Fc Gamma Receptor Glycosylation Modulates the Binding of IgG

Oct 27, 2014 - Significant heterogeneity also exists within FcγRs, resulting from polymorphisms in their extracellular domains. ... (13, 14) Less is ...
6 downloads 16 Views 6MB Size
Download PDF
Article pubs.acs.org/jpr

Fc Gamma Receptor Glycosylation Modulates the Binding of IgG Glycoforms: A Requirement for Stable Antibody Interactions Jerrard M. Hayes,*,†,# Asa Frostell,‡ Eoin F. J. Cosgrave,§ Weston B. Struwe,∥ Oscar Potter,⊥ Gavin P. Davey,† Robert Karlsson,‡ Cecilia Anneren,‡ and Pauline M. Rudd# †

School of Biochemistry & Immunology, Trinity Biomedical Sciences Institute, Trinity College, Pearse St. Dublin 2, Dublin, Ireland GE Healthcare, Rapsgatan 23, SE-75184 Uppsala, Sweden § Pharmaceutical Life Sciences Group, Waters Corporation, 34 Maple Street, Milford, Massachusetts 01757, United States ∥ Chemistry Research Laboratory, Department of Chemistry, University of Oxford, Oxford OX1 3TA, United Kingdom ⊥ Agilent Technologies, Inc., 5301 Stevens Creek Boulevard, Santa Clara, California 95051, United States # NIBRT-Glycoscience Group, NIBRT-The National Institute for Bioprocessing, Research, and Training, Foster Avenue, Blackrock, County Dublin, Ireland ‡

ABSTRACT: FcγRs play a critical role in the immune response following recognition of invading particles and tumor associated antigens by circulating antibodies. In the present study we investigated the role of FcγR glycosylation in the IgG interaction and observed a stabilizing role for receptor N-glycans. We performed a complete glycan analysis of the recombinant FcγRs (FcγRIa, FcγRIIa, FcγRIIb, FcγRIIIaPhe158/Val158, and FcγRIIIb) expressed in human cells and demonstrate that receptor glycosylation is complex and varied between receptors. We used surface plasmon resonance to establish binding patterns between rituximab and all receptors. Complex binding was observed for FcγRIa and FcγRIIIa. The IgG−FcγR interaction was further investigated using a combination of kinetic experiments and enzymatically deglycosylated FcγRI a an d FcγRIIIaPhe158/Val158 receptors in an attempt to determine the underlying binding mechanism. We observed that antibody binding levels decreased for deglycosylated receptors, and at the same time, binding kinetics were altered and showed a more rapid approach to steady state, followed by an increase in the antibody dissociation rate. Binding of rituximab to deglycosylated FcγRIIIaPhe158 was now consistent with a 1:1 binding mechanism, while binding of rituximab to FcγRIIIaVal158 remained heterogeneous. Kinetic data support a complex binding mechanism, involving heterogeneity in both antibody and receptor, where fucosylated and afucosylated antibody forms compete in receptor binding and in receptor molecules where heterogeneity in receptor glycosylation plays an important role. The exact nature of receptor glycans involved in IgG binding remains unclear and determination of rate and affinity constants are challenging. Here, the use of more extended competition experiments appear promising and suggest that it may be possible to determine dissociation rate constants for high affinity afucosylated antibodies without the need to purify or express such variants. The data described provide further insight into the complexity of the IgG−FcγR interaction and the influence of FcγR glycosylation. KEYWORDS: Glycosylation, antibody, rituximab, Fcγ receptors, surface plasmon resonance, glycan analysis

■

INTRODUCTION Immunoglobulins are an important class of glycoproteins responsible for recognizing foreign material and tumor associated antigens and communicating this information within the innate and adaptive immune systems. Following opsonization of target antigens, immunoglobulin G (IgG) coated immune complexes interact with effector cells of the innate immune system through cell surface receptors (FcγR) specific for the Fc region of the antibody molecule. FcγRs are present on platelets and a variety of innate immune cells, including monocytes, granulocytes, macrophages, B-cells, and natural killer cells, where they trigger effector responses such as neutrophil activation, macrophage phagocytosis, natural killer cell ADCC, and B-cell inhibition.1−5 In general, the receptors © 2014 American Chemical Society

can be broadly classified into three distinct families, FcγRI (CD64), FcγRII (CD32), and FcγRIII (CD16). FcγRI members bind IgG with high affinity (KD, 10−9 M), while the FcγRII and FcγRIII members bind with lower affinity (KD, 10−6−10−7 M). Both high and low affinity FcγRs bind IgG immune complexes with high avidity, but only the high affinity FcγRIa can appreciably bind monomeric IgG.6 The FcγR family can be further divided into activating (FcγRIa, FcγRIIa, and FcγRIIIa) or inhibitory (FcγRIIb) receptors, based on the immune response following the binding of IgG. Significant heterogeneity also exists within FcγRs, Received: April 24, 2014 Published: October 27, 2014 5471

dx.doi.org/10.1021/pr500414q | J. Proteome Res. 2014, 13, 5471−5485

Journal of Proteome Research

Article

Following SPR analysis we observed decreased antibody binding to deglycosylated receptors with reduced affinity. In addition, we show altered binding kinetics to deglycosylated FcγRIIIaPhe158/Val158 with a more rapid approach of steady state binding levels. We also demonstrate an increase in the antibody dissociation rate from deglycosylated receptors, suggesting a stabilizing function for glycosylation on FcγRs and the interaction with IgG. The data provide further links between receptor glycosylation, antibody binding, and carbohydratemediated stabilization of the complex IgG−FcγR interaction.

resulting from polymorphisms in their extracellular domains. Two variants of FcγRIIa exist, which differ in the amino acid in position 131 (Arg131/His131). The Arg131 substitution has increased susceptibility to SLE,7 nephropathy,8 and bacterial infections.9 Polymorphic variants also exist for FcγRIIIa with either a phenylalanine or valine residue in position 158 in the extracellular region (Phe158/Val158). Val158 results in significantly increased affinity for IgG, while the Phe158 variant is linked to autoimmune diseases, including SLE and rheumatoid arthritis (RA).10,11 Polymorphic variants also exist for FcγRIIIb (NA1 and NA2), which are believed to demonstrate further heterogeneity in glycosylation.12 The extensive glycosylation of both antibody and FcγR contribute to the properties of the IgG−FcγR interaction. Removal of the N-linked glycans at Asn297 in the IgG heavy chains causes a loss of structural integrity and decreased communication with the FcγR.13,14 Less is known about the role of FcγR glycosylation; however, recent studies have shown that elimination of the sugar moiety at Asn162 on FcγRIIIa reduces the binding affinity of afucosylated IgG by over an order of magnitude, presumably through destabilization of the IgG−FcγR interaction.15 In addition, glycosylation at Asn45 on FcγRIIIa negatively affects the binding of IgG.16 These findings implicate an active role for glycosylation in positively and negatively influencing antibody binding and downstream immune effector functions. Interestingly, the N-linked glycans of FcγRIIIa have recently been shown to form unique carbohydrate−carbohydrate interactions with the N-glycans of afucosylated IgG, explaining the increased binding and ADCC activity of antibodies lacking core-fucose.17 Furthermore, glycosylation of FcγRs is cell type specific,18 which adds further importance to how these receptors are glycosylated by cells of the immune system as this will likely play a key role in regulating the IgG−FcγR interaction and the downstream immunological response. Human FcγRs differentially bind IgG of different subclasses and a number of studies have been performed, which describe the IgG−FcγR interaction and the selectivity of the receptors. A comprehensive study by Brunhs et al.6 described the binding of polyclonal and monoclonal IgG1, IgG2, IgG3, and IgG4, as well as therapeutic anti-CD20 and anti-RhD to the complete set of human FcγRs, including polymorphic variants. More recently Visser et al. reported affinities of rituximab and a biosimilar antibody to a similar set of FcγRs.19 Other investigations have focused more on specific receptors.20,21 With varying FcγR and IgG sources and with different surface plasmon resonance (SPR) experimental set-ups, the reported binding affinities vary by up to 10-fold but, with the exception of FcγRI, are consistently in the micromolar range. The affinity for FcγRI is higher and consistently in the nanomolar range. In the present study an SPR assay format combining histidine capture of FcγRs expressed in HEK293 cells and single-cycle kinetics25 was used to investigate rituximab binding across a panel of receptors. To complement the interaction analysis and produce a more complete picture of both glycosylation and its influence on IgG binding, a detailed analysis of the glycosylation state of rituximab and each FcγR (FcγRIa, FcγRIIa, FcγRIIb, FcγRIIIaPhe158/Val158, and FcγRIIIb), including monosaccharide sequence and linkage information was performed. The function of FcγR glycosylation and its role in receptor stability and interaction with IgG was further investigated by examining the binding of rituximab to enzymatically deglycosylated FcγRIa, FcγRIIIaPhe158/Val158, and fully glycosylated forms.

