Screening and Sequencing of Glycated Proteins by Neutral Loss Scan

Anal. Chem. 2007, 79, 5991-5999

Screening and Sequencing of Glycated Proteins by Neutral Loss Scan LC/MS/MS Method Himanshu S. Gadgil,* Pavel V. Bondarenko, Michael J. Treuheit, and Da Ren*

Department of Pharmaceutics, Amgen, Inc., Thousand Oaks, California 91320

Nonenzymatic protein glycation is caused by a Schiff’s base reaction between the aldehyde groups of reducing sugars and the primary amines of proteins. A reversedphase liquid chromatography method followed by a neutral loss scan mass spectrometric method was developed for the screening of glycation in proteins. The neutral loss scan was based on a unique sugar moiety neutral loss (-162 Da) that we observed in the fragmentation spectra of glycated peptides on Q-Tof type mass spectrometers. The collision energy was optimized for this neutral loss using a glycated synthetic peptide, and 20 eV was found to be the optimum collision energy. The neutral loss scan experiment was composed of two segments. In the first segment, the glycated peptides were identified based on the signature neutral loss of 162 Da when the collision energy was elevated to 20 eV. In the second segment, the glycated peptides were selected as the parent ions and fragmented at higher collision energy to break the peptide bonds. The fragmentation spectra of the selected glycated peptides revealed both the amino acid sequences and the sites of glycation. This neutral loss scan method was used to study the glycation in human serum albumin (HSA). The glycation sites in HSA were identified based on the retention time shift of glycated peptides, the mass accuracy from the MS scan, the signature neutral loss, and MS/MS information. Using this method, we were able to identify that 31 lysine residues were partially glycated from the glycated HSA sample, which has a total of 59 lysine residues. Nonenzymatic protein glycation products are formed by the Maillard reaction between the aldehyde groups of reducing sugars and primary amines of proteins. These protein glycation products can undergo rearrangement to form Amadori products that can breakdown into carbonyl compounds. The carbonyl compounds are active enough to react with proteins to form cross-linking protein products that are the sources of advanced glycation end products (AGEs).1 The formation of AGEs is nonreversible, and they affect the structure and functionality of proteins. AGEs are * To whom correspondence should be addressed. Phone: (206) 265-8282 or (805) 313-5317. Fax: (206) 217-5529 or (805) 447-3259. E-mail: [email protected] or [email protected]. (1) Thornalley, P. J.; Battah, S.; Ahmed, N.; Karachalias, N.; Agalou, S.; BabaeiJadidi, R.; Dawnay, A. Biochem. J. 2003, 375, 581-92. 10.1021/ac070619k CCC: $37.00 Published on Web 06/16/2007

