Direct Approach for Qualitative and Quantitative Characterization of

Article pubs.acs.org/ac

Direct Approach for Qualitative and Quantitative Characterization of Glycoproteins Using Tandem Mass Tags and an LTQ Orbitrap XL Electron Transfer Dissociation Hybrid Mass Spectrometer Hongping Ye,*,† Michael T. Boyne, II,† Lucinda F. Buhse,† and John Hill‡,§ †

U.S. Food and Drug Administration, CDER, DPA, St. Louis, Missouri 63101, United States U.S. Food and Drug Administration, CDER, ONDQA/DPAII, Silver Spring, Maryland 20993, United States

‡

S Supporting Information *

ABSTRACT: The application of multiplexed isobaric tandem mass tag (TMT) labeling and an LTQ Orbitrap XL ETD (electron transfer dissociation) hybrid mass spectrometer as a direct approach for qualitative and quantitative characterization of glycoproteins is reported. Bovine fetuin was used as a model glycoprotein in this study. For online liquid chromatography− mass spectrometry (LC−MS) analysis, high-resolution, mass accurate full scan MS spectra were acquired in the Orbitrap mass analyzer followed by data-dependent tandem mass spectrometry (MS/MS) with alternating collision-induced dissociation (CID), ETD, and higher-energy collisional dissociation (HCD) scans. An additional in-source dissociation scan was used as a highly sensitive and selective detection method for eluting glycosylated peptides. By alternatively using three different dissociation methods, 23 glycoforms from all 5 corresponding glycopeptides were identified from a trypsin digest of bovine fetuin. With ETD, labile glycans were retained without any signs of carbohydrate cleavage with concurrent fragmentation of the peptide backbone. Glycosylation sites were clearly localized from the ETD fragmentation data. Glycan structure elucidation was accomplished using CID. The CID experiments generated fragment ions predominantly from cleavage of glycosidic bonds without breaking the peptide bond. Novel to this method, the TMT labeling protocol was modified and adapted for higher labeling efficiency, and a TriVersa NanoMate was used to reinfuse samples to improve ETD and HCD spectra of glycopeptides. Quantification with TMT was verified based on the HCD spectra from multiple nonglycopeptides and glycopeptides. This method can be used as a qualitative and quantitative technique for direct characterization of glycoproteins and has applicability for detection of counterfeit glycoprotein drug products.

T

nonsimilar could then be subject to additional chemical and computational analysis in order to more fully elaborate the scientific nature of the difference. For these reasons, characterization of glycoproteins is of paramount importance.5,11 Glycosylated proteins are complex molecules. The glycosylation modifications are highly labile and most often highly heterogeneous.18 Characterization of glycoproteins remains a great analytical challenge. To adequately characterize glycoproteins and compare the similarities and differences of glycoprotein drugs, one needs to know: (1) if the proteins and peptides are glycosylated, (2) the site(s) of the glycosylation and amino acid sequence, (3) the structure of the glycan, and (4) the relative amount of each glycoform at a given glycosylation site (glycopeptides). Traditionally, glycoprotein analysis involves special enrichment procedures, enzymatic removal of glycans, and separate identification of glycans and amino acid sequence. Most of those methods involve a lengthy process and may lose glycosylation site mapping information. The applications of

herapeutic proteins represent a significant and growing section of the pharmaceutical industry, many of which are glycoproteins. Glycosylation can potentiate biological activity, regulate the rate of clearance of the protein from the circulatory system and influence the potential antigenicity of the protein.1−6 Protein glycosylation is a complex post-translational modification. In a pharmaceutical manufacturing setting, external factors such as choice of bioreactor, chemical conditions in the bioreactor, nutrient addition timing and feed rates, temperature, agitation rates, etc. can affect the glycosylation process.7−10 Changes to the glycosylation pattern have the potential to significantly alter a protein’s therapeutic efficacy and can potentially give rise to life threatening adverse events.11−15 Continual monitoring and control of recombinant therapeutic proteins for changes in their glycosylation patterns is critical for drug safety and efficacy. In addition, with the recent passage of the affordable health care laws there has been considerable interest from the pharmaceutical industry in developing biosimilar therapeutic glycoproteins. In a regulatory context of demonstrating biosimilarity, the development of a method for rapidly identifying gross differences in the glycosylation pattern of a test sample and a reference sample could simplify the initial assessment of biosimilarity.16,17 Samples identified as being This article not subject to U.S. Copyright. Published 2012 by the American Chemical Society

Received: September 19, 2012 Accepted: December 18, 2012 Published: December 18, 2012 1531