■

EXPERIMENTAL SECTION

N-linked Glycan Release and 2-AB Labeling

Human Fcγ receptors used for glycosylation analysis and IgG binding studies were purchased from Sino Biological, Inc. (Beijing, China). Catalogue numbers: FcγRIa, 10256-H08H; FcγRIIa, 10374:H08H; FcγRIIb, 10259-H08H;, FcγRIIIa, 10389-H08H; FcγRIIIb, 11046-H08H. Three different batches of recombinant Fcγ receptor produced in human cells were used for analysis and replication. Each lyophilized receptor (100 μg) was reconstituted in PBS to 0.5 mg/mL, according to the manufacturer’s recommendation. N-linked glycans were released and labeled by two different methods. The SDSPAGE immobilization and release method followed the protocol described by Royle et al.22 The second glycan release method was performed in solution without immobilization of the protein. The reconstituted receptor was incubated with 100 mM DTT at 70 °C for 10 min followed by incubation with 10 mM iodoacetamide for 45 min in the dark. Glycans were released by the addition of PNGase F (0.025 U) (Prozyme, Inc.) and incubation overnight at 37 °C. The following day protein was removed by filtration through 10 kDa spin columns (Millipore) and glycans were purified by solid phase extraction using porous graphitized carbon (PGC) prior to labeling with 2-AB and purification using Phytips (Phynexus, Inc.). Hydrophilic Interaction Ultra Performance Liquid Chromatography (HILIC UPLC)

2-AB labeled glycans were prepared as 80% (v/v) acetonitrile solutions for analysis by HILIC UPLC using a Waters BEHGlycan 1.7 μm (150 mm × 2.1 mm) column and fluorescence detection (excitation at 420 nm and emission at 330 nm) as previously described.23 All separations were performed using a Waters Acquity H-Class UPLC instrument over a 30 min period at 40 °C using 50 mM ammonium formate pH 4.4 as solvent A and 100% (v/v) acetonitrile as solvent B. 2-ABlabeled dextran was used as an internal calibration standard and for the generation of glucose unit (GU) values. Retention times for 2-AB labeled dextran peaks were used to fit a fifth order polynomial distribution curve using Waters Empower 3 software, which provided the means for converting chromatographic retention times into standardized glucose unit (GU) values as previously described.22 Glycans were named according to the Oxford notation.24 Weak Anion Exchange HPLC (WAX HPLC)

WAX HPLC was used to analyze the extent of terminal sialylation of receptor N-glycans and for fractionation of charged structures. A Waters Biosuite DEAE 10 μm AXC (7.5 mm × 75 mm) column was used for separation of charged carbohydrate structures. Solvent A was 100 mM ammonium acetate pH 7 in 20% (v/v) methanol and solvent B was 20% (v/v) acetonitrile. All WAX HPLC separations were performed 5472

dx.doi.org/10.1021/pr500414q | J. Proteome Res. 2014, 13, 5471−5485

Journal of Proteome Research

Article

operated in positive ion mode with a spray voltage of −1.8 kV. Samples were run on a custom HPLC-chip, which consisted of a 500 nL porous graphitized carbon enrichment column and a 43 mm separation column packed with Tosoh TSK Gel Amide 80 (3 μm). Data-dependent acquisition was used to automatically target [M + 2H]2+ or [M + H + NH4]2+ ions of the labeled glycans for isolation in the ion trap and fragmented therein by collision induced dissociation to yield MS2 spectra.

on a Waters Alliance 2695 instrument with online Waters 4795 fluorescence detector. Carbohydrate samples were separated over a 30 min gradient. The relative proportions of terminal charged glycan structures were determined by comparison to N-linked glycans released from bovine fetuin, which contains neutral, mono-, di-, tri-, and tetra-sialylated structures. For WAX-fractionation experiments glycan pools were separated by WAX HPLC, and peaks from this analysis were manually collected. Glycans were dried, reconstituted in Milli-Q water, and analyzed by HILIC UPLC. This 2-dimensional approach was used to deconvolute the complex FcγR N-glycan pools. For nanoLC/MS/MS analysis, peaks from HILIC UPLC analysis of WAX fractionated glycan pools were manually collected, dried, and reconstituted in 10 μL of Milli-Q water.

Surface Plasmon Resonance (SPR)

All IgG−FcγR interaction experiments were performed using a Biacore T200 system (GE Healthcare, Uppsala, Sweden) with analysis temperature set to 25 °C and sample compartment temperature set to 15 °C. Recombinant extracellular domains of FcγRs expressed with C-terminal His-tags in human/ HEK293 cells were from Sino Biological, Inc.: FcγRI (NP_000557.1, Met 1-Pro 288), FcγRIIA (AAA35827, Met 1-Ile 218), FcγRIIb (NP_001002274.1, Ala 46-Pro 217), FcγRIIIa F158 (P08637-1, Met 1-Gln 208), FcγRIIIa V158 (P08637-1, Met 1-Gln 208), and FcγRIIIb (NP_000561.3, Met 1-Ser 200). FcγRs were aliquoted, stored at −70 °C, and thawed only once. Overlayed sensorgrams of rituximab binding to two different lots and to fresh and 4 month old aliquots showed no significant differences in sensorgram appearances indicating similar binding kinetics (not shown). Rituximab (MabThera) was obtained from pharmacies and stored according to the manufacturer’s instructions. Anti-His antibody (from His Capture Kit, GE Healthcare) was amine coupled in the active and reference flow cell of a CM5 chip, according to the manufacturer’s instructions. Briefly, the surface was activated by injecting a solution containing 0.2 M N-ethyl-N ¢-dimethylaminopropylcarbodiimide (EDC) and 50 mM Nhydroxysuccinimide (NHS) for 7 min. Anti-His antibody was diluted to 20 μg/mL in 10 mM Na-acetate, pH 4.5, and injected during 7 min, and the surface was then blocked by injecting 1 M ethanolamine at pH 8.5 for 7 min. Immobilization levels in the range 6000−8000 RU were used, each run having similar levels in active and reference flow cell. His-tagged FcγRs at a concentration of 0.5−1 μg/mL were injected for 60 s using a flow rate of 5 μL/min in active flow cell only, obtaining capture levels in the range 70−180 RU for kinetic data and approximately 300 RU for affinity data. A capture stabilization time of 1−3 min was applied for some FcγRs. For affinity and kinetic studies rituximab was then injected in order of increasing concentration over reference and active flow cell using five 60 s injections at 30 μL/min, applying a single cycle kinetics procedure.25 Using this procedure all concentrations were injected in sequence in order from low to high concentration and a 300 s dissociation time was added after the last rituximab injection. Rituximab concentrations in the range 1.2−300 nM were used for FcγRI, 24.7−2000 nM for FcγRIIIa, and 0.5−8 μM for FcγRIIa, FcγRIIb, and FcγRIIIb. Following each experiment both flow cells were regenerated using a 30 s injection of 10 mM glycine, pH 1.5. Blank cycles (FcγR capture + buffer injections + regeneration) were performed first, last, and upon change of FcγR subtype. Data were double referenced by first subtraction of reference flow cell and then subtraction of blank cycles. Fitting of data were performed using Biacore T200 evaluation software 2.0.

Exoglycosidase Panel Digests

Exoglycosidase arrays were used to sequence the N-glycans of human FcγRs. Using this technique, sequence, composition, and linkage specificity information was obtained. Fluorescently labeled glycans were routinely digested in 50 mM sodium acetate, pH 5.5, at 37 °C overnight in a volume of 10 μL using a selected panel of exoglycosidase enzymes. The following day enzymes were removed using 10 kDa spin filters (Pall Corp, NY, USA). Digested 2-AB-labeled glycans were analyzed by HILIC UPLC as described previously. The specificities of the exoglycosidase enzymes used are well-established and specific monosaccharides were removed as follows: terminal sialic acid in all linkages, α(2,6), α(2,3), and α(2,8), is removed with 1 mU/μL Arthrobacter ureafaciens sialidase (ABS)(Prozyme); terminal galactose monosaccharides were removed using 0.5 mU/μL bovine testes β-galactosidase (BTG)(Prozyme), which releases both β(1,3)- and β(1,4)-linked galactose; terminal Nacetylglucosamine (GlcNAc) monosaccharides in β(1,4) linkage were released with 40 mU/μL Streptococcus pneumoniae hexosaminidase (GUH)(Prozyme); core α(1,6)-fucose was selectively removed using 1 mU/μL bovine kidney α-fucosidase (BKF)(Prozyme); terminal nonreducing end fucose in α(1−3)and α(1−4)-linkages was removed using 0.004 mU/μL almond meal α-fucosidase (AMF)(Prozyme); α-linked mannose was removed using 150 mU/μL jack bean α-mannosidase (JBM)(Prozyme); and βGalNAc and βGlcNAc were removed using 50 mU/μL jack bean β-N-acetylhexosaminidase (JBH)(Prozyme). Mass Spectrometry of N-Glycans

