Article pubs.acs.org/ac
Automatic Disulfide Bond Assignment Using a1 Ion Screening by Mass Spectrometry for Structural Characterization of Protein Pharmaceuticals Sheng-Yu Huang,† Yu-Ting Hsieh,‡ Chun-Hao Chen,‡ Chao-Chi Chen,† Wang-Chou Sung,§ Min-Yuan Chou,∥ and Sung-Fang Chen*,‡ †
Mithra Biotechnology Inc., 5F, Number 79, Section 1, Sintai Fifth Road, Sijhih District, New Taipei City 221, Taiwan Department of Chemistry, National Taiwan Normal University, Number 88, Section 4, Tingchow Road, Taipei 116, Taiwan § National Institute of Infectious Diseases and Vaccinology, National Health Research Institutes, 35 Keyan Road, Zhunan, Miaoli County 350, Taiwan ∥ Biomedical Technology and Devices Research Laboratories, Industrial Technology Research Institute, 195, Section 4, Chung Hsing Road, Chutung Hsinchu 310, Taiwan ‡
S Supporting Information *
ABSTRACT: An automatic method for disulfide bond assignment using dimethyl labeling and computational screening of a1 ions with customized software, RADAR, is developed. By utilization of the enhanced a1 ions generated from labeled peptides, the N-terminal amino acids from disulfide-linked peptides can be determined. In this study, we applied this method for structural characterization of recombinant monoclonal antibodies, an important group of therapeutic proteins. In addition to a1 ion screening and molecular weight match, new RADAR is capable of confirming the matched peptide pairs by further comparing the collisioninduced dissociation (CID) fragment ions. With the N-terminal amino acid identities as a threshold, the identification of disulfide-linked peptide pairs can be achieved rapidly at a higher confidence level. Unlike most current approaches, prior knowledge of disulfide linkages or a high-end mass spectrometer is not required, and tedious work or deliberate interpretation can be avoided in this study. Our approach makes it possible to analyze unknown disulfide bonds of protein pharmaceuticals as well as their degraded forms without further protein separation. It can be used as a convenient quality examination tool during biopharmaceutical development and manufacturing processes.
S
Different approaches based on mass spectrometry for disulfide bond analysis of protein pharmaceuticals, including monoclonal antibodies, have been reported. By comparing the mass spectra of protein digests before and after reduction, possible disulfide-linked peptides of an IgG molecule can be proposed by observed molecular weight.8,9 Partial reduction/ alkylation strategy has been used for IgG2 disulfide bond characterization.10 Ion mobility mass spectrometry has been applied to separate different IgG2 isoforms with distinct disulfide bonds from each other at the hinge region.11 In recent years, tandem mass spectrometry (MS/MS or MS2) has emerged as a useful tool for disulfide bond analysis. In addition to molecular weight match of disulfide-linked peptides, MS/MS data can provide detailed information on the peptide sequences harboring these disulfide linkages. Among different MS/MS modes, ETD (electron transfer dissociation) has been applied for disulfide bond analysis because of its unique fragmentation
tructural characterization of therapeutic protein products is an important issue for the biopharmaceutical industry.1 Correct folding featuring appropriate disulfide bonds is especially of great concern because it can directly affect the efficacy and stability of protein drugs. An efficient method is required for disulfide bond investigation, not only when a therapeutic protein is newly developed but also when each lot of product is produced and its quality has to be assured. Monoclonal antibodies (mAb) represent one important category of therapeutic proteins.2,3 In the immunoglobulin G (IgG) family, multiple inter- and intrachain disulfide bonds contribute to the distinct Y-shaped structure with minor differences between each IgG subclass. Bevacizumab, an IgG1type monoclonal antibody, has been used as a therapeutic drug to treat colon, rectal, and lung cancer.4−6 Among the 16 disulfide bonds in an IgG1 molecule, there are eight and four intrachain disulfide bonds located in the heavy and light chains, respectively. For the additional four interchain disulfide bonds, two connect the heavy and light chains and the other two link up two heavy chains at the hinge region to form the dimer structure of IgG, as indicated in Figure 1. © 2012 American Chemical Society
Received: February 20, 2012 Accepted: April 20, 2012 Published: April 21, 2012 4900
dx.doi.org/10.1021/ac3005007 | Anal. Chem. 2012, 84, 4900−4906
Analytical Chemistry
Article
one enzyme may have to be used to make all peptides fall into detectable range. To date, most disulfide bond analysis approaches cannot avoid the procedures mentioned above. To develop an efficient method for minimizing laborious work would be beneficial to the biopharmaceutical industry. In this study, dimethyl labeling is applied to generate labeled peptides that exhibit enhanced a1 ions during CID fragmentation.17 For disulfide-linked peptides, multiple a1 ions can be observed due to the presence of multiple N-termini.18 Meanwhile, the identities of N-terminal amino acids can be derived from the a1 signals, which can be used as the criteria that dramatically narrow down the number of possible combinations of cysteine-containing peptides. A computational algorithm is developed to process the protein sequence as well as to read the peak list files. It can automatically match the a1 ions to the possible sequences, compare the observed m/z, and conclude the possible disulfidelinked peptides. Finally, it matches the b and y ion series as further evidence for the identification of each peptide. This approach, integrating one-pot chemical reaction with an automatic searching algorithm, provides a simple and fast method that allows unknown disulfide bonds to be solved. It can be a powerful tool to monitor lot-to-lot disulfide bond alteration and to determine disulfide bond changes under different conditions.
