Characterization of glycosylation sites for a recombinant IgG1 monoclonal antibody

Journal of Chromatography A,1218 (2011) 8140–8149Contents lists available at SciVerse ScienceDirectJournal of ChromatographyAj o u r n a l h o m e p a g e :w w w.e l s e v i e r.c o m /l o c a t e /c h r o maCharacterization of glycosylation sites for a recombinant IgG1monoclonal antibody and a CTLA4-Ig fusion protein by liquid chromatography–mass spectrometry peptide mappingJacob Bongers a ,∗,John Devincentis a ,Jinmei Fu a ,Peiqing Huang b ,David H.Kirkley a ,Kirk Leister a ,Peiran Liu b ,Richard Ludwig b ,Kathleen Rumney a ,Li Tao b ,Wei Wu b ,Reb J.Russell ba Department of Biologics Process and Product Development,Bristol-Myers Squibb Company,6000Thompson Road,East Syracuse,NY 13057,USA bP.O.Box 5400,Princeton,NJ 08543,USAa r t i c l ei n f oArticle history:Received 15April 2011Received in revised form 27August 2011Accepted 30August 2011Available online 3 September 2011Key words:Antibody CTLA-4Fc-fusionGlycosylation LC–MSPeptide mapa b s t r a c tLiquid chromatography mass spectrometry (LC–MS)peptide mapping can be a versatile technique for characterizing protein glycosylation sites without the need to remove the attached glycans as in conven-tional oligosaccharide mapping methods.In this way,both N-linked and O-linked sites of glycosylation can each be directly identified,characterized,and quantified by LC–MS as intact glycopeptides in a single experiment.LC–MS peptide mapping of the individual glycosylation sites avoids many of the limitations of preparing and analyzing an entire pool of released N-linked oligosaccharides from all sites mixed together.In this study,LC interfaced to a linear ion trap mass spectrometer (ESI-LIT-MS)were used to characterize the glycosylation of a recombinant IgG1monoclonal antibody and a CTLA4-Ig fusion pro-tein with multiple sites of N-and O-glycosylation.Samples were reduced,S-carboxyamidomethylated,and cleaved with either trypsin or endoproteinase Asp-N.Enhanced detection for minor IgG1glycoforms (∼0.1to 1.0mol%level)was obtained by LC–MS of the longer 32-residue Asp-N glycopeptide (4+pro-tonated ion)compared to the 9-residue tryptic glycopeptide (2+ion).LC–MS peptide mapping was run according to a general procedure:(1)Locate N-linked and/or O-linked sites of glycosylation by selected-ion-monitoring of carbohydrate oxonium fragment ions generated by ESI in-source collision-induced dissociation (CID),i.e.204,366,and 292Da marker ions for HexNAc,HexNAc-Hex,and NeuAc,respec-tively;(2)Characterize oligosaccharides at each site via MS and e selected ion currents (SIC)to estimate relative amounts of each glycoform;and (3)Measure the percentage of site-occupancy by searching for any corresponding nonglycosylated peptide.© 2011 Elsevier B.V. All rights reserved.1.IntroductionThe increasing importance of recombinant DNA produced mon-oclonal antibodies and other glycoproteins as therapeutic agents in recent years has placed increasing emphasis on the exten-sive characterization of their chemical structures.Glycosylation is a posttranslational structural modification of secreted proteins manufactured by mammalian cell culture that can have a sig-nificant impact on their solubility,stability,pharmacokinetics,potency,and biological activities [1–16].The N-linked glycosyla-tion at the carboxamide side-chain of a particular asparagine (Asn)∗Corresponding author at:Department of Biologics Process and Product Devel-opment,Bristol-Myers Squibb Company,6000Thompson Road,East Syracuse,NY 13057,United States.Tel.:+13154319370,fax:+13154324821.E-mail address:jacob.bongers@ (J.Bongers).residue or O-linked glycosylation at the alcohol side chain of a particular serine (Ser)or threonine (Thr)residue is only partly determined by the gene DNA sequence and corresponding pro-tein amino