LC/FLD-MS. Glycans were analyzed using a Waters Xevo G2 QTof coupled to an Acquity UPLC with a Waters BEH Glycan column (1.0 × 150 mm, 1.7 μm particle size). The analyzer was set to sensitivity mode, and data was acquired in negative mode with the following conditions: 2 kV capillary voltage, 25 V cone voltage, 4 V extraction voltage, 100 °C source temperature, 280 °C desolvation temperature, 10L/h cone gas flow, and 600L/h desolvation gas flow. Solvent A was 50 mM ammonium formate pH 4.4 , and solvent B was 100% acetonitrile. The flow rate was 30−47% buffer A for 24 min, 70% A for 3 min, and 30% A for 3 min. Injection volume was 10 μL, and samples were diluted in 70% acetonitrile prior to analysis. Data acquisition and processing were conducted with Waters MassLynx version 4.1. NanoLC/MS/MS. LC-MS(2) analysis of 2-AB labeled glycans from fractionated samples was performed using an Agilent Technologies 1200 series capillary/nano liquid chromatography system in conjunction with an Agilent ChipCube interface and a 6340 ion trap mass spectrometer (Agilent Technologies, Santa Clara, CA). The instrument was

Enzymatic Deglycosylation of FcγRIa and FcγRIIIaPhe158/Val158

Fifty micrograms of HEK293 FcγRIa and FcγRIIIaPhe158/Val158 were reconstituted in NaHCO3 (100 mM), pH 7, to a 5473

dx.doi.org/10.1021/pr500414q | J. Proteome Res. 2014, 13, 5471−5485

Journal of Proteome Research

Article

Figure 1. Human FcγRs display complex and differential glycosylation. A. Glycan analysis of Fcγ receptors expressed in Human/HEK293 cells. Glycosylation of receptors was complex and differential with 30−40 unique glycan structures for each receptor. Following enzymatic release and HILIC UPLC analysis common structures (30) were identified and are indicated by dashed vertical lines. Numbered peaks represent the structures shown in Table 1: (i) FcγRIIIa, (ii) FcγRIa, (iii) FcγRIIa, (iv) FcγRIIb, and (v) FcγRIIIb. GU values were assigned using internal 2-AB labeled dextran standards and integration using Waters Empower 3 software. Glycan structures and relative abundane for each receptor are shown in Table 1. Glycan release and analysis experiments were performed in triplicate. B. Weak anion exchange (WAX) HPLC analysis of human FcγRs shows Nglycans are neutral (S0) and monosialylated structures (S1) with minor amounts of disialylated sugars (S2). For WAX fractionation experiments individual S0 and S1 peaks were manually collected and dried for nano-LC-MS/MS analysis: (i) Bovine fetuin N-glycan standards, (ii) FcγRIa, (iii) FcγRIIa, (iv) FcγRIIb, (v) FcγRIIIa, (v) FcγRIIIb. WAX HPLC experiments were performed in triplicate. C. FcγRs contain common N-glycans with differences in relative abundance. Average peak area percentages were calculated for peaks from individual receptors (FcγRIa, FcγRIIa, FcγRIIb, FcγRIIIa, and FcγRIIIb). Thirty common peaks were identified (see panel A) and values were plotted as the average GU value and average % area from three different releases of individual receptors. This value was then used to calculate the average value across the range of receptors for each peak to show the distribution in relative abundance between different receptors. Error bars represent the standard deviation for the relative % areas. Large error bars as seen for certain peaks, e.g., GU 6.22 (Man5), signify large differences in abundance for certain N-glycans between receptors. Nglycan structures corresponding to the GU values shown can be seen in Table 1.

concentration of 1 mg/mL. For deglycosylation experiments, 5 μL PNGase F (2.5 mU/μL) was added, and the volume was increased to 100 μL and incubated at 37 °C for 5−6 h. For control deglycosylation experiments sodium bicarbonate buffer was used in place of PNGase F. Following PNGase F treatment, 1 μL of Endo F1 (17 U/ml)(QA Bio, Inc.) and 1 μL of Endo F2 (5 U/ml)(QA Bio, Inc.) were added followed by 5 μL of 250 mM sodium acetate, pH 4.5, and overnight incubation at 37 °C. Five microliters of 250 mM sodium acetate, pH 4.5, was also added to the control samples in place of Endo F1 and Endo F2. The next morning deglycosylated and control receptors were purified using 10 kDa spin filters and concentrated to 1 mg/mL. The extent of receptor deglycosylation was determined by SDS-PAGE according to Laemmli26

and by collecting the endoglycosidase released glycans and labeling with 2-AB followed by HILIC UPLC analysis.

■

RESULTS

Glycosylation of Human Fcγ Receptors Is Complex and Differential

Glycosylation of human Fcγ receptors expressed in HEK293 cells was routinely analyzed by hydrophilic interaction ultraperformance liquid chromatography (HILIC UPLC) following PNGase F release of N-glycans and fluorescent labeling. Initial glycan analysis of FcγRIa, FcγRIIa, FcγRIIb, FcγRIIIaPhe158/Val158, and FcγRIIIb revealed complex glycosylation with many overlapping peaks (see Figure 1A). 5474

dx.doi.org/10.1021/pr500414q | J. Proteome Res. 2014, 13, 5471−5485

Journal of Proteome Research

Article

Figure 2. Exoglycosidase sequencing of FcγRs identified monosaccharide composition and linkage information and showed that FcγR N-glycans were composed of common core structures. A. Exoglycosidase panel digest of FcγRIIIa. Arrows indicate the migration of major glycan peaks following monosaccharide removal: (i) undigested glycan profile, (ii) Arthrobacter ureafaciens sialidase (ABS), (iii) bovine testes β-galactosidase (BTG), (iv) bovine kidney α-fucosidase (BKF), (v) almond meal α-fucosidase (AMF), (vi) jack bean β-N-acetylhexosaminidase (JBH), and (vii) jack bean α-mannosidase (JBM). B. FcγR N-glycans contain common core structures. Common core structures were identified following exoglycosidase digestion with ABS, BTG, BKF, AMF, JBH, and JBM: (i) FcγRIa, (ii) FcγRIIa, (iii) FcγRIIb, (iv) FcγRIIIa, and (v) FcγRIIIb. Common core structures consisted of Man1 and bisecting GlcNAc residues, which were not removed by the exoglycosidase enzymes indicated above. These common structures were almost identical across the range of receptors and allowed the identification and sequencing of the receptor Nglycans. Vertical dashed lines indicate common structures. Glycan structures are shown according to the Oxford notation (see Table 1).

structures and suggested that terminal glycan moieties were present that were preventing further glycan digestion. Additional digestions with a series of enzymes including the αfucosidase AMF (almond meal α-fucosidase), which is specific for α(1,3)- and α(1,4)-linked nonreducing end terminal fucose, confirmed the presence of this outer arm modification (see Figure 2A(v)). A common glycan modification observed in kidney cells, namely, the incorporation of GalNAc residues into N-glycans, was also identified in FcγRs (see Figure 2A(vi)). A combination of enzymes was used to reveal the presence of this monosaccharide residue, specifically JBH (Jack Bean β-Nacetylhexosaminidase) and GUH. Comparison of enzyme digests with both JBH (removes both terminal β-linked GlcNAc and GalNAc) and GUH (removes only β-linked GlcNAc) revealed differences between GUH and JBH digests and the presence of terminal GalNAc structures. An approach using exoglycosidase digestions followed by mass spectrometry (LC/FLD-MS) confirmed the presence of these outer arm residues (see Figure 3A) and revealed that FcγRs contained common core structures (see Figure 2B). To complete the glycan analysis of FcγRs, WAX HPLC was used to fractionate the complex pools of glycans into neutral and monosialylated fractions, which were collected and analyzed by HILIC UPLC in a two-dimensional approach.

Glycosylation was differential with unique glycan profiles specific to receptor subtypes. Following assignment of GU values, exoglycosidase sequencing and mass spectrometry, 30 to 40 unique glycan structures of varying complexity were identified in each receptor. More than half were common structures (see Figure 1A,C for glycan distribution profile), but the relative abundance of each glycan varied both between receptors and to a lesser extent between different batches of a specific receptor, in particular Man5 which varied significantly between different batches of FcγRIIIa. Weak anion exchange (WAX) HPLC analysis revealed that each receptor was composed of predominantly neutral and monosialylated glycans with minor amounts of disialylated sugars (see Figure 1B). Sialylated structures were restricted to one sialic acid in most cases even though larger antennary structures were present. Treatment with α-sialidase (ABS) followed by WAX analysis confirmed that all charge could be accounted for by sialic acid alone. To sequence the N-glycans of FcγRs and determine their monosaccharide constituents and linkage information, an exoglycosidase panel digest was performed on the released glycans from each receptor (see Figure 2A for exoglycosidase digest panel). A standard exoglycosidase digest panel did not reduce the N-glycans of FcγRs to common Man1 core 5475