dx.doi.org/10.1021/ac3026465 | Anal. Chem. 2013, 85, 1531−1539

Analytical Chemistry

■

these methods as well as their advantages and disadvantages were extensively discussed.19,20 Recently, mass spectrometry has been used to directly characterize glycoproteins.21−28 However, these approaches either lacked site-specific glycoform quantification, amino acid sequence or were unable to provide glycosylation site mapping information. Collision-induced dissociation (CID) has limitations for determining the modification site due to the labile nature of the glycan attachment to the peptide ion. CID predominantly generates fragment ions from cleavage of glycosidic bonds without breaking amide bonds. These fragments can be used to elucidate oligosaccharide sequence21,23,29 but do not produce any peptide sequence information. In contrast to CID, electron transfer dissociation (ETD) preserves labile post-translational modifications (PTMs) and allows not only the identification of the amino acid sequence of a glycopeptide but also the assignment of its glycosylation sites.22,23,30−32 The isobaric stable isotope tagging technologies including tandem mass tags (TMT) and isobaric tags for relative and absolute quantification (iTRAQ) have become commercially available and been widely used for protein quantification.33−39 However, these applications were limited to unmodified proteins because of the following reasons. First, it is difficult to identify and characterize minor glycopeptides efficiently from samples containing large amounts of nonglycosylated peptides produced by proteolytic digestion of proteins. Second, the glycans might make the labeling reaction more complicated, and optimized labeling conditions are required to reach high labeling efficiency. Third and more importantly, due to the labile nature of the attached glycans, application of CID dissociation energy will cleave the glycosidic bonds instead of reporter ions, making quantification unfeasible. In the work done by Viner et al., single O-linked N-acetylglycesamine (O-GlcNAc) modified peptides were labeled with TMT and quantified mainly by using ETD.36 While Viner et al. demonstrated the applicability of TMT for quantification of this simple glycan structure, the quantification of complex glycoform structures have yet to be elucidated. On the other hand, because ETD is a nonergodic fragmentation process, the quantitative precision and accuracy are not as good as with CID. As was previously reported,36,38,39 quantification by HCD provides more accurate and precise results than ETD and pulsed Q dissociation (PQD). Zhang et al.40 reported the quantification of glycopeptides and nonglycopeptides for glycoprotein digests using iTRAQ plus 18O. However, the glycoform localization and quantification were lost because the glycans were cleaved by PNGase F. Recent progress in quantitative glycoproteomics has been extensively reviewed.41 Most of these methods require special glycopeptide enrichment steps before characterization. In this paper we report an improved TMT labeling protocol for glycoproteins, offering much higher labeling efficiency without acetone precipitation. For the first time,42 TMT labeled glycoprotein digests were directly qualitatively and quantitatively characterized using a combination of ETD, CID, and HCD. Bovine fetuin was used as a model glycoprotein in the study because of its well-defined sequence and glycan structure. For trypsin digest of bovine fetuin, all five glycosylated peptides were analyzed with 23 glycopeptides identified. The total protein sequence coverage was 100%. The quantification precision and accuracy of 10 glycoforms were extensively evaluated and were compared with nonglycopeptides.

Article

MATERIALS AND METHODS

Sample Preparation. The 6-plex Tandem Mass Tags reagent kit, Optima formic acid, and Optima LC/MS solvents were purchased from Fisher Scientific (Pittsburgh, PA). Mass spectrometry grade trypsin was purchased from Promega (Madison, WI). Bovine fetuin was obtained from Sigma-Aldrich (St. Louis, MO) without further purification. The isobaric labeling procedure provided by the TMT manufacturer43 was modified by increasing triethylammonium bicarbonate buffer concentration from 0.2 to 0.5 M and decreasing volume from 100 to 20 μL; denaturant (2% SDS) was reduced from 5 to 1 μL; reducing reagent was decreased from 5 μL of 200 mM to 2 μL of 50 mM tris-(2carboxyethyl)phosphine; the cysteine blocking reagent iodoacetamide (5 μL of 375 mM) was replaced with 1 μL of 200 mM methyl methanethiosulfonate (MMTS); and finally TMT reagent was dissolved in 70 μL of absolute ethanol instead of 41 μL of anhydrous acetonitrile. Specifically, 1 μL of the denaturant (containing 2% SDS) was added to each sample tube containing up to 100 μg of bovine fetuin and 20 μL of 0.5 M triethylammonium bicarbonate buffer (pH 8.5). The samples were vortexed and centrifuged before adding 2 μL of reducing reagent (50 mM tris-(2-carboxyethyl)phosphine) and were then mixed and incubated at 60 °C for 1 h. Cysteine residues were alkylated with 1 μL of 200 mM MMTS at room temperature for 10 min. To each sample tube, 2.5 μL of trypsin solution (1.0 μg/ μL) was added. Samples were vortexed and incubated at 37 °C overnight (12−16 h). Immediately before use, TMT reagents were equilibrated to room temperature and 70 μL of absolute ethanol was added to dissolve the reagents. The content of one TMT reagent vial was transferred to one sample tube and incubated at room temperature for 1 h. The labeling reactions were quenched by adding 8 μL of 5% hydroxylamine and incubating at room temperature for 15 min. The samples labeled with different tags were combined, vacuum concentrated down to 10 μL with a SpeedVac, and diluted with 50 μL of 5% formic acid before LC/MS/MS analysis. Liquid Chromatography and Mass Spectrometry. An LTQ Orbitrap XL ETD mass spectrometer (Thermo Scientific, Bremen, Germany) coupled with an Agilent 1200 series LC system consisting of a binary capillary-flow pump, a vacuum degasser, and a thermostatted autosampler (Agilent Technologies, Inc., Palo Alto, CA) was used for LC/MS/MS analysis. Isobarically labeled fetuin digests were separated with a Waters SunFire C18 column (2.1 mm × 150 mm, 3.5 μm). Typically, 2− 4 μL of sample solution was injected. Gradient elution was performed from 5 to 45% acetonitrile with 0.1% formic acid over 30 min, holding at 45% for 2 min, then immediately increasing to 95% and holding for 10 min to wash the column before column equilibration at 5% acetonitrile for the next injection. A flow rate of 0.2 mL/min was used for all analyses. As shown in the abstract graphic, in-source dissociation at 92 V generated characteristic oxonium ions at m/z 204 were further fragmented by a dedicated MS3 event for highly sensitive and selective detection of the eluting glycosylated peptides.21 For online LC/MS analysis, high-resolution, accurate-mass full scan MS spectrum was acquired in the Orbitrap mass analyzer followed by datadependent MS/MS alternating CID, HCD, and ETD scans with fragments analyzed in either the Orbitrap (CID and HCD) or ion trap (ETD) mass analyzers. Automatic gain control (AGC) target values for full-scan MS data acquisition were maintained at 5 × 105 for detection in the FT cell and 3 × 104 for detection in 1532



