Method to Differentiate Asn Deamidation That Occurred Prior to and

Anal. Chem. 2008, 80, 9491–9498

Method to Differentiate Asn Deamidation That Occurred Prior to and during Sample Preparation of a Monoclonal Antibody Georgeen Gaza-Bulseco,† Biqin Li,‡ Ashley Bulseco,† and Hongcheng Liu*,† Process Sciences Department and Chemistry Department, Abbott Bioresearch Center, Worcester, Massachusetts 01605 Asparagine (Asn) deamidation is a major source of antibody instability and micro heterogeneity. For this reason, it is critical to accurately characterize both the levels and the sites of Asn deamidation in therapeutic antibodies. Asn deamidation is normally quantified by analyzing antibodies at the peptide level by liquid chromatography-mass spectrometry. This requires denaturation, reduction, alkylation, and enzyme digestion of the antibody prior to analysis. These steps in sample preparation may directly contribute to the total levels of Asn deamidation detected. Therefore, to obtain accurate levels and sites of Asn deamidation, it is important to determine if any deamidation occurred during the sample preparation steps. However, this could be challenging because deamidation that occurred prior to and during sample preparation resulted in peptides with the same retention times and the same molecular weight increase of 1 Da. Sample preparation was carried out in 18O-water in the current study to differentiate between the two events of Asn deamidation. Using this method, deamidation that occurred during sample preparation resulted in a molecular weight increase of 3 Da instead of 1 Da. This molecular weight difference was readily detected by inspection of the isotopic peak cluster of the peptides containing the deamidation products, isoAsp and Asp residues. It enabled discrimination of deamidation that was due to analytical artifacts and thus determination of the level of deamidation that was present in the samples.

The most common mechanism of Asn deamidation at neutral to high pH is through β-elimination, in which the main chain nitrogen of the succeeding residue attacks the Asn side chain amide group to form succinimide and ammonia. Hydrolysis of the succinimide intermediate forms either isoAsp or Asp (see Scheme 1) in a molar ratio of approximately 3:1 in model peptides.4 Susceptibility of Asn residue to deamidation is dependent on multiple factors. Asn followed by glycine (Gly) is much more susceptible to deamidation at both the peptide level 5-7 and in native proteins.5 Asn deamidation is also highly dependent on protein structure.8-10 In addition, the rate of Asn deamidation is dependent on environmental conditions such as buffer composition, ionic strength, pH, and temperature.11-13 Reversed-phase liquid chromatography coupled with mass spectrometry (LC-MS) is a method that has been commonly used to detect Asn deamidation in monoclonal antibodies.14-23 LC-MS can determine the site of Asn deamidation and the quantity of isoAsp and Asp. However, the antibodies being characterized must

Stability is a key issue for the successful development of recombinant monoclonal antibodies for use as therapeutics. Monoclonal antibodies, like other proteins, are subjected to various enzymatic and non-enzymatic modifications. Among them, non-enzymatic deamidation of Asn is a major degradation pathway.1 It also contributes to recombinant monoclonal antibody heterogeneity and instability.2,3

(14) (15)

(4) (5) (6) (7) (8) (9) (10) (11) (12) (13)

(16) (17) (18)

(19) * To whom correspondence should be addressed. E-mail: hongcheng.liu@ abbott.com. Phone: 508-849-2591. Fax: 508-793-4885. † Process Sciences Department. ‡ Chemistry Department. (1) Manning, M. C.; Patel, K.; Borchardt, R. T. Pharm. Res. 1989, 6, 903–918. (2) Liu, H.; Gaza-Bulseco, G.; Faldu, D.; Chumsae, C.; Sun, J. J. Pharm. Sci. 2008, 97, 2426–2447. (3) Wang, W.; Singh, S.; Zeng, D. L.; King, K.; Nema, S. J. Pharm. Sci. 2007, 96, 1–26. 10.1021/ac801617u CCC: $40.75  2008 American Chemical Society Published on Web 11/17/2008

(20) (21) (22) (23)