dx.doi.org/10.1021/pr500414q | J. Proteome Res. 2014, 13, 5471−5485

Journal of Proteome Research

Article

Figure 3. Mass spectrometry analysis of FcγR N-glycans identified unusual and complex glycan epitopes. A. Exoglycosidase digestions of FcγRIIIa followed by mass spectrometry (FLD/LC-MS) reduced the complexity of N-glycans and confirmed the presence of bisecting GlcNAc and GalNAc structures. Released 2-AB labeled FcγRIIIa N-glycans were digested with ABS, BTG, BKF, AMF, and GUH followed by mass spectrometry, which identified common glycan structures consisting of oligomannose, bisecting GlcNAc and GalNAc structures. The top chromatogram represents the fluorescence HILIC profile with the corresponding extracted ion chromatograms of each mass and glycan composition shown below. FLD: Fluorescence detection of N-glycans separated by HILIC UPLC. EIC: Extracted ion chromatograms. B. NanoLC/MS-MS analysis of FcγR Nglycans. Fragmentation and identification of four N-glycan structures were chosen for illustrative purposes to show some of the more complex and unusual carbohydrate structures. The parent m/z shown in the inset spectra on the right-hand side corresponda to [M + NH4 + H]2+ or [M + 2H]2+ ions of glycans present in the respective fractions, and these were used to determine the likely monosaccharide compositions of the N-glycans. The larger spectra show MS2 for those parent m/z. These spectra are annotated with cartoons for the most likely B ion structures based on the knowledge that CID spectra for N-glycans are dominated by B ions resulting from cleavage on the reducing-end side of HexNAc groups. In combination, these data were used to deduce likely structures for the glycans. When interpreting this data it is important to be aware that outer-arm fucose residues may migrate from one position to another during ionization and fragmentation.41 A. F(6)Man4A1F(2)GalNAc1 (peak 16 in Figures 1 and 2 and Table 1); B. F(6)A2BG(4)2 (peak 17 in Figures 1 and 2 and Table 1); C. F(6)A2F(2)1G(4)1GalNAc1 (peak 20 in Figures 1 and 2 and Table 1); D. F(6)A2F2(2)G(4)1GalNAc1 (peak 25 in Figures 1 and 2 and Table 1).

tures containing GalNAc β-linked to antennary GlcNAc were present in approximately 19%, outer-arm fucose in approximately 10%, and bisecting GlcNAc in 13.5%. Bisecting GlcNAc was identified based on exoglycosidase digestion and confirmed by mass spectrometry (see Figures 2B and 3). This structure was shown to persist following full panel digests as there is no known exoglycosidase that will selectively remove this monosaccharide. All sialic acids were in α(2,3) linkage to galactose, and all galactose was found in β(1,4) linkages to GlcNAc. The presence of poly-N-lactosamine residues was also found in all FcγRs accounting for approximately 2%. It is also worth noting that glycosylation of FcγRIIIaPhe158 was found to be identical to FcγRIIIaVal158, which could be expected as there is only a single amino acid difference. However, this amino acid (158) is close to the asparagine residue in position 162, which is known to be glycosylated and important for the IgG interaction.15,27 The 2-AB labeled N-glycans of rituximab were

Individual peaks were collected from the HILIC analysis of neutral and monosialylated fractions and dried for mass spectrometry (nanoLC-MS/MS) analysis. Thirty of the most abundant peaks were collected and mass values and fragmentation data generated for 22 of these structures (see Figure 3B for nanoLC-MS/MS data). This approach identified approximately 83.5% of the total glycan composition (see Table 1 for N-glycans identified from FcγRs). In summary, the majority of N-glycans from FcγRs were complex (70%) with high mannose (9%) and hybrids (3%) also present. Core fucosylation was present on 67% of glycans; galactosylation was present as mainly mono- and digalactosylated structures with only a minority being capped with sialic acids. Approximately 66% of N-glycans contained no sialic acids with only 12% having one sialic acid and 4% having two sialic acids. Negligible amounts of higher sialylated structures (tri-sialylated and tetrasialylated) were found. LacdiNAc (HexNAc-HexNac) struc5476

dx.doi.org/10.1021/pr500414q | J. Proteome Res. 2014, 13, 5471−5485

Journal of Proteome Research

Article

Table 1. N-Glycans of HEK293 FcγRsa

5477

dx.doi.org/10.1021/pr500414q | J. Proteome Res. 2014, 13, 5471−5485

Journal of Proteome Research

Article

Table 1. continued

a

Glycans were identified and quantified using a combination of HILIC UPLC, exoglycosidase digestions, WAX HPLC, WAX fractionation, and mass spectrometry (MS/MS). Average GU (glucose unit) values and % areas are the mean of three individual glycan releases. Peak numbers correspond 5478

dx.doi.org/10.1021/pr500414q | J. Proteome Res. 2014, 13, 5471−5485

Journal of Proteome Research

Article

Table 1. continued to peaks shown in Figure 1. m/z values are for [M + 2H]2+ unless otherwise indicated. Glycans are named according to the Oxford notation as illustrated in the legend below the table. Where linkage information was determined by exoglycosidase sequencing, it is indicated in the glycan name, whereas when the linkage information is unsure, such as for the hybrid structures 10, 13, and 16, it has been omitted from the glycan name and the cartoon figure is a representation with structural composition only. For structures 10, 13, and 16, the mannose on the α(1,6) arm is shown linked to the mannose in the α(1,6) position for representative purposes only as Jack bean mannosidase is known to remove mannose linked in the α(1,6) and α(1,2) positions more efficiently than α(1,3).42,43

Figure 4. Rituximab binds to FcγRs with different kinetics and exhibits complex binding. SPR sensorgrams of rituximab-FcγR binding kinetics are shown: A. FcγRIa, B. FcγRIIIaVal158, C. FcγRIIIaPhe158, D. FcγRIIa, E. FcγRIIb, and F. FcγRIIIb. In A−F red solid lines represent experimental data; in A−C black dashed lines display one to one fittings. In D−F the steady state affinity fit of the sensorgram data are shown in the inset with the vertical line marking the respective obtained KD values: FcγRIIa 4.2 μM, FcγRIIb 6.9 μM, and FcγRIIIb 3.5 μM. A−C show representative sensorgrams from >10 experiments; D−F show representative sensorgrams from three separate affinity determinations. Five injections of rituximab in increasing concentration are performed in each curve, resulting in increasing binding level. Dissociation is seen starting from the end of injection of the highest concentration, at time 550 s, as well as from the end of injection of the lower concentrations.

also fully characterized by both HILIC UPLC and by mass spectrometry. Rituximab contained five major glycans: A2 (3.5%), FA2 (45%), FA2[6]G1 (26.3%), FA2[3]G1 (19.3%), and FA2G2 (1.6%) representing 95% of the total N-glycan composition.

interaction studies and was heterogeneous with respect to core fucosylation with approximately 92% of glycans containing this residue, as determined by glycan analysis. As a result, both interaction partners, FcγRs and antibody, display structural heterogeneity. Binding studies using SPR and histidine-tagged FcγRs captured to an immobilized antihistidine antibody with target antibodies injected applying a single cycle kinetics procedure were performed according to Karlsson et al.25 In brief, single cycle kinetics is a further development of the more traditionally used multicycle kinetics. In the setup of a traditional kinetics assay one antibody concentration is injected

Kinetic Analysis of Rituximab Binding to FcγRs

Rituximab is a well-characterized monoclonal antibody used for a number of clinical indications including non-Hodgkin’s lymphoma, chronic lymphocytic leukemia (CLL), and rheumatoid arthritis (RA). Rituximab was used for all 5479

dx.doi.org/10.1021/pr500414q | J. Proteome Res. 2014, 13, 5471−5485

Journal of Proteome Research

Article

Figure 5. Rituximab FcγRIIIa dissociation rate depends on interaction time and FcγR variant at saturating conditions. Rituximab was injected at saturating concentration (6 μM) over a sensor chip with captured FcγR: A. FcγRIIIaVal158 and B. FcγRIIIaPhe158. Injections were performed for 10 s (blue, dashed curve), 20 s (red, dashed curve), and 60 s (black curve) in separate cycles. Curves were overlaid and aligned at the end of injections (time = 0), and dissociation rates were visually compared.

