Research article Received: 12 August 2013
Revised: 30 September 2013
Accepted: 1 October 2013
Published online in Wiley Online Library
(wileyonlinelibrary.com) DOI 10.1002/jms.3291
Glycosylation characterization of recombinant human erythropoietin produced in glycoengineered Pichia pastoris by mass spectrometry Bing Gong,* Irina Burnina, Terrance A. Stadheim and Huijuan Li Glycosylation plays a critical role in the in vivo efficacy of both endogenous and recombinant erythropoietin (EPO). Using mass spectrometry, we characterized the N-/O-linked glycosylation of recombinant human EPO (rhEPO) produced in glycoengineered Pichia pastoris and compared with the glycosylation of Chinese hamster ovary (CHO) cell-derived rhEPO. While the three predicted N-linked glycosylation sites (Asn24, Asn38 and Asn83) showed complete site occupancy, Pichia- and CHO-derived rhEPO showed distinct differences in the glycan structures with the former containing sialylated bi-antennary glycoforms and the latter containing a mixture of sialylated bi-, tri- and tetra-antennary structures. Additionally, the N-linked glycans from Pichiaproduced rhEPO were similar across all three sites. A low level of O-linked mannosylation was detected on Pichia-produced rhEPO at position Ser126, which is also the O-linked glycosylation site for endogenous human EPO and CHO-derived rhEPO. In summary, the mass spectrometric analyses revealed that rhEPO derived from glycoengineered Pichia has a highly uniform bi-antennary Nlinked glycan composition and preserves the orthogonal O-linked glycosylation site present on endogenous human EPO and CHOderived rhEPO. Copyright © 2013 John Wiley & Sons, Ltd. Additional supporting information may be found in the online version of this article at the publisher’s web site. Keywords: N-linked glycosylation; O-linked glycosylation; Pichia pastoris; rhEPO; mass spectrometry
Introduction
1308
Human erythropoietin (EPO) is a multifunctional glycoprotein hormone with the major physiological role of regulating the level of red blood cells by promoting their proliferation, maturation and survival.[1,2] Synthesized mainly in the kidney and liver in adults, EPO comprises 165 amino acids after C-terminal processing during maturation. It contains three N-linked glycosylation sites (Asn24, Asn38 and Asn83) and a single O-linked glycan (Ser126).[3] The presence of N-linked glycosylation is critical for both in vitro and in vivo activity as PNGase F deglycosylated rhEPO demonstrated little biological activity, which was likely due to the loss of the required active conformation resulting from the removal of the core glycans.[4,5] Additionally, the actual glycosylation compositions, especially the terminal sialylation, greatly influences in vivo efficacy. While the sialylation does not contribute to EPO receptor binding, asialylated rhEPO has been shown to be nonerythropoietic in vivo. The diminishment of in vivo efficacy most likely resulted from the rapid serum clearance of EPO.[6,7] As such, efforts extending the serum half-life of rhEPO have become a focus for pursuing more efficacious rhEPO therapeutics. Darbepoetin alfa illustrates one such endeavor. In this particular case, two extra N-linked glycosylation sites were engineered at positions Asn30 and Asn88 through mutagenesis resulting in a significantly longer serum half-life and improved in vivo efficacy when compared to the native rhEPO sequence.[8] The glycosylation pattern of rhEPO can differ substantially when produced in different host cells. Currently, rhEPO is mainly produced in
J. Mass Spectrom. 2013, 48, 1308–1317
Chinese hamster ovary (CHO) cells, which shows predominantly branched and sialylated structures enriched through purification. While similar to the glycan structures from endogenous EPO, several differences exist. For instance, CHO cells lack a functional α-2,6-sialyltransferase (EC 2.4.99.1). As such, the linkage of sialic acid on CHO-derived rhEPO is solely α-2,3, while in human, both α-2,3 and α-2,6 were observed.[9] Then, the N-Glycolylneuraminic acid, a nonhuman and immunogenic structure, is present on recombinant proteins produced in CHO cells.[10] A recombinant epoetin delta (Dynepo) from an engineered human fibrosarcoma cell line HT-1080 has been approved in the European Union. It has a more human-like glycosylation profile, most likely because human cells were used for production.[11] The expression of rhEPO from wild-type yeasts including Saccharomyces cerevisiae and Pichia pastoris has been reported.[12,13] While showing both in vitro and in vivo activity, they contain typical fungal-type, high mannose glycan structures, which results in a relatively short in vivo half-life of rhEPO presumably due to the rapid clearance through mannose-binding protein (MBP) and other lectins. Recently, rhEPO expressed from a glycoengineered Pichia has been reported.[14] Through a series of genetic engineering steps, glycoengineered Pichia is capable of decorating glycoproteins
* Correspondence to: Bing Gong, GlycoFi, Biologics Discovery, Merck & Co., Inc., 16 Cavendish Court, Lebanon, NH 03766, USA. E-mail: [email protected] GlycoFi, Biologics Discovery, Merck & Co., Inc., 16 Cavendish Court, Lebanon, NH, 03766, USA
Copyright © 2013 John Wiley & Sons, Ltd.
Glycosylation of Pichia-derived rhEPO by MS with human-type bi-antennary sialylated glycans. In this work, we characterized the N- and O-linked glycosylation of glycoengineered Pichia-derived rhEPO with regard to the sitespecific N-linked glycosylation composition and the location and structure of O-linked mannosylation through mass spectrometry.
Material and experiments Materials Glycoengineered Pichia-expressed recombinant human EPO (hereafter referred to as Pichia rhEPO) was produced in-house. CHO cell expressed rhEPO (CHO rhEPO) was purchased from EMD
Millipore (Billerica, MA). Recombinant PNGase F, neuraminidase, β1-4 galactosidase and β-N-acetylglucosaminidase were from New England Biolabs (Ipswich, MA). 2,5-Dihydroxybenzoic acid and 3,5-dimethoxy-4-hydroxycinnamic acid (sinapinic acid) matrix-assisted laser desorption/ionization (MALDI) matrices were from Protea Biosciences (Morgantown, WV). Purchased from Sigma Aldrich (St. Louis, MO) were 1 M triethylammonium bicarbonate (TEAB) pH 8.5, 8 M guanidine hydrochloride (GuHCl), iodoacetamide (IAM), dithiothreitol (DTT), endoproteinase Glu-C and trypsin. InstantAB was obtained from ProZyme (Hayward, CA); 0.5 M tris(2carboxyethyl) phosphine was from Pierce (Rockford, IL); HPLC grade water, LC/MS grade water, and acetonitrile with 0.1% (v/v) formic
Figure 1. Intact molecular weight measurement of Pichia (top) and Chinese hamster ovary (CHO) (bottom) produced recombinant human erythropoietin (rhEPO) by matrix-assisted laser desorption/ionization-time-of-flight. The inset shows the LC-MS analysis of Pichia rhEPO.
J. Mass Spectrom. 2013, 48, 1308–1317
Copyright © 2013 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jms
1309
Figure 2. LC-MS analyses of recombinant human erythropoietin (rhEPO) after PNGase F and neuraminidase treatments. The PNGase F (a) and neuraminidase (c) treatments reveal that Pichia rhEPO is largely unmodified and only a small portion has O-linked glycosylation. PNGase F-treated CHO rhEPO (b) shows that the majority is modified with O-linked glycosylation and the heterogeneity of CHO rhEPO resulted from the glycans is demonstrated after neuraminidase treatment (d).
B. Gong et al. acid or 0.1% trifluoroacetic acid (TFA) were purchased from Fisher Scientific (Pittsburgh, PA); 96-well hydrophobic Immobilon-P polyvinylidene fluoride membrane plate was from Millipore (Billerica, MA); and 96-well Pall Supor® hydrophilic membrane plate was from Pall Life Sciences (Ann Arbor, MI). Glycoprotein analysis Intact CHO and Pichia rhEPO were reconstituted at 0.5 mg/ml in 20 mM sodium phosphate pH 7.4 buffer. For intact molecular weight measurement using MALDI-time-of-flight (TOF), 0.5 μl protein samples were spotted on a stainless MALDI target plate and mixed with 1 μl sinapinic acid. The samples were dried under vacuum at room temperature. All the MALDI-TOF analyses were performed on a Voyager-DE Biospectromtry workstation from Applied Biosystems (Foster City, CA). The settings for MALDITOF were as follows: accelerating voltage, 25 000 V; grid, 93%; guide wire, 0.05%; delay time, 750 ns; 100 shots per spectrum
were averaged; and data range was from 15 000 to 50 000 m/z. For the analysis using LC-MS, 10 ug rhEPO was digested with 0.5 μl PNGase F and neuraminidase at 37 °C overnight (~16 h), respectively. The digested samples were loaded onto an Agilent Q-TOF 6520 mass spectrometer coupled with an Agilent 1200 HPLC (Agilent, Santa Barbara, CA). An in-line MassPREP Micro desalt cartridge (Waters, Milford, MA) was used to remove salts from the samples prior to directing the sample to the mass spectrometer. The proteins were eluted off the cartridge using a one-step gradient: 100% buffer A (0.1% formic acid in water) flowed at 2 ml/min for 1 min, then the flow was changed to 100% buffer B (0.1% formic acid, 10% water and 90% acetonitrile) at 0.8 ml/min from 1 to 1.5 min. From 1.5 to 4 min, the flow rate was decreased to 0.5 ml/min, and the LC-MS data were collected. At the end of the gradient, a 1 min post-run of 100% buffer A at 2 ml/min was carried out to equilibrate the cartridge. The MS data were recorded in the profile mode from 500 to 3200 m/z. The dual electrospray ionization ion source was set as follows: gas
1310
Figure 3. Matrix-assisted laser desorption/ionization-time-of-flight spectra of fluorescently labeled N-linked glycans from Chinese hamster ovary (CHO) and Pichia-produced recombinant human erythropoietin. Untreated glycans from CHO (a) and Pichia (e) were monitored in the negative mode. Neuraminidase, neuraminidase/galactosidase and neuraminidase/galactosidase/hexominidase digested CHO (b, c and d), and Pichia (f, g and h) were analyzed in the positive mode.
wileyonlinelibrary.com/journal/jms
Copyright © 2013 John Wiley & Sons, Ltd.