Article

Figure 1. Base peak chromatogram and extracted ion chromatogram for TMT labeled tryptic digest of bovine fetuin. Typical full-scan MS base peak chromatogram for all modified and unmodified peptides is shown in part a, and an extracted ion chromatogram for glycopeptides of bovine fetuin digest is illustrated in part b. Part c is the MS2 mass spectra of diagnostic oxonium ion HexNAc at m/z 204. The inset of part c lists common glycan diagnostic fragment ions of N- and O-linked glycoproteins.

the linear ion trap. AGC target values for MS/MS product ion detection were 3 × 105 and 1.5 × 104 for the FT cell and linear ion trap, respectively. A resolving power of 100 000 (at m/z 400) was used for all ion detection in the FT cell. To improve ETD signals of glycopeptides for better glycosylation site mapping and HCD signals for glycopeptide quantification, a TriVersa NanoMate (Advion BioSciences, Inc., Ithaca, NY) was used to collect and reinfuse target fractions. Typically 50 and 100 scans were averaged for HCD and ETD spectra, respectively. The anion target was 9 × 104, and ETD reactions were performed for 150 ms. Data Processing. Qualitative data interpretation of glycopeptides was done manually with the help of three programs. First, Thermo Scientific’s Xtract software was used to convert the raw data into singly charged monoisotopic spectra for easy interpretation of protonated molecular ions. Second, the experimental accurate mass of the protonated molecular ion was used to search possible oligosaccharide composition using GlycoMod from the Swiss-Prot Web site.44 Finally, Northwestern University’s ProSight PTM45 was used for fragment ion prediction. The glycopeptide sequence and glycosylation sites could be verified through these fragment ions. Identification of nonglycopeptides was performed by searching the MS/MS spectra against the SwissProt database using an Internet MASCOT search engine from Matrix Science,46 or it was done using the SEQUEST search engine in Thermo Scientific’s Proteome Discoverer. The following parameters were used for search: digestion with one missed cleavage was selected; mass tolerance was 50 ppm for the precursors and 0.3 Da for the MS/ MS ions; TMT-labeled N-termini and lysines and MMTSalkylated cysteines were set as fixed modifications; oxidized methionines and TMT-labeled tyrosines were set as variable modifications.

For quantitative analysis, the TMT reagent signature peak areas for each reporter ion mass were extracted from corresponding MS/MS spectra. To validate the quantification, different amounts of fetuin were compared with 50.0 μg of fetuin. The quantification variability and precision were determined from 3 to 4 labeling experiments. In each labeling experiment, duplicates were labeled with different tags. For example, duplicate samples containing 20.0, 50.0, and 100.0 μg of fetuin each were labeled with six different TMT tags and mixed for LC/ MS/MS analysis. The signature peak areas from each set of duplicates were averaged. The protein amounts were obtained by comparing peak areas of others to ones from 50.0 μg.

■

RESULTS AND DISCUSSION Identification of Glycopeptides from Trypsin Digest of Bovine Fetuin. Figure 1a shows typical full-scan MS base peak chromatogram for all TMT labeled glycopeptides and nonglycopeptides of the bovine fetuin digest. Figure 1b is an extracted ion chromatogram for glycopeptides of the bovine fetuin digest. The in-source pseudo-MS3 survey scan at m/z 204 indicates the elution of glycopeptides. Figure 1c shows the MS2 mass spectrum of diagnostic oxonium ion HexNAc at m/z 204. The appearance of this ion pattern is the indication of glycopeptide elution. The inset of Figure 1c lists common glycan diagnostic fragment ions of N- and O-linked glycopeptides.21 Five experimental events, which were a survey scan at m/ z 204, high-resolution FT full scan, CID, ETD, and HCD, were alternatively conducted at each LC elution time point. As an example, a full scan MS spectrum at an elution time of 23 min indicated that multiple precursors with different charge states were present at the elution (Figure S-1a in the Supporting Information). After the interpretation of the spectrum with Xtract, two main glycopeptides were identified with [M + H]+ at 1533



Article

Table 1. Glycopeptides and Corresponding Glycoforms Identified by Accurate Mass for the N- and O-Linked Oligosaccharides from the TMT Labeled Tryptic Digestion of Bovine Fetuin glycopeptides with glycan composition (peptide position) b

c

a

theoretical mass [M + H]+

experimental mass [M + H]+ d

mass error (ppm)