Geiger, T.; Clarke, S. J. Biol. Chem. 1987, 262, 785–794. Wright, H. T. Crit. Rev. Biochem. Mol. Biol. 1991, 26, 1–52. Stephenson, R. C.; Clarke, S. J. Biol. Chem. 1989, 264, 6164–6170. Robinson, N. E.; Robinson, A. B. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 944–949. Clarke, S. Int. J. Pept. Protein Res. 1987, 30, 808–821. Xie, M.; Schowen, R. L. J. Pharm. Sci. 1999, 88, 8–13. Kosky, A. A.; Razzaq, U. O.; Treuheit, M. J.; Brems, D. N. Protein Sci. 1999, 8, 2519–2523. Patel, K.; Borchardt, R. T. Pharm. Res. 1990, 7, 703–711. Song, Y.; Schowen, R. L.; Borchardt, R. T.; Topp, E. M. J. Pharm. Sci. 2001, 90, 141–156. Stratton, L. P.; Kelly, R. M.; Rowe, J.; Shively, J. E.; Smith, D. D.; Carpenter, J. F.; Manning, M. C. J. Pharm. Sci. 2001, 90, 2141–2148. Kroon, D. J.; Baldwin-Ferro, A.; Lalan, P. Pharm. Res. 1992, 9, 1386–1393. Perkins, M.; Theiler, R.; Lunte, S.; Jeschke, M. Pharm. Res. 2000, 17, 1110– 1117. Wang, F.; Nakouzi, A.; Alvarez, M.; Zaragoza, O.; Angeletti, R. H.; Casadevall, A. Mol. Immunol. 2006, 43, 987–998; Epub 2005 Jul 2019. Harris, R. J.; Wagner, K. L.; Spellman, M. W. Eur. J. Biochem. 1990, 194, 611–620. Harris, R. J.; Kabakoff, B.; Macchi, F. D.; Shen, F. J.; Kwong, M.; Andya, J. D.; Shire, S. J.; Bjork, N.; Totpal, K.; Chen, A. B. J. Chromatogr., B: Biomed. Sci. Appl. 2001, 752, 233–245. Lyubarskaya, Y.; Houde, D.; Woodard, J.; Murphy, D.; Mhatre, R. Anal. Biochem. 2006, 348, 24–39; Epub 2005 Oct 2025. Chelius, D.; Rehder, D. S.; Bondarenko, P. V. Anal. Chem. 2005, 77, 6004– 6011. Wang, L.; Amphlett, G.; Lambert, J. M.; Blattler, W.; Zhang, W. Pharm. Res. 2005, 22, 1338–1349; Epub 2005 Aug 1333. Liu, H.; Gaza-Bulseco, G.; Sun, J. J. Chromatogr., B Anal. Technol. Biomed. Life Sci. 2006, 837, 35–43; Epub 2006 Apr 2027. Huang, L.; Lu, J.; Wroblewski, V. J.; Beals, J. M.; Riggin, R. M. Anal. Chem. 2005, 77, 1432–1439.

Analytical Chemistry, Vol. 80, No. 24, December 15, 2008

9491

Scheme 1. Mechanism of Asn Deamidation and the Molecular Weight Difference Used to Differentiate Asn Deamidation That Occurred in the Samples Prior to Sample Preparation from That during Sample Preparationa

a

Molecular weight increase caused by incorporation of

18

O during trypsin digestion was not shown in this diagram.

first be digested to peptides for LC-MS analysis, and deamidation can occur as an artifact during sample preparation. Sample preparation involves denaturation, reduction, alkylation, and enzyme digestion under neutral to slightly basic conditions at 37 °C. The half-life of Asn followed by Gly in 0.15 M Tris, pH 7.4 at 37 °C was reported to be between 1 to 2 days.7 Therefore, it is not surprising that a significant amount of Asn deamidation can occur during the sample preparation process. Up to 70-80% Asn deamidation was detected in proteins digested with trypsin at 37 °C for 12 h in ammonium bicarbonate.24 To obtain an accurate assessment of the sites and levels of Asn deamidation in a given sample, it is important to differentiate deamidation that occurred during sample preparation from the total detected in the sample. In a recent publication, Terashima et al.25 carried out a thermal stability study of a monoclonal antibody in 18O-water. The use of 18 O water increased the molecular weight difference of Asn deamidation from 1 to 3 Da. This facilitated the identification of Asn deamidation. In this study, sample preparation was carried out in 18O-water to take advantage of the MW difference to differentiate Asn deamidation that was present in the samples prior to sample preparation from Asn deamidation that occurred during sample preparation. The advantage of this method was demonstrated by comparison of the level of Asn deamidation of a thermally stressed recombinant monoclonal antibody to the same antibody that was stored at -80 °C. (24) Krokhin, O. V.; Antonovici, M.; Ens, W.; Wilkins, J. A.; Standing, K. G. Anal. Chem. 2006, 78, 6645–6650. (25) Terashima, I.; Koga, A.; Nagai, H. Anal. Biochem. 2007, 368, 49–60; Epub 2007 May 2017.