acid sequence.Attachment of a carbohydrate precursor to a particular Asn by the oligosaccharide transferase enzyme requires the consensus sequence,Asn-Xaa-Ser/Thr.This consensus sequence is a necessary condition but not a sufficient condition for N-glycosylation,i.e.the enzyme may or may not glycosylate a par-ticular Asn,depending on other factors.The amino acid sequence features and other factors regulating O-linked glycosylation of Ser and Thr residues are even less well understood and less predictable than for those for N-glycosylation.Moreover,after translation of the gene on the ribosome,each population of branched oligosaccharide structures or N-glycans occupying a particular site diverges by a complex series of modifications mediated by multiple glycosidase and glycotransferase enzymes.This enzymatic processing gives rise to structural microheterogeneity in the mature glycoprotein,i.e.a0021-9673/$–see front matter © 2011 Elsevier B.V. All rights reserved.doi:10.1016/j.chroma.2011.08.089J.Bongers et al./J.Chromatogr.A1218 (2011) 8140–81498141heterogeneous population of multiple oligosaccharide structures at each site of glycosylation.To further complicate matters,the occu-pancy of a particular N-linked or O-linked site may or may not be complete[16,17].Clearly,these site-specific attributes of partial occupancy and glycan microheterogeneity place high demands on analytical methods for the structural characterization of glycopro-teins.Oligosaccharide mapping,in which an entire pool of oligosac-charides is released from the protein by hydrazinolysis or,more commonly,enzymatic hydrolysis with peptidyl N-glycosidase (PNGase)or other glycosidase treatment(PNGase F,PNGase A, EndoH,etc.),is widely used for the characterization of pro-tein glycosylation[18–25].The various analytical methods for oligosaccharide mapping include high-pH anion-exchange chro-matography with pulsed amperometric detection(HPAEC-PAD) and a number of different chromatography or electrophoresis based separations,matrix-assisted laser desorption mass spec-trometry(MALDI-TOF MS)and various other MS techniques,and combinations of these including LC–MS and other hyphenated methods.In addition,the released pool of oligosaccharides can be analyzed directly or following derivatization with chromogenic,fluorogenic,or other reagents.A notable example is the sensitive analysis of low microgram amounts offluorescent derivatives of the released carbohydrates with2-aminobenzoic acid(2-AA)or2-aminobenzamide(2-AB)by either normal-phase or reversed-phase LC and LC–MS[19,20].These various established oligosaccharide mapping methods have their own relative advantages and disad-vantages as components of comprehensive analytical strategies for glycoproteins.One limitation shared by all the various oligosaccha-ride mapping methods is that the removal and pooling together of glycans results in the loss of site-specific information on oligosac-charide occupancy and structural microheterogeneity.Of course, relinquishing the site-specific microhetrogeneity information is less of a concern for IgG monoclonal antibodies[26–34]which pos-sess only a single site of glycosylation than it is for proteins with multiple sites of glycosylation.Peptide mapping is a general technique for confirming the amino acid sequences of proteins and characterizing their covalent chemical structures,including glycosylation and other posttrans-lational modifications[35–45].These peptide map methods can be used to good advantage as an alternative or in combination with oligosaccharide mapping methods when it is desirable to preserve site-specific information for proteins with multiple sites of glyco-sylation.Sample preparation involves treatment with trypsin or other proteolytic enzymes to cleave the protein into a reproducible set of peptide fragments