© 2007 American Chemical Society

known to cause diabetic complications and age-related diseases.2-4 In the biopharmaceutical industry, glycated proteins can be formed during fermentation and sometimes during formulation. Glucose and fructose, which are commonly used in the fermentation process, can cause glycation in proteins. In addition, reducing sugars such as glucose5,6 and lactose7 are sometimes used for stabilizing proteins in lyophilized formulations. Nonreducing sugars such as sucrose and trehalose used in liquid formulation can be hydrolyzed to generate reducing sugars, which can potentially cause protein glycation. Protein glycation products can be detected by various analytical methods including ion-exchange chromatography,8 capillary electrophoresis,9 and boronate affinity chromatography.10 These methods are not very specific for the detection of glycated proteins and cannot identify the sites of glycation. Recently, MS has also been used for the study of glycated proteins. Glycation leads to an addition of 162 Da in mass, which can be easily detected by mass spectrometry. Both MALDI TOF and ESI ion trap mass spectrometers have been used for the identification of glycated peptides after enzymatic digestions.11-15 High-resolution Fourier transform ion cyclotron resonance mass spectrometry has also been used to identify the molecular weights of the glycated (2) Brownlee, M.; Cerami, A.; Vlassara, H. N. Engl. J. Med. 1988, 318, 131521. (3) Bucala, R.; Cerami, A. Adv. Pharmacol. 1992, 23, 1-34. (4) Vasan, S.; Zhang, X.; Zhang, X.; Kapurniotu, A.; Bernhagen, J.; Teichberg, S.; Basgen, J.; Wagle, D.; Shih, D.; Terlecky, I.; Bucala, R.; Cerami, A.; Egan, J.; Ulrich, P. Nature 1996, 382, 275-8. (5) Li, S.; Patapoff, T. W.; Overcashier, D.; Hsu, C.; Nguyen, T. H.; Borchardt, R. T. J. Pharm. Sci. 1996, 85, 873-7. (6) Zheng, X.; Wu, S. L.; Hancock, W. S. Int. J. Pharm. 2006, 322, 13645. (7) Andya, J. D.; Maa, Y. F.; Costantino, H. R.; Nguyen, P. A.; Dasovich, N.; Sweeney, T. D.; Hsu, C. C.; Shire, S. J. Pharm. Res. 1999, 16, 350-8. (8) Al, Abed, Y.; Mitsuhashi, T.; Li, H.; Lawson, J. A.; FitzGerald, G. A.; Founds, H.; Donnelly, T.; Cerami, A.; Ulrich, P.; Bucala, R. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 2385-90. (9) Fayle, S. E.; Healy, J. P.; Brown, P. A.; Reid, E. A.; Gerrard, J. A.; Ames, J. M. Electrophoresis 2001, 22, 1518-25. (10) Brownlee, M.; Vlassara, H.; Cerami, A. Diabetes 1980, 29, 1044-7. (11) Lapolla, A.; Fedele, D.; Martano, L.; Arico’, N. C.; Garbeglio, M.; Traldi, P.; Seraglia, R.; Favretto, D. J. Mass Spectrom. 2001, 36, 370-8. (12) Lapolla, A.; Basso, E.; Traldi, P. Adv. Clin. Chem. 2005, 40, 165-217. (13) Lapolla, A.; Fedele, D.; Seraglia, R.; Traldi, P. Mass Spectrom. Rev. 2006, 25, 775-97. (14) Lapolla, A.; Fedele, D.; Reitano, R.; Arico, N. C.; Seraglia, R.; Traldi, P.; Marotta, E.; Tonani, R. J. Am. Soc. Mass Spectrom. 2004, 15, 496-509. (15) Lapolla, A.; Tubaro, M.; Fedele, D.; Reitano, R.; Arico, N. C.; Ragazzi, E.; Seraglia, R.; Vogliardi, S.; Traldi, P. Rapid Commun. Mass Spectrom. 2005, 19, 162-8.

Analytical Chemistry, Vol. 79, No. 15, August 1, 2007 5991

peptides at high mass accuracy.6,16 Lapolla et al. reported earlier that the glycated peptides have poor MS/MS in ESI ion trap instruments.14 The fragmentation spectra predominantly showed loss of water molecules from the sugar moiety and very little fragmentation across the peptide bonds. Hence, the sequences of the glycated peptide could not be determined. Due to the difficulty in getting the MS/MS information of the glycated peptides, most of the current LC/MS techniques identify the glycated peptides relying on the mass information only. In this report, we demonstrate a novel method that combines neutral loss scan, MS, and MS/MS techniques in one experiment to identify glycation from the tryptic peptides of glycated proteins. A 3-hlong, formic acid-based, reversed-phase method was also developed to separate peptides before the neutral loss scan MS analysis. The sites of glycation in the glycated proteins can be revealed by analyzing the LC/MS and LC/MS/MS peptide mapping data. Human serum albumin (HSA) was chosen for the protein glycation study with this method because glycation sites on HSA in vivo and in vitro samples have been reported before. EXPERIMENTAL SECTION Materials. High-purity HSA, TRIS, ethylenediaminetetraacetic acid (EDTA), dithiothreitol, iodoacetic acid, and hydroxylamine were purchased from Sigma-Aldrich (St. Louis, MO). Water and acetonitrile were obtained from VWR International (West Chester, PA), and formic acid was obtained from Pierce (Rockford, IL). Synthetic peptide N-acetyl WETKAETR was purchased from AnaSpec (San Jose, CA). D-Glucose was purchased from Alfa Aesar (Ward Hill, MA), and toluene was purchased from Honeywell Burdick & Jackson (Muskegon, MI). Guanidine hydrochloride was obtained from Mallinckrodt Baker (Phillipsburg, NJ), and urea was purchased from ICN (Aurora, OH). Lyophilized trypsin was obtained from Worthington Biochemical (Lakewood, NJ). Glycation of HSA. Glycation of HSA was achieved using a previous published procedure.16 Lyophilized HSA was dissolved in 10 mM phosphate buffer, pH 7.5, containing 5 mM toluene (as bacteriostatic) and 0.5 M D-glucose to a final HSA concentration of 100 mg/mL. This solution was incubated at 37 °C for 28 days to induce glycation. The HSA control sample was prepared by dissolving lyophilized HSA into a 10 mM phosphate buffer, pH 7.5, containing 5 mM toluene to a final HSA concentration of 100 mg/mL. Glycation of the synthetic peptide was achieved by incubating the peptide in a 1.0 M D-glucose pH 7.0 buffer for 1 week at 37 °C. Reduction Alkylation and Trypsin Digestion. Both the HSA control and the HSA glycated samples were diluted to 2.0 mg/ mL in 0.5 mL of pH 7.5 denaturation buffer (7.5 M guanidine hydrochloride, 0.25 M TRIS, 2 mM EDTA). Reduction was accomplished with the addition of 5.0 µL of 0.5 M dithiothreitol. HSA control and glycated samples were incubated at 45 °C for 45 min and then cooled to room temperature. Carboxymethylation was achieved with the addition of 13.0 µL of 0.5 M iodoacetic acid. The reaction was carried out in the dark for 40 min at room temperature. Excess iodoacetic acid was quenched with the addition of 7.0 µL of 0.5 M dithiothreitol. Reduced and alkylated HSA control and glycated samples were buffer exchanged to a (16) Marotta, E.; Lapolla, A.; Fedele, D.; Senesi, A.; Reitano, R.; Witt, M.; Seraglia, R.; Traldi, P. J. Mass Spectrom. 2003, 38, 196-205.