Figure 6. Deglycosylation of FcγRs shows efficient removal of N-glycans. N-glycans were enzymatically removed from FcγRIa and FcγRIIIaPhe158/Val158 using the endoglycosidases PNGase F, Endo F1, and Endo F2 without denaturing, reducing, or alkylating the proteins. A. SDS-PAGE analysis of FcγRs. Enzymatic removal of N-glycans reduced the molecular weights of FcγRIa and FcγRIIIaPhe158/Val158 from between 40 and 50 kDa to the peptide mass of 32 kDa for the FcγRIa extracellular domain and from 40 kDa to approximately 25 kDa for the FcγRIIIaPhe158/Val158 extracellular regions, showing, near complete deglycosylation. Complete deglycosylation of FcγRIIIaPhe158/Val158 may not have been achieved as SDSPAGE reveals a molecular weight of approximately 25 kDa following endoglycosidase treatment, and the peptide mass for FcγRIIIa variants is 23 kDa. In addition, multiple bands can be seen for FcγRIIIa suggesting more than one major glycoform. B. HILIC UPLC analysis showed that Nglycans were removed efficiently from native FcγRIa and FcγRIIIaPhe158/Val158 and that deglycosylation was comparable to reduced and alkylated protein: (i) FcγRIa, (ii) FcγRIa-deglycosylated, (iii) FcγRIIIaPhe158, (iv) FcγRIIIaPhe158-deglycosylated, (v) FcγRIIIaVal158, and (vi) FcγRIIIaVal158deglycosylated.

per interaction cycle, whereas in single cycle kinetics all concentrations are injected in sequence in the same cycle. Binding of rituximab to the complete set of FcγRs is presented in Figure 4. Binding profiles varied considerably depending on the receptor, ranging from stable binding to FcγRIa to rapid dissociation from FcγRIIa, FcγIIb, and FcγRIIIb receptors. For FcγRIIIaPhe158/Val158 dissociation was biphasic with rapid initial dissociation followed by slower dissociation in the late phase. This was more evident for

FcγRIIIaVal158. Data analysis (dotted lines in Figure 4) confirmed this heterogeneity, as antibody binding to FcγRIIIaPhe158/Val158 did not fit well to a one to one binding model. The interaction with FcγRIa also did not adhere to a one to one binding model, but here deviations were significant across all antibody concentrations. For both FcγRIa and FcγRIIIaVal158 the observed binding at high concentrations exceeded the binding level predicted by the fitting model. By fitting data to more complex interaction models, based on 5480

dx.doi.org/10.1021/pr500414q | J. Proteome Res. 2014, 13, 5471−5485

Journal of Proteome Research

Article

FcγRIIIaPhe158/Val158 and that deglycosylation was comparable to reduced and alkylated protein (see Figure 6B). Receptors deglycosylated using this approach allowed subsequent IgG interaction studies to be performed and the role of receptor glycosylation in stability and the binding event to be assessed. FcγRIa and FcγRIIIaPhe158/Val158 were captured to the anti-His antibody on the SPR sensor chip, and the binding of rituximab to deglycosylated receptors was compared with binding to equal amounts of unmodified and control treated receptors (Figure 7). When injecting similar concentrations of glycosylated and deglycosylated receptors, respectively, over the anti-His antibody, similar levels of receptor were captured, indicating that the His tag was available to the same extent. Furthermore, the binding curves during capture and the capture stability of glycosylated and deglycosylated receptors were comparable (not shown), indicating similarities between the preparations and suggesting that deglycosylation had not seriously affected the stability of the proteins, e.g., aggregated receptor would display a different curve shape. In the case of FcγRIa a more than 50% decrease in binding level to deglycosylated receptor was seen, compared to unmodified FcγR mainly due to visibly slower association rate (see Figure 7A). Binding of rituximab to control treated FcγRIa was also decreased, but to a lesser extent. However, binding of rituximab to control treated and unmodified FcγRIIIaPhe158/Val158 was comparable with overlapping binding curves (Figure 7B,C), indicating that the deglycosylation process itself (without enzyme) did not significantly affect binding. Binding levels obtained for FcγRIIIaVal158 (Figure 7B) were higher than those obtained with FcγRIIIaPhe158 (Figure 7C), in agreement with the documented higher affinity for FcγRIIIaVal158. Interestingly, significant changes in antibody binding to deglycosylated receptors were observed. Both deglycosylated FcγRIIIaPhe158 and FcγRIIIaVal158 affinity for rituximab decreased as shown by decreased binding levels compared to controls. Focusing first on the interaction between antibody and FcγRIIIaPhe158 (Figure 7C), antibody binding to glycosylated receptor displayed a biphasic interaction with steep association in the early association phase and slower in the late association phase. A biphasic profile is also seen during dissociation, with initially fast dissociation followed by slower dissociation. In contrast, binding to deglycosylated receptor displayed a more simple, monophasic interaction rapidly reaching equilibrium, remaining at equilibrium during the injection and then rapid dissociation back down to baseline. The binding to deglycosylated receptor allowed fitting to the steady state model in a similar manner as for FcγRIIa, FcγRIIb, and FcγRIIIb (Figure 4), obtaining an estimated KD of 1.7 μM. Similar behavior was also seen with the interaction between antibody and glycosylated/deglycosylated FcγRIIIaVal158 (Figure 7B). The difference being that the binding to deglycosylated receptor still showed some remaining heterogeneity (biphasic association and dissociation), although data almost reached equilibrium and dissociation was faster than that from the glycosylated receptor.

receptor or antibody heterogeneity or conformational change, fittings were significantly improved, but no definite conclusion could be made as to the precise binding mechanism. However, this is not surprising, as we would expect the binding to be complex and differentially modulated by the presence of differently glycosylated variants. For this reason, kinetic and affinity constants for FcγRIa and FcγRIIIaPhe158/Val158 are not reported, as the definition of these constants depend on the binding mechanism. Thus, even if steady state was reached during injection (as was not the case here) fitting equilibrium data to a one to one binding model is not appropriate. The binding behavior of rituximab to FcγRIIIaPhe158/Val158 was further investigated by injecting saturating concentrations (6 μM) of the antibody at different injection times. Binding data were aligned at the end of the injection for analysis. As shown in Figure 5 rituximab dissociated from FcγRIIIaVal158 in an injection time-dependent manner: the longer the injection time the slower the off-rate. In contrast, rituximab dissociation from FcγRIIIaPhe158 was not as clearly affected by the injection time, even though a very minor tendency similar to FcγRIIIaVal158 may be seen. The data for FcγRIIIaVal158 indicates either that the antibody−receptor complex is stabilized over time due to conformational change or that a variant of the antibody with inherent slower dissociation accumulates on the sensor surface over time. Injection time-dependent dissociation was also observed with infliximab and omalizumab (not shown) and was therefore not unique to rituximab. For FcγRIIa, FcγRIIb, and FcγRIIIb, the shapes of the binding curves obtained were very similar, as shown in Figure 4D−F. For these low affinity receptors sensorgrams were characterized by rapid approach to an equilibrium phase (steady state) followed by rapid dissociation. Rituximab affinity data from three separate determinations was fitted to a steady state model generating KD values for FcγRIIa of 4.2 μM (± 0.95), FcγRIIb 6.9 μM (± 0.67), and FcγRIIIb 3.5 (± 0.15) μM, respectively. Deglycosylation of FcγRs Influences Binding and Interaction Kinetics of Rituximab

The high affinity FcγRIa and low affinity FcγRIIIaPhe158/Val158 were enzymatically deglycosylated without denaturing, reducing, or alkylating the proteins using a combination of the endoglycosidases: PNGase F, Endo F1, and Endo F2. Appropriate deglycosylation controls were included such as FcγRIa and FcγRIIIaPhe158/Val158 treated identically to their deglycosylated forms but without the endoglycosidase enzymes. The extent of receptor deglycosylation was monitored by SDSPAGE analysis of deglycosylated receptors and by labeling the endoglycosidase released receptor N-glycans with 2-AB followed by HILIC UPLC analysis. SDS-PAGE showed that endoglycosidase treatment reduced the molecular weights of FcγRIa and FcγRIIIaPhe158/Val158 from between 40 and 50 kDa to the peptide mass of 32 kDa for the FcγRIa extracellular domain and from 40 kDa to approximately 25 kDa for the FcγRIIIaPhe158/Val158 extracellular regions, showing, near complete deglycosylation (see Figure 6A). Complete deglycosylation of FcγRIIIaPhe158/Val158 may not have been achieved as SDSPAGE reveals a molecular weight of approximately 25 kDa following endoglycosidase treatment and the peptide mass for FcγRIIIa variants is 23 kDa. In addition, multiple bands can be seen for FcγRIIIa suggesting more than one major glycoform (see Figure 6A). HILIC UPLC analysis also showed that Nglycans were removed efficiently from native FcγRIa and