amenable to analysis by LC–MS and other methods.Peptide mapping can be considered a general method-ology for identifying,characterizing,quantifying,and locating the sequence site for virtually any covalent modification to a protein, including glycosylation[35–45].Ideally,the appropriate protease or proteases are chosen so as to generate a unique well resolved peptide fragment for each glycosylation site.It is worth noting that the conventional peptide mapping and oligosaccharide map-ping approaches can be used in combination either in parallel or sequentially.For instance,it is generally feasible to chromatograph-ically isolate a pooled glycopeptide fraction specific to a particular N-linked or O-linked site and then further analyze this fraction by oligosaccharide mapping,MALDI-TOF MS,and other off-line techniques.In this way,high resolution oligosaccharide mapping data can be acquired for site-specific samples prepared and iso-lated by LC peptide mapping.Another useful experiment is tofirst deglycosylate the sample with PNGase F and then analyze the deg-lycosylated protein by peptide mapping in order to identify sites of N-glycosylation and to evaluate the percent occupancy for each N-linked site[45].Asn residues with attached N-glycans(occupied) are hydrolyzed to aspartic acid(Asp)by the PNGase F treatment and can be readily distinguished from nonglycosylated Asn residues (nonoccupied).In the current study,LC interfaced to a linear ion trap mass spectrometer(ESI-LIT-MS)was used for the site-specific peptide mapping characterization of a single site of N-linked glycosylation on a recombinant human IgG1monoclonal antibody and multiple sites of N-linked and O-linked glycosylation on a CTLA4-Ig fusion protein[46].This CTLA4-Ig construct consists of a modified IgG antibody Fc[47]fused to a high affinity variant of the extracel-lular domain of cytotoxic T lymphocyte antigen-4(CTLA4).CTLA4 (CD152)is a member of the immunoglobulin(Ig)superfamily and expressed as an extracellular receptor on the surface of activated T cells.CTLA4is similar to the T cell costimulatory protein,CD28, and both molecules bind to B7-1(CD80)and B7-2(CD86)ligands on antigen-presenting cells.CD28transmits a stimulatory signal to T-cells whereas CTLA4transmits a competitive inhibitory signal [47–51].2.Experimental2.1.Materials and reagentsRecombinant DNA derived human IgG1monoclonal antibody and CTLA4-Ig fusion protein were produced in CHO cell lines and purified in-house(Bristol-Myers Squibb,East Syracuse,NY).Dithio-threitol(DTT),and iodoacetamide were purchased from Sigma Aldrich(St.Louis,MO).All solvents including water and acetoni-trile were HPLC grade and were purchased from Burdick and Jackson(Muskegon,MI).Trifluoroacetic acid(TFA)and guanidine hydrochloride were from Pierce(Rockford,IL).Trypsin,modified sequencing grade,was purchased from Promega(Madison,WI), Catalog No.V5113.Endoproteinase Asp-N,sequencing grade,was from Roche(Indianapolis,IN),Catalog No.13874622.2.2.Sample preparationIgG1antibody was reduced,S-carboxyamidomethylated, digested with trypsin,and analyzed by RP-LC according to a stan-dard quality control method for tryptic mapping(Figs.2and3). Antibody samples(500␮g each)and a water blank were denatured by addition of8M guanidine hydrochloride buffer,pH8.0,to a final concentration of6M guanidine hydrochloride in a volume of 600␮L.Next,10␮L of175mM DTT was added and the mixture incubated at50◦C for20min to reduce the disulfide bonds.The denatured and reduced sample was then S-alkylated by the addition of10␮L of350mM iodoacetamide and incubation at 50◦C in the dark for20min.The alkylated protein was buffer-exchanged(desalted)into digestion buffer(2M urea,50mM Tris, 10mM CaCl2,10mM methionine,pH8.0)by gel-permeation on NAP-5gel columns(GE Healthcare,Catalog No.17-0853-02)by loading0.5mL sample