Anal. Chem. 2010, 82, 6363–6369
Identification of Racemization Sites Using Deuterium Labeling and Tandem Mass Spectrometry Lihua Huang,* Xiaojun Lu,† P. Clayton Gough, and Michael R. De Felippis Bioproduct Research & Development, Lilly Research Laboratories, Eli Lilly and Company, Lilly Corporate Center, Indianapolis, Indiana 46285 Racemization of amino acids is a common chemical degradation pathway observed in biopharmaceuticals and is particularly prevalent in synthetic peptides. The identification of racemized amino acid residue(s) by mass spectrometry is particularly challenging due to isobaric mass between the isomeric forms. In this paper, we present a novel methodology combining stable deuterium labeling with collisionally induced dissociation-tandem mass spectrometry (CID-MS/MS) to elucidate racemized amino acid residues in immunoglobulin samples. Immunoglobulin G subclasses IgG1, IgG2, and IgG4 samples were first stressed in protonated or deuterated buffer (pH 8 or 9) at 40 or 50 °C storage for days or weeks. These forced degraded samples were reduced, S-carbamidomethylated, and digested with trypsin in protonated solution, and the tryptic digests were then analyzed via liquid chromatography/mass spectrometry (LC-MS) or sequenced via liquid chromatography/tandem mass spectrometry (LC-MS/MS) to detect racemized peptides and elucidate the location of racemized amino acid residues. The methodology successfully identified several racemized amino acid residues in the constant region of the heavy chains of the three IgG subclasses. Although the IgG subclasses have very similar primary protein sequences, our results interestingly indicated different racemization rates for specific amino acid residues. Because of the chirality of the R-carbon, all 20 amino acids with the exception of glycine have two optical isomers, designated L- and D-enantiomers. While only L-amino acids are used by the ribosome to synthesize proteins, D-amino acids are also widely observed in a variety of human and animal species due to a slow post-translational modification (PTM) known as racemization.1-3 Although this modification appears more subtle than other types of PTMs, it might significantly affect the biological activity of peptides or proteins via a change to high-order structure.1,4 For * To whom correspondence should be addressed. E-mail: huang_lihua@ lilly.com. † Current address: One MedImmune Way, Gaithersburg, MD 20878. (1) Kreil, G. Science 1994, 266, 996–997. (2) Kreil, G. J. Biol. Chem. 1994, 269, 10967–10970. (3) Fujii, N. Biol. Pharm. Bull. 2005, 28, 1585–1589. (4) Kamatani, Y.; Minakata, H.; Iwashita, T.; Nomoto, K.; In, Y.; Doi, M.; Ishida, T. FEBS Lett. 1990, 276, 95–97. Masters, P. M.; Bada, J. L.; Zigler, J. S., Jr. Nature 1977, 268, 71–73. 10.1021/ac101348w 2010 American Chemical Society Published on Web 07/12/2010
example, it has been reported that the development of cataracts seems to coincide with the racemization of L-aspartic acids in human lenses,4 and a high proportion of D-aspartate and D-serine is found in neuritic plaque amyloids of Alzheimer’s disease.5 Since the harsh conditions used for cooking and food processing could easily trigger the conversion of L-amino acids to their D-enantiomers, racemization is also a great concern for the food industry.6,7 Because of intrinsic instability, therapeutic peptides and proteins exhibit a wide range of chemical and physical transformations that have the potential to adversely impact biological, toxicological, immunological, and pharmacological properties.8,9 Racemization is one type of chemical modification that only introduces conformational change in some R-carbons, and the reaction cannot be easily predicted by either the type or the sequence of amino acids. Therefore, reliable identification of racemized amino acid residues as part of structural characterization studies is always a very challenging task, and many approaches have been employed to address this problem.10 In the most commonly used method, protein and peptide samples are first purified and hydrolyzed by hot 6 M HCl and then derivatized prior to analysis using gas-liquid chromatographic or highperformance liquid chromatographic (HPLC) techniques.11-14 Some stereoselective D-amino acid oxidase enzymes can also react with isolated D-amino acids and yield some measurable end products (H2O2, NH3, and R-keto acid).15 Although the above two methods can quantify the D-amino acid level, neither technique can provide useful information about the sites of modification. Recently, stereoselective antibodies have emerged (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15)
Shapira, R.; Austin, G. E.; Mirra, S. S. J. Neurochem. 1988, 50, 69–74. Friedman, M. J. Agric. Food Chem. 1999, 47, 3457–3479. Friedman, M. Annu. Rev. Nutr. 1992, 12, 119–137. Shah, R. R.; Midgley, J. M.; Branch, S. K. Adverse Drug React. Toxicol. Rev. 1998, 17, 145–190. Franks, M. E.; Macpherson, G. R.; Figg, W. D. Lancet 2004, 363, 1802– 1811. McCudden, C. R.; Kraus, V. B. Clin. Biochem. 2006, 39, 1112–1130. Maroudas, A.; Bayliss, M. T.; Uchitel-Kaushansky, N.; Schneiderman, R.; Gilav, E. Arch. Biochem. Biophys. 1998, 350, 61–71. Ohtani, S. Am. J. Forensic Med. Pathol. 1998, 19, 284–287. Fujii, N.; Matsumoto, S.; Hiroki, K.; Takemoto, L. Biochim. Biophys. Acta 2001, 1549, 179–187. Fujii, N.; Takemoto, L. J.; Momose, Y.; Matsumoto, S.; Hiroki, K.; Akaboshi, M. Biochem. Biophys. Res. Commun. 1999, 265, 746–751. Caldinelli, L.; Motteran, L.; Sacchi, S.; Piubelli, L.; Boselli, A.; Mothet, J. P. In D-Amino acids: a new frontier in amino acid and protein researchsPractical methods and protocols; Konno, R., Bruckner, H., D’Aniello, A., Fisher, G., Fujii, N., Eds.; Nova Science Publishers: New York, 2006.