9492


EXPERIMENTAL SECTION Reagents. H218O (>97% purity), dithiothreitol (DTT), and iodoacetic acid were purchased from Sigma (St. Louis, MO). Acetonitrile, trifluroacetic acid (TFA), guanidine hydrochloride, and 1 N hydrochloric acid were purchased from J.T. Baker (Phillipsburg, NJ). Formic acid (FA) was purchased from EMD (Madison, WI). Trypsin was purchased from Worthington (Lakewood, NJ). Thermal Stress. A recombinant fully human monoclonal antibody of approximately 70 mg/mL was diluted to 5 mg/mL using 20 mM Tris, pH 8.0. The sample was sterilized by filtering through a 0.2 µm syringe filter (Gelman Sciences, Ann Arbor, MI). Ten milliliters of the sample was incubated at 40 °C for 1 week (stressed sample). Another 10 mL was left at -80 °C to use as a control (pH 8 control), and the original sample at approximately 70 mg/mL in formulation buffer stored at -80 °C was used as a second control. Sample Preparation. One milliliter aliquots of the stressed sample and the two controls were dried by lyophilization. The three samples were resolubilized using 6 M guanidine hydrochloride, and the concentrations were measured by UV280 nm and adjusted to 10 mg/mL. Ammonium bicarbonate, pH 7.8, DTT, and water were added to each sample to result in a final ammonium bicarbonate concentration of 100 mM, DTT at a concentration of 10 mM, and guanidine hydrochloride at a concentration of 5 M. Denaturation and reduction were carried out for 30 min at 37 °C. The samples were then alkylated using iodoacetic acid. The volumes were adjusted using Milli-Q water to give a final guanidine hydrochloride concentration of 4 M and iodoacetic acid concentra-

Table 1. Amino Acid Sequences and the Theoretical Molecular Weights of Peptide 1 and Peptide 2a amino acid sequence peptide 1 GFYPSDIAVEWESNGQPENNYK peptide 2 VVSVLTVLHQDWLNGK a

mass (Da) 2543.1 1807.0

Asn residues analyzed are in bold.

tion of 25 mM. Alkylation was allowed to proceed at 37 °C for 30 min. The samples were then diluted 4 fold using Milli-Q water and digested with trypsin at a trypsin/antibody ratio of 1:20 (w: w) at 37 °C for 4 h. A portion of the same set of samples was digested using the same procedure except the reagents were prepared in 18O-water. LC-MS. An Agilent HPLC (Santa Clara, CA) and a C18 column (Vydac, 250 mm × 1 mm i.d., 5 µm particle size, 300 A pore size) were used to separate and introduce peptides into an Agilent Q-TOF LC/MS system. Approximately 10 µg of each sample was loaded at 98% mobile phase C (0.02% TFA, 0.08% FA in Milli-Q water) and 2% mobile phase D (0.02% TFA, 0.08% FA in acetonitrile), then eluted by increasing mobile phase D from 2 to 35% in 140 min. IonSpray voltage was set at 4200 V. Source temperature was set at 75 °C and m/z was scanned from 250 to 2000. RESULTS AND DISCUSSION The aim of this study was to differentiate between Asn deamidation that was present in the sample from Asn deamidation that occurred during sample preparation. As mentioned previously, Asn residues that are followed by a Gly residue are highly susceptible to non-enzymatic deamidation.5-7 Asn residues in such a sequence context was also shown to be susceptible to deamidation during sample preparation.24 The recombinant fully human monoclonal antibody used in this study contains two Asn residues followed by a Gly residue. The tryptic peptides containing these Asn residues are referred to as peptide 1 and peptide 2 as shown in Table 1 along with their monoisotopic molecular weights. Peptide 1 contains a total of three Asn residues which are referred to as Asn1, Asn2, and Asn3 from the N- to the C-terminus of this peptide. The deamidation products were named accordingly. The primary focus of this study was to analyze Asn deamidation of the two Asn residues that are followed by Gly residues, which as discussed in later sections showed significant levels of deamidation caused by sample preparation. In addition, it was interesting to observe that deamidation of Asn2 in peptide 1 was present at higher levels in the control and in the stressed sample prior to sample preparation, while relatively lower levels were introduced during sample preparation. Asn3 was not included in the study because it did not deamidate to a detectable level. When analyzed by LC-MS, no difference was observed between the two controls (-80 °C vs pH 8); therefore, we focused on analyzing only the pH 8 control and the stressed sample. Using 18O-water in the sample preparation steps enabled differentiation between Asn deamidation that was present in the samples from deamidation that occurred during sample preparation by monitoring the differential increase in the molecular weights of the resulting isoAsp and Asp containing peptides. The