5265.35590 6083.63630 6448.76850 6739.86390 7030.95930

5265.35078 6083.62400 6448.76049 6739.84959 7030.93900

0.97 2.02 1.24 2.12 2.89

4509.96700 5166.19460 5457.29000

4509.96889 5166.19123 5457.27644

0.42 0.65 2.48

4152.70910 4808.93670 5100.03210

4152.70086 4808.92641 5100.03823

1.98 2.14 1.20

4341.13330 5159.41370 5450.50910 5815.64126 6106.73666 6397.83210

4341.13429 5159.42397 5450.49699 5815.65063 6106.73122 6397.83220

0.23 1.99 2.22 1.61 0.89 0.02

6889.48940 7180.58480 7545.71700 7836.81240 8201.94460 8493.04000

6889.47126 7180.57749 7545.69648 7836.80351 8201.93074 8493.02532

2.63 1.02 2.72 1.13 1.69 1.73

c

RPTGEVYDIEIDTLETT CHV LDPTPLA N CSVR (AA 54−85) Hex3HexNAc3Neu5Ac1 e Hex5HexNAc4Neu5Ac2 e Hex6HexNAc5Neu5Ac2 e Hex6HexNAc5Neu5Ac3 e Hex6HexNAc5Neu5Ac4 b KLcCPDcCPLLAPLaNDSR (AA 126−141) e Hex5HexNAc4Neu5Ac2 e Hex6HexNAc5Neu5Ac3 Hex6HexNAc5Neu5Ac4 b c L CPDcCPLLAPLaNDSR (AA 127−141) e Hex5HexNAc4Neu5Ac2 e Hex6HexNAc5Neu5Ac3 Hex6HexNAc5Neu5Ac4 b VVHAVEVALATFaNAESNGSYLQLVEISR (AA 142−169) e Hex3HexNAc3 Hex5HexNAc4Neu5Ac1 Hex5HexNAc4Neu5Ac2 Hex6HexNAc5Neu5Ac2 e Hex6HexNAc5Neu5Ac3 e Hex6HexNAc5Neu5Ac4 b VTcCTLFQTQPVIPQPQPDGAEAEAPaSAVPDAAGPaTP a SAAGPPVASVVVGPSVVAVPLPLHR (AA 228−288) e Hex1HexNAc1Neu5Ac1 Hex1HexNAc1Neu5Ac2 e Hex2HexNAc2Neu5Ac2 e Hex2HexNAc2Neu5Ac3 e Hex3HexNAc3Neu5Ac3 e Hex3HexNAc3Neu5Ac4

a Glycosylation sites. bTMT labeling sites. cAlkylation sites. dThe experimental mass [M + H]+ was calculated using Xtract. eGlycopeptides confirmed by ETD.

m/z 6739.87 and 8201.96 (Figure S-1b,c in the Supporting Information). With a single MS run, CID, ETD, and HCD spectra were acquired automatically for all major precursors. For this known glycoprotein, bovine fetuin, the identification of nonglycopeptides was done by comparing experimentally produced accurate masses with those of known peptides. The task was also performed by a database search using SEQUEST in Proteome Discoverer and Internet MASCOT search engine from Matrix Science to demonstrate that the search can be done with mixed MS/MS data of glycopeptides and nonglycopeptides. All nonglycopeptides were unambiguously identified. In this study, accurate mass matching was the main strategy to identify glycopeptides. Specifically, a list of glycopeptides with all possible glycoforms was established considering masses of the cysteine protecting group and isotopic labeling tag. Next, the masses of all protonated peptide ions were calculated using Xtract and compared with those theoretically predicted. Then, the peptide sequence and glycan composition were determined by this accurate mass comparison. Finally, the peptide and glycan sequences were verified with CID, ETD, and HCD spectra. Using this strategy, various glycoforms of each glycosylated peptide were tentatively assigned based on their accurate masses. The results listed in Table 1 represents the detailed site-specific glycan heterogeneity . The mass accuracy for all the glycoforms identified was better than 3 ppm using the monoisotopic masses of the singly charged species. The observed forms were in good agreement with previously reported.21,28 Although the relative

abundance of a few O-linked glycoforms was not greater than 3% of the base peak detected, the excellent mass accuracy resulted in high confidence in the assigned chemical compositions for these glycoforms. As many as 23 glycoforms (not including isomers for identical composition) were identified from the tryptic digest of bovine fetuin using direct LC/MS/MS analysis. In addition to the survey scan and high resolution/mass accuracy spectrum acquired for identification of the glycoforms, oligosaccharide sequencing was accomplished during the parallel data dependent MS/MS events. For example, Figure 2a shows a raw full-scan MS/MS spectrum of the 6+ charge state precursor ion of a peptide (AA 54−85) glycoform at m/z 1124 detected with the Orbitrap mass analyzer. Figure 2b shows the same CID spectrum as that shown in Figure 2a but interpreted with Xtract and displayed with monoisotopic masses. All peaks were assigned to corresponding fragments. All measured masses were within 2 ppm of their theoretical values, which allowed for unambiguous assignment of the glycan sequence. The term “glycan sequence” refers to monosaccharide connecting order and does not imply α/β linkages. Isomeric glycopeptides with a slightly different sequence of a glycan moiety could be differentiated by CID MS/MS analysis. As shown in Figure 2b, a minor isomer with HexNAc1Neu5Ac1 conjugate was identified from the peak at 5718.50 (5427.41 + 291.10) Da with an indicated glycan structure. This type of conjugate was previously reported for bovine fetuin.47 This conjugate assignment was further supported with ion at 948.33 1534