and collecting1.0mL eluate.The buffer-exchanged samples(1mL)were digested with20␮g each of trypsin(1:50wt:wt enzyme to substrate)at38◦C for5h after which the digest was acidified with25␮L1.0M HCl to halt the enzyme activity.A separate100␮L portion of the above reduced and S-carboxyamidomethylated antibody was also digested with 6.6␮g of endoproteinase Asp-N(1:150wt:wt enzyme to substrate) at37◦C for18.5h and then halted with2.5␮L of1.0M HCl.CTLA4-Ig fusion protein was reduced,S-carboxyamidomethylated,and digested with trypsin exactly according to a standard quality control method for tryptic peptide mapping.Samples consisting of100␮L of20mg/mL CTLA4-Ig (2mg)were each mixed with550␮L of denaturing buffer(8M guanidine hydrochloride,50M Tris/HCl,pH8.0),reduced with 35␮L of200mM DTT for20min at37◦C,and S-alklylated with8142J.Bongers et al./J.Chromatogr.A 1218 (2011) 8140–8149Fig.1.The IgG1model (A)is that of Padlan [52],a composite of Protein Data Dank (PDB)X-ray crystal structures,2ig2and 1fc2[53]with theoretical hinge.The glycosylation consists of a pair of neutral complex-type biantennary N-glycans (A2G1F)enclosed in the central cavity of the Fc domain.The CTLA4-Ig fusion protein model (B)shows PDB crystal structures of human CTLA4homodimer (PDB 1i85)[70]and human Fc (1fc2)[53].Models of four sialylated A2S2G2F complex-type biantennary N-glycans were added to the CTLA4arms using GlyProt [55]with arbitrary orientations and bond angles.N-Glycans are highly flexible and usually exist in a multitude of rapidly converting conformations [54,55].Complex type biantennary N-glycans are named using the abbreviation “AaSsGgF ”where “a ”is the number of antennae linked to the trimannosyl core,“s ”the number of sialic acid residues (NeuAc),“g ”the number of galactose residues (Gal),and “F ”indicates core-fucosylation (Fuc).O-linked glycans in the hinge-region of CTLA4-Ig are not shown in these models.The CTLA4-Ig tryptic peptides containing glycosylation sites are indicated,i.e.T5,T7,T9,and T14.38.5␮L 400mM iodoacetamide for 20min at room temperature in the dark.Removal of reagents and buffer-exchange into digestion buffer (50mM Tris,10mM CaCl 2,pH 7.6)was performed on NAP-5Sephadex G-25gel columns (GE Healthcare Catalog No.17-0853-01)by loading 0.5mL sample and collecting 1.0mL eluate.Digestions were performed with 50␮L of 0.5mg/mL trypsin per 1mL of 1.4mg/mL substrate at 37◦C for 120min and halted with 20␮L 1.0M HCl.2.3.LC–MS peptide mappingLC–MS was run on a Waters 2695Alliance LC or a Waters Acquity UPLC (Milford,MA)interfaced with a ThermoFinnigan LTQ-XL lin-ear ion-trap mass spectrometer (San Jose,CA)with electrospray ionization acquiring sequential MS full-scans with data-dependent acquisition of MSMS and high resolution zoom-scans for abun-dant ions.The LC–MS data were acquired in positive ion mode in a mass to charge ratio (m /z )range of 200–2000,capillary tem-perature of 275◦C,and a source voltage of 5000V.Tandem mass spectra (MSMS)data were acquired using 15–35%normalized col-lision energy,0.25activation Q and 100ms activation time.LC–MS runs for the CTLA4-Ig fusion protein were also made using in-source CID (SID =96V,m /z 140–380Da scans)for monitoring of carbohy-drate oxonium fragment ions [35–38].IgG1antibody tryptic digests were analyzed by C18reversed-phase HPLC–MS run on a Waters 2695Alliance LC equipped with a 2487UV detector,set at 214and 280nm,and interfaced to the above LTQ-XL linear ion-trap mass spectrometer.Injections of 100␮L each were made onto a Grace Vydac 218TP52column (2.1mm ×250mm),40◦C,at 0.2mL/min,with eluents (A)0.10%TFA/H 2O and (B)0.09%TFA/95%CH 3CN,and gradient:2%B isocratic for 4min,2–16%B from 5to 30min,16–19.5%B from 30to 60min,19.5–23%B from 60to 84min,23–29%B from 84to 102min,29–30%B from 102to 118min,and 30–42%B from 118to 143min,42–100%B from 143to 