J. Mass Spectrom. 2013, 48, 1308–1317
Glycosylation of Pichia-derived rhEPO by MS temp, 350 °C; drying gas, 13 l/min; nebulizer, 45 psig; fragmentor, 150 V; skimmer, 65 V; Oct1 RF VPP, 750 V; and Vcap, 3500 V. The MS spectra were analyzed using MassHunter software.
gradient was set for 30% buffer A from 0 to 1 min, then to 54% buffer A in 10 min. The settings for the FLR were excitation at 278 nm, emission at 344 nm, post-translational modification gain at 1 and data rate at 80 pts/s.
N-linked glycosylation analysis N-linked glycans were released and fluorescently labeled following a method described by Burnina et al.[15] Briefly, 20 μg Pichia and CHO rhEPO in 50 μl, 20 mM sodium phosphate pH 7.4 buffer was denatured and reduced by mixing with 150 μl, 10 M GuHCl, 10 mM tris(2-carboxyethyl) phosphine and incubation at 50 °C for 10 min in a 96-well hydrophobic Immobilon-P polyvinylidene fluoride membrane plate. The plate was subsequently washed three times with 200 μl TEAB pH 8.5 buffer. Added was 1 μl PNGase F in 50 μl TEAB pH 8.5 buffer, and the deglycosylation proceeded at 50 °C for 30 min. Then the released glycans in TEAB pH 8.5 buffer were labeled with 5 μl pre-constituted instantAB at 37 °C for 5 min. The labeled glycans were cleaned-up on a 96-well Pall Supor® hydrophilic membrane plate. Deposited on a MALDI target plate and mixed with 1 μl dihydroxybenzoic acid was 0.5 μl of labeled glycans. For glycans without neuraminidase treatment, they were analyzed in the negative mode. For the samples subject to neuraminidase treatment, they were analyzed in the positive mode. The settings for MALDI-TOF were as follows: accelerating voltage, 20 000 V; grid, 90.5% for the positive mode and 94% for the negative mode; guide wire, 0.01% for the positive mode and 0.08% for the negative mode; delay time, 100 ns for the positive mode and 225 ns for the negative mode; 100 shots per spectrum were averaged; and mass range was from 1000 to 5000 Da. Quantitative glycan analysis was performed on a Waters Acquity UPLC with a fluorescence detector (FLR-UPLC) and an Acquity BEH glycan column (2.1 × 100 mm). Mixed with 50 μl, 90% acetonitrile was 1 μl of labeled glycans; 10 μl was injected for analysis. The analyses were performed under the following conditions: column compartment temperature, 60 °C; sample manager temperature, 10 °C; flow rate, 0.4 ml/min; buffer A was 100 mM formate pH 4.6; and buffer B was acetonitrile. The
Peptide mapping For trypsin digestion, Pichia rhEPO (150 μg) was denatured and reduced in 6 M GuHCl and 20 mM DTT at 50 °C for 1 h. The sample was then cooled to room temperature and alkylated by incubating with 50 mM IAM for 45 min. The excess IAM was consumed by incubation with 5 mM DTT at room temperature for 10 min. All these chemicals were removed by buffer exchanging the sample to 25 mM ammonium bicarbonate pH 7.8 through a 5 KDa MW cutoff Millipore spin column. Trypsin (5 μg) was added to the sample, and the digestion proceeded at 37 °C overnight (~16 h). For Glu-C digestion, 300 μg Pichia rhEPO was buffer exchanged to 25 mM ammonium bicarbonate pH 7; 8. 10 μg Glu-C and 5 μl 100% acetonitrile were added to the sample for digestion, and the final volume was adjusted to 100 μl with 25 mM ammonium bicarbonate pH 7.8. The digestion proceeded at 25 °C overnight (~16 h). For PNGase F treatment, the Glu-C digestion was boiled for 20 min, and 10 μl was taken out and treated with 0.2 μl PNGase F after it was cooled to room temperature. Tryptic peptide mapping (10 μg trypsin digested rhEPO) was performed on a Thermo LTQ mass spectrometer with the Survey HPLC and Advion Triversal Nanomate. The column was a 250 × 4.6 mm Phenomenex Jupiter® Proteo column. Buffer A and buffer B were 0.1% formic acid in water and acetonitrile, respectively. The flow rate of HPLC was 1000 μl/min, and 300 nl/min was diverted to the mass spectrometer through the Nanomate. The HPLC gradient was 2–35% buffer B in 70 min, then to 98% in 10 min. Column temperature was maintained at 30 °C. The mass spectrometer settings were spray voltage (kV), 2.0; capillary Temp, 115; capillary voltage, 5; tube lens (V), 77; scan range, 200–2000 m/z; normalized collision energy, 25%; and activation time, 30 ms.
J. Mass Spectrom. 2013, 48, 1308–1317
Copyright © 2013 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jms
1311
Figure 4. Glu-C peptide mapping of Pichia recombinant human erythropoietin. The shifted peaks between before (a) and after (b) PNGase F digestion reveal the positions of glycopeptides, except for the Asn38 glycopeptide, which is resistant to PNGase F digestion. The peaks containing N-linked glycans are noted with retention time and highlighted with relevant sites.
B. Gong et al. Glu-C digested rhEPO (10 μg) was analyzed on the Thermo LTQ Orbitrap XL mass spectrometer with Waters Acquity UPLC system and a BEH C18 (1.7 μm, 2.1 × 50 mm) column. Buffer A was 0.1% TFA in water, and buffer B was 0.1% TFA in acetonitrile. The flow rate was 300 μl/min. The gradient was 3% buffer B for 5 min, then to 30% in 30 min. The settings for the mass spectrometer were sheath gas (arb), 30; auxgas flow rate, 5; sweep gas flow rate, 2; spray voltage (kV), 4.5; capillary Temp, 275; capillary voltage, 47; tube lens (V), 90; scan range, 300–2000 m/z; normalized collision energy, 35%; and activation time, 30 ms. Full scan using FTMS
with a resolution of 30 000 was used for the full scan, and the ion trap was used for product ion detection.
Results and discussion Glycoprotein analysis The molecular weight measurement of the intact rhEPO from CHO and Pichia was first carried out on a MALDI-TOF mass spectrometer. An initial attempt of analyzing CHO rhEPO on LC-MS
1312
Figure 5. The spectra of the glycopeptides of Asn24(a), Asn38 (b) and Asn83 (c) show that both A2 and A1 glycan structures are present and the former is the major species.
wileyonlinelibrary.com/journal/jms
Copyright © 2013 John Wiley & Sons, Ltd.
J. Mass Spectrom. 2013, 48, 1308–1317
Glycosylation of Pichia-derived rhEPO by MS was not successful, as the heterogeneity of the glycosylation, to the most part the sialylation, prevented us from acquiring any resolvable spectrum. As depicted in Fig. 1, Pichia rhEPO shows a much better resolved MALDI-TOF spectrum than CHO rhEPO. Three distinct peaks representing various levels of sialylation are clearly presented. As the predicted protein mass for rhEPO is ~18 KDa, the glycans account for ~26% of the total mass with an average N-glycan mass of ~2 KDa. For CHO rhEPO, a single broad peak centered ~29 KDa was observed. The broadness reflects the heterogeneity of the glycosylation. Glycans account for ~38% of the total mass of CHO rhEPO with an average mass of ~3.3 KDa. The greater homogeneity of the glycosylation on Pichia rhEPO enabled us to analyze the intact molecule using LC-MS (the inset of Fig. 1). It is also apparent that there is a clear discrepancy in the level of sialylation when Pichia rhEPO was analyzed by MALDI-TOF versus LC-MS. From the LC-MS analysis, Pichia rhEPO is almost completely fully sialylated with other minor species. In contrast, while the MALDI-TOF spectra do show that the fully sialylated species is the majority, however, the species with partial sialylation are also substantial. This difference could be caused by the different ionization efficiencies of differently glycosylated rhEPO in these two techniques. A more accurate measurement of the glycan composition of Pichia rhEPO was carried out using fluorescently labeled glycans and is presented in the following section. Pichia and CHO rhEPO treated with PNGase F and neuraminidase were analyzed using LC-MS. Figure 2a and b shows the deconvoluted spectra of PNGase F deglycosylated Pichia- and CHO-derived rhEPO. A main predominant peak of 18 239 Da was observed in the spectra of PNGase F-treated Pichia rhEPO and matches the calculated mass of rhEPO protein (Fig. 2a). Additionally, a very minor peak of 18 563 Da implying the potential O-linked mannosylation is also present.
In the deconvoluted spectrum of PNGase F-treated CHO rhEPO (Fig. 2b), two major peaks with masses of 18 896 and 19 187 Da represent the rhEPO modified with mono- and di- sialylated O-linked glycans, respectively. The observed peak of 18 239 Da indicates that a minor portion of CHO rhEPO was not O-glycosylated, which is different from Pichia rhEPO where only a small fraction is modified by O-linked glycosylation. Figures 2c and d show the deconvoluted spectra of neuraminidase digested Pichia and CHO rhEPO. A single main peak of 23 106 Da appears for Pichia rhEPO (Fig. 2c), which corresponds to the loss of six sialic acids (two sialic acid residues from each of the three bi-antennary oligosaccharides) from the intact rhEPO mass of 24 854 Da observed (the inset in Fig. 1). As evident in Fig. 2d, the glycosylation of CHO rhEPO remains heterogenous even after the sialic acids are removed. At this stage, it is quite difficult to determine the actual structures that each peak represents because of the interference due to O-linked glycosylation. N-linked glycosylation analysis The N-linked glycosylation of rhEPO was further characterized first by sequential glycosidase treatments of released glycans. To accurately quantify the relative amounts of various glycan structures, we used a fluorescence tag, instantAB from ProZyme, to label the released N-linked glycans. Figure S1 shows the labeled glycans analysis using FLR-UPLC. The glycans are predominantly A1 and A2 (the nomenclature of these glycans are shown in Fig. S1), and they account for ~8% and 92%, respectively. In addition to enable accurate quantitative analysis, fluorescence tags have been found to improve the ionization efficiency during the mass spectrometry analysis of glycans by rendering the otherwise hydrophilic glycans more hydrophobic.