Figure 1. Typical structure of IgG1-type antibody.
pattern which leads to the observation of individual molecular weight of peptides linked by disulfide bonds.12 With the aid of sodium dodecyl sulfide−polyacrylamide gel electrophoresis (SDS−PAGE) separation, ETD followed by CID (collisioninduced dissociation) MS2 or MS3 was used for analyzing disulfide linkages of several therapeutic proteins, including their scrambled forms under heat stress.13 In spite of the abundant information that MS2 spectra can provide, it remains a tedious job to manually select and assign thousands of spectra generated from one liquid chromatography (LC)-MS/MS run because typical database search engines are not designed for analyzing dipeptides. Several computational programs have been developed to assist the annotation of typical CID spectra from disulfide-linked peptides. “MassMatrix” was introduced to analyze MS/MS data of disulfide-linked peptides by considering the fragment ions of backbone cleavage from both peptides.14 “DBond” further takes cysteine thioaldehyde (−2 Da), cysteine persulfide (+32 Da), and dehydroalanine (−34 Da), three sets of fragment ions that are specific for disulfide-linked peptides, into account.15 Computational algorithms alone can help to interpret the spectra of disulfide bonds; however, without an outstanding screening mechanism, it may take a long time to process all spectra. Moreover, the notable CID products generated from double cleavage of these disulfide-linked peptides have not been covered yet.16 In addition to complicated annotation of MS/MS spectra, another challenge is to deal with a protein molecule which contains numerous cysteine residues that can cause different disulfide bond possibilities. For bevacizumab, as an example, trypsin digestion should generate 15 different cysteinecontaining peptides theoretically. Due to the steric hindrance of the disulfide bonds within the IgG molecule of bevacizumab, it is possible to generate 42 or 76 cysteine-containing peptides in the case of one or two missed cleavages allowed under nonreducing conditions. If prior knowledge of disulfide structure is not available, or the determination of all disulfide bonds including scrambled ones from degraded protein is desired, generation of a huge mass list considering all possible peptide combinations cannot be avoided. If all possible charge states are taken into account, a table of m/z representing all possible disulfide-linked peptides can be obtained. Manual inspection of all mass spectra and MS/MS data is necessary to figure out the correct disulfide structure, which might be the most tedious process, not to mention the fact that more than
■
EXPERIMENTAL SECTION Materials. All solvents are chromatography-grade or better. Acetonitrile (ACN) was purchased from J. T. Baker (Center Valley, PA). Sequencing-grade modified trypsin, endoproteinase Glu-C (Staphylococcus aureus protease V8), and endoproteinase Lys-C were obtained from Promega (Madison, WI), New England Biolabs, and Roche (Indianapolis, IN), respectively. Magic C18AQ for precolumn was purchased from Michrom Bioresources (Auburn, CA). C18 for main column was purchased from Macherey-Nagel (Duren, Germany). Thermolysin, sodium cyanoborohydride, formaldehyde-d2, Nethylmaleimide (NEM), sodium acetate, triethylammonium bicarbonate (TEAB) and formic acid (FA) were purchased from Sigma−Aldrich (St. Louis, MO). Enzymatic Digestion and Dimethyl Labeling. The recombinant monoclonal antibody bevacizumab (Avastin, Genentech, CA) was diluted with 50 mM TEAB, pH 7, with or without 8 M urea, and then NEM was added at 5 mM to block free cysteines at room temperature for 30 min. Enzymatic