Article

Figure 2. CID spectrum of 6+ charge state precursor ion at m/z 1124 of TMT labeled bovine fetuin triantennary peptide (AA 54−85). Part a is a raw CID spectrum, and part b is the same spectrum interpreted with Xtract. A minor glycan isomer was identified, and the glycan structure was presented at m/z 5718.5 Da.

ion, all possible peptide masses (excluding glycan masses), labeling reagent mass, and cysteine protecting reagent mass were alternatively subtracted from or added to the experimental singly charged monoisotopic mass of an investigating ion. The resulting number was input as experimental mass into GlycoMod Tool to search for the best matching glycan composition and other modifications such as phosphate and sulfate. By using this strategy, many nonisobaric labeled or less labeled species were found. The presence of a pair of peaks at 22.45 and 23.03 min indicates the incomplete labeling before optimization (Figure S-2 in the Supporting Information). This peptide contains a lysine residue. Two TMT tags should be labeled with one on the lysine and other on the terminal amine. The peak at 22.45 min is the partially labeled portion of the glycopeptide peak at 23 min. The presence of the peaks at 31.33 and 34.80 min is another example with the peak at 31.33 min as a nonisobaric labeled one. It was determined that the labeling efficiency was about 70% without modification of the labeling procedure. When the TMT isobaric reagent protocol was followed, the acetone precipitation step was omitted because it increased quantification variability between samples. The excess cysteine protecting reagent iodoacetamide might interfere with isobaric labeling to cause low labeling efficiency. Many protocol modification attempts were conducted to improve labeling efficiency. An optimized protocol was achieved through modifications as described in the Sample Preparation section. With all these modifications, the labeling efficiency was enhanced to greater than 95%. Data listed in Table S-1 in the Supporting Information indicates the improvement of labeling efficiency after optimization. This improvement made both quantification and identification more confident because the reporter signals would not only be used for quantification but also to determine if a certain glycoform was present.

(657.23 + 291.10) Da and additional fragments (not labeled in Figure 2) at 5921.58 (5630.49 + 291.10) Da and 6374.73 (6083.64 + 291.10) Da. It was shown that CID MS/MS experiments predominantly yielded the fragment ions from the cleavage of glycosidic bonds without breaking amide bonds (Figure 2). Therefore, for glycopeptides CID can only be used for glycan sequencing but not for glycosylation site determination and amino acid sequencing. HCD experiments generated ions from both amide bond breaking and glycosidic bond cleavage. Figure 3 shows HCD spectra of the same glycopeptide (AA 54−85) as described in Figure 2. Changing the collision energy with HCD resulted in different sets of observed fragment ions. At high energies, more amide bond breakage was observed with b- and y-ions formed and TMT reporter groups were released for quantification (Figure 3a). On the other hand, at lower energy settings, glycosidic bond cleavage was preferred (Figure 3b). Most fragments in Figure 3b were resulting from glycosidic bond cleavage. These fragments were comparable with ones formed with CID for the same glycopeptide as shown in Figure 2 and could be used for glycan structure elucidation. Few b- and y-ions were also formed as labeled in Figure 3b. The inset of Figure 3b shows an expansion of the singly charged region for better view of the b-ions. Because CID fragmentation only resulted in glycosidic bond cleavage, these b- and y-ions from HCD spectra provided additional information and helped to confirm the amino acid sequence of glycopeptides. In this study, HCD was mainly used for glycan isoform quantification as described in the separate section below. TMT Labeling Efficiency. An indirect mass matching strategy was used to identify unpredicted glycopeptides or other modifications. For an unknown, high abundance molecular 1535



Article

Figure 3. HCD spectra of 6+ charge state precursor ion at m/z 1124 of TMT labeled bovine fetuin triantennary peptide (AA 54−85). Changing the instrument’s collision energy with HCD resulted in different sets of ions. The spectrum (a) was acquired at high energy settings; more amide bond breakage was observed producing b- and y-ions, and TMT reporter groups were released for quantification. The spectrum (b) was acquired at lower energy settings, and glycosidic bond cleavage was preferred. The inset of part b is an expanded singly charged region for a better view of the b-ions. See Figure 4 for the peptide sequence.

Figure 4. ETD spectrum of 6+ charge state precursor ion at m/z 1124 of TMT labeled bovine fetuin triantennary peptide (AA 54−85). A large group of c-ions and z•-ions were detected, the glycosylation site was clearly identified at position 81, and the peptide sequence was identified with a series of c/z• ions. Not all fragments were labeled on the figure. See Figure 2 for the monosaccharide symbols.

Glycosylation Site Mapping by Using ETD Spectra. The primary strength of ETD in proteomic analysis is the ability to efficiently sequence peptides with labile PTMs.30 However, it was found that with online ETD in which a few microscans were averaged it was very difficult to obtain meaningful spectra for glycopeptides. Most ETD spectra were noisy with very low

intensity especially for glycopeptides with m/z greater than 1000. For the low concentrations of glycopeptides eluted from the column, more than just a few microscans are needed to achieve high enough quality of the ETD spectra for glycosylation site mapping and peptide sequencing. For this purpose, elution fractions were collected every 15 s with a TriVersa NanoMate 1536