Table 1. Glycopeptides identified through Glu-C peptide mapping Glycosylation site
Asn24
Peptide sequence
Deglycosylated peptide
*
AENITTGCEHCSLNE
AENITTGCAEHCSLNE **
AENITTGCAEHCSLNE Asn38
NITVPDTKVNFYAWKRME
Asn83
AVLRGQALLVNS AVLRGQALLVNSSQPWEPLQLHVD AVLRGQALLVNSSQPWEPLQ LHVDKAVSGLRS # AVLRGQALLVNSSQPWE PLQ LHVDKAVSGLRS # AVLRGQALLVNSSQPWE PLQLHVD AVLRGQALLVNSSQPWEPLQLHVD KAVSGLRSLTTLLRALGAQKE
Glycopeptide
Retention time (min)
Detected mass (Da)
Retention time (min)
Detected mass (Da)
8.4
1627.7
7.3
11
1689.68
10.0
13.5
1889.88
12.3
2212.55
17.0
14.9 20.8 21
1240.74 2671.44 3470.94
12.7 19.1 19.7
23.2
3451.92
21.8
24.5 27.8
2653.44 4865.76
22.7 26.6
3542.94 3834.48 3603.39 3894.51 3803.58 4094.70 4126.83 4417.92 3445.53 4877.25 5383.64 5674.72 5367.54 5656.71 4858.23 6781.48 7071.56
Glycan
A1 A2 A1 A2 A1 A2 A1 A2 A2 A2 A1 A2 A1 A2 A2 A1 A2
J. Mass Spectrom. 2013, 48, 1308–1317
Copyright © 2013 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jms
1313
The observed peptide masses and their corresponding glycans are based on the observed masses except for Asn38 for which the peptide mass is based on the sequence, as the aglycosylated peptide was not observed because of its resistance to PNGase F digestion. Modifications at Cys residues (noted with *) in the Asn24 and deamidation on the Asn83 (noted with #) peptides were detected and resulted in the different elution times for these peptides.
B. Gong et al. InstantAB dye from ProZyme requires a mild labeling condition that helps in preserving the glycan structures especially those labile forms such as sialylated glycoforms. However, one caveat to this approach is that instantAB labeled glycans gain a mass addition of 162 Da, which coincides with the mass of a single hexose. Therefore, a high labeling efficiency is generally required to assign the labeled glycan structure using the mass spectrometry reliably. The spectra of the labeled glycans are shown in Fig. 3. Figure 3a–d shows the N-linked glycans from CHO rhEPO, which shows a heterogeneous mixture of glycoforms. The spectra in Fig. 3a and e were obtained by operating MALDI-TOF in the negative mode in order to reveal the sialylation more clearly, while those in Fig. 3b–g were from the positive mode analysis. A mixture of sialylated bi-, tri- and tetra-antennary structures is observed. We assigned the major peaks with potential structures solely based on the observed masses, and the actual linkages might be different from what is depicted here. For example, the sialic acids in the 2900.48 ion could be on any two of the three branches, not necessarily the two labeled. Additionally,
there are Galactose-N-acetylglucosylamine (LacNAc) extensions observed in Fig. 3b and c where a series of 365 Da (LacNAc) additions is present between the main peaks after neuraminidase and neuraminidase/galactosidase treatments. Like the sialic acids, the LacNAc extensions could be on any of the branches.[16] After treatment with neuraminidase, galactosidase and N-acetylglucosaminidase, a fucosylated core Man 3 structure was observed (Fig. 3d). Figure 3e–h shows the spectra of the N-linked glycans released from Pichia rhEPO. As shown in Fig. 3e, the glycans primarily comprise the fully sialylated bi-antennary structure A2 with a minor species of partially sialylated bi-antennary A1, which is consistent with the results we obtained from FLR-UPLC analysis. After digestion with neuraminidase (Fig. 3f), the structures collapsed to G2 as expected. Further digested with neuraminidase/galactosidase and neuraminidase/galactosidase/N-acetylglucosaminidase, the glycans were converted to G0 and Man 3 (Fig. 3g and h), respectively. All of these glycan structures on Pichia rhEPO are afucosylated.
1314
Figure 6. MS2 collision-induced-dissociation spectra reveal the breakage after Ser104. The MS2 spectra of 1157.68 m/z (a) identify it as the peptide AVLRGQALLVNSSQPWEPLQLHVDKAVSGLRSer104 and 707.70 m/z (b) as Leu105TTLLRALGAQKE.
wileyonlinelibrary.com/journal/jms
Copyright © 2013 John Wiley & Sons, Ltd.
J. Mass Spectrom. 2013, 48, 1308–1317
Glycosylation of Pichia-derived rhEPO by MS Besides the macro-heterogeneity, the N-linked glycans on CHO rhEPO show site-wise micro-heterogeneity.[17,18] We analyzed the site-specific glycosylation of Pichia rhEPO through peptide mapping with Glu-C digestion. Figure 4 shows the UV absorbance chromatograms of Glu-C digested Pichia rhEPO treated with and without PNGase F. The peaks shifted upon PNGase F treatment and highlighted with retention time indicate the presence of glycopeptides. Among the shifted peaks, the glycopeptides containing Asn24 were identified at 7.3, 10 and 12.3 min on the chromatogram of the sample with only Glu-C treatment (Fig. 4a), and the corresponding deglycosylated peptides were identified at 8.4, 11.1 and 13.5 min on the chromatogram of the sample treated with both Glu-C and PNGase F (Fig. 4b). The Asn83 containing glycopeptides were identified at 12.7, 19.1, 19.7, 21.8, 22.7 and 26.6 min in Fig. 4a, and their corresponding deglycosylated peptides appeared at 14.9, 20.8, 21.0, 23.2, 24.5 and 27.8 min in Fig. 4b. In the shift peaks, we did not observe any peptides harboring Asn38. Instead, the peptide, N38ITVPDTKVNFYAWKRME, was identified at 17 min in both Fig. 4a and b. The glycans at Asn38 peptides were resistant to PNGase F digestion as the glycans are attached to the very first amino acid on its N-terminus. It is well known that the N-linked glycans at the asparagine, which is on either the N- or C-terminus of proteins or peptides, are resistant to PNGase F digestion.[19]
The full scan spectra of the major glycopeptides eluted at 10 min (Asn24), 17 min (Asn38) and 19.7 min (Asn83) in Fig. 4a are shown in Fig. 5 to demonstrate the attached glycan structures. The ions of 1202.13 and 1299.50 in Fig. 5a represent the triply charged Asn24 glycopeptides with A1 and A2 structure, respectively. The full scan spectrum of the glycopeptide containing Asn38 is demonstrated in Fig. 5b. The ions of 1376.61 and 1473.64 are the A1 and A2 glycan modified glycopeptides with three charges. The ions of 1032.71 and 1105.48 are the A1 and A2 modified glycopeptides with four charges. The MS2 spectra of 1473.64 and 1376.61 ions are shown in Fig. S2. Figure 5c shows the spectrum of the glycopeptides contains Asn83. Both A2 and A1 modified glycopeptides with three and four charges were presented. Based on the MS signals, A2 modified glycopeptides account for the majority of glycopeptides for the three sites, which is in good agreement with the results of FLR-UPLC analysis of the fluorescently labeled glycans. The observed glycopeptides and their associated glycans were listed in Table 1. During the analysis of Glu-C peptide mapping, the glycopeptides containing Asn24 and Asn83 were observed in multiple peaks (Fig. 4 and Table 1). After examining the spectra, we found that they were due to the presence of Cys modifications and Ser-related cleavages that occurred, most likely, during the sample preparation for Glu-C peptide mapping, as we did not observe any of those
J. Mass Spectrom. 2013, 48, 1308–1317
Copyright © 2013 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jms
1315
Figure 7. MS2(a) and MS3(b) collision-induced-dissociation spectra of the ion of 895 m/z show the O-linked mannosylation at Ser126. The product ion of 733 m/z in the MS2 was used as the parent ion in the MS3 experiment. The product ions indicating the location of the O-linked glycosylation at Ser126 are highlighted with underline.
B. Gong et al. modifications and cleavages in the tryptic peptide mapping and intact protein analysis. The modifications of Cys residues were found on the Asn24 peptides, and cleavages after Ser were observed on Asn83 peptides. Figure 6a shows the MS2 spectra of Asn83 glycopeptide (AVLRGQALLVN83SSQPWEPLQLHVDKAVSGLRS104) with a C-terminal Ser, and the MS2 spectrum of its neighboring C-terminal peptide (L105TTLLRALGAQKE) is demonstrated in Fig. 6b. Another Ser cleavage was also observed on this peptide (AVLRGQALLVN83S). These Ser cleavages might indicate the potential labile positions in the EPO molecule.
the mammalian mucin-type O-glycosylation starts. Our analysis of Pichia-expressed recombinant human granulocyte colonystimulating factor also showed that the O-glycosylation site used by human and CHO cells was also utilized by glycoengineered Pichia pastoris for O-mannosylation (Gong B et al., manuscript submitted). It seems that the protein itself plays a more fundamental role than the expression host does in determining the O-glycosylation sites.