5992

Analytical Chemistry, Vol. 79, No. 15, August 1, 2007

pH 7.5 digestion buffer (0.1 M TRIS, 1.0 M urea, and 20 mM hydroxylamine) using a NAP-5 column (GE Healthcare, Piscataway, NJ). Lyophilized trypsin was dissolved in water to a final concentration of 0.5 mg/mL. Digestion was started with the addition of 4 µL of trypsin solution to the reduced, alkylated, and buffer-exchanged samples to achieve a 1:50 enzyme to substrate ratio. Digestion was carried at 37 °C for 4 h. The final digest was quenched with the addition of 5 µL of 20% formic acid. Reversed-Phase LC. Reversed-phase separation of HSA tryptic digests was carried out on an Agilent 1100 series CapLC system (Agilent, Santa Clara, CA) equipped with a BioSuite PA-B C18 2 × 250 mm column (Waters, Milford, MA). Twenty micrograms of protein was injected onto the column. Solvent A consisted of 0.1% aqueous formic acid, and solvent B included acetonitrile and 0.1% formic acid. The temperature was maintained at 50 °C, and the flow rate was at 0.2 mL/min. A linear gradient from 0 to 40% B was run over 195 min. ESI-MS/MS. Mass spectrometric analysis was carried out on a Q-Tof micro (Waters) tandem mass spectrometer equipped with an electrospray ion source. The Q-Tof micro was operated in positive ion mode. The capillary and cone voltages were set at 3000 and 25 V, respectively. The desolvation and source temperatures were set at 350 and 100 °C, respectively. All other voltages were optimized to provide maximum signal intensity. For the neutral loss MS experiment, the following parameters were used: neutral loss peak selection high energy was 20 eV and the low energy was 5 eV; neutral loss masses used were 81.026 and 54.018; MS survey was stopped when the intensity of individual ion rising above 10 counts/s; MS/MS was stopped after 8 s regardless of the ion intensity; charge-state peak selection was chosen on doubly and triply charged ions; collision energies were defined using charge-state recognition based on the charge-state profile provide in the MassLynx software. Data Processing. Raw data files acquired from the Q-Tof micro were processed using ProteinLynx Global Server (PLGS) 2.2 software (Waters). The deisotoping and centroiding peak width was set at four channels, and the TOF resolution was set at 5000. Peptide Mass Fingerprint plus fragment ion search mode was selected for processing the LC/MS/MS data. A FASTA format database was built using the published HSA amino acid sequence and was used for the databank search. Identical data processing parameters were used for the HSA control sample peptide map and the glycated HSA peptide map in the PLGS software. RESULTS AND DISCUSSION Due to the lower ionization efficiency of the glycated peptides in mass spectrometers, instead of using trifluoroacetic acid-based mobile phases, a formic acid-based, reversed-phase LC method was developed to improve the MS signal of the glycated peptides. This reversed-phase method was used to separate the tryptic peptides of HSA samples. Figure 1a and Figure 1b show the reversed-phase mass chromatograms of the tryptic peptides of the glycated HSA sample and the HSA control sample, respectively. All the peaks that are unique to the glycated HSA sample are labeled with stars in Figure 1a. When nonenzymatic glycation occurs in proteins, the lysine residues being glycated are protected by the sugar moieties from digestion with lysine-specific enzymes such as trypsin or endoproteinase Lys-C. Hence, new peptides