145min,100%B isocratic from 145to 146.1min,100–2%B from 146.1to 146.2min,and 100%B isocratic from 146.2to 166min.The flow rate was 0.2mL/min from 0to 145min,0.2to 0.3mL/min from 145to 145.1min,and 0.3mL/min from 145.1to 166min.A rotary switching-valve was used to divert the 2M urea in the digest away from the ESI source for the first 5min of the run.IgG1antibody trypsin and endoproteinase Asp-N digests were analyzed by C18reversed-phase UP LC–MS run on Waters Acquity UPLC instrument equipped with a photodiode array UV detec-tor,set at 214and 280nm,and interfaced to the above LTQ-XL linear ion-trap mass spectrometer.Injections of 50␮L each were made onto a Waters BEH300C18column (Part No.186003687),1.7␮m,2.1mm ×150mm,45◦C,at 0.2mL/min,with eluents (A)0.02%TFA/H 2O and (B)0.02%TFA/95%CH 3CN.The tryptic map-ping LC gradient was:1%B isocratic for 6min,1–35%B from 6to 80min,35–100%B from 80to 92min,100%B isocratic from 92to 95min,100–1%B from 95to 96min,and 1%B isocratic from 96to 110min,with a flow rate of 0.2mL/min throughout.The endopro-teinase Asp-N mapping LC gradient was:1%B isocratic for 6min,1–50%B from 6to 80min,50–100%B from 80to 92min,100%B isocratic from 92to 95min,100–1%B from 95to 96min,and 1%B isocratic from 96to 110min,with a flow rate of 0.2mL/min throughout.A rotary switching-valve was used to divert the 2M urea in the digest away from the ESI source for the first 3min of the runs.CTLA4-Ig tryptic digests were analyzed by C18reversed-phase HPLC–MS peptide mapping run on a Waters 2695Alliance LC equipped with a 2487UV detector,set at 215and 280nm,and interfaced to the above LTQ-XL linear ion-trap mass spectrom-eter.Injections of 25␮L were run on a Grace Vydac 218MS52column (2.1mm ×250mm),at 0.2mL/min,45◦C,with eluents (A)0.02%TFA/H2O and (B)0.02%TFA/95%CH 3CN,and gradient:1.7%B isocratic for 4min,1.7–30%B from 4to 64min,30–50%B from 64to 80min,and 80–100%B from 80to 90min,100%B iso-cratic from 90to 94min,100–1.7%B from 94to 95min,and 1.7%B isocratic from 95to 110min,with a flow rate of 0.2mL/min throughout.J.Bongers et al./J.Chromatogr.A 1218 (2011) 8140–81498143Fig.2.Summary of ESI linear ion-trap LC–MS tryptic peptide mapping results for a recombinant IgG1monoclonal antibody (CHO cell culture derived)showing (A)full-scale 214nm UV and MS base-peak chromatograms;(B)selected ion-current (SIC)chromatograms for the doubly protonated ions of the three main Fc domain tryptic glycopeptides (A2G0F,A2G1F,and A2G2F)and nonglycosylated heavy chain (NGHC);(C)mass spectrum summed across the 24.0–25.2min glycopeptide region with main peak assignments,(D)expanded view of 280nm UV trace of glycopeptides region showing peak area quantitation of mole percent NGHC,and (E)MS and MSMS spectra of the NGHC tryptic peptide.3.Results and discussion3.1.General procedures for LC–MS peptide mapping of glycosylation sitesComplete sequence coverage for the antibody was obtained by overlapping trypsin and endoproteinase Asp-N LC–MS pep-tide maps.The respective peptides can be readily identified by their observed monoisotopic molecular weights determined via on-line high resolution mass spectrometry and their sequences confirmed by tandem mass spectra of their collision induced disso-ciation (CID)fragmentation patterns.Samples were fully reduced with DTT,S-carboxyamidomethylated (S-alkylation of all cysteine residues),and cleaved with either trypsin or endoproteinase Asp-N and then analyzed by LC–MS according to the following general scheme:LC–MS peptide mapping was run according to a general procedure:(1)Locate N-linked and/or O-linked sites of glycosy-lation by single-ion-monitoring (SIM)of carbohydrate oxonium fragment ions generated by ESI in-source collision-induced disso-ciation (CID),i.e.204,366,and 292Da marker ions for HexNAc,HexNAc-Hex,and NeuAc,respectively [35–38];(2)Characterize