O-linked glycosylation analysis
The glycosylation of rhEPO produced in glycoengineered Pichia was characterized using MALDI-TOF and LC-MS. The majorities of N-linked glycans are highly sialylated bi-antennary structures. The glycosylation pattern among the three N-linked glycosylation sites is highly consistent. A small fraction of Pichia rhEPO is modified with O-linked mannosylation, which was determined to be on the Ser126 and consists primarily as a mannobiose structure. This study demonstrates the utility of mass spectrometry in analyzing both N- and O-linked glycosylation on the glycoproteins from glycoengineered Pichia pastoris.
1316
O-linked glycosylation has been identified at Ser126 on both endogenous and mammalian cell-produced recombinant EPO. A low level of O-linked mannosylation was observed on Pichia rhEPO through LC-MS analysis (18 563 Da in Fig. 2a). In order to determine whether the same Ser126 site is occupied on Pichia rhEPO, a tryptic peptide mapping was performed. Figure S3 showed the base peak chromatogram of LC-MS analysis of tryptic Pichia rhEPO. As glycosylation is a labile modification and not stable during the fragmentation process in the mass spectrometer, the LTQ mass spectrometer was operated in a neutral loss mode with the neutral masses set for 162, 81 and 54 Da. The entire rhEPO sequence was covered (the peptides and their corresponding retention times were listed in Table S1). The peptides containing the N-linked glycan sites were identified at 43.73 and 47.32 min and mostly modified with A2 glycans as described in the previous section. An MS3 event was triggered by a neutral loss of 162 Da from the ion of 895.23 m/z. Figures 7a and b show the MS2 and MS3 spectra of the 895.23 ion, respectively. Based on the MS3 spectra (Fig. 7b, the parent ion is 733 m/z), it was identified as the peptide, EAIS120PPDAAS126AAPLR, which contains two Ser residues, Ser120 and Ser126. The 895.23 ion was doubly charged, and the neutral loss of 162 Da suggested that there are two mannoses on this peptide. As a labile modification, the O-linked mannoses were preferably removed in the MS2 spectra. The product ions still carrying the modification are presumably weak if observed. Indeed, the ions of 695.57(y11 + 3242+), 738.96 (y12 + 3242+), 1080.51(y8 + 324), 1389.64(y11 + 324) and 1476.57 (y12 + 324) with two mannoses attached and 1314.61 (y12 + 162) with one mannose left have much weaker intensity compared to their unmodified counterparts (533.30, 576.92, 756.56, 1065.53, 1152.62). Even though there are two Ser residues (120 and 126) in the peptide, the presence of 695.57(y11 + 3242+), 1080.51 (y8 + 324) and 1389.64(y11 + 324) indicates that the two mannoses, in a linear configuration, are solely located at Ser126. Additionally, the same peptide modified with a single mannose was also observed (Fig. S4). The product ions with the mannose attached, 847.32 (y7 + 162), 918.29(y8 + 162), 1130.62(y10 + 162), 1227.52 (y11 + 162), 1243.26(b12 + 162) and 1314.59(y12 + 162), are observed and the presence of 847.32 (y7 + 162) 918.29(y8 + 162), 1130.62(y10 + 162) and 1227.52(y11 + 162) suggests that the single mannose is also located at Ser126. It is of interest to note that the same Ser126 modified with O-linked glycosylation on endogenous and CHO expressed recombinant human EPO is also the O-mannosylation site for the glycoengineered Pichia-derived rhEPO. Even though O-linked mannosylation has been reported on mammalian cell-derived proteins, such as antibodies and endogenous proteins,[20,21] Pichia has a distinctly different O-glycosylation pathway initiating in ER instead of the Golgi where
wileyonlinelibrary.com/journal/jms
Conclusions
Acknowledgements The authors would like to acknowledge the support from the Strain Development, Purification, Analytical, High Throughput Screening and Fermentation groups at GlycoFi.
References [1] S. E. Graber, S. B. Krantz. Erythropoietin and the control of red cell production. Annu. Rev. Med. 1978, 29, 51. [2] D. Fliser, F. H. Bahlmann, K. deGroot, H. Haller. Mechanisms of Disease: erythropoietin – an old hormone with a new mission? Nat. Rev. Cardio. 2006, 3, 563. [3] P. H. Lai, R. Everett, F. F. Wang, T. Arakawa, E. Goldwasser. Structural characterization of human erythropoietin. J. Biol. Chem. 1986, 261, 3116. [4] M. Takeuchi, S. Takasaki, M. Shimada, A. Kobata. Role of sugar chains in the in vitro biological activity of human erythropoietin produced in recombinant Chinese hamster ovary cells. J. Biol. Chem. 1990, 265, 12127. [5] E. Tsuda, G. Kawanishi, M. Ueda, S. Masuda, R. Sasaki. The role of carbohydrate in recombinant human erythropoietin. Eur. J. Biochem. 1990, 188, 405. [6] S. Erbayraktar, G. Grasso, A. Sfacteria, Q. Xie, T. Coleman, M. Kreilgaard, L. Torup, T. Sager, Z. Erbayraktar, N. Gokmen, O. Yilmaz, P. Ghezzi, P. Villa, M. Fratelli, S. Casagrande, M. Leist, L. Helboe, J. Gerwein, S. Christensen, M. A. Geist, L.Ø. Pedersen, C. Cerami-Hand, J. P. Wuerth, A. Cerami, M. Brines. Asialoerythropoietin is a nonerythropoietic cytokine with broad neuroprotective activity in vivo. Proc. Natl. Acad. Sci. U. S. A. 2003, 100, 6741. [7] R. J. Darling, U. Kuchbhotla W. Glaesner, R. Micanovic, D. R. Witcher, J. Beals. Glycosylation of erythropoietin affects receptor binding kinetics: role of electrostatic interactions. Biochem. 2002, 41, 14524. [8] J. C. Egrie, J. K. Browne. Development and characterization of a novel erythropoiesis stimulating proteins (NESP). Nephrol. Dial. Transplant. 2001, 16(Suppl 3), 3. [9] M. Takeuchi, S. Takasaki, H. Miyazaki, T. Kato, S. Hoshi, N. Kochibe, A. Kobata. Comparative study of the asparagine-linked sugar chains of human erythropoietin purified from urine and the culture medium of recombinant Chinese hamster ovary cells. J. Biol. Chem. 1988, 263, 3657. [10] P. Hossler, S. F. Khattak, Z. J. Li. Optimal and consistent protein glycosylation in mammalian cell culture. Glycobiology 2009, 19, 936. [11] E. Llop, R. Gutiérrez-Gallego, J. Segura, J. Mallorquí, J. A. Pascual. Structural analysis of the glycosylation of gene-activated erythropoietin (epoetin delta, Dynepo). Anal. Biochem. 2008, 383, 243.
Copyright © 2013 John Wiley & Sons, Ltd.
J. Mass Spectrom. 2013, 48, 1308–1317
Glycosylation of Pichia-derived rhEPO by MS [12] S. Elliott, J. Giffin, S. Suggs, E. P. Lau, A. R. Banks. Secretion of glycosylated human erythropoietin from yeast directed by the α-factor leader region. Gene 1989, 79, 167. [13] C. Ciaccio, A. Gambacurta, G. De Sanctis, D. Spagnolo, C. Sakarikou, G. Petrella, M. Coletta. rhEPO (recombinant human eosinophil peroxidase): expression in Pichia pastoris and biochemical characterization. Biochem. J. 2006, 395, 295. [14] J. Nett, S. Gomathinayagam, S. R. Hamilton, B. Gong, R. C. Davidson, M. Du, D. Hopkins, T. Mitchell, M. R. Mallem, A. Nylen, S. S. Shaikh, N. Sharkey, G. C. Barnard, V. Copeland, L. Liu, R. Evers, Y. Li, P.M. Gray, R. B. Lingham, D. Visco, G. Forrest, J. DeMartino, T. Linden, T. I. Potgieter, S. Wildt, T. A. Stadheim, M. d’Anjou, H. Li, N. Sethuraman. Optimization of erythropoietin production with controlled glycosylation-PEGylated erythropoietin produced in glycoengineered Pichia pastoris. J. Biotechnol. 2012, 157, 198. [15] I. Burnina, E. Hoyt, H. Lynaugh, H. Li, B. Gong. A cost-effective platebased sample preparation approach for antibody N-glycan analysis. J. Chromatogr. A 2013, 1307, 201. [16] J. Bones, N. McLoughlin, M. Hilliard, K. Wynne, B. L. Karger, P. M. Rudd. 2D-LC analysis of BRP 3 erythropoietin N-glycosylation using anion exchange fractionation and hydrophilic interaction UPLC reveals long poly-N-acetyl lactosamine extensions. Anal. Chem. 2011, 83, 4154.
[17] R. S. Rush, P. L. Derby, D. M. Smith, C. Merry, G. Rogers, M. F. Rohde, V. Katta. Microheterogeneity of Erythropoietin Carbohydrate Structure. Anal. Chem. 1995, 67, 1452. [18] M. Ohta, N. Kawasaki, S. Hyuga, M. Hyuga, T. Hayakawa. Selective glycopeptide mapping of erythropoietin by on-line highperformance liquid chromatography–electrospray ionization mass spectrometry. J. Chromatogr. A 2001, 910, 1. [19] R. A. Dwek, C. J. Edge, D. J. Harvey, M. R. Wormald, R. B. Parekh. Analysis of glycoprotein-associated oligosaccharides. Annu. Rev. Biochem. 1993, 62, 65. [20] T. Martinez, D. Pace, L. Brady, M. Gerhart, A. Balland. Characterization of a novel modification on IgG2 light chain: evidence for the presence of O-linked mannosylation. J. Chromatogra. A. 2007, 1156, 183. [21] T. Sasaki, H. Yamada, K. Matsumura, T. Shimizu, A. Kobata, T. Endo. Detection of O-mannosyl glycans in rabbit skeletal muscle α-dystroglycan. Biochim. Biophys. Acta 1998, 1425, 599.
Supporting information Additional supporting information may be found in the online version of this article at the publisher’s web site.
