be analyzed to identify isomers.25,26 In addition to these cases, MS/MS techniques may be combined with low and high energy CID to generate diagnostic fragment ions that can be used to differentiate between βAsp and Asp.27-29 Recently ECD/ETD methods were applied to the differentiation between βAsp and Asp, since these ionization techniques produce c- and z-ions, which do not disrupt the side chains of amino acid residues; these techniques also generate a diagnostic ion, z-56, and they have been reported to be useful in differentiating between the isomeric residues.30,31 Furthermore, βAsp may be quantified using the ETD method.32 However, in CID experiments, careful operation of the overall experimental procedures is necessary to reproducibly generate the diagnostic ions. ECD methods based on FT-MS may require a level of skill such that the methods might not be suitable for high through-put analysis and may not be acceptable for a wide number of researchers in biochemical or medical fields. In addition, the higher instrumentation costs may be prohibitive for many laboratories. Furthermore, there are still unclear issues of fragmentation rules in analyzing the results of ECD/ETD33-35 (23) Sadakane, Y.; Yamazaki, T.; Nakagomi, K.; Akizawa, T.; Fujii, N.; Tanimura, T.; Kaneda, M.; Hatanaka, Y. J. Pharm. Biomed. Anal. 2003, 30, 1825– 1833. (24) Johnson, R. S.; Martin, S. A.; Biemann, K.; Stults, J. T.; Watson, J. T. Anal. Chem. 1987, 59, 2621–2625. (25) Yamagaki, T.; Suzuki, H.; Tachibana, K. J. Am. Soc. Mass Spectrom. 2006, 17, 67–74. (26) Yamagaki, T.; Nakanishi, H. J. Mass Spectrom. 2000, 35, 1300–1307. (27) Lehmann, W. D.; Schlosser, A.; Erben, G.; Pipkorn, R.; Bossemeyer, D.; Kinzel, V. Protein Sci. 2000, 9, 2260–2268. (28) Gonza´lez, L. J.; Shimizu, T.; Satomi, Y.; Betancourt, L.; Besada, V.; Padrón, G.; Orlando, R.; Shirasawa, T.; Shimonishi, Y.; Takao, T. Rapid Commun. Mass Spectrom. 2000, 14, 2092–2102. (29) Carr, S. A.; Hemling, M. E.; Bean, M. F.; Roberts, G. D. Anal. Chem. 1991, 63, 2802–2824. (30) Cournoyer, J. J.; Pittman, J. L.; Ivleva, V. B.; Fallows, E.; Waskell, L.; Costello, C. E.; O’Connor, P. B. Protein Sci. 2005, 14, 452–463. (31) O’Connor, P. B.; Cournoyer, J. J.; Pitteri, S. J.; Chrisman, P. A.; McLuckey, S. A. J. Am. Soc. Mass Spectrom. 2006, 17, 15–19. (32) Cournoyer, J. J.; Lin, C.; Bowman, M. J.; O’Connor, P. B. J. Am. Soc. Mass Spectrom. 2007, 18, 48–56. (33) Coon, J. J. Anal. Chem. 2009, 81, 3208–3215.


