Article pubs.acs.org/ac
Rapid Identification of Low Level Glycation Sites in Recombinant Antibodies by Isotopic Labeling with 13C6-Reducing Sugars Jennifer Zhang,*,† Taylor Zhang,† Lihua Jiang,‡ Daniel Hewitt,† YungFu Huang,† Yung-Hsiang Kao,† and Viswanatham Katta† †
Protein Analytical Chemistry, Genentech Inc., 1 DNA Way, South San Francisco, California 94080, United States Thermo Fisher Scientific, 355 River Oaks Parkway, San Jose, California 95134, United States
‡
S Supporting Information *
ABSTRACT: Recombinant antibodies exhibit low levels of glycation from exposure to reducing sugars during production. As the glycation sites are typically distributed across the entire antibody, the levels at any one site are low and it becomes difficult to detect them in the conventional peptide maps. A model antibody was subjected to forced glycation by incubating with a high concentration of a 1:1 mixture of 12C6/13C6 reducing sugars with the assumption that the same sites in the native antibody will be glycated but to a lower extent. This approach simplified the detection of glycated tryptic peptide elution in the LC/MS analysis by giving a unique signature of two molecular ions with equal intensity and differing by 6.018 Da. An in-house developed script automatically processed large data files to generate a list of such peptide mass pairs. The high mass accuracy of the Orbitrap allowed us to assign the sequences unambiguously by comparison with all possible glycated peptide masses. This sequence list was subsequently used to verify their presence/absence in the digest of the native antibody. This work flow enabled rapid and confident identification of site-specific glycation even when levels are below 0.5%. We found the glycation sites to be distributed across the entire antibody studied.
R
degradation rates.4,17 Depending on the sites of glycation, these modifications could in turn impact the efficacy, stability, and toxicity. In extreme cases, glycated proteins can undergo additional oxidative-type reactions resulting in advanced glycation end products (AGEs).17,18 These products are highly reactive and can become cross-linked, structurally altered, and ultimately aggregated.19 Although typically at very low levels, glycation of recombinant antibodies (referred as mAbs) purified from mammalian cell cultures has been detected frequently in biotechnology products.4−6 Since regulatory filings require detailed characterization of the structural and chemical heterogeneity of mAbs,20 developing more sensitive methods for detecting site-specific glycation is becoming increasingly important. A number of approaches have been published for characterization of glycated proteins. These methods are mainly based on bottom-up work flows including the enrichment and isolation of glycated proteins and/or peptides with boronate affinity chromatography (BAC) followed by tandem mass spectrometry analysis.4,5,21−23 The main problem of bottom-up analysis of glycated peptides is that the collision-induced dissociation (CID)-MS/MS spectrum is dominated by the high abundance of ions corresponding to various degrees of neutral-
ecombinant antibodies are often expressed in mammalian cells that are grown in media containing various nutrients including high concentrations of reducing sugars such as glucose, a major energy source for the cells. Sometimes other sugars such as galactose are also added to modulate the glycosylation pathway.1 Typical batch mode production may involve growth of the cells for 10−14 days, long enough for a portion of the secreted antibodies to be exposed to reducing sugars under conditions suitable for a process known as nonenzymatic glycation.2 In this process, the reducing sugars react with amino groups (on either the lysine side chains or the N-terminus of the protein) to form Schiff bases and undergo Amadori rearrangement to form a more stable ketoamine product.3 It has been previously reported that the sugar feeds during upstream processes may result in glycation of antibodies.4−6 Alternatively, protein glycation may occur during subsequent formulation, long-term storage, or clinical administration steps, where sugars are commonly used as excipients in liquid or lyophilized formulations of therapeutic proteins.7−13 For example, sucrose-containing formulations have been reported to undergo accelerated degradation resulting from Maillard-like reactions.14−16 Although sucrose is a nonreducing disaccharide, it can hydrolyze into reducing sugars glucose and fructose at low pH and elevated temperature. The modification of therapeutic proteins by reducing sugars can change their net charge, target binding capacities, and © 2012 American Chemical Society