strategy is outlined in Scheme 1. Asn deamidation that was present in the samples prior to sample preparation resulted in isoAsp and Asp containing peptides with a molecular weight increase of 1 Da from the conversion of the amide group to a carboxyl group. On the other hand, Asn deamidation that occurred during sample preparation resulted in isoAsp and Asp containing peptides with a molecular weight increase of 3 Da, 1 Da from the conversion of the amide group to a carboxyl group and 2 Da from incorporation of 18O into the newly formed carboxyl group when the succinimide intermediate was hydrolyzed. There is one potential limiting factor for the use of this method. It has been demonstrated previously that trypsin digestion in 18O-water resulted in the incorporation of more than one 18O into the C-terminal carboxyl group of the newly generated peptide because of the ability of the peptides to act as a pseudo trypsin substrate.26 Addition of more than one 18 O could result in a heterogeneous population of isotopic peaks, which would complicate spectrum interpretation. However, incorporation of only one 18O was observed in the current study. It was observed that incorporation of one or two 18O into the newly generated C-terminal carboxyl groups is highly dependent on the peptide itself and the composition of the buffer (study ongoing). Thus, at the time of LC-MS analysis, peptides that contained the original Asn increased in molecular weight by 2 Da. Peptides that contained isoAsp or Asp generated during thermal stress increased by 3 Da. Peptides with isoAsp and Asp generated during sample preparation increased in molecular weight by 5 Da. The 2 Da molecular weight difference between isoAsp and Asp in the samples before and after sample preparation can be used to differentiate between baseline and sample preparation derived Asn deamidation. Level of isoAsp and Asp in Peptide 1. Identification of the Deamidation Sites. Typical extracted ion chromatograms (EIC) of peptide 1 from the pH 8 control and the stressed sample, which were obtained from digestion of the antibody in Milli-Q water, are shown in Figure 1. Sample preparation in 18O-water resulted in the same EIC profiles (data not shown). Four peaks in the expected m/z range of doubly charged peptide 1 were observed. A molecular weight of 2543.1 Da, which is in agreement with the theoretical molecular weight of the peptide (Table 1), was observed for peak b. Thus, peak b contained the peptide with all of its original Asn residues. A molecular weight increase of 1 Da was observed in peaks a, c, and d, which suggested that these peaks contained peptide 1 with deamidation products, isoAsp or Asp. The order of elution of isoAsp, Asn, and Asp under similar chromatographic conditions has been well documented with isoAsp eluting first, Asn in the middle, and Asp last.15,20,21,24 Therefore, peak a contained peptide 1 with one isoAsp, and peaks c and d contained peptide 1 with one Asp. MS/MS spectra of peaks a-d from the pH 8 control were employed to localize the specific sites of deamidation (Figure 2). The observed molecular weights of y ions and the theoretical y ions were compared. Peak a contained y ions with the calculated molecular weights up to y8, while y9 showed a molecular weight increase of 1 Da, which indicated that peak a contained isoAsp mainly from deamidation of Asn 1. Peak b contained peptide 1 with the original Asn residues because all of the predicted y ion molecular weights were (26) Schnolzer, M.; Jedrzejewski, P.; Lehmann, W. D. Electrophoresis 1996, 17, 945–953.