digestion was performed in TEAB at 37 °C overnight with a trypsin/protein ratio of 1:50 (w/w) or thermolysin/protein ratio of 1:25 (w/w) or endoproteinase Lys-C/protein ratio of 1:50 (w/w). Tryptic peptides were further digested with endoproteinase Glu-C of 1:20 (w/w) enzyme/peptide ratio at 37 °C overnight. The resulting three samples, trypsin plus GluC digest, thermolysin digest, and Lys-C digest, were diluted twice with 0.1 M sodium acetate, pH 5, and subsequently for dimethyl labeling. To perform dimethyl labeling, 5 μL of 4% (w/v) formaldehyde-d2 was added to 200 μL of protein digest, followed by the addition of 5 μL of 600 mM sodium cyanoborohydride at room temperature for 30 min. Labeled peptide mixtures are ready for LC-MS/MS analysis. LC−MS Analysis. LC-MS/MS analysis was performed on either an ABI QSTAR XL mass spectrometer (Applied Biosystems, MA) with Agilent 1100 HPLC system (Agilent, CA) or a Waters CapLC system coupled to a Q-TOF 2 mass spectrometer. In the former system, the modified nanoLC 4901
dx.doi.org/10.1021/ac3005007 | Anal. Chem. 2012, 84, 4900−4906
Analytical Chemistry
Article
Figure 2. Flow chart of this study.
comprises a lab-made flow split inject system with a capillary main column (75 μm i.d., 10 cm·L, 3 μm, 100 Å) and a capillary precolumn (100 μm i.d., 2 cm·L, 5 μm, 200 Å). System setup was similar to that of Goodlett and co-workers.19 A microtee was placed between precolumn and main column; it also connected to a metal union (high voltage applied) through a short capillary (100 μm i.d., 3 cm) to induce ion spray voltage via liquid junction. Approximately 1 μg of each peptide sample was loaded with 98% buffer A (2% ACN, 0.1% FA in Milli-Q water) and 2% buffer B (0.1% FA, 2% Milli-Q water in ACN) at a flow rate of 10 μL/min. Peptides were eluted from the column with a linear gradient of increasing buffer B from 2% to 60% in 90 min with a flow rate of 200 nL/min. The IDA (information-dependent acquisition) was performed at the m/z range of 350−1600 for MS (m/z 300−500 for thermolysin digest due to smaller cysteinyl peptides generated) and 50− 2000 for MS/MS. Data Analysis. Analyst QS 1.1 (Applied Biosystems, Foster City, CA) was used to process peak lists from raw data (default precursor charge state from 2+ to 5+, reject spectra if less than 10 peaks, deisotope, centroid all MS2 spectra) obtained from QSTAR XL. MassLynx 4.0 was used to produce peak lists from raw data (subtract 30%, smooth 3/2 Savitzky Golay, and center three channels 80% centroid) obtained from Waters Q-TOF 2. New version of custom-made software RADAR18 (rapid assignment of disulfide linkage via a1 ion recognition, now open for free trial at http://www.mass-solutions.com.tw/) was used to screen a1 ions and search for the corresponding molecular weight for disulfide bond assignment. Both MGF files (Applied Biosystems) and PKL files (Waters) can be read by RADAR. The search was performed by use of the known bevacizumab sequence (http://www.drugbank.ca/) as reference database with parameters including the following: arginine, lysine, and glutamic acid were selected when trypsin plus Glu-C was employed; cleaved except proline with one or two maximum missed cleavages; lysine was selected with one maximum missed cleavage when Lys-C was employed; cleavage at the Ntermini of isoleucine, leucine, methionine, phenylalanine, tryptophan, and valine was selected for thermolysin experiment with one or two maximum missed cleavages. D-labeled,
intensity ratio cutoff 10%, a1 tolerance ±0.01 Da, mass tolerance ±0.2 Da, and maximum chain number 2 were selected for RADAR search. Higher mass tolerance was used for peptides larger than 5000 Da.