Article

Table 2. Quantitative Evaluation for Glycopeptides from the TMT Labeled Tryptic Digestion of Bovine Fetuin protein amount used (μg) glycopeptides

a

a

20.0

40.0

80.0

100.0

19.2 4.66% −4.02%

38.3 2.68% −4.26%

73.4 6.17% −8.31%

95.1 1.93% −4.87%

[M + H]+ 6448 (AA 54−85) Hex6HexNAc5Neu5Ac2

amount found (μg) RSD error



18.7 9.72% −6.70%

40.6 9.25% 1.54%

80.8 9.96% 1.03%

94.3 3.67% −5.68%



18.6 3.16% −7.22%

39.9 6.89% −0.22%

78.9 4.58% −1.32%

92.9 7.26% −7.14%



18.2 9.06% −9.04%

41.4 7.89% 3.62%

75.8 0.74% −5.25%

96.6 11.63% −3.43%



19.3 4.50% −3.34%

40.9 4.12% 2.31%

76.1 9.26% −4.93%

95.0 5.95% −4.96%



19.5 6.11% −2.58%

37.3 5.78% −6.75%

79.2 11.83% −0.95%

108.7 2.56% 8.68%



20.3 6.99% 1.51%

38.2 2.59% −4.48%

79.5 11.34% −0.63%

108.4 7.62% 8.40%



19.9 2.37% −0.68%

39.9 0.63% −0.19%

76.3 0.72% −4.60%

98.3 4.13% −1.72%



20.5 5.59% 2.31%

41.4 3.90% 3.43%

79.7 8.76% −0.43%

98.4 7.26% −1.55%



20.1 6.36% 0.72%

40.7 3.88% 1.73%

78.4 9.96% −2.04%

103.4 11.48% 3.44%

Experimental amount was obtained using 50 μg of fetuin as reference; the results were based on four labeling experiments.

glycopeptides and the averaging of numerous microscans improved the ETD spectrum quality, the glycopeptides at very low abundance could have good ETD spectra and be identified with this method. Precursor ion intensity of 6 digital counts was sufficient for an informative ETD spectrum. The glycopeptides listed in Table 1 and marked “#” were confirmed by ETD spectra. In general, lower m/z ions were preferred for ETD. It was hard to obtain high-quality ETD spectrum for an ion with m/z greater than 1300 Da. The method was suitable for both N- and O-linked glycopeptides. Figure S-3 in the Supporting Information shows an example of an ETD spectrum in which three O-glycosylation sites were determined with corresponding glycans identified. Some other approaches have been reported for the determination of the glycosylation site,19 such as MS3 or MS4 CID of target glycopeptides. However, the ETD fragmentation provides more direct and clear mapping information especially when multiple potential glycosylation sites exist. Combining peptide sequence information from ETD and HCD spectra with

while chromatograms, full-scan MS, and MS/MS were recorded. Since the glycopeptides’ elution can be detected by an in-source pseudo-MS3 survey scan at m/z 204 as described in the previous section, those target fractions were infused using the TriVersa. Typically, 100 microscans were averaged for a good quality ETD spectrum. The supplemental activation was turned on with the energy setting at 6−10 and the charge state setting at 6−8. For a glycopeptide with multiple precursors, better ETD spectra were obtained for precursors with lower m/z values. Figure 4 shows an example of ETD spectrum of the 6+ charge state precursor at m/z 1124 of bovine fetuin triantennary peptide (AA 54−85). For this glycopeptide, a large group of c ions and z• ions was detected. The glycosylation site was clearly determined at position 81 and the peptide sequence was identified with a series of c/z• ions. The mass of these c/z• ions were confirmed with ProSight PTM.45 With ETD technology, the labile glycan was retained without signs of carbohydrate cleavage. Since the HPLC separation and TriVersa fraction collection enriched the 1537



Article

fragments which allow for complete structural elucidation of the glycopeptides. HCD provides supplemental information for both peptide sequence and glycan structure. TMT labeled HCD can be used for glycopeptide quantification. Since the samples and reference standard were isobarically labeled and mixed for analysis, it eliminated the errors caused by variation in instrumental performance over time and potential differences in ionization efficiency. The combination of high resolution/ accurate mass and multiple dissociation techniques (CID, ETD, HCD) on the LTQ Orbitrap XL ETD are indispensible for comprehensive structural glycopeptide analysis within a single LC/MS run. This method can be used to qualitatively and quantitatively characterize up to six samples in a single experiment. It has the potential to monitor the efficiency and control of the fermentation process, monitor manufacturing process irregularities, counterfeit, or adulterated API. Because TMT tags offer the ability to perform differential profiling of multiple test samples, test samples and reference materials, or test samples and calibration controls, this technique is also a useful method to demonstrate the similarity between approved and biosimilar products. In addition, because all peptides are labeled, any difference in amino acid sequence and glycan composition can be quickly identified by reporter signal intensity. The method could also be used as a screening procedure for quality control purposes.