■

DISCUSSION The binding of IgG antibodies with FcγRs is a complex interaction, further complicated by the significant heterogeneity that exists in both the antibody and the receptor due to the presence of differently glycosylated species. The interaction of rituximab with glyco-analyzed FcγRs was studied to analyze the influence of FcγR glycosylation on antibody binding. Structural 5481

dx.doi.org/10.1021/pr500414q | J. Proteome Res. 2014, 13, 5471−5485

Journal of Proteome Research

Article

HEK293/kidney cells28−32 but relatively uncommon to human N-glycans. Some of these LacdiNAc structures were also capped with sialic acids, and others contained outer-arm fucose residues to form the GalNAc equivalent of Lewis B type structures. It is unclear what effect these unusual carbohydrate epitopes have on IgG binding, and it is unlikely that they would be found on natural receptors; however, Zeck et al. report differences in binding due to the larger HEK293 type glycans in comparison to smaller CHO glycans.27 Previous studies have reported on the glycosylation of recombinant FcγRs, even though limited information is available regarding the glycosylation state of the natural forms. We have previously performed extensive analysis of human FcγRs expressed in NS0 cells and found extensive murine specific and immunogenic carbohydrate epitopes such as gal-α(1−3)-gal and N-glycolyneuraminic acid.33 Zeck et al.27 performed a study of site-specific glycosylation of FcγRIIIa expressed in HEK293 and CHO cells. Similar to what we have seen, large amounts of LacdiNAc structures and outer-arm fucosylation was found. In addition, the authors report on the sugars found at the Asn162 residue: the site closest to the IgG binding site and the site where the glycan involved in carbohydrate−carbohydrate interactions with IgG resides.1517 Both FcγRIIa and FcγRIIIb have been expressed in baby hamster kidney cells (BHK),20,21 displaying multiantennary structures containing up to four GlcNAc residues with minimal sialic acid capping. Interestingly, Zeck et al. also report batch to batch variations in FcγRIIIaVal158 receptor glycosylation but identified three abundant outer arm fucosylated glycans on the Asn162 residue that were implicated in receptor/IgG binding. They also demonstrate heterogeneity in binding to the receptor by different IgG glycoforms. Ferrara et al. expressed FcγRIIIaVal158 receptor variants with reduced glycan complexity, including a receptor with 100% oligomannose type sugars to investigate IgG/FcγR binding. They, however, did not use heterogeneous IgG glycoforms, but rather antibody forms that were either fucosylated or afucosylated. Close inspection of their binding data reveals lower heterogeneity for the receptor variants with reduced complexity in glycan expression. We identified several high mannose glycans in our preparations similar to Ferrara et al. but fewer outer arm fucosylated glycans than Zeck et al. A common feature for all investigators, however, is that IgG/receptor binding is heterogeneous. These results indicate that the specificity of glycan binding can be low, and this may partly explain the heterogeneity of binding that is observed. Zeck and Ferrara also focused on FcγRIIIaVal158, whereas we also investigated binding to FcγRIIIaPhe158. IgG binding to FcγRIIIaPhe158 is of lower affinity but remains complex. Finally, FcγRIa is also interesting as deglycosylation of the receptor appears to lower the binding of IgG but with less impact on binding kinetics and only slightly reduced heterogeneity. While these experiments continue to shed light on the binding mechanisms involved, the exact nature of the sugars involved in receptor and IgG binding is still unclear, and the challenge to determine individual rate and affinity constants remains. With heterogeneity in both receptor and antibody, at least eight rate constants have to be determined to describe the interaction. Interaction of therapeutic antibodies with recombinant FcγRs is widely used to study the specifity, affinity, and effectiveness of antibodies. Numerous studies report using HEK293 expressed receptors in IgG binding experiments.27,34,35 Here, we introduce the use of single cycle kinetics for these types of

Figure 7. Deglycosylation of FcγRs results in altered binding kinetics of rituximab: A. FcγRI, B. FcγRIIIaVal158, and C. FcγRIIIaPhe158. In each panel, rituximab binding to equal amounts of captured unmodified glycosylated FcγR, glycosylation control FcγR, and deglycosylated FcγR is shown. From top to bottom, the sensorgrams in each panel represent unmodified FcγR (black), FcγR deglycosylation control (blue), and deglycosylated FcγR (red). The deglycosylation control in each case is receptor-treated identically to the deglycosylated receptor but without the deglycosylating enzymes (PNGase F, EndoF1, and EndoF2). Five injections of rituximab in increasing concentration are performed in each curve, resulting in increasing binding level. Dissociation is seen starting from the end of injection of the highest concentration, at time 550 s, as well as from the end of injection of the lower concentrations.

characterization using techniques such as HILIC separation of fluorescently labeled N-glycans, WAX analysis, WAX fractionation, and mass spectrometry, including MS/MS analysis revealed a close link between the carbohydrates of different FcγRs and more than 50% were structures common to all FcγRs examined. The most striking feature of FcγR glycosylation was the extensive presence of LacdiNAc residues (19%), a feature not uncommon to proteins expressed in 5482

dx.doi.org/10.1021/pr500414q | J. Proteome Res. 2014, 13, 5471−5485

Journal of Proteome Research

Article

FcγRIIIaVal158, time-dependent dissociation was observed, indicating either that the antibody−receptor complex stabilized over time due to conformational change or that a variant of the antibody with inherent slower dissociation accumulated on the sensor surface over time. Knowing that approximately 8% of the rituximab antibody preparation used here was afucosylated, these results therefore suggest that a large part of the heterogeneity of rituximab binding to FcγRIIIaVal158 is related to heterogeneity in the antibody. In contrast, the variation in dissociation rate observed for rituximab binding to FcγRIIIaPhe158 with time was minimal, however, with similar tendency as for FcγRIIIaVal158. Cardarelli et al. show that there is indeed binding of afucosylated antibody also to FcγRIIIaPhe158 but to a far less extent than to FcγRIIIaVal158.38 Thus, the relatively smaller complexity in rituximab binding observed for FcγRIIIaPhe158 could be explained by independent binding of rituximab to different receptor forms, i.e., by receptor heterogeneity and to a lesser extent with heterogeneity in the antibody. The injection time experiments shown in Figure 5 demonstrate slower dissociation for prolonged injection times consistent with accumulation of the high affinity afucosylated rituximab. This phenomenon may be exploited as a means to determine the relative content of afucosylated antibody in different batches, where higher content of afucosylated antibody may be linked to the higher response at a specific time after the end of injection. Another extension of this approach can be to prolong injection times until no further stabilization of binding is observed. At this point the receptor will have the highest occupancy of the high affinity variant, and a dissociation rate constant for the afucosylated version may be obtained from this data. Data from injection time experiments and experiments using deglycosylated receptors are clearly supportive of each other and demonstrate that both receptor and rituximab heterogeneity is reflected in the observed kinetics. It has previously been demonstrated that nonfucosylated IgG binds much tighter in particular to FcγRIIIaVal158,15 and more recently Zeck et al. demonstrated that when nonfucosylated IgG1 was used as the analyte, heterogeneity in binding to FcγRIIIaVal158 was eliminated.27 The observation that rituximab interactions with deglycosylated receptors become more homogeneous is consistent with FcγR heterogeneity. When rituximab binding to deglycosylated FcγRIa was analyzed, we observed a significant drop in the antibody binding level but no apparent change in the dissociation rate. With deglycosylated FcγRIIIa receptors, a similar, albeit less pronounced, decrease in binding levels was observed, but interaction kinetics were changed and rituximab dissociated more quickly from the receptors. Interestingly, the observed heterogeneity of rituximab binding to deglycosylated FcγRIIIaPhe158 was eliminated when the receptor was deglycosylated, and here, steady state was reached. This observation adds further strength to the impact of receptor glycosylation on the heterogeneity observed and its influence on the IgG−FcγR interaction. However, heterogeneity was still observed for rituximab binding to deglycosylated FcγRIIIaVal158. A number of important studies have implicated receptor glycosylation as an important contributor to the IgG−FcγR interaction. The vast majority of data comes from FcγRIIIa due to its role in ADCC and its importance to the biopharmaceutical industry. N-linked glycans have been shown to positively and negatively regulate IgG binding, and recently, crystallization of FcγRIIIa with afucosylated anti-CD20 antibody

binding studies. Observing interactions and complex formation that does not dissociate completely between injections allows studies over an extended period of time, resulting in higher resolution of the data. Furthermore, in combination with capture of the FcγR, the single cycle kinetic approach saves FcγR reagent and time compared with the more traditional kinetic assay setup where only one antibody concentration is injected per interaction cycle. The two kinetic procedures result in similar kinetic constants.25 We selected the anti-CD20 monoclonal antibody rituximab/mabthera as a suitable model antibody for comparison of binding behavior for different FcγRs. Consistent with previous reports,6,19,27,36 affinity to FcγRs varied depending on the receptor. The highest affinity was to FcγRIa and lower affinity to the remaining receptors. KD values of 4.2, 6.9, and 3.5 μM were obtained for FcγRIIa, FcγRIIb, and FcγRIIIb, respectively, following fitting to a steady state model, reflecting the low affinity nature of these receptors. Bruhns et al.6 performed an extensive study presenting data for human IgG subclasses, as well as for anti-CD20 antibodies interacting with HEK293 expressed FcγRs. Our rituximab affinity data agree well with their anti-CD20 data for FcγRIIa, FcγRIIb, and FcγRIIIb with data in the 0.5−7 μM range for these receptors. Visser et al. also present SPR affinity data for rituximab binding to the different FcγRs with reported KD values for FcγRIIa, FcγRIIb, and FcγRIIIb all in the range 2.4− 12.8 μM.19 Taking into account the different SPR assay setups used (Bruhns and Visser amine coupled FcγRs, whereas the FcγRs used here were immobilized via the N-terminal His-tag) as well as possible differences in receptor qualities, reported affinities for these receptors are consistent. In our study, binding of rituximab to FcγRs was characterized by a rapid approach to steady state followed by fast dissociation and the sensorgrams contained little kinetic data. To our knowledge single cycle kinetics sensorgrams for rituximab binding to FcγRs have not previously been published. The kinetic analysis of rituximab binding to FcγRIa and FcγRIIIaPhe158/Val158 was complex. When data were fitted assuming a one-to-one binding mechanism, large deviations were observed, in particular for FcγRIa and FcγRIIIaVal158, whereas rituximab binding to FcγRIIIaPhe158 displayed a fit closer to a one-to-one binding mechanism. More complex interaction models (conformational change, FcγR heterogeneity, antibody heterogeneity, and avidity models) gave better fitting, but it was not possible to define a precise interaction mechanism. It is likely that both receptor and antibody heterogeneity is reflected in the observed kinetics, particularly fucosylated and nonfucosylated antibody glycoforms together with FcγR heterogeneity. As already mentioned, the binding mechanism is complex, and it therefore cannot be described by a single set of rate constants and by one unique affinity constant. A minimum of four rate constants and two affinity constants would be required to describe the data. To avoid overinterpretation or simplification of these interactions, we decided not to report values on kinetic and affinity constants. Affinities reported in the literature vary significantly due to the fitting difficulties and differences in experimental setup, as well in the antibodies and FcγRs used. To investigate the binding mechanism in more detail we performed an interaction study in which rituximab was allowed to bind to FcγRIIIa at saturating concentrations during 10, 20, or 60 s, after which the dissociation was compared; see Karlsson and Falt.37 Clear differences between the morphologic variants of FcγRIIIa were evident (see Figure 5). With 5483