■
RESULTS AND DISCUSSION Principle of Our Method. Figure 2 depicts the study principle of our method. The target protein is first treated with N-ethylmaleimide to block free cysteines. Several enzymes can be applied to generate cysteine-containing peptides with appropriate lengths. Afterward, formaldehyde-d2 and sodium cyanoborohydride are used to perform dimethyl labeling on Nterminal amino acids and ε-amino groups of lysine residues. All steps described above can be carried out in one pot and the resulting labeled peptides are subjected to LC-MS/MS analysis with ESI-Q/TOF directly without further buffer exchange or desalting. The peak list data are then analyzed by custom-made software RADAR. Dimethyl-labeled peptides provide enhanced a1 ion signals in CID spectra. RADAR screens target a1 ions if a1 ions match the N-terminal amino acids on the list of cysteinecontaining peptides and then further search for MW combination match, otherwise ignoring it. If the MW matches, RADAR will move on to look for the corresponding b/y ions to confirm the sequence assignment. Typical ions that are commonly found in CID fragmentation of disulfide-linked peptides, b/y −2 Da (cysteine thioaldehyde), b/y +32 Da (cysteine persulfide), and b/y −34 Da (dehydroalanine), are all taken into account as well. With the identities of N-terminal amino acids as a prefilter, the computational processing time is dramatically decreased compared with those algorithms that directly search for all possible peptide combinations with molecular weights and fragment ions. The presented approach is applied for the structural analysis of two monoclonal antibodies: bevacizumab (Avastin) and trastuzumab (Herceptin); the amino acid sequences of the two antibodies are illustrated in Figure S-13 in Supporting Information. These two protein molecules, with minor sequence differences, share the typical IgG1 structure shown in Figure 1. There are four interchain and 12 intrachain disulfide bonds (total of 16) from two identical pairs of heavy 4902
dx.doi.org/10.1021/ac3005007 | Anal. Chem. 2012, 84, 4900−4906
Analytical Chemistry
Article
Table 1. Observed Peptide Pairs Linked by Disulfide Bondsa location
obsd peptide pairs
CH1, CH150−CH206
*ICN, *LGC
CH1, CH150-CH206
*LGCL, *ICN
CH2, CH267−CH327
*TPEVTCVVVDVSHEDPE, *C*K
CH2, CH267−CH327
*VTCVVVDVSHEDPEV*K, *C*K
CH3, CH373−CH431
*NQVSLTCLV*K, *WQQGNVFSCSVMHE
CH3, CH373−CH431
*LTC, *VFSCS
VH, CH22−CH96
*LSCAASGYTFTNYGMNWVR, *AEDTAVYYCA*K
CL, CL134−CL194
d-labeled a1 ions
obsd m/z
enzyme (missed cleavage)
L 118.15, I 118.15 L 118.15, I 118.15
351.68 (2+)
thermolysin (0)
408.24 (2+)
thermolysin (1)
T 106.12, C 106.06 V 104.14, C 106.06
733.38 (3+)
trypsin + Glu-C (2)
533.30 (4+)
trypsin + Glu-C (2)
N 119.11, W 191.15 L 118.15, V 104.14
713.15 (4+)
trypsin + Glu-C (0)
470.23 (2+)
thermolysin (1)
L 118.15, A 76.11
867.68 (4+)
trypsin + Glu-C (1)
*SGTASVVCLLNNFYPR, *VYACE
S 92.10, V 104.14
796.09 (3+)
trypsin + Glu-(0)
CL, CL134−CL194
*VVC, *VYACEVTHQG
thermolysin (1)
*VVC, *VYACE
104.14, 104.14 104.14, 104.14
496.57 (3+)
CL, CL134−CL194
V V V V
483.24 (2+)
thermolysin (1)
VL, CL23−CL88
*VTITCSASQDISNYLNWYQQ*KPG*K, *DFATYYCQQYSTVPWTFGQGT*K
V 104.14, D 120.09
1099.53 (5+)
H−L interchain, CH226-CL214
*SCD*K, *SFNRGEC
453.26 (3+)
Lys-C (0)
hinge, CH232−CH232, CH235− CH235
*THTCPPCPAPE, *THTCPPCPAPE
S: 92.10 S: 92.10 T 106.12, T 106.12
788.68 (3+)
trypsin + Glu-C (0)
a
trypsin + Glu-C (0)
Asterisks indicate labeling sites.