Analytical Chemistry, Vol. 82, No. 15, August 1, 2010
6363
as powerful tools for the study of amino acid racemization.16-18 However, their general application in racemization site mapping is limited by the fact that antibodies only bind to specific epitopes. Edman degradation is often combined with chiral derivatization to identify racemization sites in peptides.19 However, this method is less useful for a complex mixture sample. Since the conversion from L-amino acids to D-enantiomers never introduces any change in mass, the application of MS for elucidation of racemized amino acids in protein and peptides is comparatively scarce. In one strategy, tandem MS (MS/MS) is employed to fragment synthesized peptides with one amino acid residue replaced by its D-enantiomer, and the acquired fragmentation ion pattern is compared with the original one to measure the impact of the substitution. For example, Adams et al. employed electron capture dissociation (ECD) to probe the tertiary structure of Trp-cage, a fast-folding protein with 20 residues, and found that the abundances of some fragment ions changed significantly when one Tyr residue in the protein was substituted by its D-enantiomer.20 Further development led to a quantitative methodology that can accurately measure the D-amino acid content in diastereomeric peptide mixtures.21 In addition, similar procedures were also established based on collisionally induced dissociation (CID) and matrix-assisted laser desorption/ionization (MALDI).21-23 No matter which mass spectrometry technique was employed, these approaches had to synthesize peptides with different D-amino acid substitution and establish the prior knowledge of the relationship between the shift in fragment ion abundance and D-amino acid content. In this work, we describe a novel procedure based on the combination of deuterium labeling and CID-tandem MS that is able to identify potential racemization sites in large proteins. To demonstrate the application of the methodology, three IgG antibodies were exposed to stressed conditions in deuterated solution and digested by Lys-C or/and trypsin, and the fragments containing racemized residues were analyzed by HPLC-MS or HPLC-MS/MS. EXPERIMENTAL SECTION Materials. IgG1, IgG2, and IgG4 antibodies were produced at Eli Lilly and Company (Indianapolis, IN). Deuterium oxide (D, 99.9%) was purchased from Cambridge Isotope Laboratories, Inc. (Andover, MA). Standard 50 mM potassium phosphate buffer, pH 8 or 100 mM potassium borate buffer 9, was purchased from Fisher Scientific (Pittsburgh, PA). Endoproteinase Lys-C was purchased from Wako Chemical USA, Inc. (Richmond, VA). Trypsin was purchased from Promega Corporation (Madison, WI). Dithiothreitol (DTT) and 8.0 M guanidine HCl were purchased (16) Bonde, M.; Qvist, P.; Fledelius, C.; Riis, B. J.; Christiansen, C. Clin. Chem. 1994, 40, 2022–2025. (17) Bonde, M.; Fledelius, C.; Qvist, P.; Christiansen, C. Clin. Chem. 1996, 42, 1639–1644. (18) Rosenquist, C.; Fledelius, C.; Christgau, S.; Pedersen, B. J.; Bonde, M.; Qvist, P.; Christiansen, C. Clin. Chem. 1998, 44, 2281–2289. (19) Iida, T.; Santa, T.; Toriba, A.; Imai, K. Biomed. Chromatogr. 2001, 15, 319– 327. (20) Adams, C. M.; Kjeldsen, F.; Zubarev, R. A.; Budnik, B. A.; Haselmann, K. F. J. Am. Soc. Mass Spectrom. 2004, 15, 1087–1098. (21) Adams, C. M.; Zubarev, R. A. Anal. Chem. 2005, 77, 4571–4580. (22) Serafin, S. V.; Maranan, R.; Zhang, K.; Morton, T. H. Anal. Chem. 2005, 77, 5480–5487. (23) Sachon, E.; Clodic, G.; Galanth, C.; Amiche, M.; Ollivaux, C.; Soyez, D.; Bolbach, G. Anal. Chem. 2009, 81, 4389–4396.