Received: November 10, 2011 Accepted: February 10, 2012 Published: February 10, 2012 2313
dx.doi.org/10.1021/ac202995x | Anal. Chem. 2012, 84, 2313−2320
Analytical Chemistry
Article
antibody digest were probed for the presence of these glycated peptides. We found that the majority of the lysines that were found to be glycated in the labeling experiment are also glycated in the native mAb though at much lower levels. The site-specific glycation level was estimated on the basis of the peak area ratio from extracted ion chromatogram of the modified peptide versus the unmodified. The new approach described here enabled glycation characterization and quantification in antibodies by a simple and sensitive process compared to the tedious data mining described by previous methods.4−6,15,32
loss (i.e., H2O, 2H2O, 3H2O, 4H2O, 3H2O + HCHO, and C6H10O5) and exhibits very limited and weak peptide backbone fragmentation.24,25 To obtain sequence information of glycated peptides, several mass spectrometric ion activation modes have been used, including electron transfer dissociation (ETD)26−28 or electron capture dissociation (ECD),29 CID in datadependent MS3 and pseudo-MS3 approaches (neutral loss scanning and multistage activation, respectively),30,31 and higher energy collisional dissociation (HCD).32 Nevertheless, in these approaches, the identification of peptide sequences is mainly achieved by automatic database search, which is greatly impacted by the quality of the product ion spectra. In our experience, the spectral quality varies substantially among peptides. The scoring values are typically low because of the poor quality of the tandem mass spectra, especially for the larger glycated peptides present at low concentration. This may reduce the confidence of peptide identification and require additional confirmatory experiments or, in some cases, glycated peptides to be completely missed. Considering that a mAb is a relatively high molecule weight protein (150k Da) and the glycation level of mAbs is known to be very low in most of cases, the current approaches limit the capability to fully characterize mAb glycation. Recently, Sanchez et al. 32 developed a method for quantitative analysis of glycated proteins in human plasma by labeling of proteins via [13 C 6 ] glucose incubation to discriminate the glycation extent under in vitro from in vivo, which are glycated with [12C6] glucose. Values of the ratio between the peak areas of the in vivo ([12C6]glucose-glycated peptides) and in vitro ([13C6]glucose-glyated peptides) from extracted ion chromatograms can be used to compare the glycation activity at different sites. However, the identification of each glycated peptide was still based on tandem mass spectrometry (HCD-MS2 and CID-MS3 neutral loss scan) and a database search using a very complex searching process. In our experience, both activation modes still have limitation on providing extensive sequence information on low abundance of glycated peptides, thus posing a challenge for data analysis. Separately, Stefanowicz et al.33 published a method for detection of glycation sites of proteins such as ubiquitin and human serum albumin by labeling the protein with an equimolar mixture of glucose and [13C6] glucose. The digested protein was subjected to HPLC fraction collection depending on the protein size. Each fraction was subsequently infused for ESI-MS analysis. An Excel spreadsheet generated from mass spectrum was used to find the pair of ions differing by 6 Da, which corresponds to the glycated peptides after isotopic labeling. The aim of this report is to detail a streamlined methodology to fully characterize and quantify low levels of glycation