6385

despite an advantage of analysis for post-translational modifications. While most MS methods in previous reports have utilized electrospray ionization combined with MS/MS techniques, there have been only a few reports of applying MALDI to isomer differentiation. MALDI ionization generates primarily singly charged ions, providing the advantage of quick surveys over a wide mass range. The ions generated by MALDI are often subjected to PSD analysis or TOF/TOF to obtain structural information, including amino acid sequences. In particular, the sequence information obtained by PSD is easy to explain, because most of the cleavages occur at peptide bonds, resulting in generation of b- and y-ions. Furthermore, “seamless” PSD on a curved field reflectron can focus all of the fragment ions entering the reflectron on a focal point concurrently, giving the advantage of more sensitive detection of fragment ions compared to other reflectrons, in principle.36,37 Some stable isotope labeling techniques utilizing MALDI in conjunction with TOF/TOF are widely accepted to perform quantitative analysis,38-40 indicating that MALDI is applicable not only for qualitative analysis but also for quantitative analysis. The use of PSD in the differentiation of isomeric carbohydrates has also been reported, and the possibility of carrying out a semiquantitative analysis of carbohydrate using PSD was postulated.41 The analytical procedure of PSD is composed of two steps, selection of a precursor and increasing the laser attenuation. A primary benefit of seamless PSD on a curved field reflectron MALDI-TOF MS is its very simple ion optics.36,37 In this paper, we report the differentiation and semiquantitative analysis of a βAsp residue in R-crystallin using PSD on a curved field reflectron. Results of experiments utilizing synthetic peptides containing βAsp corroborate the presence of a diagnostic Cterminal fragment ion, yl-n+1 - 46, which was previously reported as an applicable ion in the differentiation of a βAsp using ESI-MS.28 Furthermore, PSD revealed that Asp isomerization causes significant changes in the intensities of fragment ions, which are responsible for cleavages at the N- and C-terminal sides of the isomeric residues. Of note, these significant changes were observed not only in the peptides that include an arginine residue at their C-terminus but also in peptides that do not, while it was reported that the diagnostic ion was observed only in peptides that include an arginine residue at their C-terminus.28 Moreover, substitutions of residues adjacent to the Asp/βAsp resulted in the same types of the changes as were seen on their unsubstituted counterparts, indicating that the fragmentation in PSD is widely (34) Zubarev, R. A.; Horn, D. M.; Fridriksson, E. K.; Kelleher, N. L.; Kruger, N. A.; Lewis, M. A.; Carpenter, B. K.; McLafferty, F. W. Anal. Chem. 2000, 72, 563–573. (35) Iavarone, A. T.; Paech, K.; Williams, E. R. Anal. Chem. 2004, 76, 2231– 2238. (36) Cornish, T. J.; Cotter, R. J. Rapid Commun. Mass Spectrom. 1993, 7, 1037– 1040. (37) Cornish, T. J.; Cotter, R. J. Rapid Commun. Mass Spectrom. 1994, 8, 781–785. (38) Matsuo, E.; Watanabe, M.; Kuyama, H.; Nishimura, O. J. Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2009, 877, 2607–2614. (39) Ong, S. E.; Mann, M. Nat. Chem. Biol. 2005, 1, 252–262. (40) Ross, P. L.; Huang, Y. N.; Marchese, J. N.; Williamson, B.; Parker, K.; Hattan, S.; Khainovski, N.; Pillai, S.; Dey, S.; Daniels, S.; Purkayastha, S.; Juhasz, P.; Martin, S.; Bartlet-Jones, M.; He, F.; Jacobson, A.; Pappin, D. J. Mol. Cell. Proteomics 2004, 3, 1154–1169. (41) Yamagaki, T.; Suzuki, H.; Tachibana, K. J. Mass Spectrom. 2006, 41, 454– 462.