6364
Analytical Chemistry, Vol. 82, No. 15, August 1, 2010
from Pierce Biotechnology (Rockford, IL). All other chemicals were reagent grade and commercially available. Methods. Preparation of the Stressed Immunoglobulin Gamma (IgG) Antibodies. To prepare the buffers used for stress studies, sodium phosphate or potassium borate buffer, pH 8 or 9 solution was first lyophilized and reconstituted in the same volume of H2O or D2O. IgG antibodies lyophilized powders (100 µg) were then reconstituted in nondeuterated or deuterated buffers to a concentration of 1.0 mg/mL and incubated at 40 °C for 4 weeks for pH 8 samples or at 50 °C for 5 days for pH 9 samples. To fully reduce antibodies and allow the following enzyme digestion, these stressed IgG samples were lyophilized and reconstituted in 10 µL of 250 mM Tris buffer, pH 8, containing 6 M guanidine · HCl. Following this treatment, 2 µL of 500 mM DTT was added and the solutions were incubated at 37 °C for 1 h. The fully reduced antibodies were alkylated by adding 5 µL of 50 mg/mL iodoacetamide and then diluted to 200 µL with water. Enzyme Digestion. The fully reduced and alkylated antibodies were first treated with Lys-C in a 1:50 enzyme-to-antibody ratio (weight/weight) and incubated at 37 °C for 2 h, and then trypsin was added into the mixture in the same ratio. After another 2 h of incubation, the overall digestion was quenched by adding 2 µL of 50% glacial acetic acid. Reversed-Phase HPLC Chromatography. Prior to MS analysis, all fragments were separated using a Waters Acquity UPLC system (Waters, Milford, MA) equipped with a Waters Acquity UPLC BEH C18 column (2.1 × 150 mm, 1.7 µm) (Waters). The flow rate was 250 µL/min, and the column was maintained at 60 °C. Two mobile phases were prepared having the following composition: A, 0.05% aqueous trifluoroacetic acid (TFA) solution, and B, 0.04% TFA in acetonitrile. Chromatographic separation was carried out using the following gradient: 0.0% B in 2 min, 0-5%B in 4 min, 5-35% B in 84 min, 35-90% B in 1 min, 90% B for 3 min, and then rapidly returned to 0.0% B. MS Analysis. All MS experiments were performed on a Waters Synapt HDMS mass spectrometer (Waters, Milford, MA) in positive electrospray ionization mode. The instrument settings were capillary voltage of 3.2 kV, sampling cone voltage of 30.0 V, mass range of 200-2000 Da, desolvation temperature of 300 °C, cone gas flow at 30 L/h and desolvation gas flow at 700 L/h. To perform MS/MS experiments, different precursor ions were selected and fragmented with suitable trap collision energy based on their m/z value. All raw data were analyzed with Masslynx V4.1 (Waters). The average mass of each product ion was calculated according to the isotopic distribution of the singly charged ion peak. RESULTS AND DISCUSSION Racemization typically occurs at a very slow rate, and D-amino acids often accumulate within years during aging of proteins. Nevertheless, the modification can be influenced by some factors, such as protein structure, temperature, and pH. Here, an IgG antibody was incubated at 40 °C and pH 8 or 50 °C and pH 9 to accelerate the process, and some racemized residues were clearly observed after 5 days. When the conversion of L-amino acids to their D-forms occurs in a H2O solution, no change in mass will be detected. In an effort to differentiate the racemized residues, we stressed the antibody sample in a deuterated solution, so
Scheme 1. Proposed Pathway of Amino Acid Racemization in Nondeuterated (A) and Deuterated (B) Solutions
each inverted residue would incorporate a deuterium atom instead of a hydrogen atom, which gives rise to an increment in the corresponding residue mass of 1 Da (see Scheme 1 for
the principle). Following enzyme digestion and LC-MS analysis, we easily identified several fragments containing racemized amino acids. It should be recognized that during the labeling process,
Figure 1. LC-MS analysis of IgG4 heavy chain 17-30 peptide of the constant region. Extracted ion chromatograms (m/z 712-716) from total ion chromatograms for nondeuterated (a) and deuterated samples (b). The three peptide diastereomers are denoted as isomer I, isomer II, and isomer III. MS spectra of isomer I (c), II (d), and III (e) from the nondeuterated sample and MS spectra of isomer I (f), II (g), and III (h) from the deuterated sample. Analytical Chemistry, Vol. 82, No. 15, August 1, 2010
6365
Figure 2. CID tandem MS spectra (a, b) and the expansion of the 825-1150 m/z section (c, d) of the HC17-30 peptide isomer III of the constant region of the IgG4 heavy chain. The interpretation of this product ion spectrum is summarized at the top of the figure.