of a recombinant antibody. Intact mass analysis after deglycosylation was first used to assess the extent of glycation at the protein level. The antibody was then glycated in vitro by incubating with a 1:1 mixture of 12C6/13C6 reducing sugars (glucose or galactose). The resulting mixtures along with the native antibody were subjected to trypsin digestion prior to LC/MS/MS analysis on an Orbitrap mass spectrometer. To increase the data analysis speed, a script was developed that processes the deconvoluted full mass scans and automatically selects the pairs of masses that differ by 6.018 Da. This mass list was used to identify the potential glycation sites by matching these values with the predicted glycated peptide masses from the known sequence of the antibody. LC/MS data of native
■
EXPERIMENTAL SECTION Materials and Reagents. The recombinant humanized monoclonal antibody A (rhuMAb A) used in this study was produced from a Chinese hamster ovary (CHO) cell line and purified at Genentech Inc., South San Francisco, CA. The purified rhuMAb A in formulation buffer was stored at −70 °C prior to analysis. Trypsin was purchased from Roche Applied Science (Indianapolis, IN). Peptide/N-glycosidase F (PNGase F) was from New England BioLabs Inc. (Ipswich, MA). α-D-12C6glucose, α-D-13C6-glucose, α-D-12C6-galactose, and α-D-13C6galactose were procured from Sigma-Aldrich (St. Louis, MO). PNGase F Deglycosylation of rhuMAb. The rhuMAb A sample was diluted to 1 mg/mL with 50 mM Tris buffer, pH 7.5. PNGase F was added to the sample at an enzyme-tosubstrate ratio of 1:20 (w/w). The digestion was performed at 37 °C for overnight (∼15 h). Liquid Chromatography Electrospray Time-of-Flight Mass Spectrometry (LC-ESI-TOF) Analysis of rhuMAb after Deglycosylation. The molecular mass analysis of the deglycosylated rhuMAb A was performed by an Agilent 6210 electrospray-ionization Time-of-Flight (ESI-TOF) mass spectrometer following the previously described method.34 In Vitro Glycation of rhuMAb. The rhuMAb A in formulation buffer was glycated in vitro by incubating with a mixture of 12C6-glucose (0.5 M) and 13C6-glucose (0.5 M) or 12 C6-galactose (0.5 M) and 13C6-galactose (0.5 M) in a water bath at 37 °C for 72 h. Another set of rhuMAb A samples was also incubated with 1.0 M 12C6-glucose and 1.0 M 12C6galactose separately at 37 °C for 72 h for in vitro glycation. The in vitro glycated rhuMAb A samples were purified by a HiTrap protein A column (1 mL, GE Healthcare Bio-Sciences, Uppsala, Sweden) per the manufacturer’s instructions. They were then buffer exchanged into formulation buffer using a NAP-5 column (GE Healthcare, Buckinghamshire, UK). Reduction, Alkylation, and Tryptic Digestion. The rhuMAb A and purified in vitro glycated rhuMAb A samples were subjected to reduction, alkylation, and tryptic digestion. The detailed method was described previously.34 The digests were stored at −70 °C prior to analysis. Liquid Chromatography Tandem Mass Spectrometry (LC/MS/MS) Analysis of Tryptic Peptides. The tryptic peptides were separated by a reversed-phase HPLC in conjunction with a LTQ Orbitrap mass spectrometer (Thermo Scientific, Bremen, Germany). The HPLC and mass spectrometry conditions for the LC/MS/MS analysis of the peptides are the same as in the previous publication.34 Data Analysis. The full scans of the entire LC/MS/MS data set of the 12C6/13C6 in vitro glycated rhuMAb A digest were deconvoluted into neutral monoisotopic masses using the Qual Browser Xtract software (Thermo Scientific, Bremen, 2314
dx.doi.org/10.1021/ac202995x | Anal. Chem. 2012, 84, 2313−2320
Analytical Chemistry
Article
Figure 1. Deconvoluted mass spectrum of rhuMAb A after PNGase F treatment. The predicted average mass of the deglycosylated species without glycation is 145 563.6 Da. The peak at +162 Da is most likely attributed to one glycation of the antibody. The peak at +324 Da corresponds to either two glycations and/or VHS extension of the N-terminus of antibody.
Figure 2. Deconvoluted mass spectrum of (a): rhuMAb A after in vitro glycation with 12C6-glucose; (b): rhuMAb A after in vitro glycation with 12C6galactose. Both samples were deglycosylated by PNGaseF prior to MS analysis.