9493

Figure 1. Extracted ion current chromatograms (EIC) of peptide 1 from the pH 8 control (pH 8 control) and the thermally stressed sample (thermal stress). The peaks are labeled as a, b, c, d, which corresponded to peptide 1 containing isoAsp, Asn, Asp2, and Asp1 as discussed in the text.

observed. This was in agreement with the earlier observation made from the MS spectrum. In peak c, there was a 1 Da molecular weight increase between y3 and y4, which localized the deamidation site to Asn 2. While, the 1 Da increase in peak d occurred between y8 and y9 and thus peak d contained Asp 1 from the deamidation of Asn 1. MS/MS spectra confirmed that the sites of deamidation in peaks a, c, and d from the thermally stressed sample were the same as observed in the pH 8 control. In summary, peak a contained mainly isoAsp1 and peak b contained the original peptide 1. Peaks c and d contained Asp2 and Asp1, respectively, in both the pH 8 control and the stressed sample. Level of Deamidation in the pH 8 Control. Three steps of calculations were required to determine the level of deamidation that was present in the pH 8 control. In step 1, the relative peak areas of isoAsp, Asp 1, and Asp 2 were calculated by dividing each individual EIC peak area by the sum of all peak areas (peaks a-d). The percentage obtained represented the total levels of isoAsp and Asp from Asn deamidation that was present in the sample and Asn deamidation that occurred during sample preparation. In step 2, the isotopic peak distribution and intensities were used to determine the proportion of Asn deamidation that was not due to sample preparation. In step 3, the percentage obtained from step 1(total) was multiplied with the percentage obtained from step 2 (prior to sample preparation) to arrive at the level of isoAsp and Asp that was present in the sample prior to sample preparation. First, the relative peak areas of isoAsp, Asp1, and Asp2 determined using EIC from Figure 1a are shown in Table 2 (control, step 1). These values represented the levels of isoAsp and Asp from Asn deamidation that occurred prior to and during sample preparation, which cannot be differentiated because of the same retention times and molecular weight increase. 9494


Second, the proportion of isoAsp and Asp from Asn deamidation that occurred prior to sample preparation was calculated using the mass spectra of peaks a-d obtained from sample preparation in 18O-water (Figure 3). The monoisotopic molecular weight of the peptide with Asn was 2545.1 Da as calculated from the doubly charged monoisotopic peak (Figure 3, Asn), which was 2 Da higher than the theoretical molecular weight of 2543.1 Da. The 2 Da increase in MW was due to the addition of 18O-water into the carboxyl group formed during peptide generation. Two different isotopic distributions of peaks were observed in the mass spectra of peaks a, c, and d (Figure 3). The first distribution series started with m/z of 1274.07, while the second started with m/z of 1275.07. The monoisotopic molecular weights corresponding to these m/z values were 2546.1 and 2548.1 Da, which were 1 and 3 Da higher than the monoisotopic molecular weight of the peptide with an Asn residue. Thus, m/z of 1274.07 corresponded to deamidation in the samples that was present prior to sample preparation, while m/z at 1275.07 was due to deamidation that occurred during sample preparation. It should be mentioned that the peak at m/z of 1275.07 overlapped with the third peak of the m/z of 1274.07 series. Nevertheless, the peak intensity at m/z of 1275.07 was much higher than the peak intensity at m/z of 1274.07 in the mass spectra from peaks a, c, and d. This result indicated that the majority of Asn deamidation that was detected in peptide 1 of the pH control occurred during sample preparation. The relative percentage of isoAsp and Asp from deamidation that occurred prior to sample preparation was calculated using the peak intensities of m/z at 1274.07 and m/z at 1275.07 taking into account the overlap of the peaks from the two isotopic distribution series. The overlap was calculated using the peak distribution in the mass spectrum of peptide 1 containing all of its original Asn residues