6386


applicable to analysis of any amino acid sequence containing βAsp. Using PSD, we successfully obtained the βAsp content % in R-crystallin from cataract patients, and the results were consistent with HPLC analysis of the same samples. EXPERIMENTAL PROCEDURES Materials. R-Cyano-4-hydroxycinnamic acid (CHCA) was obtained from Shimadzu GLC (Kyoto, Japan). A sequence grade trypsin was purchased from Promega Corp. (Madison, WI). Analytical grade solvents and buffers were used for all experiments. All peptides for external calibrants were purchased from Sigma-Aldrich (St. Louis, MO). Human cataract lenses were kindly provided by Kanazawa Medical University. Synthesis of Asp- and βAsp-Containing Peptides. The peptides were synthesized by Fmoc (9-fluorenylmethoxycarbonyl) solid phase chemistry. Fmoc-L-Asp(OtBu)-OH, Fmoc-D-Asp(OtBu)OH, Fmoc-L-Asp-OtBu, and Fmoc-D-Asp-OtBu were used as building blocks to synthesize LR-, DR-, Lβ-, and Dβ-isomers, respectively. The coupling reaction was carried out using each Fmoc amino acid (5 equiv), benzotriazole-1-yl-oxy-tris-pyrrolidino-phosphonium hexafluorophosphate (PyBOP, 5 equiv), 1-hydroxybenzotriazole (HOBt) (5 equiv), and N-methylmorpholine (7.5 equiv) in dimethylformamide (DMF). The N-termius Fmoc group was deblocked with 20% piperidine in DMF. The cleavage of the peptide from resin and protective groups was treated with a cocktail containing 90% trifluoroacetic acid (TFA), 5% 1,2-ethanedithiol, and 5% thioanisole. The cleavage of the peptide contains Arg from resin and protective groups and was carried out with 82.5% TFA, 5% water, 5% thioanisole, 3% ethylmethylsulfide, 2.5% 1,2-ethanedithiol, and 2% thiophenol. The peptides were analyzed by HPLC to confirm their purities, all of which were more than 95%. All peptides were prepared at 1 mg/mL in distilled water containing 0.1% TFA and subjected to MS. Preparation of rA Crystallin and Its Digest. Human lenses with a cataract were homogenized and suspended in 20 mM Tris/ HCl (pH 7.8) including 150 mM NaCl. After the centrifugation, the supernatant fraction of the suspension was subjected to gel filtration chromatography to separate R-crystallin and other proteins, and fractions of the constituents were collected. RA- and RB-crystallin were separated from the R-crystallin fraction by reversed phase HPLC. The final concentration of the RA-crystallin was estimated as more than 2 to 3 pmol/µL. A fraction of RAcrystallin was dried in a speed vac and then dissolved in 100 mM ammonium bicarbonate. RA-Crystallin was digested with trypsin for 20 h at 37 °C in 0.1 M Tris HCl buffer (20 mM CaCl2, pH 7.6), at enzyme to substrate ratio of 1/100 (w/w). The digested protein was desalted by mono Tip C18 (GL Science, Tokyo, Japan) and then subjected to MS analysis. Mass Spectrometry. Seamless PSD on a curved field reflectron (CFR) was performed on an AXIMA-Performance (Shimadzu Biotech) in positive ion mode. Five mg/mL of CHCA in 50% acetonitrile aqueous solution including 0.1% TFA was used for MS measurements. The matrix solution (0.5 µL) was mixed with the same amount of analyte on a stainless steel MALDI plate. The instrument was operated at 20 keV acceleration voltage in the m/z range from 10 to 5000, and a pulsed extraction functioning to improve a mass resolution was set to m/z 2500. The TOF analyzer was calibrated using the following external calibrants: a dimer of CHCA ([2M + H]+; 379.09), human angiotensin II ([M + H]+;

Table 1. Synthetic Peptides Subjected to PSD Analysis and Ratios of Remarkable Fragment Ions

1046.54), and human ACTH7-38 ([M + H]+; 3657.89). For PSD, the resolution for isolating precursors was set to 150. For quantitative analysis, a prior measurement of a synthetic peptide containing βAsp was performed to find an appropriate laser attenuation for PSD. While attenuations less than a threshold value resulted in obvious but a little change on the yl-n:yl-n+1 ratio as increasing the attenuation, the changes of the ratio, caused by the attenuations above the threshold value, reached a plateau. Finally, the adequate attenuation was judged by an absolute intensity of yl-n, which was always detected with a high intensity, and a signal-to-noise ratio around this fragment ion. A nitrogen laser with the identical attenuation was used for a series of PSD experiments. All MS data were acquired

and analyzed with MALDI-MS software (Kratos/Shimadzu Biotech). A set of peak processing parameters including peak smoothing and baseline subtraction was carefully chosen, resulting in adequate PSD spectra for quantitative analysis. All PSD spectra were processed using the same set of parameters. Relative Quantification of βAsp by Reversed Phase HPLC. Relative quantification of the βAsp content was performed with an LC-10AD system (Shimadzu Corp., Kyoto, Japan). Separation of Asp and its isoform was achieved using a C18 column (CAPCELL UG 80, 3.0 mm id, 250 mm length, Shiseido, Tokyo, Japan). A linear gradient of acetonitrile including 0.1% TFA was set from 0% to 40% over 170 min with a 0.5 mL/minute flow rate. RP-HPLC showed a clear separation of the peptides containing Analytical Chemistry, Vol. 82, No. 15, August 1, 2010

6387

Figure 2. PSD spectra of various T6s. Overall view of the spectra, which are normalized to precursor ions (A), and the expanded regions, which are normalized to y7 (m/z 747) (B). Note that y7 and y8 are C-terminal ions cleaved at C-terminal and N-terminal of Asp/βAsp.