Germany). The S/N threshold was set at 50. All the other parameters used default values in the Xtract program. The output of deconvolution for each scan is an Excel file. All the Excel files from one LC/MS/MS data were saved in one folder and were submitted to the script written in Python language (see the Supporting Information for the script). The script looks for the mass pairs that differ by 6.018 ± 0.01 Da in each Excel file and generates a combined list. The output of the script processing is shown in the Supporting Information section. General Protein/Mass Analysis for Windows (GPMAW, Lighthouse Data, Denmark) was used to generate the predicted peptide mass list by in silico digestion for the given protein
sequence. The parameters were selected as follows: digestion enzyme, trypsin; partials, 2; mass type, monoiso.
■
RESULTS AND DISCUSSION
Discovery of rhuMAb Glycation by Intact Mass Analysis after Deglycosylation. Since the molecular mass of an antibody increases by 162 Da for glycation at each site, one can get a picture of the extent of glycation by measuring the intact molecular mass profile. However, the natural glycosylation at Asn-297 site on each heavy chain also results in heterogeneity of species, some of which differ by 162 Da in molecular mass. This complication can be eliminated by deglycosylating the antibody with PNGase-F. Figure 1 shows the deconvoluted mass spectrum of intact rhuMAb A acquired 2315
dx.doi.org/10.1021/ac202995x | Anal. Chem. 2012, 84, 2313−2320
Analytical Chemistry
Article
on ESI-TOF MS after PNGase F treatment. A major peak at 145564.7 Da was observed. The predicted average mass is 145 563.6 Da for the deglycosylated rhuMAb A after taking into account the conversion of the previously glycosylated Asparagine to Aspartic acid. The difference (∼1 Da) between the observed and predicted values is within the measurement error (∼50 ppm) for peaks in this mass range. The peak at +162 Da from the main peak is most likely attributed to nonenzymatic glycation of the protein backbone since PNGase F treatment did not eliminate this species. The peak at +324 Da can be attributed either to two glycations and/or to the VHS (Val-His-Ser) extension of the N terminus of either the light chain or the heavy chain, which resulted from the nonspecific cleavage of the signal peptidase during biosynthesis of the antibody.35 There are a total of 92 lysine residues and 2 N-terminal amines as potential sites for glycation, and the nonezymatic nature of glycation makes all of them equally possible sites. Occasionally, certain lysines are reported to be preferred sites for glycation.5 LC/MS analysis of the reduced rhuMAb A after deglycosyaltion (data not shown) suggested both the light chain and heavy chain are glycated to similar extents. Identification of Glycation Sites on rhuMAb by Isotopic Labeling Using 13C6-Reducing Sugars. To gain information on the site-specific glycation in the antibody, we developed a rapid method to identify glycation sites of recombinant antibodies using isotopic labeling with 13C6reducing sugars based on the assumption that the mechanism of glycation of antibodies in cell culture is similar to that under in vitro conditions. Glycation of rhuMAb under in Vitro Conditions. Typically, the cell culture media is maintained at approximately millimolar concentrations of reducing sugar (glucose and galactose) during the cell culture process. It has been previously shown that in vitro glycation experiments by incubating rhuMAb with glucose will force the antibody to form glycation product extensively.4,5 A previous report5 indicated that, regardless of the method of glycation (in vitro or under native conditions), the glycation hot spot of the rhuMAb examined was at the same site. In this present work, we evaluated the glycation level of rhuMAb A under in vitro conditions with glucose and galactose, using intact mass analysis after PNGase F treatment. Figure 2 shows the deconvoluted mass spectra of these two samples acquired by ESI-TOF. The high degree of glycation of rhuMAb A after incubation with glucose and galactose was evidenced by the presence of species containing up to seven 162-Da additions. It is also shown that the glycation level resulting from galactose incubation is higher than that from glucose incubation because the distribution shifted to higher mass side in the galactose-treated sample, similar to what Ledesma-Osuna et al.36 previously reported. Work flow for Detecting Glycated Peptides by Isotopic Labeling. As the glycation sites are typically distributed across the entire antibody, the levels at any one site are low and it becomes difficult to detect them in the conventional HPLC-UV peptide map. Even with LC/MS/MS, it is difficult to identify the elution of glycated peptides because the tandem mass spectra do not provide enough sequence information under CID conditions. An accepted procedure is to use forced glycation samples to obtain better quality MS/MS data for confident assignment of the glycation sites, with the assumption that the same sites will be glycated but at higher levels.11,12
On the basis of this assumption, we propose an approach to utilize the in vitro glycated samples to first determine the preferred glycation sites of rhuMAb A and then to search for these peptides in the digest of a native sample to determine the glycation sites that are present at much lower level. The complete work flow is illustrated in Figure 3. The principle of
Figure 3. Work flow for automatic detection of glycated peptides by isotopic labeling with 13C6-glucose and LC/MS analysis.