Figure 2. Partial MS/MS spectra of peaks a, b, c, and d of peptide 1 from the pH 8 control. Partial amino acid sequence of peptide 1 with the calculated y ions is shown above the spectra. The spectra indicated that peak a contained isoAsp from deamidation of Asn1, peak b contained peptide 1 with its original Asn residues, peak c contained Asp2, and peak d contained Asp1. The y ions indicating the site of the 1 Da molecular weight increase are in bold. Table 2. Calculation of the Percentage of IsoAsp and Asp in Peptide 1 relative percentage samples and calculation steps

determination method

peak a (isoAsp)

peak b (Asp2)

peak c (Asp1)

step 1. step 2. step 3.

pH 8 Control EIC in Figure 1 19.5 spectra in Figure 3 9.4 step 1 × step 2 1.8

1.6 50.4 0.8

5.2 13.9 0.7


Stressed Sample EIC in Figure 4 22.4 spectra in Figure 5 49.4 step 1 × step 2 11.0

9.9 90.6 9.0

3.9 37.7 1.5

(Figure 3, Asn), where the first peak (m/z, 1273.57) had a similar intensity as the third peak (m/z, 1274.57). Thus, similar intensities of the first and the third peaks of the same m/z series in the respective mass spectrum containing isoAsp or Asp were assumed. Therefore, the intensity of the peak at m/z 1275.07 was used as the total amount of isoAsp or Asp detected. The intensity of the peak at m/z of 1274.07 represented the amount of isoAsp and Asp that were present in the sample. The proportion of isoAsp and

Figure 3. Mass spectra of doubly charged ions of peptide 1 containing isoAsp (peak a), Asn (peak b), Asp2 (peak c), or Asp1 (peak d) from the pH 8 control.

Asp that was not due to sample preparation was calculated by dividing the peak intensity of m/z of 1274.07 by the peak intensity of m/z of 1275.57, which, as shown in Table 2 (control, step 2), was 9.4% isoAsp, 50.4% Asp 2, and 13.9% Asp 1. This result suggested that most of the signal detected for isoAsp and Asp1 was from deamidation that occurred during sample preparation, while a significant level of Asp2 was present in the sample prior to sample preparation. Third, the final percentage of isoAsp and Asp that were present in the pH 8 control sample was calculated using the total percentage detected from step 1 and the percentage that occurred prior to sample preparation from step 2. This resulted in 1.8% isoAsp, 0.8% Asp 2, and 0.7% Asp 1 (control, step 3 in Table 2). Analysis of the Thermally Stressed Sample. The level of Asn deamidation in the stressed sample was determined by following the same procedure as discussed in the previous section. Briefly, in the first step, the relative percentage of peaks corresponding to isoAsp, Asp 2, and Asp 1 was determined using EIC peak areas (Figure 1B), which resulted in 22.4% isoAsp, 9.9% Asp2, and 3.9% Asp 1 (Table 2, stressed, step 1). In the second step, mass spectra Analytical Chemistry, Vol. 80, No. 24, December 15, 2008

9495

Table 3. Calculation of the Percentage of IsoAsp and Asp in Peptide 2 relative percentage samples and calculation steps

Figure 4. Mass spectra of doubly charged ions of peptide 1 containing isoAsp (peak a), Asn (peak b), Asp1(peak c), or Asp2 (peak d) from the stressed sample.

Figure 5. Extracted ion current chromatograms (EIC) of peptide 2 from the pH 8 control (pH 8 control) and the thermally stressed sample (thermal stress). The peaks are labeled as a, b, and c, which corresponded to peptide 2 containing isoAsp, Asn, and Asp2 as discussed in the text. 9496


determination method

peak a (isoAsp)


pH 8 Control EIC in Figure 1 spectra in Figure 3 step 1 × step 2


Stressed Sample EIC in Figure 4 8.9 ± 0.1 spectra in Figure 5 11.7 step 1 × step 2 1.0