the βAsp and normal Asp residues, and the βAsp content was calculated from the ratio of the chromatographic peak areas of the both peptides. RESULTS AND DISCUSSION Diagnostic Ion and Specific Ratios between C-Terminal Ions. Synthetic peptides analyzed by PSD are listed in Table 1. Almost all peptide sequences corresponded to portions of RA- and βB2-crystallin. The PSD spectra of peptides containing βAsp were compared with the corresponding spectra for their counterparts that contained Asp at the same position. For instance, four varieties of peptide T6, including different stereoisomers and the isomeric residues, were designed and analyzed by PSD (Figure 2). While y8 - 46 (m/z 816) was observed in the spectrum of T6-Lβ and T6-Dβ, both of which contained a βAsp, no intensity of y8 - 46 was observed in the spectra of T6-LR and T6-DR, which contained an Asp at the same position. This y8 - 46 signal apparently corresponds to a specific fragment ion, and it indicates the presence of βAsp. In fact, it was previously reported that a yl-n+1 - 46 was a diagnostic C-terminal fragment ion and that it could be used to differentiate the βAsp.28 Hereafter, the specific ions are represented as, yl-n, yl-n+1, and yl-n+1 - 46, where “l ” and “n” represent the length of the peptide and the position of the Asp/βAsp residue, 6388


respectively. As summarized in Table 1, while the ratio y7:y8 46:y8 (m/z 747, 816, and 862) in the PSD spectra of T6-Lβ and T6-Dβ was 0.72:0.04:0.24, the ratio in the spectra of T6-LR and T6-DR was 0.92:0:0.08. Since the relative ratio y7:y8 reflects the susceptibility of the peptide to undergo conversion to βAsp, the ratios are also applicable to differentiating βAsp as well as the y8 - 46. Furthermore, the ratios may be used to discern the presence of βAsp, as well as its position, because y7 and y8 are C-terminal ions cleaved at the C-terminal and N-terminal side of Asp (or βAsp), respectively. On the other hand, there was no informative fragment ion in the spectra of any of the T6 peptides to differentiate between the D- and L-stereoisomers of Asp and βAsp. Although the diagnostic ion was reported using low energy CID, there were quite a few subsequent reports where the ion was not detected. This was probably due to technical hurdles in setting appropriate conditions to obtain reproducibility. In contrast, the diagnostic ion and the specific ratios were highly reproducible in our PSD spectra. It has also been reported that while yl-n+1 - 46 was observed in tryptic peptides in which the arginine residues were mostly located at the C-terminus, it was not observed in other peptides in which no strong basic residue was located at the C-terminus.28

Thus, an influence of the C-terminal basicity on fragmentation patterns is expected in PSD analysis. To further explore this issue, additional synthetic peptides, in which Lys residues were located at the C-terminus or at both Nand C-termini, were analyzed using the PSD (Figure 3). Although a signal intensity of yl-n+1 - 46 was not observed in these PSD spectra and so could not be used to identify βAsp, specific ratios, yl-n:yl-n+1, were clearly observed in the PSD spectra, which enabled us to identify both the presence and the position of the βAsp. For instance, when four isopeptides of βB2 were analyzed (Figure 3A), while y5:y6 (m/z 628 and 743) in the Dβ-Asp containing peptide was 0.37:0.63, this ratio in the other DR-asp containing peptide was 0.71:0.29. Although no y6 - 46 was observed in the spectra, a great change of the y5:y6 ratio in βB2 enabled us to differentiate between βAsp and Asp. A N-terminal b2-ion (m/z 215) was also observed in the peptides that included βAsp. However, intensities of these b2-ions were so weak that it was not useful for differentiation (data not shown). Similar to βB2, in the analysis of peptide RA, the y12: y13 (m/z 1399 and 1514) ratio in the βAsp containing peptides was significantly different from this ratio in peptides containing an Asp (Figure 3B). For instance, while the ratio in the peptide containing LR-Asp was 0.76:0.24, the ratio in the peptides containing Lβ-Asp was 0.46:0.54. In addition, b6:b7 (m/z 748, 863) ratio in the spectra was different for those containing Asp or βAsp (Figure 3C). Since a Lys residue is located at the N-terminus, the b-ions contain basic residues, which can aid the b-ions charge retention. Because absolute intensities of the b6 and b7 were not intense, it could be difficult to apply these ratios to the differentiation between βAsp and Asp. It has been noted that the significant changes were observed not only in the peptides that included an arginine residue at their C-terminus but also in peptides that do not, while the diagnostic ion was observed using ESI-CID-MS/MS only for peptides that included an arginine residue at their C-terminus.28 Hence, PSD using a CFR is more applicable to this type of differentiation than ESI-MS/MS. Peptides with molecular weights from 800 Da to over 2200 Da (sequence length; 7-19), were subjected to the experiments, and the presence and the position of βAsp were determined successfully. MALDI ionization generates singly charged ions, and especially in the case of tryptic digests, the charge is localized at the C-terminal arginine/lysine. The PSD method can utilize this charge localization to produce simple and explainable fragmentation patterns. On the other hand, the ETD/ECD has an essential advantage to analyze post-translational modifications of proteins, while the method still involves unclear issues on a fragmentation rule, such as an additional gain/loss of proton on c- and z-ions,33,34 an effect of number of charge.35 PSD could be a complementary technique to ETD/ECD, as is the case of ETD/ECD and CID.33 In this view, it might be an advantage that the PSD method is applicable to analyze longer sequences than those investigated in this report, although success in such studies could depend on the amount of peptide analyzed. Substitutions at the C- and N-Terminal Side of Asp Residues. While there have been some reports of the differentiation of βAsp using low energy CID, earlier studies indicated that PSD was unsuitable for differentiation between βAsp and Asp.28