Figure 3. MS2 spectrum of m/z 796.09 (3+) derived from trypsin plus GluC experiment showing peptide pair *SGTASVVCLLNNFYPR, *VYACE linked by the disulfide bond CL134−CL194. The enhanced a1 ion signals, 92.10 and 104.14, indicate Ser and Val at the two N-termini, respectively. Due to the significant a1 ion enhancement, the m/z range above the a1 ions is increased by 4 times for clear annotation.
4903
dx.doi.org/10.1021/ac3005007 | Anal. Chem. 2012, 84, 4900−4906
Analytical Chemistry
Article
Figure 4. (A) RADAR search result of disulfide bond analysis of bevacizumab via trypsin + GluC. New features marked 1−6 are discussed in the text. (B) An illustration of the pop-up window showing matched b/y ions from the MS2 spectrum of the peptide pair *SGTASVVCLLNNFYPR, *VYACE.
and light chains. Twelve individual domains (CH1, CH2, CH3, VH, VL, CL, hinge region, H−L interchain) are generated from enzymatic digestion and assigned accordingly. To make all cysteinyl peptides fall into the detectable range, three sets of enzymestrypsin plus GluC, Lys-C, and thermolysinwere used for disulfide linkage analysis. The observed peptide pairs of bevacizumab were summarized in Table 1. In the trypsin plus
Glu-C experiment, there are six disulfide-linked peptide pairs identified that correspond to seven disulfide bonds. In the thermolysin experiment, there are five disulfide-linked peptides observed that correspond to three disulfide bonds. In the Lys-C experiment, there is one disulfide-linked peptide observed containing one disulfide bond that cannot be found in the other two enzyme digestions. It is worth noting that all disulfide 4904
dx.doi.org/10.1021/ac3005007 | Anal. Chem. 2012, 84, 4900−4906
Analytical Chemistry
Article
peptide pair, the match could be a false positive. (5) If RADAR gets a match according to the N-terminal amino acid and molecular weight match, it will further compare the fragment ions in the peak list with the assigned peptide sequences. The tolerance window as well as intensity threshold can be selected in “Fragment tolerance” and “Fragment intensity cutoff” options. The “Matched_ion” column shows the numbers of matched b and y ions. For example, y13b15 and y5b5 for peptide m/z 796.09 in Figure 4A indicate that there are 13 y ions and 15 b ions that match the first peptide sequence, SGTASVVCLLNNFYPR, and five y ions and five b ions that match the second peptide sequence, VYACE. If one doubleclicks on each matched entry, a new window will pop up and show the exact bn/yn ion assigned to the peptide sequences as illustrated in Figure 4B. Not only bn/yn ions commonly found in CID spectra but also disulfide-specific fragments such as bn/ yn − 2 Da (cysteine thioaldehyde), bn/yn + 32 Da (cysteine persulfide), and bn/yn − 34 Da (dehydroalanine) are considered by RADAR. (6) Finally, each result can be outputted as csv format for customized data processing. The peak list files generated from the LC-MS/MS raw data of bevacizumab analysis via trypsin plus GluC, thermolysin, and Lys-C digestion were directly imported to RADAR. After target a1 ion screening and molecular weight match, RADAR automatically compared the b/y ions on the CID spectra for the matched entries. In total there were 13 disulfide-linked peptide pairs that correspond to nine unique disulfide bonds identified. The highest a1 ion for each entry can be designated to one of the N-terminal amino acids of the assigned peptides. Abundant b/y ions observed from each identification increase the confidence of disulfide bond assignment. As a proof of concept, we manually validated all the matched spectra to make sure that correct results were obtained. In a routine analysis or a lot-to-lot quality control step, as long as the columns including “highest a1”and “matched ions” are carefully checked, the RADAR results can be used directly/confidently without manual inspection. The same approach was also applied to analyze another monoclonal antibody drug, trastuzumab (Herceptin). The RADAR results obtained from different enzyme experiments are illustrated in Figures S-9 to S-12 in Supporting Information.