oligosaccharides at each site via MS and e selected ion currents (SIC)to estimate relative amounts of each glycoform;and (3)Measure the percentage of site-occupancy by searching for any corresponding nonglycosylated peptide.3.2.LC–MS tryptic peptide mapping of an IgG1monoclonal antibodyIgG antibodies usually have a single N-linked glycosylation site in the heavy chain sequence yielding a pair of N-linked oligosac-charides enclosed in a central cavity in between the two C H 2constant-region Ig subunits of the Fc domain of the mature homod-imeric antibody structure as shown in Fig.1[4–6,52–55].The Fc N-glycans are predominantly neutral (nonsialylated)bianten-nary oligosaccharides with zero,one,or two terminal galactose units (A2G0F,A2G1F,and A2G2F).Since the Fc N-glycans are sand-wiched in between the two C H 2constant-region Ig subunits,these oligosaccharides obviously play an integral role in shaping the three dimensional structure of the Fc and modulating effector functions such as binding to the high affinity Fc ␥R receptor and mediating antibody dependent cellular cytotoxicity (ADCC)and complement fixation [4–7].For example,glycoengineering to produce afucosyl Fc N-glycans has been used to make monoclonal antibody thera-peutics with greatly enhanced ADCC [56,57].Conversely,it has been shown that the Fc N-glycosylation of IgG antibody drugs has essen-tially no effect on their pharmacokinetics and serum half-lives [11].The low microheterogeneity and lack of effect on PK clearance are reasonable considering the peculiar disposition of these N-glycans which are almost entirely enclosed and inaccessible inside the Fc domain.8144J.Bongers et al./J.Chromatogr.A 1218 (2011) 8140–8149parison of ESI linear ion-trap LC–MS glycopeptide mapping results for a recombinant IgG1monoclonal antibody obtained with (A)trypsin versus (B)endoproteinase Asp-N using the same digest concentrations,injection volumes,and MS instrument conditions.Selected ion currents are shown for the three major Fc N-glycans (A2G0F,A2G1F,and A2G2F),two minor sialylated glycoforms (A2S1G2F and A2S2G2F),and a minor oligomannose type glycoform (M5).A more comprehensive list of N-glycans observed by LC–MS Asp-N peptide mapping is given in Table 1.Peptide mapping is routinely used for confirming expres-sion of the correct amino acid sequence and characterizing the covalent chemical structures of antibodies and other recombi-nant proteins [40–43].We have found that complete amino acid sequence coverage for recombinant antibodies can usually be obtained by overlapping trypsin and endoproteinase Asp-N pep-tide maps.The respective peptides can be readily identified by their observed monoisotopic molecular weights determined via on-line high resolution mass spectrometry and their sequences confirmed by tandem mass spectra of their collision induced dis-sociation (CID)fragmentation patterns.The other well known IgG1posttranslational modifications,characteristic of IgG1antibodies,were observed for the heavy chain including N-glycosylation at the conserved Fc region Asn-Xaa-Ser/Thr consensus site,cyclization of N-terminal glutamine to pyroglutamate,and proteolytic removal of the C-terminal lysine residue.The LC–MS tryptic peptide map of the reduced/S-alkylated IgG1mAb in Fig.2was obtained according to a standard quality con-trol testing assay and the flow from the UV detector directed into a linear ion-trap mass spectrometer.In addition to provid-ing information on the amino acid sequence and covalent chemical structure of the protein,this LC–MS peptide map contains a con-siderable amount of information on the glycosylation.The three main N-glycans (A2G0F,A2G1F,and A2G2F)at the single N-linkedglycosylation site are typical for the Fc domain of an IgG1anti-body.Minor amounts of the sialylated glycans,A2S1G2F (∼0.4%)and A2S2G2F (∼0.3%)were