8.2 12.8 1.0

peak c (Asp) 7.6 14.8 1.1 8.3 ± 0.6 13.8 1.1

of peptides containing isoAsp, Asn, and Asp (Figure 4) obtained from the stressed sample by carrying out sample preparation in 18 O-water were used to determine the proportion of isoAsp and Asp that was not due to sample preparation. Calculation using the peak intensity of m/z of 1274.07 and m/z of 1275.57 resulted in 49.4% isoAsp, 90.6% Asp2, and 37.7% Asp1 (Table 2, stressed, step 2). In contrast to the pH 8 control, higher percentages of isoAsp and Asp detected were present prior to sample preparation, which indicated that thermal stress resulted in a significant increase in the level of Asn deamidation. However, in agreement with the result from the pH 8 control, the majority of Asp 2 was present prior to sample preparation, while isoAsp and Asp1 were formed prior to and during sample preparation. In the third step, the percentage from step 1 was multiplied by the percentage from step 2, which resulted in 11% isoAsp, 9.0% Asp2, and 1.5% Asp1 (Table 2, stressed, step 3), which represented the levels of isoAsp, Asp2, and Asp1 that were present in the stressed sample. The MS/MS analysis suggested that peak a of the pH 8 control and the stressed sample contained isoAsp mainly from deamidation of Asn1. However, the MS/MS spectra could be misleading as most of the isoAsp detected was due to deamidation that occurred during sample preparation, where deamidation of Asn1 contributed more than deamidation of Asn 2. Thus, the isoAsp signal, resulting mainly as an artifact from deamidation of Asn 1 during sample preparation masked the isoAsp signal from deamidation of Asn2. The ratio of isoAsp to Asp is approximately 3 to 1 as determined using model peptides.4 Assuming a 3 to 1 ratio, the percentage of isoAsp1 was calculated by taking the percentage of Asp1 and multiplying it by 3 resulting in 4.5% (1.5% × 3) of isoAsp, which should have resulted from deamidation of Asn1. Thus, the remaining 6.5% (11%-4.5%) of isoAsp should come from deamidation of Asn 2. Experimental determination of the levels of isoAsp from deamidation of Asn 1 or Asn 2 requires a separation of the peptides containing isoAsp, which was not achieved in the current study. However, it was reasonable to assume that the majority of isoAsp that was in the sample was due to deamidation of Asn2 prior to sample preparation. Finally, the level of isoAsp, Asp1, and Asp 2 that was caused by thermal stress was calculated based on the level of deamidation in the pH 8 control and the stressed sample. This was achieved by subtracting the levels of isoAsp, Asp2, and Asp1 that were present in the pH 8 control from that of the stressed sample. The final percentage was 9.2% isoAsp, 8.2% Asp2, and 0.8% Asp1.

Figure 6. Mass spectra of doubly charged ions of peptide 2 containing isoAsp (peak a), Asn (peak b), or Asp (peak c) as labeled in the figure from the pH 8 control (left, ph 8 control) and the thermally stressed sample (right, thermal stress).

Level of isoAsp and Asp in Peptide 2. The EIC of peptide 2 using Milli-Q water is shown in Figure 5. Digestion in 18O-water resulted in a similar EIC profile (data not shown). Peaks a, b, and c contained peptide 2 with isoAsp, Asn, or Asp, respectively, as determined by MS/MS results (data not shown) and the expected elution order of the peptides. The relative percentage of peptides containing isoAsp and Asp was calculated by following the same procedure employed for the analysis of peptide 1 and is summarized in Table 3. Mass spectra of peptide 2 from the control and the stressed sample obtained from digestion in 18O-water are shown in Figure 6. Inspection of the mass spectra indicated that the majority of isoAsp and Asp detected were from Asn deamidation that occurred during sample preparation. The monoisotopic molecular weight of peptide 2 with its original Asn residue was 1809.0 Da as calculated from the doubly charged monoisotopic peak (Figure 6, Asn), which was 2 Da higher than the theoretical molecular weight of 1807.0 Da. The 2 Da increase in the MW was due to the addition of 18O into the carboxyl group formed during trypsin digestion. As with peptide 1, two different isotopic distributions of peaks were observed in the isoAsp and Asp mass spectra (Figure 6, peaks a and c). The first distribution series started with m/z of 906.01, while the second started with m/z of 907.01. The monoisotopic molecular weights corresponding to these m/z