this approach is based on the differential labeling with isotopic sugars under in vitro conditions. In our in vitro glycation experiment, equimolar mixtures of 12C6/13C6 reducing sugar (glucose or galactose) are used thus producing the Amadori products (Ketosamine) with both 12C- and 13C-labeled glycation sites. The difference in monoisotopic mass between the 12C labeled and 13C labeled peptides will be 6 × (13C−12C) = 6 × (13.003−12.000) = 6.018 Da. In addition, the experimental design would result in these two species having equal abundances. In the subsequent peptide map analysis, a unique pair of ions that differ by 6.018 Da with similar intensity should be present on any peptide that contains a glycated Lys residue. This pattern can be readily recognized in the high resolution spectra even by visual inspection. Typically, different charge states are present for the same peptide. The difference between 12C and 13C labeling will be 3.009 m/z for doubly charged ions and 2.006 m/z for triply charged ions, respectively. Figure 4 shows one of the MS scans obtained in which two triply charged glycated peptides were observed. The doublet signals are 797.43/799.43 and 1167.22/1169.23 m/z units with a mass shift of 6 Da. This sample underwent incubation of the native rhuMAb A with equalmolar of “light” and “heavy” glucose; thus, the intensities of MS signals corresponding to 2316
dx.doi.org/10.1021/ac202995x | Anal. Chem. 2012, 84, 2313−2320
Analytical Chemistry
Article
Figure 4. Example of one MS scan during the elution of glycated tryptic peptides generated from the work flow. The signals corresponding to glycated peptides are enlarged to show the doublet pattern with 1:1 intensity and differing by 6.018 Da.
mostly likely background ions resulting from electrospray ionization. One way to screen them is to check if the same pair of masses is also present in the native rhuMAb A digest. Since the native sample did not undergo isotopic labeling, the presence of a doublet signal in both native and glycated sample cannot be real glycation. We found that there are a few doubly charged ion pairs in the list, for example, m/z 670.7976/ 673.8062 and m/z 679.2555/682.2641 in Table S-2 and m/z 770.8560/773.8642 and m/z 992.4904/995.4984 in both Tables S-1 and S-2 (Supporting Information), present in both native and glycated sample. We believe these doublet ions correspond to a single potassium adduct and a doubly sodiated adduct of same peptide, respectively. More specifically, the parent peptide resulting in doublet ions at m/z 670.7976/ 673.8062, m/z 679.2555/682.2641, m/z 770.8560/773.8642, and m/z 992.4904/995.4984 correspond to the doubly protonated ions at m/z 651.8246, 660.2825, 751.8829, and 973.5182, respectively. The mass difference between the singly potassiated and the doubly sodiated adduct happens to be (2 × 22.9898) − 1.0078 − 38.9637= 6.0081 Da, very close to mass difference resulting from 12C6/13C6 differential labeling. Additionally, we can utilize the unique fragmentation pattern of MS/ MS spectrum showing multiple neutral loss of water from the precursor ion for the glycated peptides to confirm the glycation on peptides from the list. After these screening steps, the number of real glycated peptides decreased to 19 for glucosetreated and 28 for galactose-treated samples, respectively. In
these two differentially labeled peptides were essentially the same. Data Processing to Automatically Detect Glycated Peptides. In order to expedite the detection of isotopically labeled glycated peptides, we have developed a work flow (see Figure 3) to process the high resolution LC/MS/MS data. We first used Qual Browser Xtract software to deconvolute all the Orbitrap high resolution full scans into neutral monoisotopic masses. In this way, each full scan spectrum was converted into an Excel file that contains the deconvoluted mass at each corresponding m/z. The multiple charge states of each peptide were deconvoluted into one single mass. Then, all these Excel files were submitted to a script, developed in-house using Python language that looks for the characteristic mass difference (6.018 ± 0.01 Da) and generates an Excel file listing all such pairs and their charge states and scan numbers in the original spectrum. The mass pairs correspond to the potentially glycated peptides. The detection sensitivity of this approach can be controlled by choosing a different signal-tonoise ratio (S/N) threshold when performing deconvolution by Xtract. Typically, S/N at 50 is sufficient to detect the glycated peptides with very high sensitivity. The output for LC/MS data of in vitro glycated with 12C6/13C6 glucose and galactose are shown in Tables S-1 and S-2 in the Supporting Information section, respectively. In this example for rhuMAb A, the Excel files contain 25 and 42 pairs of masses that are potentially glycated for glucose and galactose treated samples, respectively. It is noted that, among these masses, the singly charged ions are 2317