Figure 1. Total ion chromatograms (TICs) for the peptides generated from a tryptic digest for the HSA control and the glycated HSA sample. (a) TIC of the glycated HSA sample peptide map; (b) TIC of the HSA control sample peptide map. Peak tops are labeled with retention times. Stars in (a) indicated the differences between the HSA control and the glycated HSA sample.

with a noncleaved glycated lysine are formed in the glycated HSA peptide map. Glycated peptides are known to lose water molecules from the glycated lysine residues under low-energy collision activation in ion trap mass spectrometers.14 In the fragmentation spectra of glycated peptides on a Q-Tof mass spectrometer, we observed the loss of the hexose moiety (162 Da) from the glycated peptides in addition to the consecutive water losses at moderate collision energies. Because this sugar moiety loss is unique to the glycated peptides, we designed a neutral loss scan experiment to screen the peptide maps of glycated proteins based on this signature neutral loss to identify the glycated peptides. The identified glycated peptides were then fragmented using standard MS/MS parameters to get the sequence as well as the glycation site information. Due to its selectivity on the glycated peptides, this method is capable of detecting and sequencing the low-intensity glycated peptides coeluting with the nonglycated peptides in the glycated HSA peptide maps. In order to get the optimized collision energy that generates the entire sugar moiety neutral loss from the glycated peptides, we performed MS/MS experiments at different collision energies on a glycated synthetic peptide with the sequence of N-acetyl WETKAETR. The N-terminus of the peptide was blocked by an acetyl group to prevent the formation of the N-terminal glycation, which is not predominant in glycated protein peptide maps. The m/z value of 612.8 for the doubly charged glycated N-acetyl WETKAETR ion was selected as the parent ion in the MS/MS experiment. This parent ion was subjected to collisionally activated dissociation at different collision energies, and the resulting MS/ MS spectra are shown in Figure 2. When 20 eV was used as the

collision energy (Figure 2c), the fragmentation spectrum of the glycated N-acetyl WETKAETR peptide showed the maximum neutral loss of the sugar moiety (162 Da). When the collision energy was increased to 30 eV (Figure 2a), the MS/MS spectrum predominantly showed b and y ions. Even though the optimum collision energy might vary for different glycated peptides to generate the sugar moiety loss, our study showed that 20-eV collision energy is sufficient to cause the neutral loss of the whole sugar moiety from most of the glycated peptides at their preferred charge states. This optimized collision energy value was used to design the neutral loss experiment for the screening of glycation in proteins. The flow chart of the neutral loss scan experiment used in this report is shown in Figure 3. All the sections shown in the figure were incorporated into one experimental setup. In the first section, all the tryptic peptides eluted from the reversed-phase column underwent low collision energy MS scan. The collision energy used in this section was 5 eV in order to prevent the fragmentation of the peptides. In the second section, the collision energy was elevated moderately to 20 eV to generate the signature neutral loss of 162 Da specifically in the glycated peptides in the MS scan mode. The value for this moderately elevated collision energy of 20 eV was chosen based on the optimization experiments described in Figure 2. In both MS scan modes, the quadrupole was set in the rf mode to pass all the peptide ions through the collision cell. Only the ions with a neutral loss of 162 Da observed in the 20-eV MS scan were selected in the quadrupole and were subjected to MS/MS experiments at higher collision energies to yield a series of b or y ions that contained the peptide sequence information. If the ion with the signature Analytical Chemistry, Vol. 79, No. 15, August 1, 2007

5993

Figure 2. MS/MS spectra for the glycated peptide: N-acetyl WETK*AETR. (a) 30-eV collision energy; (b) 25-eV collision energy; (c) 20-eV collision energy; (d) 15-eV collision energy. The signature neutral loss (162 Da) has the highest intensity in (c) when 20-eV collision energy was used.

Figure 3. Flow chart of the neutral loss scan experimental setup. All the sections in this flow chart were incorporated into one experiment.

neutral loss was not found, the mass spectrometer returned to the low-energy MS scan mode to start the next cycle. As a result of the above experimental setup, the mass spectrometer generated four ion channels of MS and MS/MS data. Figure 4 shows the four ion channels from a tryptic peptide map of the glycated HSA sample. Figure 4a is the MS scan using 5-eV collision energy, and Figure 4b is the MS scan using 20-eV collision energy. When the ions with the signature neutral losses were found at the MS scan using 20-eV collision energy, the MS/ MS mode was triggered. The mass spectrometer was set to monitor two ions simultaneously in the MS/MS mode to include two coeluting glycated peptides. Figure 4c and Figure 4d are the two MS/MS channels for the selected glycated peptides. Most of the glycated peptides from the glycated HSA peptide maps were presented as either doubly or triply charged ions in the mass spectrometer. Hence, the neutral loss values of m/z 81.026 and 54.018 were used, which represent the sugar moiety 5994