dx.doi.org/10.1021/pr500414q | J. Proteome Res. 2014, 13, 5471−5485

Journal of Proteome Research

Article

components of the lymphoid system. Immunology 1971, 21 (6), 967− 81. (2) Anderson, C. L.; Shen, L.; Eicher, D. M.; Wewers, M. D.; Gill, J. K. Phagocytosis mediated by three distinct Fc gamma receptor classes on human leukocytes. J. Exp. Med. 1990, 171 (4), 1333−45. (3) Young, J. D.; Ko, S. S.; Cohn, Z. A. The increase in intracellular free calcium associated with IgG gamma 2b/gamma 1 Fc receptorligand interactions: role in phagocytosis. Proc. Natl. Acad. Sci. U.S.A. 1984, 81 (17), 5430−4. (4) Titus, J. A.; Perez, P.; Kaubisch, A.; Garrido, M. A.; Segal, D. M. Human K-natural killer-cells targeted with hetero-cross-linked antibodies specifically lyse tumor-cells invitro and prevent tumor-growth invivo. J. Immunol. 1987, 139 (9), 3153−3158. (5) Kilchherr, E.; Schumaker, V. N.; Phillips, M. L.; Curtiss, L. K. Activation of the first component of human complement, C1, by monoclonal antibodies directed against different domains of subcomponent C1q. J. Immunol 1986, 137 (1), 255−62. (6) Bruhns, P.; Iannascoli, B.; England, P.; Mancardi, D. A.; Fernandez, N.; Jorieux, S.; Daeron, M. Specificity and affinity of human Fcgamma receptors and their polymorphic variants for human IgG subclasses. Blood 2009, 113 (16), 3716−25. (7) Duits, A. J.; Bootsma, H.; Derksen, R. H.; Spronk, P. E.; Kater, L.; Kallenberg, C. G.; Capel, P. J.; Westerdaal, N. A.; Spierenburg, G. T.; Gmelig-Meyling, F. H.; et al. Skewed distribution of IgG Fc receptor IIa (CD32) polymorphism is associated with renal disease in systemic lupus erythematosus patients. Arthritis Rheum. 1995, 38 (12), 1832−6. (8) Haseley, L. A.; Wisnieski, J. J.; Denburg, M. R.; MichaelGrossman, A. R.; Ginzler, E. M.; Gourley, M. F.; Hoffman, J. H.; Kimberly, R. P.; Salmon, J. E. Antibodies to C1q in systemic lupus erythematosus: characteristics and relation to Fc gamma RIIA alleles. Kidney Int. 1997, 52 (5), 1375−80. (9) Sanders, L. A.; van de Winkel, J. G.; Rijkers, G. T.; VoorhorstOgink, M. M.; de Haas, M.; Capel, P. J.; Zegers, B. J. Fc gamma receptor IIa (CD32) heterogeneity in patients with recurrent bacterial respiratory tract infections. J. Infect. Dis. 1994, 170 (4), 854−61. (10) Wu, J.; Edberg, J. C.; Redecha, P. B.; Bansal, V.; Guyre, P. M.; Coleman, K.; Salmon, J. E.; Kimberly, R. P. A novel polymorphism of FcgammaRIIIa (CD16) alters receptor function and predisposes to autoimmune disease. J. Clin. Invest. 1997, 100 (5), 1059−70. (11) Nieto, A.; Caliz, R.; Pascual, M.; Mataran, L.; Garcia, S.; Martin, J. Involvement of Fcgamma receptor IIIA genotypes in susceptibility to rheumatoid arthritis. Arthritis Rheum. 2000, 43 (4), 735−9. (12) Kimberly, R. P.; Tappe, N. J.; Merriam, L. T.; Redecha, P. B.; Edberg, J. C.; Schwartzman, S.; Valinsky, J. E. Carbohydrates on human Fc gamma receptors. Interdependence of the classical IgG and nonclassical lectin-binding sites on human Fc gamma RIII expressed on neutrophils. J. Immunol 1989, 142 (11), 3923−30. (13) Jefferis, R.; Lund, J.; Pound, J. D. IgG-Fc-mediated effector functions: molecular definition of interaction sites for effector ligands and the role of glycosylation. Immunol Rev. 1998, 163, 59−76. (14) Allhorn, M.; Olin, A. I.; Nimmerjahn, F.; Collin, M. Human IgG/Fc gamma R interactions are modulated by streptococcal IgG glycan hydrolysis. PLoS One 2008, 3 (1), e1413. (15) Ferrara, C.; Stuart, F.; Sondermann, P.; Brunker, P.; Umana, P. The carbohydrate at FcgammaRIIIa Asn-162. An element required for high affinity binding to non-fucosylated IgG glycoforms. J. Biol. Chem. 2006, 281 (8), 5032−6. (16) Shibata-Koyama, M.; Iida, S.; Okazaki, A.; Mori, K.; KitajimaMiyama, K.; Saitou, S.; Kakita, S.; Kanda, Y.; Shitara, K.; Kato, K.; Satoh, M. The N-linked oligosaccharide at Fc gamma RIIIa Asn-45: an inhibitory element for high Fc gamma RIIIa binding affinity to IgG glycoforms lacking core fucosylation. Glycobiology 2009, 19 (2), 126− 34. (17) Ferrara, C.; Grau, S.; Jager, C.; Sondermann, P.; Brunker, P.; Waldhauer, I.; Hennig, M.; Ruf, A.; Rufer, A. C.; Stihle, M.; Umana, P.; Benz, J. Unique carbohydrate-carbohydrate interactions are required for high affinity binding between FcgammaRIII and antibodies lacking core fucose. Proc. Natl. Acad. Sci. U.S.A. 2011, 108 (31), 12669−74.

revealed a unique interaction consisting of a carbohydrate− carbohydrate interface, which is weakened when the antibody is core-fucosylated.15−17,39 We and others believe that Fcγ receptor glycosylation and IgG binding are intricately linked, and therefore, the glycan structural analysis described here and its influence on antibody interactions is necessary to provide a more complete picture of the interaction. We believe that the data presented provides further compelling evidence for the role of FcγR N-glycans in the interaction with IgG and the immune response. To our knowledge this is the first report that shows that the binding interaction with antibody is modulated and stabilized by receptor glycosylation and that antibody binds with reduced affinity and dissociates quicker from the receptor in the absence of carbohydrates, making it less potent. It remains a challenge to fully characterize the impact of FcγR glycosylation and the sugars involved, particularly in their natural forms by cells such as monocytes, B-cells, and NK cells and their role in the immune response. Several important papers by Kimberly et al. and Edberg et al. have addressed this question, typically using traditional glycoprotein analysis techniques such as lectin binding assays.12,18 Several critical attributes were described such as the cell type specific glycosylation of FcγRIIIa by monocytes and NK cells and differential ligand binding to these receptors.40 This intriguing discovery suggests that different cells of the innate immune system will use IgG differently depending on the glycosylation status of the receptor. This has implications for activation and control of the immune response based on the cells’ own glycosylation machinery and its preference to glycosylate in a cell-type specific manner. Edberg et al. also comment on the somewhat ubiquitous presence of high mannose type glycans associated with FcγRs and suggest that they may play a role in opsonin-dependent receptor ligation, a key feature in phagocytic uptake of bacterial pathogens displaying lectintype adhesins in fimbriated Escherichia coli, Salmonella typhimurium, and Pseudomonas aeruginosa. This is particularly fascinating as it may be the case that FcγRs have unknown functions, such as in nonclassical receptor ligation. It is also highly likely that glycosylation of receptors other than FcγRs will play a similar role in many other protein−receptor and protein−ligand interactions.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Phone: (353) 1 896 3527. Fax: (353) 1 677 2400. Notes

The authors declare the following competing financial interest(s): A.F., R.K., and C.A. are employees of GE Healthcare; E.C. is an amployee of Waters Corporation; and O.P. is an employee of Agilent Corporation.