dx.doi.org/10.1021/ac202995x | Anal. Chem. 2012, 84, 2313−2320
Analytical Chemistry
Article
Figure 5. An example of paired tryptic peptides with 6.018 Da difference selected by the script: (a) in the in vitro glycated rhuMAb A with 1:1 mixture of 12C6/13C6 glucose; (b) same pairs were detected in the in vitro glycated rhuMAb A with 1:1 mixture of 12C6/13C6 galactose; (c) the presence of the same glycated peptide at m/z 1261.2592 (+3) was observed in the native rhuMAb A. The peak retention times in all three samples are similar.
seen in Figure 5c. It can be concluded that the peptide corresponding to this ion is glycated both natively and in vitro. Next, the accurate mass obtained by the Orbitrap can be utilized to identify the corresponding glycated peptide sequence directly. The neutral monoisotopic mass after deconvolution by Xtract for this ion is 3780.7512 Da. After subtracting the glucose mass of 162.0528 Da, the remaining mass at 3618.6984 Da will be the monoisotopic mass of the tryptic peptide. From the known amino acid sequence for rhuMAb A, the peptide sequence can be readily deduced on the basis of the trypsin specificity and mass accuracy. Herein, GPMAW program was used to generate the tryptic peptides and their corresponding monoisotopic mass lists, allowing two missed cleavages of trypsin. By comparing the theoretical mass with observed mass at 3618.6984 Da, the peptide sequence can only be VDNALQSGNSQESVTEQDSK20DSTYSLSSTLTLSK (theoretical mass at 3618.7020 Da) within the ±5 ppm mass tolerance window. Since trypsin does not cleave the lysine residue that was glycated, the glycated site can only be at Lys20. We can also estimate the approximate percentage of glycation at the corresponding lysine residue. For example, for glycated peptide VDNALQSGNSQESVTEQDSK20DSTYSLSSTLTLSK, the relative level of glycation at Lys-20 was calculated on the basis of the area of the extracted ion chromatogram of the ketoamine-modified peptide at m/z 1261.2592 divided by the total peak area of the ketoamine and
Tables S-1 and S-2 (Supporting Information), these real glycated peptides are distinguished by red color. Verification and Quantitation of Glycated Peptides in Native rhuMAb A Digest. After screening the false positive candidates, the remaining mass list can be used to determine the sequences of the observed glycated peptides by taking advantage of accurate mass capabilities to match all the potential glycated peptide masses based on the theoretical protein sequence. GPMAW was used to generate all the predicted peptide masses to match the observed glycated peptide mass after subtracting 162.0528 Da (the mass addition due to glycation). The Excel output derived from in vitro glycated samples using 12C/13C reducing sugars are used to verify if the same glycation sites occur in native mAbs by the presence or absence of the same masses in the tryptic digest of native rhuMAb A. Figure 5a shows an example that illustrates the signature ions that are labeled with 12C- and 13C-glucose. The monoisotopic mass with 12C-labeling is at m/z 1261.2576 (+3), and the monoisotopic mass with 13C-labeling is at m/z 1263.2643 (+3). Their mass difference is 6.0201 Da, which is in good agreement with the predicted value at 6.0180 Da. The same signature ions are also observed in the sample that is incubated with 12C/13Cgalactose in vitro (see Figure 5b). When we extract the 12Clabeled monoistopic mass (m/z 1261.2576) in the LC/MS data of the tryptic digest of native rhuMAb A with a ±5 ppm window, only one peak appears at a similar retention time as 2318
dx.doi.org/10.1021/ac202995x | Anal. Chem. 2012, 84, 2313−2320
Analytical Chemistry
Article
showed higher levels of glycation of BSA with D-galactose compared to D-glucose.36 Residue Lys-63 in the heavy chain is the only potential glycation site contained in the complementarity-determining regions (CDRs). Our data indicated that this site was not glycated in the native rhuMAb A, but it could be susceptible to glycation under in vitro conditions when incubating with 1 M reducing sugars. However, no difference was observed in biological activity between the native and glucose-glycated rhuMAb A under in vitro.