the glycan structure information from the CID spectrum leads to complete characterization of glycopeptides. Quantification of Glycopeptides and Glycoproteins with TMT Labeling and HCD. The HCD spectra acquired through online alternative CID/HCD/ETD scans of TMT labeled peptides provided high-quality reporter signals for nonglycopeptides. The intensity of these reporter signals could be used for relative quantification of unmodified portions and comparison of amino acid sequences. However, the online HCD spectra of TMT labeled glycopeptides did not provide any reasonable reporter group signals. In most cases, there was no reporter group signal at all. This made the quantification of different glycoforms unfeasible. Lack of reporter ions was caused by glycosidic bonds fragmenting first rather than amide bonds and reporter ions (C−C bonds) which require more energy. More reporter ions were generated by increasing the collision energy, but signals were inconsistent by averaging only a few scans with online HCD. As with ETD’s weak signal issue, high quality and consistent reporter signals were obtained with the help of the TriVersa NanoMate. The target glycopeptide fractions were infused using the TriVersa, and 50−100 microscans were averaged for a high quality HCD spectrum and consistent reporter signals. Setting charge state at a lower level, typically at 1, instead of setting collision energy at a higher level resulted in more intense and consistent reporter signals. For the experiments described here, the collision energy was set at 35. In addition, because of the low reporter ion fragmentation efficiency, the MS2 isolation width, typically between 10 and 20 m/z, should be wide enough to cover all isotopic ions and metal adducts of precursors. For a given glycopeptide, the precursors with lower m/z yielded more intense reporter ion signals. The signal intensity of precursors also contributed to the quality of reporter ion signals. Isolating target precursors prior to performing HCD improved the reporter signals. The inset of Figure 3a is an example of the reporter region for m/z 964 of glycopeptide (AA 54−85). The ratio of the labeled protein amount was 2:5:10:2:5:10. To verify this quantification technique, samples with different amounts of fetuin were labeled with TMT tags, the relative quantification to a designated reference sample with 50.0 μg of fetuin was examined. Typically, two levels with duplicates and two references with 50.0 μg of fetuin each were labeled with 6 different TMT tags and mixed for LC/MS/MS analysis. Peak areas of reporter ions for nonglycopeptides and glycopeptides were extracted and results from duplicates were averaged and compared with the average peak area of 50.0 μg fetuin to obtain the relative amount of proteins. Labeling for the same amount of fetuin was repeated for a total of four times. Table 2 shows the results from ten glycopeptides. For comparison, results from six nonglycopeptides are listed in Table S-2 in the Supporting Information. The results from nonglycopeptides and glycopeptides were comparable with % RSD less than 10 in most cases. The quantification errors were less than 9% for all examined peptides.

■

ASSOCIATED CONTENT

S Supporting Information *

Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Address: U.S. Food & Drug Administration, Division of Pharmaceutical Analysis, 1114 Market Street, Room 1002, St. Louis, MO 63101. Phone: (314)539-2174. Fax: (314)539-2113. E-mail: [email protected]. Present Address §

1921 Cheyenne Dr., Carrollton, TX 75010.

Notes

The findings and conclusions in this article have not been formally disseminated by the Food and Drug Administration and should not be construed to represent any agency determination or policy. The authors declare no competing financial interest.

■

REFERENCES

(1) Brooks, S. A. Mol. Biotechnol. 2004, 28, 241−255. (2) Warnock, J. N.; Al-Rubeai, M. Biotechnol. Appl. Biochem. 2006, 45, 1−12. (3) Mukovozov, I.; Sabijicl, T.; Hortelano, G.; Ofosu, F. A. Thromb. Haemostasis 2008, 99, 874−882. (4) Haller, C. A.; Cosenza, M. E.; Sullivan, J. T. Clin. Pharmacol. Ther. 2008, 84, 624−627. (5) Kawasaki, N.; Itoh, S.; Hashii, N.; Takakura, D.; Qin, Y.; Huang, X.; Yamaguchi, T. Biol. Pharm. Bull. 2009, 32, 796−800. (6) Rudd, P. M.; Elliott, T.; Cresswell, P.; Wilson, I. A.; Dwek, R. A. Science 2001, 291, 2370−2376. (7) Senger, R. S.; Karim, M. N. Biotechnol. Prog. 2003, 19, 1199−1209. (8) Cabrera, G.; Cremata, J. A.; Valdes, R.; Garcia, R.; Gonzalez, Y.; Montesino, R.; Gomez, H.; Gonzalez, M. Biotechnol. Appl. Biochem. 2005, 41, 67−76. (9) Kunkel, J. P.; Jan, D. C.; Butler, M.; Jamieson, J. C. Biotechnol. Prog. 2000, 16, 462−470.

■

CONCLUSIONS The method reported here, including sample preparation, MS/ MS analysis, and the data process strategy, provides a valuable approach for characterization and quantification of glycoproteins and other modified protein pharmaceuticals. As a soft fragmentation technique, ETD preserves labile glycans and facilitates the identification of both the peptide of interest and its site of modification. CID and ETD generate complementary 1538