also found as well as an M5oligoman-nose structure (∼2.0%),and nonglycosylated heavy chain (NGHC)of ∼1.0%,i.e.,an occupancy of ∼99.0%at this asparagine.These minor amounts of nonglycosylation and the immature oligomannose gly-coforms are of interest as possible indicators of cell viability during upstream cell culture processes [2].LC–MS peptide mapping of the antibody with trypsin and endo-proteinase Asp-N are compared in Fig.3and very similar amounts were obtained for this set of glycoforms.The Asp-N glycopeptide (32residues)has multiple basic residues for ionization (protona-tion)and thus tends to give higher signal-to-noise for the selected ion currents (4+ions).Enhanced detection for minor IgG1glyco-forms (∼0.1to 1.0mol%level)was obtained by LC–MS of the longer 32-residue endoproteinase Asp-N glycopeptide (4+protonated ion)compared to the 9-residue tryptic glycopeptide (2+ion).Oppo-site effects on reversed-phase retention times were observed for presence of an additional galactose (hydrophilic)versus sialic acid (hydrophobic)for both the tryptic glycopeptides and the Asp-N glycopeptides.In addition to the mass spectral peak assignments shown in Figs.2and 3,the various glycoforms identified by LC–MS Asp-N mapping are also listed in Table 1.The relative amounts of the glycopeptides as estimated by LC–MS selected ion currentJ.Bongers et al./J.Chromatogr.A 1218 (2011) 8140–81498145Fig.4.LC–MS tryptic map of a CHO cell culture derived recombinant CTLA4-Ig fusion protein showing 215nm UV chromatogram,MS base-peak chromatogram,and selected ion-current chromatograms (SIC’s)for the carbohydrate oxonium fragment marker-ions,N-acetyl hexoseamine (HexNAc,m /z =204Da),hexose N-acetyl hexoseamine disaccharide (HexNAc-Hex,m /z =366Da),and sialic acid (NeuAc,m /z =292Da)generated by in-source collision-induced dissociation (CID).The boxed-in regions indicate glycopeptides corresponding to the three specific N-linked sites and one O-linked site of glycosylation on CTLA4-Ig as shown in Fig.1.Table 1Glycopeptides found for recombinant IgG1monoclonal antibody by LC–MS endoproteinase Asp-N mapping.N-glycan a FormulaFormulaMW calc b (Da)m /z calc 4+(Da)m /z obs 4+(Da)Ret time (min)SIC rel heightRel (%)aglyco C161H258N48O52–3697.9019925.4828925.572.80.8nd A1G0C203H327N51O82C42H69N3O304793.29851199.33191199.771.60.950.20A1G0F C209H337N51O86C48H79N3O344939.35651235.84641236.271.4 4.010.86A2G0C211H340N52O87C50H82N4O354996.37781250.10171250.171.4 5.23 1.13A2G0F C217H350N52O91C56H92N4O395142.43571286.61621286.971.232269.29A2G1F C223H360N52O96C62H102N4O445304.48851327.12941327.571.110221.95A2G2F C229H370N52O101C68H112N4O495467.54141367.89261367.871.011.2 2.41A2S1G1F C234H377N53O104C73H119N5O525596.58401400.15331399.971.7 2.230.48A2S1G2F C240H387N53O109C79H129N5O575758.63681440.66651440.471.6 1.570.34A2S2G2F C251H404N54O117C90H146N6O656049.73191513.44031513.272.1 1.230.26M5C207H334N50O87C46H76N2O354914.32471229.58851229.871.38.81 1.90M6C213H344N50O92C52H86N2O405076.37751270.10171270.071.2 1.530.33M7C219H354N50O97C58H96N2O455238.43041310.61491310.771.2 2.880.62M8C225H364N50O102C64H106N2O505400.48321351.12811351.071.20.820.18M9C231H374N50O107C70H116N2O555563.53601391.89131391.871.00.230.05Sum464.69100aComplex type biantennary N-glycans are named using the abbreviation “AaSsGgFB ”where “a ”is the number of antennae linked to the trimannosyl core,“s ”the number of sialic acid residues (NeuAc),“g ”the number of galactose residues (Gal),and “F ”indicates core-fucosylation (Fuc).Oligomannose type N-glycans are named “Mm ”where m is the number of mannose residues.The amino acid sequence of the Asp-N peptide is DGVEVHNAKTKPREEQYNSTYRVVSVLTVLHQ.bCalculated masses are for the most abundant isotope.peak heights in the Asp-N map were in close agreement with rela-tive amounts found for the corresponding glycoforms