1317
J. Mass Spectrom. 2013, 48, 1308–1317
Copyright © 2013 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jms
Revised: 30 September 2013
Accepted: 1 October 2013
Published online in Wiley Online Library
(wileyonlinelibrary.com) DOI 10.1002/jms.3291
Glycosylation characterization of recombinant human erythropoietin produced in glycoengineered Pichia pastoris by mass spectrometry Bing Gong,* Irina Burnina, Terrance A. Stadheim and Huijuan Li Glycosylation plays a critical role in the in vivo efficacy of both endogenous and recombinant erythropoietin (EPO). Using mass spectrometry, we characterized the N-/O-linked glycosylation of recombinant human EPO (rhEPO) produced in glycoengineered Pichia pastoris and compared with the glycosylation of Chinese hamster ovary (CHO) cell-derived rhEPO. While the three predicted N-linked glycosylation sites (Asn24, Asn38 and Asn83) showed complete site occupancy, Pichia- and CHO-derived rhEPO showed distinct differences in the glycan structures with the former containing sialylated bi-antennary glycoforms and the latter containing a mixture of sialylated bi-, tri- and tetra-antennary structures. Additionally, the N-linked glycans from Pichiaproduced rhEPO were similar across all three sites. A low level of O-linked mannosylation was detected on Pichia-produced rhEPO at position Ser126, which is also the O-linked glycosylation site for endogenous human EPO and CHO-derived rhEPO. In summary, the mass spectrometric analyses revealed that rhEPO derived from glycoengineered Pichia has a highly uniform bi-antennary Nlinked glycan composition and preserves the orthogonal O-linked glycosylation site present on endogenous human EPO and CHOderived rhEPO. Copyright © 2013 John Wiley & Sons, Ltd. Additional supporting information may be found in the online version of this article at the publisher’s web site. Keywords: N-linked glycosylation; O-linked glycosylation; Pichia pastoris; rhEPO; mass spectrometry
Introduction
1308
Human erythropoietin (EPO) is a multifunctional glycoprotein hormone with the major physiological role of regulating the level of red blood cells by promoting their proliferation, maturation and survival.[1,2] Synthesized mainly in the kidney and liver in adults, EPO comprises 165 amino acids after C-terminal processing during maturation. It contains three N-linked glycosylation sites (Asn24, Asn38 and Asn83) and a single O-linked glycan (Ser126).[3] The presence of N-linked glycosylation is critical for both in vitro and in vivo activity as PNGase F deglycosylated rhEPO demonstrated little biological activity, which was likely due to the loss of the required active conformation resulting from the removal of the core glycans.[4,5] Additionally, the actual glycosylation compositions, especially the terminal sialylation, greatly influences in vivo efficacy. While the sialylation does not contribute to EPO receptor binding, asialylated rhEPO has been shown to be nonerythropoietic in vivo. The diminishment of in vivo efficacy most likely resulted from the rapid serum clearance of EPO.[6,7] As such, efforts extending the serum half-life of rhEPO have become a focus for pursuing more efficacious rhEPO therapeutics. Darbepoetin alfa illustrates one such endeavor. In this particular case, two extra N-linked glycosylation sites were engineered at positions Asn30 and Asn88 through mutagenesis resulting in a significantly longer serum half-life and improved in vivo efficacy when compared to the native rhEPO sequence.[8] The glycosylation pattern of rhEPO can differ substantially when produced in different host cells. Currently, rhEPO is mainly produced in
J. Mass Spectrom. 2013, 48, 1308–1317
Chinese hamster ovary (CHO) cells, which shows predominantly branched and sialylated structures enriched through purification. While similar to the glycan structures from endogenous EPO, several differences exist. For instance, CHO cells lack a functional α-2,6-sialyltransferase (EC 2.4.99.1). As such, the linkage of sialic acid on CHO-derived rhEPO is solely α-2,3, while in human, both α-2,3 and α-2,6 were observed.[9] Then, the N-Glycolylneuraminic acid, a nonhuman and immunogenic structure, is present on recombinant proteins produced in CHO cells.[10] A recombinant epoetin delta (Dynepo) from an engineered human fibrosarcoma cell line HT-1080 has been approved in the European Union. It has a more human-like glycosylation profile, most likely because human cells were used for production.[11] The expression of rhEPO from wild-type yeasts including Saccharomyces cerevisiae and Pichia pastoris has been reported.[12,13] While showing both in vitro and in vivo activity, they contain typical fungal-type, high mannose glycan structures, which results in a relatively short in vivo half-life of rhEPO presumably due to the rapid clearance through mannose-binding protein (MBP) and other lectins. Recently, rhEPO expressed from a glycoengineered Pichia has been reported.[14] Through a series of genetic engineering steps, glycoengineered Pichia is capable of decorating glycoproteins
* Correspondence to: Bing Gong, GlycoFi, Biologics Discovery, Merck & Co., Inc., 16 Cavendish Court, Lebanon, NH 03766, USA. E-mail: [email protected] GlycoFi, Biologics Discovery, Merck & Co., Inc., 16 Cavendish Court, Lebanon, NH, 03766, USA
Copyright © 2013 John Wiley & Sons, Ltd.
Glycosylation of Pichia-derived rhEPO by MS with human-type bi-antennary sialylated glycans. In this work, we characterized the N- and O-linked glycosylation of glycoengineered Pichia-derived rhEPO with regard to the sitespecific N-linked glycosylation composition and the location and structure of O-linked mannosylation through mass spectrometry.
Material and experiments Materials Glycoengineered Pichia-expressed recombinant human EPO (hereafter referred to as Pichia rhEPO) was produced in-house. CHO cell expressed rhEPO (CHO rhEPO) was purchased from EMD
Millipore (Billerica, MA). Recombinant PNGase F, neuraminidase, β1-4 galactosidase and β-N-acetylglucosaminidase were from New England Biolabs (Ipswich, MA). 2,5-Dihydroxybenzoic acid and 3,5-dimethoxy-4-hydroxycinnamic acid (sinapinic acid) matrix-assisted laser desorption/ionization (MALDI) matrices were from Protea Biosciences (Morgantown, WV). Purchased from Sigma Aldrich (St. Louis, MO) were 1 M triethylammonium bicarbonate (TEAB) pH 8.5, 8 M guanidine hydrochloride (GuHCl), iodoacetamide (IAM), dithiothreitol (DTT), endoproteinase Glu-C and trypsin. InstantAB was obtained from ProZyme (Hayward, CA); 0.5 M tris(2carboxyethyl) phosphine was from Pierce (Rockford, IL); HPLC grade water, LC/MS grade water, and acetonitrile with 0.1% (v/v) formic
Figure 1. Intact molecular weight measurement of Pichia (top) and Chinese hamster ovary (CHO) (bottom) produced recombinant human erythropoietin (rhEPO) by matrix-assisted laser desorption/ionization-time-of-flight. The inset shows the LC-MS analysis of Pichia rhEPO.
J. Mass Spectrom. 2013, 48, 1308–1317
Copyright © 2013 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jms
1309
Figure 2. LC-MS analyses of recombinant human erythropoietin (rhEPO) after PNGase F and neuraminidase treatments. The PNGase F (a) and neuraminidase (c) treatments reveal that Pichia rhEPO is largely unmodified and only a small portion has O-linked glycosylation. PNGase F-treated CHO rhEPO (b) shows that the majority is modified with O-linked glycosylation and the heterogeneity of CHO rhEPO resulted from the glycans is demonstrated after neuraminidase treatment (d).
B. Gong et al. acid or 0.1% trifluoroacetic acid (TFA) were purchased from Fisher Scientific (Pittsburgh, PA); 96-well hydrophobic Immobilon-P polyvinylidene fluoride membrane plate was from Millipore (Billerica, MA); and 96-well Pall Supor® hydrophilic membrane plate was from Pall Life Sciences (Ann Arbor, MI). Glycoprotein analysis Intact CHO and Pichia rhEPO were reconstituted at 0.5 mg/ml in 20 mM sodium phosphate pH 7.4 buffer. For intact molecular weight measurement using MALDI-time-of-flight (TOF), 0.5 μl protein samples were spotted on a stainless MALDI target plate and mixed with 1 μl sinapinic acid. The samples were dried under vacuum at room temperature. All the MALDI-TOF analyses were performed on a Voyager-DE Biospectromtry workstation from Applied Biosystems (Foster City, CA). The settings for MALDITOF were as follows: accelerating voltage, 25 000 V; grid, 93%; guide wire, 0.05%; delay time, 750 ns; 100 shots per spectrum
were averaged; and data range was from 15 000 to 50 000 m/z. For the analysis using LC-MS, 10 ug rhEPO was digested with 0.5 μl PNGase F and neuraminidase at 37 °C overnight (~16 h), respectively. The digested samples were loaded onto an Agilent Q-TOF 6520 mass spectrometer coupled with an Agilent 1200 HPLC (Agilent, Santa Barbara, CA). An in-line MassPREP Micro desalt cartridge (Waters, Milford, MA) was used to remove salts from the samples prior to directing the sample to the mass spectrometer. The proteins were eluted off the cartridge using a one-step gradient: 100% buffer A (0.1% formic acid in water) flowed at 2 ml/min for 1 min, then the flow was changed to 100% buffer B (0.1% formic acid, 10% water and 90% acetonitrile) at 0.8 ml/min from 1 to 1.5 min. From 1.5 to 4 min, the flow rate was decreased to 0.5 ml/min, and the LC-MS data were collected. At the end of the gradient, a 1 min post-run of 100% buffer A at 2 ml/min was carried out to equilibrate the cartridge. The MS data were recorded in the profile mode from 500 to 3200 m/z. The dual electrospray ionization ion source was set as follows: gas
1310
Figure 3. Matrix-assisted laser desorption/ionization-time-of-flight spectra of fluorescently labeled N-linked glycans from Chinese hamster ovary (CHO) and Pichia-produced recombinant human erythropoietin. Untreated glycans from CHO (a) and Pichia (e) were monitored in the negative mode. Neuraminidase, neuraminidase/galactosidase and neuraminidase/galactosidase/hexominidase digested CHO (b, c and d), and Pichia (f, g and h) were analyzed in the positive mode.
wileyonlinelibrary.com/journal/jms
Copyright © 2013 John Wiley & Sons, Ltd.