Figure 3. PSD spectra of peptide βB2, normalized to y5 (m/z 628) or y6 (m/z 743) (A). PSD spectra of peptide RA, normalized to the most intense peak in the range from m/z 1380 to 1550 (B). The same peptide RA, normalized to the most intense peak in the range from m/z 725 to 885 (C).

However, as shown in our results for peptides T6, βB2, and RA, it appears that the PSD using a curved field reflectron can easily detect the isomerization, e.g., an insertion of an atypical Analytical Chemistry, Vol. 82, No. 15, August 1, 2010

6389

Figure 4. Schematic drawing of fragmentations around Asp/βAsp. A formation indicating preferential cleavage at the C-terminal side of Asp (A, Reprinted with permission from ref 42. Copyright 1993 American Chemical Society), and a conceivable formation for βAsp (B). A proposed pathway to form yl-n+1 - 46 (C, Reprinted with permission from ref 27. Copyright 2000 The Protein Society).

methylene group to a backbone chain, by detecting specific fragment ions cleaved at Asp and βAsp. A preferential cleavage at the C-terminal side of Asp in the gas phase has been of interest since early studies of MALDI ionization.42 A cyclic formation, which occurs after an intramolecular interaction between an acidic side chain and a peptide backbone, was proposed to account for the cleavable peptide bond on the C-terminal side of Asp in the MALDI process (Figure 4A).42 In fact, this cyclic formation prior to the C-terminal fragmentation must be responsible for the dominant yl-n in the PSD spectra of the peptides containing Asp. Isomerization of an Asp to a βAsp results in the insertion of an atypical methylene group into the peptide backbone. Therefore, one may deduce that since the side chain of βAsp is shorter than that of Asp, it could not interact as readily with a longer backbone chain. Therefore, a cyclic formation around βAsp might be less stable than a linear formation in the gas phase. For this reason, the probability of a cleavage at yl-n+1 in the peptides containing βAsp could be relatively high (Figure 4B). To explain the formation of yl-n+1 - 46, a fragmentation pathway for βAsp containing peptides was proposed (shown in Figure 4C).28 According to this proposal, it is deduced that since an intermediate five-membered ring in a fragment pathway of the (42) Yu, W.; Vath, J. E.; Huberty, M. C.; Martin, S. A. Anal. Chem. 1993, 65, 3015–3023.

6390


βAsp containing peptides is unstable, rearrangement to form yl-n+1 - 46 occurs immediately. In contrast, the frequency of a mechanism involving yl-n+1 - 46 in the Asp containing peptides is likely to be very low, because of the relatively higher stability of their intermediate cyclic state in the gas phase. Regardless of which fragment pathways described above are correct, substitutions at neighboring residues of the isomeric residues will not affect propensities to form a yl-n+1 - 46 and a yl-n+1 during PSD. Although we cannot definitively identify detailed mechanisms leading to the formation of these C-terminal ions, it is necessary to investigate effects of substitutions at the N- and C-terminal sides of the isomeric residues. To investigate the influence of adjacent amino acids on the fragmentation patterns, analogs of peptide T6, in which a leucine and a serine residue were substituted to arginines, were synthesized and subjected to PSD analysis. Substitution at a leucine residue on the N-terminal side of βAsp to an arginine caused a great change in the y7:y8 - 46:y8 relative to the unsubstituted counterpart. The ratio 0.72:0.04:0.24 (m/z 748, 817, and 863) in T6-Lβ (Figure 2B) was changed to 0.39:0.04:0.57 in T6LR-Lβ (Figure 5A) due to this substitution, indicating that the position at yl-n+1 in T6LR-Lβ became more cleavable than yl-n. This substitution also caused a change in fragmentation of the Asp containing peptides. For instance, in comparison with T6-LR and