Analytical Chemistry, Vol. 79, No. 15, August 1, 2007

neutral losses from the doubly and triply charged glycated peptides. Occasionally, quadruply charged glycated peptides were also selected with these parameters because when a quadruply charged glycated peptide has two glycation sites on it, the m/z of the two sugar moiety losses is also 81.026. This helped in the identification of bigger peptides with double glycation sites. Panels a and b in Figure 5 are the partial mass spectra generated from the peaks eluting at 67.57 min in the mass chromatograms of Figure 4b and a, respectively. A collision energy of 20 eV was used in the MS scan shown in Figure 5a, and a collision energy of 5 eV was used in the MS scan shown in Figure 5b. The reversed-phase peak at 67.57 min was mainly composed of a doubly charged peptide ion with an m/z value of 815.86. As demonstrated in Figure 5b, at a collision energy of 5 eV, there was little fragmentation of the peptide. However, at a collision energy of 20 eV (Figure 5a), the peptide showed up to four consecutive losses of water molecules. Importantly, the ion with

Figure 4. Four scanning channels from the neutral loss scan experiment for the glycated HSA tryptic peptide map. (a) Total ion chromatogram at 5-eV collision energy; (b) total ion chromatogram at 20-eV collision energy; (c) MS/MS scan channel I; (d) MS/MS scan channel II. Retention times are labeled on the peak tops in (a) and (b). Retention times and parent ion m/z values are labeled on the peak tops in (c) and (d).

the complete loss of the sugar moiety (162 Da) from the peptide was also observed. This neutral loss indicates that the peptide is a glycated peptide, which led to the triggering of the MS/MS portion of the experiment. Figure 6 shows the deconvoluted MS/MS spectrum of the m/z 815.86 ion stored in MS/MS channel I eluting at 67.57 min in Figure 4c. The raw MS/MS spectrum was processed using Waters MaxEnt 3 software to obtain the deconvoluted MS/MS spectrum, which is shown in Figure 6. The collision energy used in the MS/ MS experiment for the selected ion is 34.7 eV, which was selected by the charge recognition using the collision energy profile provided by MassLynx software. Unlike the fragmentation of this ion at 20-eV collision energy, the peptide backbone fragmentation ions are the dominant species at a collision energy of 34.7 eV. A series of y ions ranging from y1 to y11 were identified in the deconvoluted MS/MS spectrum. The sequence of the peptide was identified as VTK475*CCTESLVNR with glycation occurring on lysine residue 475. Three and four water molecule losses were found in the deconvoluted MS/MS spectrum of the VTK475*CCTESLVNR peptide. We also identified the neutral losses of 84.056 Da from the glycated peptide parent ion, the y11 ion, and the y10 ion. This neutral loss could be the partial dehydrated glucose with an empirical formula of C4H4O2 (84.021 Da). All the glycation sites in the HSA samples could have been identified manually as demonstrated above. Due to the complexity of the neutral loss scan LC/MS and LC/MS/MS data, the Waters PLGS 2.2 software was used for our data analysis after some customization. First, noise reduction, centroiding, and deisotoping parameters were set up in the processing parameters section of

the PLGS software to treat the raw MS and MS/MS data. Second, a modification named Glycation, which has an addition of 162.053 Da on the lysine side chains was defined in the modifier tool of PLGS 2.2. Third, the amino acid sequence of HSA was used to build a FASTA format database for the databank searching. In this study, peptide mass fingerprint plus fragment ion search (PMF + fragment ion search) type databank search, which included both the MS and the MS/MS information from the neutral loss scan experiments, were used to identify and sequence the glycated peptides. In the databank search workflow, carboxymethyl and the glycation were considered as two possible modifications. The carboxymethyl modification was set as the fixed modification in the databank search because the HSA samples were completely reduced and alkylated. The glycation modification was set as a variable modification to avoid false identifications of glycated peptides. The neutral loss scan data of the HSA control sample peptide map was first analyzed using the PLGS software. Two glycation sites were identified even though the HSA control sample was not exposed to any reducing sugar. This indicates that the HSA protein sample provided by the vendor was partially glycated. The databank search result showed that partial glycation occurred on K205 in the peptide CASLQK205*FGER and K525 in the peptide K525*QTALVELVK. Interestingly, K525 was reported earlier as the major site of nonenzymatic glycation of HSA in vivo.17,18 However, nonenzymatic glycation on lysine residue 205 has not been (17) Garlick, R. L.; Mazer, J. S. J. Biol. Chem. 1983, 258, 6142-6. (18) Shaklai, N.; Garlick, R. L.; Bunn, H. F. J. Biol. Chem. 1984, 259, 38127.