J. Mass Spectrom. 2013, 48, 1308–1317
Glycosylation of Pichia-derived rhEPO by MS temp, 350 °C; drying gas, 13 l/min; nebulizer, 45 psig; fragmentor, 150 V; skimmer, 65 V; Oct1 RF VPP, 750 V; and Vcap, 3500 V. The MS spectra were analyzed using MassHunter software.
gradient was set for 30% buffer A from 0 to 1 min, then to 54% buffer A in 10 min. The settings for the FLR were excitation at 278 nm, emission at 344 nm, post-translational modification gain at 1 and data rate at 80 pts/s.
N-linked glycosylation analysis N-linked glycans were released and fluorescently labeled following a method described by Burnina et al.[15] Briefly, 20 μg Pichia and CHO rhEPO in 50 μl, 20 mM sodium phosphate pH 7.4 buffer was denatured and reduced by mixing with 150 μl, 10 M GuHCl, 10 mM tris(2-carboxyethyl) phosphine and incubation at 50 °C for 10 min in a 96-well hydrophobic Immobilon-P polyvinylidene fluoride membrane plate. The plate was subsequently washed three times with 200 μl TEAB pH 8.5 buffer. Added was 1 μl PNGase F in 50 μl TEAB pH 8.5 buffer, and the deglycosylation proceeded at 50 °C for 30 min. Then the released glycans in TEAB pH 8.5 buffer were labeled with 5 μl pre-constituted instantAB at 37 °C for 5 min. The labeled glycans were cleaned-up on a 96-well Pall Supor® hydrophilic membrane plate. Deposited on a MALDI target plate and mixed with 1 μl dihydroxybenzoic acid was 0.5 μl of labeled glycans. For glycans without neuraminidase treatment, they were analyzed in the negative mode. For the samples subject to neuraminidase treatment, they were analyzed in the positive mode. The settings for MALDI-TOF were as follows: accelerating voltage, 20 000 V; grid, 90.5% for the positive mode and 94% for the negative mode; guide wire, 0.01% for the positive mode and 0.08% for the negative mode; delay time, 100 ns for the positive mode and 225 ns for the negative mode; 100 shots per spectrum were averaged; and mass range was from 1000 to 5000 Da. Quantitative glycan analysis was performed on a Waters Acquity UPLC with a fluorescence detector (FLR-UPLC) and an Acquity BEH glycan column (2.1 × 100 mm). Mixed with 50 μl, 90% acetonitrile was 1 μl of labeled glycans; 10 μl was injected for analysis. The analyses were performed under the following conditions: column compartment temperature, 60 °C; sample manager temperature, 10 °C; flow rate, 0.4 ml/min; buffer A was 100 mM formate pH 4.6; and buffer B was acetonitrile. The
Peptide mapping For trypsin digestion, Pichia rhEPO (150 μg) was denatured and reduced in 6 M GuHCl and 20 mM DTT at 50 °C for 1 h. The sample was then cooled to room temperature and alkylated by incubating with 50 mM IAM for 45 min. The excess IAM was consumed by incubation with 5 mM DTT at room temperature for 10 min. All these chemicals were removed by buffer exchanging the sample to 25 mM ammonium bicarbonate pH 7.8 through a 5 KDa MW cutoff Millipore spin column. Trypsin (5 μg) was added to the sample, and the digestion proceeded at 37 °C overnight (~16 h). For Glu-C digestion, 300 μg Pichia rhEPO was buffer exchanged to 25 mM ammonium bicarbonate pH 7; 8. 10 μg Glu-C and 5 μl 100% acetonitrile were added to the sample for digestion, and the final volume was adjusted to 100 μl with 25 mM ammonium bicarbonate pH 7.8. The digestion proceeded at 25 °C overnight (~16 h). For PNGase F treatment, the Glu-C digestion was boiled for 20 min, and 10 μl was taken out and treated with 0.2 μl PNGase F after it was cooled to room temperature. Tryptic peptide mapping (10 μg trypsin digested rhEPO) was performed on a Thermo LTQ mass spectrometer with the Survey HPLC and Advion Triversal Nanomate. The column was a 250 × 4.6 mm Phenomenex Jupiter® Proteo column. Buffer A and buffer B were 0.1% formic acid in water and acetonitrile, respectively. The flow rate of HPLC was 1000 μl/min, and 300 nl/min was diverted to the mass spectrometer through the Nanomate. The HPLC gradient was 2–35% buffer B in 70 min, then to 98% in 10 min. Column temperature was maintained at 30 °C. The mass spectrometer settings were spray voltage (kV), 2.0; capillary Temp, 115; capillary voltage, 5; tube lens (V), 77; scan range, 200–2000 m/z; normalized collision energy, 25%; and activation time, 30 ms.
J. Mass Spectrom. 2013, 48, 1308–1317
Copyright © 2013 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jms
1311
Figure 4. Glu-C peptide mapping of Pichia recombinant human erythropoietin. The shifted peaks between before (a) and after (b) PNGase F digestion reveal the positions of glycopeptides, except for the Asn38 glycopeptide, which is resistant to PNGase F digestion. The peaks containing N-linked glycans are noted with retention time and highlighted with relevant sites.
B. Gong et al. Glu-C digested rhEPO (10 μg) was analyzed on the Thermo LTQ Orbitrap XL mass spectrometer with Waters Acquity UPLC system and a BEH C18 (1.7 μm, 2.1 × 50 mm) column. Buffer A was 0.1% TFA in water, and buffer B was 0.1% TFA in acetonitrile. The flow rate was 300 μl/min. The gradient was 3% buffer B for 5 min, then to 30% in 30 min. The settings for the mass spectrometer were sheath gas (arb), 30; auxgas flow rate, 5; sweep gas flow rate, 2; spray voltage (kV), 4.5; capillary Temp, 275; capillary voltage, 47; tube lens (V), 90; scan range, 300–2000 m/z; normalized collision energy, 35%; and activation time, 30 ms. Full scan using FTMS
with a resolution of 30 000 was used for the full scan, and the ion trap was used for product ion detection.
Results and discussion Glycoprotein analysis The molecular weight measurement of the intact rhEPO from CHO and Pichia was first carried out on a MALDI-TOF mass spectrometer. An initial attempt of analyzing CHO rhEPO on LC-MS
1312
Figure 5. The spectra of the glycopeptides of Asn24(a), Asn38 (b) and Asn83 (c) show that both A2 and A1 glycan structures are present and the former is the major species.
wileyonlinelibrary.com/journal/jms
Copyright © 2013 John Wiley & Sons, Ltd.
J. Mass Spectrom. 2013, 48, 1308–1317
Glycosylation of Pichia-derived rhEPO by MS was not successful, as the heterogeneity of the glycosylation, to the most part the sialylation, prevented us from acquiring any resolvable spectrum. As depicted in Fig. 1, Pichia rhEPO shows a much better resolved MALDI-TOF spectrum than CHO rhEPO. Three distinct peaks representing various levels of sialylation are clearly presented. As the predicted protein mass for rhEPO is ~18 KDa, the glycans account for ~26% of the total mass with an average N-glycan mass of ~2 KDa. For CHO rhEPO, a single broad peak centered ~29 KDa was observed. The broadness reflects the heterogeneity of the glycosylation. Glycans account for ~38% of the total mass of CHO rhEPO with an average mass of ~3.3 KDa. The greater homogeneity of the glycosylation on Pichia rhEPO enabled us to analyze the intact molecule using LC-MS (the inset of Fig. 1). It is also apparent that there is a clear discrepancy in the level of sialylation when Pichia rhEPO was analyzed by MALDI-TOF versus LC-MS. From the LC-MS analysis, Pichia rhEPO is almost completely fully sialylated with other minor species. In contrast, while the MALDI-TOF spectra do show that the fully sialylated species is the majority, however, the species with partial sialylation are also substantial. This difference could be caused by the different ionization efficiencies of differently glycosylated rhEPO in these two techniques. A more accurate measurement of the glycan composition of Pichia rhEPO was carried out using fluorescently labeled glycans and is presented in the following section. Pichia and CHO rhEPO treated with PNGase F and neuraminidase were analyzed using LC-MS. Figure 2a and b shows the deconvoluted spectra of PNGase F deglycosylated Pichia- and CHO-derived rhEPO. A main predominant peak of 18 239 Da was observed in the spectra of PNGase F-treated Pichia rhEPO and matches the calculated mass of rhEPO protein (Fig. 2a). Additionally, a very minor peak of 18 563 Da implying the potential O-linked mannosylation is also present.
In the deconvoluted spectrum of PNGase F-treated CHO rhEPO (Fig. 2b), two major peaks with masses of 18 896 and 19 187 Da represent the rhEPO modified with mono- and di- sialylated O-linked glycans, respectively. The observed peak of 18 239 Da indicates that a minor portion of CHO rhEPO was not O-glycosylated, which is different from Pichia rhEPO where only a small fraction is modified by O-linked glycosylation. Figures 2c and d show the deconvoluted spectra of neuraminidase digested Pichia and CHO rhEPO. A single main peak of 23 106 Da appears for Pichia rhEPO (Fig. 2c), which corresponds to the loss of six sialic acids (two sialic acid residues from each of the three bi-antennary oligosaccharides) from the intact rhEPO mass of 24 854 Da observed (the inset in Fig. 1). As evident in Fig. 2d, the glycosylation of CHO rhEPO remains heterogenous even after the sialic acids are removed. At this stage, it is quite difficult to determine the actual structures that each peak represents because of the interference due to O-linked glycosylation. N-linked glycosylation analysis The N-linked glycosylation of rhEPO was further characterized first by sequential glycosidase treatments of released glycans. To accurately quantify the relative amounts of various glycan structures, we used a fluorescence tag, instantAB from ProZyme, to label the released N-linked glycans. Figure S1 shows the labeled glycans analysis using FLR-UPLC. The glycans are predominantly A1 and A2 (the nomenclature of these glycans are shown in Fig. S1), and they account for ~8% and 92%, respectively. In addition to enable accurate quantitative analysis, fluorescence tags have been found to improve the ionization efficiency during the mass spectrometry analysis of glycans by rendering the otherwise hydrophilic glycans more hydrophobic.