Figure 6. PSD spectra of mixtures composed of peptide T6-LR and -Lβ, normalized to y7 (m/z 747) in the range from m/z 720 to 880. The content % of T6-Lβ is 10, 30, 50, 70, and 90% from lower to upper.

Figure 5. PSD spectra of peptide T6LR, normalized to the most intense peak in the range from m/z 730 to 890 (A), peptide T6SR, normalized to y7 (m/z 817) in the range from m/z 805 to 965 (B).

T6LR-LR (Figures 2B and 5A), while the ratio y7:y8 - 46:y8 in the former peptide was 0.92:0:0.08, the ratio in the latter peptide was 0.80:0:0.20. However, despite these changes, the relative intensity of y8 in T6LR-Lβ was higher than that in T6LR-LR, indicating that the ratio is applicable to differentiation of the βAsp. Substitution at the residue on the N-terminal side of βAsp caused additional changes in N-terminal fragment ions (data not shown). In T6LR-Lβ and -Dβ, a previously unobserved peak corresponding to b3 + H2O was detected, while the intensity of b4 increased significantly relative to T6LR-LR and -DR. This b3 + H2O was reported as an N-terminal diagnostic ion, along with yl-n+1 - 46.28 The b4 could easily be formed due to an influence of the basic arginine residue, which was substituted from a hydrophobic leucine residue. On the other hand, substitutions at the C-terminal sides of isomeric residues were responsible for a modest change in the relative intensities of y7, y8 - 46, and y8 (m/z 817, 886, and 932 in Figure 5B). The y8 - 46 and y8 were not observed in T6SR-LR and -DR, and the relative intensities of these ions in T6SR-Lβ and -Dβ were a little less than those seen in their unsubstituted counterparts (T6Lβ and -Dβ in Figure 2). Importantly, in addition to the N-terminal substitutions, the relative intensity of y8 is still applicable to the differentiation. Not all of the substitutions caused significant changes in the diagnostic ion, yl-n+1 - 46, in analog peptides

containing βAsp. Although one could speculate that the frequency of forming yl-n+1 is almost independent of the substitutions, it is difficult to discuss it due to the weak intensities of yl-n+1. All of the substitution experiments indicated that fragmentation by PSD is very susceptible to the isomeric form, even when a strong basic residue is located near the Asp/βAsp. Furthermore, there was no overall change of fragment ions intensities, except for the remarkable ions cleaved at the N- and C-terminal sides of the isomeric residues. This simple result is expected to be widely applicable to quantitative analysis of peptides containing βAsp. Quantitative Analysis of βAsp in a Protein. Peptides T6-LR and T6-Lβ were mixed at various ratios and then subjected to PSD analysis, to investigate if quantitative analysis of the βAsp without any separation is possible using the PSD. Five mixtures composed of T6-LR and T6-Lβ were prepared with T6-Lβ content, namely, βAsp content, ranging from 10% to 90%. As shown in Figure 6, the intensities of y8 and y8 - 46 relative to y7 increased as the content of T6-Lβ became higher, clearly indicating an increase of βAsp content in the mixtures. Although y8 - 46 is definitely a diagnostic ion indicating the presence of βAsp, we avoided adopting this ion for quantification, because its absolute intensity is relatively weak. It is likely that a calibration curve based on the intensity of this ion would introduce a large error into estimations of the βAsp content. The observed ratio y7:y8 is plotted versus the βAsp content in Figure 7. An analysis of the observed plots using the least-squares approach produced an almost linear relationship between the observed and the βAsp content, with R2 values greater than 0.95, strongly indicating that the relationship can be utilized to estimate the βAsp content with the ratio, y7:y8. Notably, this linear relationship was obtained using the PSD measurement, which involves very simple measurement procedures, specifically isolation of a precursor ion and operating an attenuation of a laser. However, a previous report indicated that PSD measurements are not useful for analysis of isomerization.28 The only difference between the current experiment and the previous one is the architecture of the reflectron. Because of the simple ion optics in a curved field reflectron, in which there are not any metastable suppressors or stepwise elevations of voltage, Analytical Chemistry, Vol. 82, No. 15, August 1, 2010