Analytical Chemistry, Vol. 79, No. 15, August 1, 2007

5995

Figure 5. Zoomed-in regions of the MS spectra acquired for the peak at retention time 67.57 min from Figure 4c and Figure 4d. (a) Mass spectrum at 20-eV collision energy; (b) mass spectrum at 5-eV collision energy. The doubly charged peptide ion is labeled as [M + 2H]2+; the ions with 162 and 84 Da neutral losses as well as consecutive water losses are labeled accordingly.

previously reported. Since the HSA used in this study was purified from human serum, the K205 residue could be another site in HSA that is heavily susceptible to glycation in vivo. This is the first report that identifies glycation on K205 in HSA, and the glycation on this site could be important for the understanding of the physiological significance during diabetic completion and also for developing methods for the early screening of diabetes. The identification of the glycation sites on the HSA control sample indicated that the neutral loss scan method has high sensitivity and can be used for glycation site screening in the in vivo protein samples. Glycation on the amino acid residues K205 and K525 were also observed in the glycated HSA sample but with higher intensities. The PLGS databank search results for the glycated HSA tryptic digest neutral loss scan LC/MS and LC/MS/MS data using the identical processing parameters is shown in Table 1. Twenty-three glycated peptides and five nonglycated peptides were identified by the PLGS software. Although the neutral loss scan experiment targeted only the glycated peptides, five nonglycated peptides were selected and sequenced. The reason for the nonglycated peptides being selected by the neutral loss scan screening is that the threshold for the signature neutral loss was set at a low value to ensure the selection of the low-intensity glycated peptides. The low threshold setting could also lead to a false identification during the PLGS analysis if a nonglycated peptide had a minor neutral loss of 162 Da. Because the amino acid sequence of the nonglycated peptide was identified by the PLGS software, it did not lead to the false identification of a glycated peptide. By selecting the 5996 Analytical Chemistry, Vol. 79, No. 15, August 1, 2007

mass chromatograms of the glycated HSA peptide map at retention times where the glycated peptides were identified, some of the glycated peptides coeluted with nonglycated peptides were also observed, and as a result, no apparent difference was observed between the chromatograms of the tryptic peptide maps of the HSA control sample and the glycated HSA sample. If we only relied on the reversed-phase separation, we could not have identified these glycated peptides coeluted with the nonglycated peptides. After labeling all the glycated peptides peaks identified by PLGS software on the chromatogram of the glycated HSA peptide map, we observed that there were still some new peaks in the glycated HSA tryptic map that were not labeled. These new peaks are caused by glycation because they only exist in the glycated HSA sample but not in the HSA control sample. These unassigned peaks in the glycated HSA sample could have been missed by either the neutral loss scan experiment or by the PLGS software data analysis. By examining the MS scans at 20-eV collision energies, all of the unassigned glycated peptide peaks showed the glycated peptide fragmentation pattern that is similar to that of Figure 5a and most of the peptides show the signature neutral loss of 162 Da. But the unassigned peptide peaks did not have the MS/ MS scan information. The prerequisite for the triggering of the MS/MS function is the observation of the ion with a neutral loss of 162 Da in the 20-eV collision energy MS spectrum. If the signature loss ion already exists in the 5-eV collision energy MS spectrum, it will not be counted as the signature ion in the 20-eV

Figure 6. Deconvoluted MS/MS spectrum for the glycated peptide VTK475*CCTESLVNR from the glycated HSA tryptic digest neutral loss scan experiment. K475 is identified as the glycation site. The fragmentation ions are labeled accordingly on the peak tops. Table 1. PLGS Databank Searching Results for the Neutral Loss Scan LC/MS and LC/MS/MS Data of the Glycated HSA Peptides from the Tryptic Digest Rt (min)

m/z

charges

mass error ppm)

peptide sequence

33.34 54.18 51.88 54.57 64.65 65.26 67.57 68.48 69.95 71.04 74.31 76.49 78.63 82.11 91.43 121.64 122.64 125.63 127.59 136.95 138.52 142.18 143.14 147.05 154.98 162.71

782.847 750.780 555.796 680.794 729.870 538.753 815.862 858.400 855.915 679.813 913.874 694.832 609.327 645.876 749.572 736.112 1044.691 1131.023 894.933 906.999 832.463 921.477 1002.023 822.430 1003.027 1278.932