in the tryptic map.However,the smaller amounts of minor glycoforms in the ∼0.1to 1.0%range that were below the limits of detection for thetryptic map were readily identified by Asp-N mapping.The rela-tive amounts of the various N-glycans estimated by LC–MS Asp-N peptide mapping selected ion currents in Table 1compare rea-sonably closely with those obtained independently by analyses of8146J.Bongers et al./J.Chromatogr.A 1218 (2011) 8140–8149Fig.5.Mass spectra summed across the four boxed-in regions of the LC–MS tryptic map for the recombinant CTLA4-Ig fusion protein in Fig.4corresponding to the three N-linked sites and one O-linked site of glycosylation on the CTLA4-Ig fusion protein molecule in Fig.1.Major peaks in the spectra are labeled with oligosaccharide structure assignments.Amino acid sequences are shown for the glycosylated tryptic peptides T5and T7(N-linked,CTLA4domain),T9(O-linked,hinge region),and T14(N-linked Fc domain)which correspond to residues 39-83,94-128,133-158,and 203-211of the CTLA4-Ig,respectively.oligosaccharides released from this antibody and similar recom-binant DNA derived monoclonal antibodies via treatment with PNGase F.3.3.LC–MS tryptic peptide mapping of a CTLA4-Ig fusion proteinThe CTLA4-Ig fusion protein shown in Fig.1consists of the extracellular portion of a disulfide-linked homodimeric glycopro-tein receptor,human cytotoxic T-lymphocyte antigen-4(CTLA4)[46–51],fused to a modified human Fc domain [47].CTLA4is a member of the CD28family of receptor/ligands involved in T-cell costimulation [48–51].CTLA4-Ig has two N-linked sites per CTLA4domain,in addition to the above Fc site,plus O-linked glycosyla-tion in the hinge region.The CTLA4domain is an immunoglobulin V (IgV)type of fold and the N-glycosylation sites at Asn 76and Asn 108are contained within tryptic peptides,T5and T7,respectively.Asn 76is located in beta-strand E of the ABED beta-sheet and the Asn 108N-glycosylation site is located in the C-terminal G strand of the CC’FG beta-sheet on the opposite face of the IgV sandwich.The “MYPPPYY loop”connecting beta-strands E and F (residues 97–103)is the key receptor-binding interface for the interaction of the CTLA4homodimer with the B7-1and B7-2ligands on antigen presenting cells.Note that,unlike the atypical case of the N-glycans constrained inside the Fc domain cavity,the N-glycans attachedto the CTLA4domain represent the more usual case of oligosac-charides projecting out from the surface of the protein into the surrounding medium (Fig.1).The four glycopeptide regions in the tryptic map can be visu-alized via single-ion monitoring of in-source CID carbohydrate marker ions in Fig.4.For the CTLA4-Ig fusion protein,the sim-plest of the three N-linked sites is that on the Fc domain T14peptide containing the three neutral biantennary complex-type core-fucosylated N-glycans (A2G0F,A2G1F,and A2G2F)which are typical for virtually all recombinant IgG1antibodies produced by mammalian cell culture.In contrast to the rather simple population of Fc N-glycans,the two N-glycosylation sites on the CTLA4domain show highly diverse microheterogeneous populations of sialylated and neutral forms of bi-,tri-,and tetra-antennary complex-type core-fucosylated glycans with variable numbers of terminal galac-tose (Fig.5).The T7N-glycans contain significantly higher levels of galactose and sialic acid,overall,versus the T5site.The T7glycopeptides are partially resolved into distinct regions on this RP-LC system (Fig.6).The first envelope within the range of approximately 67.8–68.8min consists primarily of the T7glycopep-tide with monosialylated biantennary N-glycans,A2S1G2F and A2S1G1F.The next region from approximately 68.0–70.0min con-tains T7with the disialyated biantennary N-glycan,A2S2G2F,and lesser amounts of T7with the disialylated and trisialylated species,。