6391

Figure 7. Plots of ratios between yl-n and yl-n+1 versus βAsp contents. Open diamond and open reversed triangle are y7 and y8 in Figure 6. Open circle and open triangle are y6 and y7 in Figure 8C). Lines exhibit results of least-squares analysis of the plots. R2 values are 0.988 for y7 and y8 in Figure 6, and 0.953 for y6 and y7 in Figure 8C. Filled triangle and filled square display relative βAsp content estimated from the fitted line for y6 and y7.

seamless PSD performed using a CFR can collect all fragment ions generated by PSD concurrently.37 This property could be very suitable for the analysis of isomerizations. On the other hand, although differentiation of the isomeric residues using mass spectrometry was reported in CID measurements originally, it was pointed out that overall intensities of fragment ions in CID experiments could fluctuate, due to conversions of Asp to βAsp. In this case, estimates of βAsp content % will be erroneous, so experimental conditions including a collision gas, a gas pressure, and a collision offset voltage, should be chosen carefully. A careful optimization of the experimental conditions for CID was attempted previously, and a ratio between complementary b- and y-ions adjacent to Asp/βAsp residues and another ratio between the immonium ions derived from the residues were used for differentiation and quantitative analysis of the βAsp27. In this report, the b/y ratio in the peptides containing βAsp was smaller than in those containing Asp, which suggests that the relative probability for the generation of yl-n+1 increased in the former peptides as well as our results. However, it was also reported that while the ratios of the immonium ions indicated a linear relationship versus βAsp contents, the b/y ratios increased exponentially versus βAsp contents, in contrast to the linear relationship between the ratio, yl-n:yl-n+1, and βAsp contents in our PSD experiments, suggesting that it could be hard to utilize the ratio between different ion series, which were generated in different pathways as the authors supposed. Meanwhile, selecting the correct experimental conditions to obtain adequate PSD spectra involves only one step, choosing an appropriate laser attenuation, and it is necessary to apply an identical laser attenuation to both analyte and calibrant 6392


mixtures, which are composed of peptides containing RAsp and βAsp. Finally, a feasibility study was carried out in which the βAsp content in RA-crystallin derived from human cataract lenses was estimated using PSD. This isomerization in the protein has been studied intensively, and an Asp in peptide T18, which is a tryptic peptides of this protein, is known to be frequently transformed to a βAsp. As well as the procedure for making the calibration curve using peptide T6, synthetic T18-LR and -Lβ were utilized to make a calibration curve to obtain the βAsp content (Figure 7). RACrystallin was digested with trypsin, then MS analysis of a mixture of tryptic peptides was performed, and the MS spectra of the peptides were assigned to an amino acid sequence of RA-crystallin (Figure 8A,B). Peptide T18 in the mixture was isolated as a precursor ion; PSD of this peptide was performed (Figure 8C), and the βAsp content was estimated from the calibration curve (Figure 7). The PSD experiments to the identical sample were performed five times, and the obtained βAsp contents were averaged. Concurrently, the digest solution was applied to reversed phase HPLC to quantify the relative abundance of Asp/ βAsp. Since separation of the four synthetic peptides containing LR-, DR-, Lβ-, and Dβ-Asp had been achieved using the HPLC, in which two βAsp were separated at each retention time prior to the RAsp and two RAsp were eluted at almost the same retention time, the obtained βAsp content by PSD was validated by comparison with the value from the HPLC. While the averaged βAsp content obtained using the PSD method was 33% ± 14%, the one obtained using HPLC was 22%. These βAsp contents were close to each other within experimental error. It is very important that the content can be determined using the PSD method without

Figure 8. MS spectrum of a tryptic digest of RA-crystallin (A), amino acid sequence of the protein (B), PSD spectra, normalized to y6 (m/z 684) in the range from m/z 660 to 820 (C). In (C), upper spectrum was obtained from the peptide T18 in (A), and lower three spectra were obtained from the mixtures composed of peptide T6-LR and -Lβ. The βAsp contents in the mixtures were 10, 30, and 50%, respectively.