Table 1. Glycopeptides identified through Glu-C peptide mapping Glycosylation site
Asn24
Peptide sequence
Deglycosylated peptide
*
AENITTGCEHCSLNE
AENITTGCAEHCSLNE **
AENITTGCAEHCSLNE Asn38
NITVPDTKVNFYAWKRME
Asn83
AVLRGQALLVNS AVLRGQALLVNSSQPWEPLQLHVD AVLRGQALLVNSSQPWEPLQ LHVDKAVSGLRS # AVLRGQALLVNSSQPWE PLQ LHVDKAVSGLRS # AVLRGQALLVNSSQPWE PLQLHVD AVLRGQALLVNSSQPWEPLQLHVD KAVSGLRSLTTLLRALGAQKE
Glycopeptide
Retention time (min)
Detected mass (Da)
Retention time (min)
Detected mass (Da)
8.4
1627.7
7.3
11
1689.68
10.0
13.5
1889.88
12.3
2212.55
17.0
14.9 20.8 21
1240.74 2671.44 3470.94
12.7 19.1 19.7
23.2
3451.92
21.8
24.5 27.8
2653.44 4865.76
22.7 26.6
3542.94 3834.48 3603.39 3894.51 3803.58 4094.70 4126.83 4417.92 3445.53 4877.25 5383.64 5674.72 5367.54 5656.71 4858.23 6781.48 7071.56
Glycan
A1 A2 A1 A2 A1 A2 A1 A2 A2 A2 A1 A2 A1 A2 A2 A1 A2
J. Mass Spectrom. 2013, 48, 1308–1317
Copyright © 2013 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jms
1313
The observed peptide masses and their corresponding glycans are based on the observed masses except for Asn38 for which the peptide mass is based on the sequence, as the aglycosylated peptide was not observed because of its resistance to PNGase F digestion. Modifications at Cys residues (noted with *) in the Asn24 and deamidation on the Asn83 (noted with #) peptides were detected and resulted in the different elution times for these peptides.
B. Gong et al. InstantAB dye from ProZyme requires a mild labeling condition that helps in preserving the glycan structures especially those labile forms such as sialylated glycoforms. However, one caveat to this approach is that instantAB labeled glycans gain a mass addition of 162 Da, which coincides with the mass of a single hexose. Therefore, a high labeling efficiency is generally required to assign the labeled glycan structure using the mass spectrometry reliably. The spectra of the labeled glycans are shown in Fig. 3. Figure 3a–d shows the N-linked glycans from CHO rhEPO, which shows a heterogeneous mixture of glycoforms. The spectra in Fig. 3a and e were obtained by operating MALDI-TOF in the negative mode in order to reveal the sialylation more clearly, while those in Fig. 3b–g were from the positive mode analysis. A mixture of sialylated bi-, tri- and tetra-antennary structures is observed. We assigned the major peaks with potential structures solely based on the observed masses, and the actual linkages might be different from what is depicted here. For example, the sialic acids in the 2900.48 ion could be on any two of the three branches, not necessarily the two labeled. Additionally,
there are Galactose-N-acetylglucosylamine (LacNAc) extensions observed in Fig. 3b and c where a series of 365 Da (LacNAc) additions is present between the main peaks after neuraminidase and neuraminidase/galactosidase treatments. Like the sialic acids, the LacNAc extensions could be on any of the branches.[16] After treatment with neuraminidase, galactosidase and N-acetylglucosaminidase, a fucosylated core Man 3 structure was observed (Fig. 3d). Figure 3e–h shows the spectra of the N-linked glycans released from Pichia rhEPO. As shown in Fig. 3e, the glycans primarily comprise the fully sialylated bi-antennary structure A2 with a minor species of partially sialylated bi-antennary A1, which is consistent with the results we obtained from FLR-UPLC analysis. After digestion with neuraminidase (Fig. 3f), the structures collapsed to G2 as expected. Further digested with neuraminidase/galactosidase and neuraminidase/galactosidase/N-acetylglucosaminidase, the glycans were converted to G0 and Man 3 (Fig. 3g and h), respectively. All of these glycan structures on Pichia rhEPO are afucosylated.
1314
Figure 6. MS2 collision-induced-dissociation spectra reveal the breakage after Ser104. The MS2 spectra of 1157.68 m/z (a) identify it as the peptide AVLRGQALLVNSSQPWEPLQLHVDKAVSGLRSer104 and 707.70 m/z (b) as Leu105TTLLRALGAQKE.
wileyonlinelibrary.com/journal/jms
Copyright © 2013 John Wiley & Sons, Ltd.
J. Mass Spectrom. 2013, 48, 1308–1317
Glycosylation of Pichia-derived rhEPO by MS Besides the macro-heterogeneity, the N-linked glycans on CHO rhEPO show site-wise micro-heterogeneity.[17,18] We analyzed the site-specific glycosylation of Pichia rhEPO through peptide mapping with Glu-C digestion. Figure 4 shows the UV absorbance chromatograms of Glu-C digested Pichia rhEPO treated with and without PNGase F. The peaks shifted upon PNGase F treatment and highlighted with retention time indicate the presence of glycopeptides. Among the shifted peaks, the glycopeptides containing Asn24 were identified at 7.3, 10 and 12.3 min on the chromatogram of the sample with only Glu-C treatment (Fig. 4a), and the corresponding deglycosylated peptides were identified at 8.4, 11.1 and 13.5 min on the chromatogram of the sample treated with both Glu-C and PNGase F (Fig. 4b). The Asn83 containing glycopeptides were identified at 12.7, 19.1, 19.7, 21.8, 22.7 and 26.6 min in Fig. 4a, and their corresponding deglycosylated peptides appeared at 14.9, 20.8, 21.0, 23.2, 24.5 and 27.8 min in Fig. 4b. In the shift peaks, we did not observe any peptides harboring Asn38. Instead, the peptide, N38ITVPDTKVNFYAWKRME, was identified at 17 min in both Fig. 4a and b. The glycans at Asn38 peptides were resistant to PNGase F digestion as the glycans are attached to the very first amino acid on its N-terminus. It is well known that the N-linked glycans at the asparagine, which is on either the N- or C-terminus of proteins or peptides, are resistant to PNGase F digestion.[19]
The full scan spectra of the major glycopeptides eluted at 10 min (Asn24), 17 min (Asn38) and 19.7 min (Asn83) in Fig. 4a are shown in Fig. 5 to demonstrate the attached glycan structures. The ions of 1202.13 and 1299.50 in Fig. 5a represent the triply charged Asn24 glycopeptides with A1 and A2 structure, respectively. The full scan spectrum of the glycopeptide containing Asn38 is demonstrated in Fig. 5b. The ions of 1376.61 and 1473.64 are the A1 and A2 glycan modified glycopeptides with three charges. The ions of 1032.71 and 1105.48 are the A1 and A2 modified glycopeptides with four charges. The MS2 spectra of 1473.64 and 1376.61 ions are shown in Fig. S2. Figure 5c shows the spectrum of the glycopeptides contains Asn83. Both A2 and A1 modified glycopeptides with three and four charges were presented. Based on the MS signals, A2 modified glycopeptides account for the majority of glycopeptides for the three sites, which is in good agreement with the results of FLR-UPLC analysis of the fluorescently labeled glycans. The observed glycopeptides and their associated glycans were listed in Table 1. During the analysis of Glu-C peptide mapping, the glycopeptides containing Asn24 and Asn83 were observed in multiple peaks (Fig. 4 and Table 1). After examining the spectra, we found that they were due to the presence of Cys modifications and Ser-related cleavages that occurred, most likely, during the sample preparation for Glu-C peptide mapping, as we did not observe any of those
J. Mass Spectrom. 2013, 48, 1308–1317
Copyright © 2013 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jms
1315
Figure 7. MS2(a) and MS3(b) collision-induced-dissociation spectra of the ion of 895 m/z show the O-linked mannosylation at Ser126. The product ion of 733 m/z in the MS2 was used as the parent ion in the MS3 experiment. The product ions indicating the location of the O-linked glycosylation at Ser126 are highlighted with underline.
B. Gong et al. modifications and cleavages in the tryptic peptide mapping and intact protein analysis. The modifications of Cys residues were found on the Asn24 peptides, and cleavages after Ser were observed on Asn83 peptides. Figure 6a shows the MS2 spectra of Asn83 glycopeptide (AVLRGQALLVN83SSQPWEPLQLHVDKAVSGLRS104) with a C-terminal Ser, and the MS2 spectrum of its neighboring C-terminal peptide (L105TTLLRALGAQKE) is demonstrated in Fig. 6b. Another Ser cleavage was also observed on this peptide (AVLRGQALLVN83S). These Ser cleavages might indicate the potential labile positions in the EPO molecule.
the mammalian mucin-type O-glycosylation starts. Our analysis of Pichia-expressed recombinant human granulocyte colonystimulating factor also showed that the O-glycosylation site used by human and CHO cells was also utilized by glycoengineered Pichia pastoris for O-mannosylation (Gong B et al., manuscript submitted). It seems that the protein itself plays a more fundamental role than the expression host does in determining the O-glycosylation sites.