bonds on bevacizumab can be solved automatically with this approach. The results regarding MS/MS characteristics and method automation are discussed in the following sections. Tandem Mass Spectrometric Data of Disulfide-Linked Peptide Pairs. Dimethyl labeling is known as a complete, fast, and economical chemical reaction and has been used for quantitative protein analysis.20−22 The addition of 32 Da mass difference for each derivatized site (except 16 Da for proline) is expected when formaldehyde-d2 is used for reductive amination on peptide N-termini and Lys residues. The characteristic a1 ions can be found on the CID spectra of such labeled peptides. When formaldehyde-d2 is used, an a1 tolerance window of ±0.02 Da can distinguish all 20 amino acids except for Ile and Leu. If a disulfide-linked cysteine is located at the N-terminus, there is a 2 Da deduction in MW of a1 ion because the thiol group at the cysteine residue of disulfide peptide is transformed to thiolaldehyde moiety under CID fragmentation. Figure 3 illustrates the CID-MS2 spectrum of m/z 796.09 (3+), the peptide pair *SGTASVVCLLNNFYPR, *VYACE linked by an interchain disulfide bond, CL134−CL194, on the constant region of bevacizumab light chain. Two distinct a1 ions, 92.10 and 104.14, indicate Ser and Val at the two Ntermini, respectively. Using these clues, RADAR is able to quickly locate the disulfide linkage at the two peptides, SGTASVVCLLNNFYPR and VYACE, with the molecular weight match (2385.08 Da). The remaining b and y ions further confirmed the assignment of the peptide sequences. Other MS/MS data regarding all the disulfide-linked peptides of bevacizumab are shown in the Supporting Information. Automatic Disulfide Bond Assignment by RADAR. The aim of this study is to develop an automatic method that is suitable for routine disulfide bond analysis. The intent is to avoid trying all molecular weight combinations and performing manual interpretation of each spectrum. Figure 4A shows the layout of new RADAR as well as the result obtained from bevacizumab analysis via trypsin plus GluC digestion. The work principle of RADAR and its basic functions have been described and discussed previously.18 In this updated version, new features are included to facilitate the automatic disulfide bond analysis: (1) The input of protein sequence information can be achieved either by using a plain text file or by direct typing. The “peptide sequence” option offers the possibility to import peptide lists manually, which is useful for non-enzyme peptide studies. (2) To generate cysteinyl peptide lists, RADAR now supports experiments with enzymes such as thermolysin and endoproteinase Asp-N that cleave at the N-terminal side of amino acids. As illustrated in Figure S-8 in Supporting Information, the cleavage sites of thermolysin (I, L, M, F, W, and V) can be selected in both “Enable” and “Before” columns. This new function is useful because combining different enzymes to hydrolyze peptides into detectable mass range is a general strategy for comprehensive disulfide bond analysis. It helps to separate the cysteine residues onto different peptides to avoid ambiguous assignment. This flexible option can also be played around when the enzyme specificity is questionable. (3) New columns including mass (experimental), mass (calculated), and Δmass are added. They indicate the molecular weight derived from experiment, theoretical molecular weight, and the difference between these two, respectively. (4) The “highest a1” for each entry is shown. Instead of showing the highest a1 ion among the target a1 ions, it is defined as the highest a1 from 20 amino acids. If the highest a1 ion does not represent one of the N-terminal amino acids from the matched
■
CONCLUSIONS The approach presented here is simple and convenient without the need for prior knowledge of disulfide bonds, minimizing the spectra inspection effort significantly. By combination of the identities of N-terminal amino acids via a1 ions and an automatic search algorithm, the confidence in disulfide bond assignment is greatly enhanced. The whole search for one peak list takes only seconds because the a1 ion screening is used as the first criterion that already eliminates many possibilities. It is then not necessary to compare all the molecular weight combinations of all cysteinyl peptides and try all the b/y ions for peptide sequence assignment. We believe this proposed method can be extremely helpful for routine structural characterization of protein products and can facilitate the progress of the biopharmaceutical industry.