carrying out a prior HPLC analysis. The possible reasons for the difference between two quantifications are both difficulties on separation of racemic/isomeric Asp by HPLC and the relatively large error in the PSD analysis of the digest. In addition to close retention times of racemic/isomeric Asp, a shallow gradient curve for the separation results in such a broadened peak width on the HPLC chromatograph that it is difficult to recognize peak areas assigned to LR-, Lβ-, DR-, and DβAsp, respectively, in the case of low amounts. On the other hand, while the statistical errors for the synthetic peptides used as calibrants were 4.1-6.5%, the error for the tryptic peptide was larger. Since a variability of the intensities of the fragment ions in the tryptic peptide analysis was wider than that in the synthetic peptides analysis, this error was responsible for the inhomogeneous coaggregation of the tryptic peptide and the matrix on the MALDI plate, which might be attributed to the protein digest itself. Despite the remaining questions for the PSD analysis, our results indicated that the error was sufficiently low to utilize the measurement for a semiquantitative analysis of βAsp content, as well as a previous report of

carbohydrates.41 Since there are few experimental factors to be optimized, due to the simple ion optics of a CFR, it is a very easy operation to carry out these types of PSD experiments. CONCLUSIONS The differentiation and semiquantitative analysis of βAsp content in a R-crystallin using PSD performed on a curved field reflectron was demonstrated. The relative content of βAsp in peptides was successfully estimated from a unique ratio, yl-n:yl-n+1, in a tryptic peptide of R-crystallin from cataract patients. The content was consistent with the estimate obtained by reversed phase HPLC analysis that was performed independently. The yl-n:yl-n+1 obtained by PSD using a CFR was a sensitive indicator of isomerization, because the frequencies of cleavages at the N- and C-terminal sides of the isomeric residues in the PSD method were highly dependent on the isomerization state. Of note, this characteristic fragment pattern that reflects the isomerization state was observed not only in the peptides that include an arginine residue at their C-terminus but also in those Analytical Chemistry, Vol. 82, No. 15, August 1, 2010

6393

that do not. Since the diagnostic ion, yl-n+1 - 46, was observed only in the peptides that included an arginine residue at their C-terminus, as in a previous report, the characteristic fragment pattern in the PSD spectrum could be more applicable to the analysis of βAsp than the diagnostic ion. In addition, our reproducible results of analyzing synthetic and tryptic peptides containing βAsp corroborate the usability of this diagnostic ion, which had not been examined in terms of this differentiation until now, despite a previous report using low-energy CID in ESI-MS. Utilizing both the diagnostic ion and the fragment pattern, the PSD method could be applied to various kinds of peptides containing βAsp. Although the mechanisms of the cleavages at both sides of the isomeric residues, as well as the formation of the diagnostic ion, are still unclear, the PSD method using a CFR was shown to be useful to analyze the isomerization. The substitutions at amino acids adjacent to the isomeric residues resulted in the same propensities of the yl-n:yl-n+1 as those seen for their unsubstituted counterparts. Namely, regardless of what amino acids are adjacent to the βAsp, the yl-n:yl-n+1 could be applicable to semiquantitative analysis of the isomeric residues. Interestingly, only a substitution at the N-terminal side of the βAsp caused a great increase of yl-n+1 relative to yl-n, whereas substitutions at the C-terminal side of the βAsp and at both the N- and C-terminal

6394


sides of the Asp caused modest changes. This result strongly suggests that an intermediate cyclic formation prior to fragmentation around the βAsp may differ from the one around the Asp or that a linear formation could be preferable around the βAsp. The isomerization of Asp is of interest in relation with degenerative diseases, and the potential unveiling of clinically relevant connections between some pathological states, such as cataracts, and the isomerization content is eagerly anticipated. Mass spectrometry is very sensitive in detecting isomeric residues, and the method described here takes advantage of the simple operation of PSD. The semiquantitative analysis using this method is a quick and efficient method with which to investigate disease states and mechanisms. It is likely that this method can be applied to automate multisample analysis as well. ACKNOWLEDGMENT This work was supported, in part, by a research grant from the Ministry of Education, Science, Sports, and Culture of Japan. The authors thank Matthew Openshaw for helpful discussions.

Received for review February 3, 2010. Accepted June 15, 2010. AC100310X

Differentiation and Semiquantitative Analysis of an Isoaspartic Acid in

Recommend Documents