O-linked glycosylation analysis
The glycosylation of rhEPO produced in glycoengineered Pichia was characterized using MALDI-TOF and LC-MS. The majorities of N-linked glycans are highly sialylated bi-antennary structures. The glycosylation pattern among the three N-linked glycosylation sites is highly consistent. A small fraction of Pichia rhEPO is modified with O-linked mannosylation, which was determined to be on the Ser126 and consists primarily as a mannobiose structure. This study demonstrates the utility of mass spectrometry in analyzing both N- and O-linked glycosylation on the glycoproteins from glycoengineered Pichia pastoris.
1316
O-linked glycosylation has been identified at Ser126 on both endogenous and mammalian cell-produced recombinant EPO. A low level of O-linked mannosylation was observed on Pichia rhEPO through LC-MS analysis (18 563 Da in Fig. 2a). In order to determine whether the same Ser126 site is occupied on Pichia rhEPO, a tryptic peptide mapping was performed. Figure S3 showed the base peak chromatogram of LC-MS analysis of tryptic Pichia rhEPO. As glycosylation is a labile modification and not stable during the fragmentation process in the mass spectrometer, the LTQ mass spectrometer was operated in a neutral loss mode with the neutral masses set for 162, 81 and 54 Da. The entire rhEPO sequence was covered (the peptides and their corresponding retention times were listed in Table S1). The peptides containing the N-linked glycan sites were identified at 43.73 and 47.32 min and mostly modified with A2 glycans as described in the previous section. An MS3 event was triggered by a neutral loss of 162 Da from the ion of 895.23 m/z. Figures 7a and b show the MS2 and MS3 spectra of the 895.23 ion, respectively. Based on the MS3 spectra (Fig. 7b, the parent ion is 733 m/z), it was identified as the peptide, EAIS120PPDAAS126AAPLR, which contains two Ser residues, Ser120 and Ser126. The 895.23 ion was doubly charged, and the neutral loss of 162 Da suggested that there are two mannoses on this peptide. As a labile modification, the O-linked mannoses were preferably removed in the MS2 spectra. The product ions still carrying the modification are presumably weak if observed. Indeed, the ions of 695.57(y11 + 3242+), 738.96 (y12 + 3242+), 1080.51(y8 + 324), 1389.64(y11 + 324) and 1476.57 (y12 + 324) with two mannoses attached and 1314.61 (y12 + 162) with one mannose left have much weaker intensity compared to their unmodified counterparts (533.30, 576.92, 756.56, 1065.53, 1152.62). Even though there are two Ser residues (120 and 126) in the peptide, the presence of 695.57(y11 + 3242+), 1080.51 (y8 + 324) and 1389.64(y11 + 324) indicates that the two mannoses, in a linear configuration, are solely located at Ser126. Additionally, the same peptide modified with a single mannose was also observed (Fig. S4). The product ions with the mannose attached, 847.32 (y7 + 162), 918.29(y8 + 162), 1130.62(y10 + 162), 1227.52 (y11 + 162), 1243.26(b12 + 162) and 1314.59(y12 + 162), are observed and the presence of 847.32 (y7 + 162) 918.29(y8 + 162), 1130.62(y10 + 162) and 1227.52(y11 + 162) suggests that the single mannose is also located at Ser126. It is of interest to note that the same Ser126 modified with O-linked glycosylation on endogenous and CHO expressed recombinant human EPO is also the O-mannosylation site for the glycoengineered Pichia-derived rhEPO. Even though O-linked mannosylation has been reported on mammalian cell-derived proteins, such as antibodies and endogenous proteins,[20,21] Pichia has a distinctly different O-glycosylation pathway initiating in ER instead of the Golgi where
wileyonlinelibrary.com/journal/jms
Conclusions
Acknowledgements The authors would like to acknowledge the support from the Strain Development, Purification, Analytical, High Throughput Screening and Fermentation groups at GlycoFi.
References [1] S. E. Graber, S. B. Krantz. Erythropoietin and the control of red cell production. Annu. Rev. Med. 1978, 29, 51. [2] D. Fliser, F. H. Bahlmann, K. deGroot, H. Haller. Mechanisms of Disease: erythropoietin – an old hormone with a new mission? Nat. Rev. Cardio. 2006, 3, 563. [3] P. H. Lai, R. Everett, F. F. Wang, T. Arakawa, E. Goldwasser. Structural characterization of human erythropoietin. J. Biol. Chem. 1986, 261, 3116. [4] M. Takeuchi, S. Takasaki, M. Shimada, A. Kobata. Role of sugar chains in the in vitro biological activity of human erythropoietin produced in recombinant Chinese hamster ovary cells. J. Biol. Chem. 1990, 265, 12127. [5] E. Tsuda, G. Kawanishi, M. Ueda, S. Masuda, R. Sasaki. The role of carbohydrate in recombinant human erythropoietin. Eur. J. Biochem. 1990, 188, 405. [6] S. Erbayraktar, G. Grasso, A. Sfacteria, Q. Xie, T. Coleman, M. Kreilgaard, L. Torup, T. Sager, Z. Erbayraktar, N. Gokmen, O. Yilmaz, P. Ghezzi, P. Villa, M. Fratelli, S. Casagrande, M. Leist, L. Helboe, J. Gerwein, S. Christensen, M. A. Geist, L.Ø. Pedersen, C. Cerami-Hand, J. P. Wuerth, A. Cerami, M. Brines. Asialoerythropoietin is a nonerythropoietic cytokine with broad neuroprotective activity in vivo. Proc. Natl. Acad. Sci. U. S. A. 2003, 100, 6741. [7] R. J. Darling, U. Kuchbhotla W. Glaesner, R. Micanovic, D. R. Witcher, J. Beals. Glycosylation of erythropoietin affects receptor binding kinetics: role of electrostatic interactions. Biochem. 2002, 41, 14524. [8] J. C. Egrie, J. K. Browne. Development and characterization of a novel erythropoiesis stimulating proteins (NESP). Nephrol. Dial. Transplant. 2001, 16(Suppl 3), 3. [9] M. Takeuchi, S. Takasaki, H. Miyazaki, T. Kato, S. Hoshi, N. Kochibe, A. Kobata. Comparative study of the asparagine-linked sugar chains of human erythropoietin purified from urine and the culture medium of recombinant Chinese hamster ovary cells. J. Biol. Chem. 1988, 263, 3657. [10] P. Hossler, S. F. Khattak, Z. J. Li. Optimal and consistent protein glycosylation in mammalian cell culture. Glycobiology 2009, 19, 936. [11] E. Llop, R. Gutiérrez-Gallego, J. Segura, J. Mallorquí, J. A. Pascual. Structural analysis of the glycosylation of gene-activated erythropoietin (epoetin delta, Dynepo). Anal. Biochem. 2008, 383, 243.
Copyright © 2013 John Wiley & Sons, Ltd.
J. Mass Spectrom. 2013, 48, 1308–1317
Glycosylation of Pichia-derived rhEPO by MS [12] S. Elliott, J. Giffin, S. Suggs, E. P. Lau, A. R. Banks. Secretion of glycosylated human erythropoietin from yeast directed by the α-factor leader region. Gene 1989, 79, 167. [13] C. Ciaccio, A. Gambacurta, G. De Sanctis, D. Spagnolo, C. Sakarikou, G. Petrella, M. Coletta. rhEPO (recombinant human eosinophil peroxidase): expression in Pichia pastoris and biochemical characterization. Biochem. J. 2006, 395, 295. [14] J. Nett, S. Gomathinayagam, S. R. Hamilton, B. Gong, R. C. Davidson, M. Du, D. Hopkins, T. Mitchell, M. R. Mallem, A. Nylen, S. S. Shaikh, N. Sharkey, G. C. Barnard, V. Copeland, L. Liu, R. Evers, Y. Li, P.M. Gray, R. B. Lingham, D. Visco, G. Forrest, J. DeMartino, T. Linden, T. I. Potgieter, S. Wildt, T. A. Stadheim, M. d’Anjou, H. Li, N. Sethuraman. Optimization of erythropoietin production with controlled glycosylation-PEGylated erythropoietin produced in glycoengineered Pichia pastoris. J. Biotechnol. 2012, 157, 198. [15] I. Burnina, E. Hoyt, H. Lynaugh, H. Li, B. Gong. A cost-effective platebased sample preparation approach for antibody N-glycan analysis. J. Chromatogr. A 2013, 1307, 201. [16] J. Bones, N. McLoughlin, M. Hilliard, K. Wynne, B. L. Karger, P. M. Rudd. 2D-LC analysis of BRP 3 erythropoietin N-glycosylation using anion exchange fractionation and hydrophilic interaction UPLC reveals long poly-N-acetyl lactosamine extensions. Anal. Chem. 2011, 83, 4154.
[17] R. S. Rush, P. L. Derby, D. M. Smith, C. Merry, G. Rogers, M. F. Rohde, V. Katta. Microheterogeneity of Erythropoietin Carbohydrate Structure. Anal. Chem. 1995, 67, 1452. [18] M. Ohta, N. Kawasaki, S. Hyuga, M. Hyuga, T. Hayakawa. Selective glycopeptide mapping of erythropoietin by on-line highperformance liquid chromatography–electrospray ionization mass spectrometry. J. Chromatogr. A 2001, 910, 1. [19] R. A. Dwek, C. J. Edge, D. J. Harvey, M. R. Wormald, R. B. Parekh. Analysis of glycoprotein-associated oligosaccharides. Annu. Rev. Biochem. 1993, 62, 65. [20] T. Martinez, D. Pace, L. Brady, M. Gerhart, A. Balland. Characterization of a novel modification on IgG2 light chain: evidence for the presence of O-linked mannosylation. J. Chromatogra. A. 2007, 1156, 183. [21] T. Sasaki, H. Yamada, K. Matsumura, T. Shimizu, A. Kobata, T. Endo. Detection of O-mannosyl glycans in rabbit skeletal muscle α-dystroglycan. Biochim. Biophys. Acta 1998, 1425, 599.
Supporting information Additional supporting information may be found in the online version of this article at the publisher’s web site.
1317
J. Mass Spectrom. 2013, 48, 1308–1317
Copyright © 2013 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jms