Article

(10) Hossler, P.; Khattak, S. F.; Li, Z. J. Glycobiology 2009, 19, 936− 949. (11) Liszewski, K. Genet. Eng. News 2005, 25, 22−22. (12) Dwek, R. A.; Butters, T. D.; Platt, F. M.; Zitzmann, N. Nat. Rev. Drug Discovery 2002, 1, 65−75. (13) Helenius, A.; Aebi, M. Science 2001, 291, 2364−2369. (14) Okuyama, N.; Ide, Y.; Nakano, M.; Nakagawa, T.; Yamanaka, K.; Moriwaki, K.; Murata, K.; Ohigashi, H.; Yokoyama, S.; Eguchi, H.; Ishikawa, O.; Ito, T.; Kato, M.; Kasahara, A.; Kawano, S.; Gu, J.; Taniguchi, N.; Miyoshi, E. Int. J. Cancer 2006, 118, 2803−2808. (15) Kyselova, Z.; Mechref, Y.; Al Bataineh, M. M.; Dobrolecki, L. E.; Hickey, R. J.; Vinson, J.; Sweeney, C. J.; Novotny, M. V. J. Proteome Res. 2007, 6, 1822−1832. (16) Reid, C. Q.; Tait, A.; Baldascini, H.; Mohindra, A.; Racher, A.; Bilsborough, S.; Smales, C. M.; Hoare, M. Biotechnol. Bioeng. 2010, 107, 85−95. (17) Chow, S. C.; Endrenyi, L.; Lachenbruch, P. A.; Yang, L. Y.; Chi, E. Biosimilars 2011, 1, 13−26. (18) Van den Steen, P.; Rudd, P. M.; Dwek, R. A.; Opdenakker, G. Crit. Rev. Biochem. Mol. Biol. 1998, 33, 151−208. (19) Schiel, J. E. Anal. Bioanal. Chem. 2012, 404, 1141−1149. (20) Lazar, I. M.; Lazar, A. C.; Cortes, D. F.; Kabulski, J. L. Electrophoresis 2011, 32, 3−13. (21) Peterman, S. M.; Mulholland, J. J. J. Am. Soc. Mass Spectrom. 2006, 17, 168−179. (22) Chalkley, R. J.; Thalhammer, A.; Schoepfer, R.; Burlingame, A. L. Proc. Natl. Acad. Sci. U.S.A. 2009, 106, 8894−8899. (23) Alley, W. R., Jr.; Mechref, Y.; Novotny, M. V. Rapid Commun. Mass Spectrom. 2009, 23, 161−170. (24) Wu, S. L.; Huhmer, A. F.; Hao, Z.; Karger, B. L. J. Proteome Res. 2007, 6, 4230−4244. (25) Schiel, J. E.; Au, J.; He, H. J.; Phinney, K. W. Anal. Bioanal. Chem. 2012, 403, 2279−2289. (26) Schiel, J. E.; Lowenthal, M. S.; Phinney, K. W. J. Mass Spectrom. 2011, 46, 649−657. (27) Hua, S.; Nwosu, C. C.; Strum, J. S.; Seipert, R. R.; An, H. J.; Zivkovic, A. M.; German, J. B.; Lebrilla, C. B. Anal. Bioanal. Chem. 2012, 403, 1291−1302. (28) Ritchie, M. A.; Gill, A. C.; Deery, M. J.; Lilley, K. J. Am. Soc. Mass Spectrom. 2002, 13, 1065−1077. (29) Morelle, W.; Canis, K.; Chirat, F.; Faid, V.; Michalski, J. C. Proteomics 2006, 6, 3993−4015. (30) Mikesh, L. M.; Ueberheide, B.; Chi, A.; Coon, J. J.; Syka, J. E.; Shabanowitz, J.; Hunt, D. F. Biochim. Biophys. Acta 2006, 1764, 1811− 1822. (31) Kjeldsen, F.; Giessing, A. M.; Ingrell, C. R.; Jensen, O. N. Anal. Chem. 2007, 79, 9243−9252. (32) Housley, M. P.; Rodgers, J. T.; Udeshi, N. D.; Kelly, T. J.; Shabanowitz, J.; Hunt, D. F.; Puigserver, P.; Hart, G. W. J. Biol. Chem. 2008, 283, 16283−16292. (33) Bantscheff, M.; Schirle, M.; Sweetman, G.; Rick, J.; Kuster, B. Anal. Bioanal. Chem. 2007, 389, 1017−1031. (34) Ong, S. E.; Mann, M. Nat. Chem. Biol. 2005, 1, 252−262. (35) Ye, H.; Hill, J.; Kauffman, J.; Han, X. Anal. Biochem. 2010, 400, 46−55. (36) Viner, R. I.; Zhang, T.; Second, T.; Zabrouskov, V. J. Proteomics 2009, 72, 874−885. (37) Dayon, L.; Hainard, A.; Licker, V.; Turck, N.; Kuhn, K.; Hochstrasser, D. F.; Burkhard, P. R.; Sanchez, J. C. Anal. Chem. 2008, 80, 2921−2931. (38) Phanstiel, D.; Zhang, Y.; Marto, J. A.; Coon, J. J. J. Am. Soc. Mass. Spectrom. 2008, 19, 1255−1262. (39) Bantscheff, M.; Boesche, M.; Eberhard, D.; Matthieson, T.; Sweetman, G.; Kuster, B. Mol. Cell. Proteomics 2008, 7, 1702−1713. (40) Zhang, S.; Liu, X.; Kang, X.; Sun, C.; Lu, H.; Yang, P.; Liu, Y. Talanta 2012, 91, 122−127. (41) Zhang, Y.; Yin, H.; Lu, H. Glycoconj J 2012, 29, 249−258. (42) This work was first reported as a poster presented at AAPS Annual Meeting 2011, in Washington, DC.

(43) http://www.piercenet.com/instructions/2162073.pdf. (44) http://ca.expasy.org/tools/glycomod (accessed October 5, 2012). (45) https://prosightptm2.northwestern.edu (accessed October 5, 2012). (46) www.matrixscience.com (accessed August 10, 2012). (47) Huang, Y.; Mechref, Y.; Novotny, M. V. Anal. Chem. 2001, 73, 6063−6069.

1539


Direct Approach for Qualitative and Quantitative Characterization of

Recommend Documents