2 2 2 2 2 2 2 2 2 2 2 2 2 2 4 4 4 2 4 2 2 3 2 3 4 4

7.9 -3.7 -11.4 10.6 17.4 -14.5 8.8 -6.5 -2.1 6.2 3.5 1.7 -3.8 5.7 8.3 -9.5 -2.2 -13.7 -29.4 -23.0 -26.8 -29.3 -34.8 -30.0 -41.0 -36.5

VGSKCCKHPEAK TCVADESAENCDK LKCASLQK DVCKNYAEAK LAKTYETTLEK NECFLQHK VTKCCTESLVNR QEPERNECFLQHK LKECCEKPLLEK CASLQKFGER YKAAFTECCQAADK FKDLGEENFK KYLYEIAR KQTALVELVK LAKTYETTLEKCCAAADPHECYAK LVRPEVDVMCTAFHDNEETFLKK VHTECCHGDLLECADDRADLAKYICENQDSISSK EFNAETFTFHADICTLSEK TCVADESAENCDKSLHTLFGDKLCTVATLR AEFAEVSKLVTDLTK FYAPELLFFAKR QNCELFEQLGEYKFQNALLVR EQLKAVMDDFAAFVEK ATKEQLKAVMDDFAAFVEK VFDEFKPLVEEPQNLIKQNCELFEQLGEYK VFDEFKPLVEEPQNLIKQNCELFEQLGEYKFQNALLVR

173.53 173.72

840.810 946.787

3 4

-91.9 -54.2

MPCAEDYLSVVLNQLCVLHEK ALVLIAFAQYLQQCPFEDHVKLVNEVTEFAK

collision energy MS spectrum. Thus, the MS/MS scan will not occur. Because the bond connecting the sugar moiety of the glycated peptide is more fragile compared to the peptide bond,

glycated K residue no. 436 or 439 nonglycated 199 317 351 nonglycated 475 nonglycated 276 and 281 205 162 12 137 525 351 136 262 nonglycated 73 233 159 402 545 541 and 545 378 and 389 378 and 389 and 402 nonglycated 41

some of the glycated peptides already generate ions with the loss of the whole sugar moiety even in the 5-eV collision energy MS experiment. With the threshold for the signature loss detection Analytical Chemistry, Vol. 79, No. 15, August 1, 2007

5997

Table 2. Identified Glycation Sites from the Glycated HSA in Vitro Sample Using the Neutral Loss Scan Peptide Mapping Methoda Rt (min)

identified by

m/z

charges

mass error (ppm)

peptide sequence

32.36 33.34 43.12 44.22 51.88 54.57 64.65 67.57 69.95 71.04 74.31 76.49 78.63 82.11 91.43 100.22 95.28 109.22 116.66 120.82 121.64 122.64 127.59 130.17 136.95 138.52 140.72 142.18 143.14 143.88 143.88 147.05 152.77 154.98 156.62 159.33 162.71

manually PLGS manually manually PLGS PLGS PLGS PLGS PLGS PLGS PLGS PLGS PLGS PLGS PLGS manually manually manually manually manually PLGS PLGS PLGS manually PLGS PLGS manually PLGS PLGS manually manually PLGS manually PLGS manually manually PLGS

964.530 782.847 788.858 713.671 555.796 680.794 729.870 815.862 855.915 679.813 913.874 694.832 609.327 645.876 749.572 702.317 652.379 931.733 699.045 903.409 736.112 1044.691 894.933 977.928 906.999 832.463 1120.500 921.477 1002.023 910.936 1439.364 822.430 1338.026 1003.027 821.429 1043.180 1278.932

1 2 2 3 2 2 2 2 2 2 2 2 2 2 4 3 2 3 3 3 4 4 4 2 2 2 4 3 2 4 3 3 3 4 3 3 4

1.7 7.9 0.1 -0.9 -11.4 10.6 17.4 8.8 -2.1 6.2 3.5 1.7 -3.8 5.7 8.3 11.5 -2.0 5.4 -21.4 -3.3 -9.5 -2.2 -29.4 -18.2 -23.0 -26.8 -4.3 -29.3 -34.8 -36.8 -27.5 -30.0 -34.7 -41.0 -45.8 -34.5 -36.5