■

ACKNOWLEDGMENTS This work was supported by the National Institute for Bioprocessing, Research, and Training (NIBRT), GE Healthcare, and the Industrial Development Authority (IDA) Ireland.

■

REFERENCES

(1) Chan, P. L.; Sinclair, N. R. Regulation of the immune response. V. An analysis of the function of the Fc portion of antibody in suppression of an immune response with respect to interaction with 5484

dx.doi.org/10.1021/pr500414q | J. Proteome Res. 2014, 13, 5471−5485

Journal of Proteome Research

Article

immunoglobulin Fc fragment (Fcab). Arch. Biochem. Biophys. 2012, 526 (2), 154−8. (36) Maenaka, K.; van der Merwe, P. A.; Stuart, D. I.; Jones, E. Y.; Sondermann, P. The human low affinity Fcgamma receptors IIa, IIb, and III bind IgG with fast kinetics and distinct thermodynamic properties. J. Biol. Chem. 2001, 276 (48), 44898−904. (37) Karlsson, R.; Falt, A. Experimental design for kinetic analysis of protein-protein interactions with surface plasmon resonance biosensors. J. Immunol. Methods 1997, 200 (1−2), 121−33. (38) Cardarelli, P. M.; Rao-Naik, C.; Chen, S.; Huang, H.; Pham, A.; Moldovan-Loomis, M. C.; Pan, C.; Preston, B.; Passmore, D.; Liu, J.; Kuhne, M. R.; Witte, A.; Blanset, D.; King, D. J. A nonfucosylated human antibody to CD19 with potent B-cell depletive activity for therapy of B-cell malignancies. Cancer Immunol. Immunother. 2010, 59 (2), 257−65. (39) Mizushima, T.; Yagi, H.; Takemoto, E.; Shibata-Koyama, M.; Isoda, Y.; Iida, S.; Masuda, K.; Satoh, M.; Kato, K. Structural basis for improved efficacy of therapeutic antibodies on defucosylation of their Fc glycans. Genes Cells 2011, 16 (11), 1071−80. (40) Edberg, J. C.; Barinsky, M.; Redecha, P. B.; Salmon, J. E.; Kimberly, R. P. Fc gamma RIII expressed on cultured monocytes is a N-glycosylated transmembrane protein distinct from Fc gamma RIII expressed on natural killer cells. J. Immunol. 1990, 144 (12), 4729−34. (41) Harvey, D. J.; Mattu, T. S.; Wormald, M. R.; Royle, L.; Dwek, R. A.; Rudd, P. M. ″Internal residue loss″: rearrangements occurring during the fragmentation of carbohydrates derivatized at the reducing terminus. Anal. Chem. 2002, 74 (4), 734−40. (42) Tomiya, N.; Lee, Y. C.; Yoshida, T.; Wada, Y.; Awaya, J.; Kurono, M.; Takahashi, N. Calculated two-dimensional sugar map of pyridylaminated oligosaccharides: elucidation of the jack bean alphamannosidase digestion pathway of Man9GlcNAc2. Anal. Biochem. 1991, 193 (1), 90−100. (43) Dohi, K.; Isoyama-Tanaka, J.; Misaki, R.; Fujiyama, K. Jack bean alpha-mannosidase digestion profile of hybrid-type N-glycans: effect of reaction pH on substrate preference. Biochimie 2011, 93 (4), 766−71.

(18) Edberg, J. C.; Kimberly, R. P. Cell type-specific glycoforms of Fc gamma RIIIa (CD16): differential ligand binding. J. Immunol. 1997, 159 (8), 3849−57. (19) Visser, J.; Feuerstein, I.; Stangler, T.; Schmiederer, T.; Fritsch, C.; Schiestl, M. Physicochemical and functional comparability between the proposed biosimilar rituximab GP2013 and originator rituximab. BioDrugs 2013, 27 (5), 495−507. (20) Takahashi, N.; Yamada, W.; Masuda, K.; Araki, H.; Tsukamoto, Y.; Galinha, A.; Sautes, C.; Kato, K.; Shimada, I. N-glycan structures of a recombinant mouse soluble Fcgamma receptor II. Glycoconjugate J. 1998, 15 (9), 905−14. (21) Takahashi, N.; Cohen-Solal, J.; Galinha, A.; Fridman, W. H.; Sautes-Fridman, C.; Kato, K. N-glycosylation profile of recombinant human soluble Fcgamma receptor III. Glycobiology 2002, 12 (8), 507− 15. (22) Royle, L.; Radcliffe, C. M.; Dwek, R. A.; Rudd, P. M. Detailed structural analysis of N-glycans released from glycoproteins in SDSPAGE gel bands using HPLC combined with exoglycosidase array digestions. Methods Mol. Biol. 2006, 347, 125−43. (23) Ahn, J.; Bones, J.; Yu, Y. Q.; Rudd, P. M.; Gilar, M. Separation of 2-aminobenzamide labeled glycans using hydrophilic interaction chromatography columns packed with 1.7 microm sorbent. J. Chromatogr. B: Anal. Technol. Biomed. Life Sci. 2010, 878 (3−4), 403−8. (24) Harvey, D. J.; Merry, A. H.; Royle, L.; Campbell, M. P.; Dwek, R. A.; Rudd, P. M. Proposal for a standard system for drawing structural diagrams of N- and O-linked carbohydrates and related compounds. Proteomics 2009, 9 (15), 3796−801. (25) Karlsson, R.; Katsamba, P. S.; Nordin, H.; Pol, E.; Myszka, D. G. Analyzing a kinetic titration series using affinity biosensors. Anal. Biochem. 2006, 349 (1), 136−47. (26) Laemmli, U. K. Cleavage of structural proteins during assembly of head of bacteriophage-T4. Nature 1970, 227 (5259), 680−&. (27) Zeck, A.; Pohlentz, G.; Schlothauer, T.; Peter-Katalinic, J.; Regula, J. T. Cell type-specific and site directed N-glycosylation pattern of FcgammaRIIIa. J. Proteome Res. 2011, 10 (7), 3031−9. (28) Van den Eijnden, D. H.; Neeleman, A. P.; Van der Knaap, W. P.; Bakker, H.; Agterberg, M.; Van Die, I. Novel glycosylation routes for glycoproteins: the lacdiNAc pathway. Biochem. Soc. Trans. 1995, 23 (1), 175−9. (29) Do, K. Y.; Do, S. I.; Cummings, R. D. Differential expression of LacdiNAc sequences (GalNAc beta 1−4GlcNAc-R) in glycoproteins synthesized by Chinese hamster ovary and human 293 cells. Glycobiology 1997, 7 (2), 183−94. (30) Smith, P. L.; Skelton, T. P.; Fiete, D.; Dharmesh, S. M.; Beranek, M. C.; MacPhail, L.; Broze, G. J., Jr.; Baenziger, J. U. The asparaginelinked oligosaccharides on tissue factor pathway inhibitor terminate with SO4−4GalNAc beta 1, 4GlcNAc beta 1,2 Mana alpha. J. Biol. Chem. 1992, 267 (27), 19140−6. (31) Hiyama, J.; Weisshaar, G.; Renwick, A. G. The asparagine-linked oligosaccharides at individual glycosylation sites in human thyrotrophin. Glycobiology 1992, 2 (5), 401−9. (32) Sato, T.; Gotoh, M.; Kiyohara, K.; Kameyama, A.; Kubota, T.; Kikuchi, N.; Ishizuka, Y.; Iwasaki, H.; Togayachi, A.; Kudo, T.; Ohkura, T.; Nakanishi, H.; Narimatsu, H. Molecular cloning and characterization of a novel human beta 1,4-N-acetylgalactosaminyltransferase, beta 4GalNAc-T3, responsible for the synthesis of N,N′diacetyllactosediamine, galNAc beta 1−4GlcNAc. J. Biol. Chem. 2003, 278 (48), 47534−44. (33) Cosgrave, E. F.; Struwe, W. B.; Hayes, J. M.; Harvey, D. J.; Wormald, M. R.; Rudd, P. M. N-linked glycan structures of the human Fcgamma receptors produced in NS0 cells. J. Proteome Res. 2013, 12 (8), 3721−37. (34) Ha, S.; Ou, Y.; Vlasak, J.; Li, Y.; Wang, S.; Vo, K.; Du, Y.; Mach, A.; Fang, Y.; Zhang, N. Isolation and characterization of IgG1 with asymmetrical Fc glycosylation. Glycobiology 2011, 21 (8), 1087−96. (35) Kainer, M.; Antes, B.; Wiederkum, S.; Wozniak-Knopp, G.; Bauer, A.; Ruker, F.; Woisetschlager, M. Correlation between CD16a binding and immuno effector functionality of an antigen specific 5485

dx.doi.org/10.1021/pr500414q | J. Proteome Res. 2014, 13, 5471−5485