■
ASSOCIATED CONTENT
S Supporting Information *
Thirteen figures showing MS/MS spectra of disulfide-linked peptides of bevacizumab identified by RADAR; RADAR search result of disulfide bond analysis of bevacizumab with 4905
dx.doi.org/10.1021/ac3005007 | Anal. Chem. 2012, 84, 4900−4906
Analytical Chemistry
Article
thermolysin; RADAR search results of disulfide bond analysis of trastsuzumab with trypsin + Glu-C, thermolysin, Lys-C, and trypsin; and amino acid sequences of bevacizumab and trastuzumab. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail
[email protected]; fax +886-2-29324249. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This work was supported by grants from the National Science Council of Taiwan, under Contracts NSC 98-2113-M-003-007MY2 and NSC 100-2113-M-003-002-MY2, and by Small Business Innovation Research (SBIR, Taiwan) Program 3Z990095.
■
REFERENCES
(1) Srebalus Barnes, C. A.; Lim, A. Mass Spectrom. Rev. 2007, 26, 370. (2) Harris, R. J.; Shire, S. J.; Winter, C. Drug Dev. Res. 2004, 61, 137. (3) Owens, S. M. Drug Dev. Res. 2004, 61, 107. (4) Ferrara, N.; Hillan, K. J.; Gerber, H. P.; Novotny, W. Nat. Rev. Drug Discovery 2004, 3, 391. (5) Ferrara, N.; Hillan, K. J.; Novotny, W. Biochem. Biophys. Res. Commun. 2005, 333, 328. (6) Kerr, D. J. Nat. Clin. Pract. Oncol. 2004, 1, 39. (7) Gorman, J. J.; Wallis, T. P.; Pitt, J. J. Mass Spectrom. Rev. 2002, 21, 183. (8) Zhang, W.; Marzilli, L. A.; Rouse, J. C.; Czupryn, M. J. Anal. Biochem. 2002, 311, 1. (9) Mhatre, R.; Woodard, J.; Zeng, C. H. Rapid Commun. Mass Spectrom. 1999, 13, 2503. (10) Martinez, T.; Guo, A.; Allen, M. J.; Han, M.; Pace, D.; Jones, J.; Gillespie, R.; Ketchem, R. R.; Zhang, Y. L.; Balland, A. Biochemistry 2008, 47, 7496. (11) Bagal, D.; Valliere-Douglass, J. F.; Balland, A.; Schnier, P. D. Anal. Chem. 2010, 82, 6751. (12) Wu, S. L.; Jiang, H. T.; Lu, Q. Z.; Dai, S. J.; Hancock, W. S.; Karger, B. L. Anal. Chem. 2009, 81, 112. (13) Wang, Y.; Lu, Q. Z.; Wu, S. L.; Karger, B. L.; Hancock, W. S. Anal. Chem. 2011, 83, 3133. (14) Xu, H.; Zhang, L. W.; Freitas, M. A. J. Proteome Res. 2008, 7, 138. (15) Choi, S.; Jeong, J.; Na, S.; Lee, H. S.; Kim, H. Y.; Lee, K. J.; Paek, E. J. Proteome Res. 2010, 9, 626. (16) Clark, D. F.; Go, E. P.; Toumi, M. L.; Desaire, H. J. Am. Soc. Mass Spectrom. 2011, 22, 492. (17) Hsu, J. L.; Huang, S. Y.; Shiea, J. T.; Huang, W. Y.; Chen, S. H. J. Proteome Res. 2005, 4, 101. (18) Huang, S. Y.; Wen, C. H.; Li, D. T.; Hsu, J. L.; Chen, C.; Shi, F. K.; Lin, Y. Y. Anal. Chem. 2008, 80, 9135. (19) Yi, E. C.; Lee, H.; Aebersold, R.; Goodlett, D. R. Rapid Commun. Mass Spectrom. 2003, 17, 2093. (20) Huang, S. Y.; Tsai, M. L.; Wu, C. J.; Hsu, J. L.; Ho, S. H.; Chen, S. H. Proteomics 2006, 6, 1722. (21) Hsu, J. L.; Huang, S. Y.; Chow, N. H.; Chen, S. H. Anal. Chem. 2003, 75, 6843. (22) Ji, C. J.; Guo, N.; Li, L. J. Proteome Res. 2005, 4, 2099.
4906
dx.doi.org/10.1021/ac3005007 | Anal. Chem. 2012, 84, 4900−4906