unmodified tryptic peptides (see Figure S-1 in the Supporting Information section). For this particular peptide, for example, the peak area of peptide VDNALQSGNSQESVTEQDSK20 was used as the unmodified tryptic peptide. Since trypsin does not cleave the Lys residues that are glycated, a drawback of this quantitation method is that we have to assume the ionization efficiency of the glycated peptides and its corresponding unmodified tryptic peptide to be the same. However, trypsin was chosen for digestion because, compared to other enzymes like Glu-C and Asp-N, the resulting glycated peptides contain the least number of Lys residues, thus making the assignment of glycation site more straightforward. The presence of multiple Lys residues in peptides from other enzyme digest would make both the site-specific identification and quantitation difficult. Table S-3 (in the Supporting Information section) lists all the glycated peptides identified using this new approach. The table includes the observed m/z for each glycated peptide, the assigned peptide sequence, and the mass errors compared to the predicted m/z as well as the percentage of glycation at each site for both native and in vitro glycated rhuMAb A. Glycation of the N-terminus of both the light chain and heavy chain was not observed. In most of cases, the peptide only contains one Lys residue, suggesting that only this glycation site is possible. However, in a number of cases, there are more than one Lys residue present in the glycated peptide sequence. This is because the presence of glucose can interfere with trypsin’s access to nearby Lys residues, resulting in an increased number of missed cleavage sites in glycated samples. Under CID experiments, it is difficult to identify the specific glycation site by MS/MS due to the preferential loss of Amadori adduct first; therefore, all the Lys residues listed are considered to be potential glycation sites The possibility of two adjacent Lys residues being glycated was also considered. By changing the mass difference from 6.018 to 12.036 Da of paired peaks in the script, we can easily determine if there is any paired peptide differing by 12 Da, which corresponds to the doubly glycated peptide. We found that none of the doubly glycated peptides was detected in both of the in vitro glycated rhuMAb A samples. Theoretically, the procedure we developed should allow us to detect the potential advanced glycation end products (AGEs) that contain the glucose degradation products. However, under our current in vitro glycation conditions (incubating with 1.0 M glucose for 72 h at 37 °C), the AGEs were not detected in our study. Several low levels of semitryptic peptides were detected, all corresponding to glycation at Lys-152 (see Table S-3, Supporting Information). Similar semitryptic products of the unmodified peptide HC aa153-aa215, cleaved at the same residues, were also present in the native rhuMAb tryptic digest, suggesting that the semitryptic products were not induced by the process of glycation but due to the atypical specificity of trypsin on this peptide. In most cases, the same glycated peptides were present in both native and in vitro glycated samples, though at different levels. The peptides observed with low levels of glycation in the in vitro glycated samples are typically found not to be glycated in the native sample. The identified glycation sites are widely distributed across the model antibody studied, most of them are at levels