NLGKVGSK VGSKCCKHPEAK NLGKVGSKCCK NLGKVGSKCCKHPEAK LKCASLQK DVCKNYAEAK LAKTYETTLEK VTKCCTESLVNR LKECCEKPLLEK CASLQKFGER YKAAFTECCQAADK FKDLGEENFK KYLYEIAR KQTALVELVK LAKTYETTLEKCCAAADPHECYAK ADLAKYICENQDSISSK KLVAASQAALGL LVNEVTEFAKTCVADESAENCDK SLHTLFGDKLCTVATLR EFNAETFTFHADICTLSEKER LVRPEVDVMCTAFHDNEETFLKK VHTECCHGDLLECADDRADLAKYICENQDSISSK TCVADESAENCDKSLHTLFGDKLCTVATLR AVMDDFAAFVEKCCK AEFAEVSKLVTDLTK FYAPELLFFAKR SHCIAEVENDEMPADLPSLAADFVESKDVCKNYAEAK QNCELFEQLGEYKFQNALLVR EQLKAVMDDFAAFVEK SHCIAEVENDEMPADLPSLAADFVESKDVCK RPCFSALEVDETYVPKEFNAETFTFHADICTLSEK ATKEQLKAVMDDFAAFVEK LVRPEVDVMCTAFHDNEETFLKKYLYEIAR VFDEFKPLVEEPQNLIKQNCELFEQLGEYK NYAEAKDVFLGMFLYEYAR DVCKNYAEAKDVFLGMFLYEYAR VFDEFKPLVEEPQNLIKQNCELFEQLGEYKFQNALLVR

173.72 176.45 186.13

PLGS manually manually

946.787 1006.587 897.265

4 4 4

-54.2 -90.1 -80.9

ALVLIAFAQYLQQCPFEDHVKLVNEVTEFAK FKDLGEENFKALVLIAFAQYLQQCPFEDHVK DLGEENFKALVLIAFAQYLQQCPFEDHVK

glycated K residue no. 432 436 or 439 432 and 436 432, 436, 439 199 317 351 475 276 and 281 205 162 12 137 525 351 262 574 51 73 519 136 262 73 557 233 159 313 and 317 402 545 313 500 541 and 545 136 and 137 378 and 389 323 317 and 323 378 and 389 and 402 41 12 and 20 20

a The glycation sites identified by the PLGS software are marked as PLGS in the “identified by” column; the glycations sites identified manually are marked as “manually” in the same column.

being set at a low value, when the mass spectrometer detected a low-intensity ion with a 162 Da neutral loss in the 5-eV collision energy MS scan, it will ignore the higher intensity signature ion at 20-eV collision energy MS scan. The missing MS/MS information of some of the glycated peptides led to the missed identification in the PLGS software. In order to identify the unassigned glycated peptide peaks, we manually compared the low- and high-energy MS scan and searched for the signature neutral loss in the 20-eV collision energy MS scan. The glycated peptide fragmentation pattern was also considered at the same time. If the consecutive water molecule losses and the 162 Da neutral loss were found in the 20-eV collision energy scan, the peptide was considered as a glycated peptide. Table 2 shows the final result of the glycated peptide identification in the glycated HSA peptide map sample including both glycated peptides identified by the PLGS software and by manual data analysis. 5998

Analytical Chemistry, Vol. 79, No. 15, August 1, 2007

A total of 31 glycation sites were identified by the neutral loss scan method, which is more than previously reported for a glycated HSA sample prepared using the same protocol.14 The results generated using the neutral loss scan method were checked with the mass accuracy in the 5-eV collision energy MS scan. As shown in Table 2, most of the mass errors on the identified glycated peptides are within the 40 ppm range from their theoretical molecular weights. CONCLUSIONS In this study, we demonstrated a neutral loss scan RPHPLCMS method, which uses the signature loss of the reducing hexose moiety to screen for the glycated peptides and uses both MS and MS/MS information to identify peptide sequences as well as the sites of glycation. This method adds an additional dimension to the MS-only method, and it can detect even a small amount of glycated peptide coeluting with the unmodified peptide. This could be the reason for the dramatic increase in the number of

glycationsites identified in the glycated human serum albuminsample. We have observed that 31 out of 59 lysine residues in a glycated HSA in vitro sample were partially glycated. Two glycation sites in the HSA control sample on lysine residues 205 and 525, which could be the glycation sites of HSA in vivo, were also detected. The PLGS software, which is a proteomics data analysis tool, was used for the automated data analysis of the neutral scan LC/MS and LC/MS/MS data. With some customization, PLGS was able to successfully identify the glycated peptides in the glycated HSA peptide map. This neutral loss scan method showed high selectivity and sensitivity for the study of protein

glycation and could be used for complex in vivo sample glycation analysis. ACKNOWLEDGMENT The authors thank Dr. David Brems for helpful discussions and guidance.

Received for May 16, 2007.

review

March

28,

2007.

Accepted

AC070619K

Analytical Chemistry, Vol. 79, No. 15, August 1, 2007

5999