the protons attached to nitrogen, oxygen, and other heteroatoms would also experience deuterium exchange. However, these deuterium atoms are more labile than the ones attached to carbon atoms and are expected to exchange back to protons during the enzyme digestion step as confirmed by the mass spectrometry results. On the basis of LC-MS analysis for the tryptic digest of reduced and S-carbamidomethylated IgG4, several tryptic peptides were found to elute at different times with an identical mass (isomer) for the pH 8 or pH 9 stressed samples in H2O solutions. Among them, two tryptic peptides in the constant region of the heavy chain contained a significant amount of isomer (>10%) when the IgG4 sample was stressed in pH 8 solution at 40 °C for 4 weeks. The two peptides have the sequences: 17STSESTAALGCLVK-30 (HC17-30) and 31-DYFPEPVTVSWNSGALTSGVHTFPAVLQSSGLYSLSSVVTVPSSSLGTK-79 (HC3179). The former peptide was eluted at three positions, 43.0, 43.5 (unmodified), and 44.0 min, whereas the latter peptide was eluted at the two positions, 87.10 and 87.34 min (unmodified). Two types of major degradation pathways in protein or peptide produce isomers, i.e., the isomerization of an Asp to IsoAsp and the racemization of the L to D-enantiomer of an amino acid residue. When using D2O in place of H2O for the 6366
Analytical Chemistry, Vol. 82, No. 15, August 1, 2010
stressed studies, peptide masses found in these isomer peaks were one or more Dalton(s) higher compared with unmodified peptide masses. The peptide, HC17-30, was observed to be most heavily racemized after the stress treatment. This peptide has a theoretical molecular mass of 1422.70 Da, with an S-carbamidomethylated cysteine residue. Under the separation conditions described in the Experimental Section, this fragment is originally eluted at 43.6 min and monitored as a doubly charged ion at m/z 712.36. After 4 weeks storage at 40 °C under pH 8 H2O solution conditions, the extracted ion chromatogram for the m/z range of 712-715 clearly showed three peaks at 43.0, 43.5, and 44.0 min (Figure 1a), and careful inspection of the mass spectra (Figure 1c-1e) revealed that all of these displayed the identical isotopic pattern. Since the peptide does not contain any Asp residues, isomers were most likely produced by racemization of specific amino acid residues resulting in the formation of diastereomers. Two isomer peaks obtained for the peptide indicated the appearance of at least two additional diastereomers. To simplify the discussion, we adopted the following nomenclature for the three diastereomeric peptides and denote them as isomer I, isomer II, and isomer III as shown in Figure 1a. When the IgG4 antibody was similarly stressed in deuterated solution, almost an identical extracted ion
Figure 3. CID tandem MS spectra (a, b) and the expansions of the noted m/z section of HC17-30 peptide isomer I of the constant region of IgG4 heavy chain. The interpretation of this product ion spectrum is summarized at the top of the figure.
chromatogram was observed (Figure 1b), but the isotopic peak profiles of isomers I and III shifted to 1 Da higher due to the incorporation of deuterium (Figure 1f,h). The isotopic peak profile of isomer II shifted to higher mass due to partial deuterium incorporation (Figure 1g). Diastereomers of peptides can be generally separated by reversed-phase chromatography; although it may be difficult to completely separate all diastereomers when multiple amino acid residues are racemized. To further identify racemized residues, deuterium atoms in three D-isomers were localized using conventional CID tandem MS. MS spectra produced from isomer III are first discussed. As illustrated in Figure 2, the same y9 ions were observed regardless of whether the precursor ions were deuterated, so no deuterium was incorporated in the C-terminal segment, 22-TAALGCLVK-30. Furthermore, the deuterium labeling process appeared to introduce an increment of 1 Da for all isotopic peaks of the y10 or larger y ion and left the monoisotopic ion with very low abundance, which might be due to the tailing of intact peak (isomer II) (Figure 1h). The above results clearly indicate the presence of one deuterium atom localized to the amino acid at position 21 and leads to the interpretation that isomer III is the HC17-30 peptide sequence with a racemized Ser.
As shown in Figure 3, the tandem MS spectra of isomer I displayed a different distribution of deuterium incorporation from isomer III. The y11 or smaller y ions were the same whether the precursor ions were deuterated or not, so no deuterium was incorporated into the C-terminal segment, 20-ESTAALGCLVK-30. However, the deuterium labeling process appeared to introduce an increment of