values were 1810.0 and 1812.0 Da, which were 1 and 3 Da higher than the monoisotopic molecular weight of the peptide with an Asn residue. Thus the first distribution series represented deamidation that occurred prior to sample preparation, while the second distribution series represented deamidation that occurred during sample preparation with the overlap of the third peak from the first series. Taking into account the overlap of the two m/z distribution series, the proportion of isoAsp and Asp that was not due to sample preparation was calculated to be less than 15%. The levels of isoAsp and Asp that were in the pH 8 control and the stressed samples were 1.0% and 1.1%, respectively. Furthermore, subtracting the level of isoAsp and Asp in the pH 8 control from that of the stressed sample resulted in no detectable level of Asn deamidation, which suggested that thermal stress did not cause deamidation at this Asn site. It was interesting to compare deamidation of the two Asn residues that were followed by a Gly residue with the Asn residue in peptide 1 that was not followed by a Gly residue. Susceptibility of the Asn residues that were followed by Gly in peptides 1 and 2 has been reported in several humanized recombinant monoclonal antibodies.19-21 However, as determined in the current study, none of these susceptible Asn residues was deamidated to a significant level. They were, however, highly susceptible to deamidation during sample preparation. Furthermore, the Asn Analytical Chemistry, Vol. 80, No. 24, December 15, 2008

9497

residue that was not followed by Gly was more susceptible to deamidation than the two Asn residues that were followed by Gly under thermal stress. The difference in susceptibility can be explained from the structural point of view, where the two Asn residues that were followed by Gly residues were located in β-sheet structures, while Asn2 in peptide 1 was located in a loop structure.27 This result suggested that although the amino acid on the C-terminal side of Asn has been known to have a greater effect on the rate of Asn deamidation at the peptide level, the threedimensional structure is more important in defining the susceptibility of Asn residues in native proteins. A similar strategy using 18O-water was employed to determine the level of Asn deamidation caused by sample preparation using calmodulin, ribonuclease A and lysozyme as model proteins.28 They found that a 4 h trypsin digestion process generated no additional Asn deamidation. This was not the case with the two Asn residues followed in this study where significant levels of Asn deamidation due to sample preparation were already detected at 4 h. The difference could be due to different proteins and different experimental conditions in the two studies. In addition, the employment of a sample of recombinant monoclonal antibody after thermal stress in the current study, which provided a significant level of Asn deamidation before sample preparation for LC-MS, demonstrated the power of using 18O-water to differentiate between Asn deamidation that occurred in the sample prior to and during sample preparation. (27) Lefranc, M. P.; Pommie, C.; Kaas, Q.; Duprat, E.; Bosc, N.; Guiraudou, D.; Jean, C.; Ruiz, M.; Da Piedade, I.; Rouard, M.; Foulquier, E.; Thouvenin, V.; Lefranc, G. Dev. Comp. Immunol. 2005, 29, 185–203. (28) Li, X.; Cournoyer, J. J.; Lin, C.; O’Connor, P. B. J. Am. Soc. Mass Spectrom. 2008, 19, 855–864; Epub 2008 Mar 2005.

9498


CONCLUSION Determination of the levels and sites of Asn deamidation is critical for understanding the heterogeneity and stability of recombinant monoclonal antibodies. Significant levels of Asn deamidation can occur during sample preparation for LC-MS analysis, which needs to be differentiated from the total amount of Asn deamidation detected. 18O-water was used in the reagents for sample preparation, which provided a means by MW difference to differentiate Asn deamdiation that was present in the samples prior to sample preparation from Asn deamidation that occurred during sample preparation. This enabled the determination of Asn residues that were resistant to deamidation in the native state but susceptible to deamidation in a more denatured state during sample preparation. Furthermore, a control is not necessary when using this method. This is highly advantageous since there are many situations where no controls are available. Application of this method will allow a comparison between deamidation results in monoclonal antibodies among different laboratories since trypsin digestion time or buffer composition will no longer be a factor. ACKNOWLEDGMENT The authors would like to thank Czeslaw H. Radziejewski, Gary J. Welch, and Peter Moesta for their support.

Received for review July 31, 2008. Accepted October 27, 2008. AC801617U

Method to Differentiate Asn Deamidation That Occurred Prior to and

Recommend Documents