Enhancing activity of FcαRI-bispecific antibodies using glycoengineering

Abstract

Macrophages and natural killer (NK) cells can effectively kill tumor cells in the presence of anti-cancer IgG monoclonal antibodies (mAbs), but neutrophils are less effective. We previously showed that IgG1 bispecific antibodies (BsAb), which target the IgA Fc receptor (FcαRI, CD89) and a tumor associated antigen induce effective neutrophil recruitment and tumor cell killing in vivo. Here we investigated if the efficacy of an anti-EGFR (CetuximAb)/FcαRI-bispecific antibody could be further improved by implementing glycoengineering of the IgG-Fc, aimed at increasing FcγRIIIa/b binding and/or complement activity. Fc afucosylation was introduced to enhance antibody-dependent cellular cytotoxicity (ADCC) by FcγRIIIa on NK/macrophages, which can also reduce neutrophil-mediated ADCC through their GPI-linked FcγRIIIb. Fc galactylation was found to enhance antibody hexamerization and thereby complement dependent cytotoxicity (CDC). Low fucosylated BsAbs moderately increased NK cell-mediated tumor cell killing, but did not affect neutrophil-mediated tumor cell killing nor phagocytosis by macrophages. Glycoengineering of these EGFR-specific BsAb, which normally are devoid of CDC-activity, did not enable their complement activities. In conclusion, glycoengineered FcαRI BsAbs increased ADCC by NK cells but had little effect on neutrophil or macrophage mediated tumor killing.

Introduction

Monoclonal antibody therapy is a promising strategy to treat cancer, as they link tumor cells with the immune system.^1,2 Most clinically used mAbs are of the IgG isotype,^3–5 which interact with different components of the immune system. IgG antibodies can activate complement⁶ but also interact effectively with FcγR (FcγRI, FcγRIIa, FcγRIIb, FcγRIIIa and FcγRIIIb), which are expressed on effector immune cells.^5,7,8 Antibodies can induce various mechanisms, including (1) complement- dependent cytotoxicity (CDC), (2) antibody-dependent cellular cytotoxicity (ADCC) by natural killer (NK) cells (via FcγRIIIa, CD16a),⁹ and (3) antibody-dependent cellular phagocytosis (ADCP) by macrophages (via FcγRI, CD64; FcγRIIa, CD32a; and FcγRIIIa).^10,11 In addition, triggering of the GPI-linked FcγRIIIb, expressed solely on neutrophils, can lead to stronger bacterial and phagocytosis of platelets, mediated by FcγRIa and FcγRIIa, by diverting signaling from these transmembrane receptors.^12–14

In addition to macrophages and NK cells, neutrophils have shown to mediate tumor cell killing in the presence of mAbs, via a process referred to as trogocytosis.^15–17 Enhanced trogocytosis and tumor cell elimination was induced after targeting neutrophils via FcαRI (CD89), which has been suggested as an alternative potential therapeutic target,^15,18,19 especially as targeting FcαRI on neutrophils initiated leukotriene B4 release resulting in neutrophil recruitment.²⁰ It was demonstrated that FcαRI bispecific antibodies (F(ab′)₂ anti-FcαRI × F(ab′)₂ Her2/Neu) promoted neutrophil recruitment and killing of breast cancer colonies in vitro. This also led to secretion of pro-inflammatory cytokines, such as IL-1β and TNF-α by neutrophils.¹⁵ However, both F(ab′)₂ × F(ab′)₂ FcαRI BsAb and IgA have short-half life in vivo, which may hamper clinical development and application.

To tackle these limitations, we recently developed a BsAb targeting a tumor associated antigen and FcαRI with a functional IgG1 Fc tail.²¹ FcαRI BsAb recruited and activated neutrophils via FcαRI, and binding to FcγR additionally induced ADCC or ADCP by NK cells or macrophages in vitro, respectively. Moreover, in vivo treatment with anti-FcαRI BsAbs delayed tumor development. Tumors were, however, not eliminated completely, indicating that efficacy needs to be improved.

It was previously shown that glycoengineering can improve effector functions of antibodies.^22,23 While the removal of fucose increases affinity to FcγRIIIa and FcγRIIIb, galactosylation of human IgG1 enhances its capacity to form hexamers on IgG-opsonized surfaces, and thereby to bind and activate the first component of the classical complement pathway: C1q.^24–29 In addition, combined afucosylation and galactosylation induced slightly enhanced FcγRIII affinity compared to afucosylation only.^24,30,31 Enhanced FcγRIIIa-binding leads to increased ADCC by NK cells.^24,25,32,33 Afucosylated antibodies also triggered enhanced phagocytosis by THP-1 cells in vitro and by Kupffer cells in vivo.^34,35 IgG-Fc fucosylation stimulated polymorphonuclear neutrophils (PMN) activation which led to induction of phagocytosis.^13,14,36 Purified PMNs induced enhanced phagocytosis of opsonized chronic lymphocytic leukemia (CLL) cells in the presence of low fucose antibodies compared to wild type (WT) antibodies.³⁶ In another study it was however shown that lack of fucose decreased ADCC by PMN.¹² Unlike fucosylation and galactosylation, bisection or sialylation of the IgG-Fc has no or negligible effect on either ADCC or CDC, with the strongest effect being a slight negative effect of sialylation on FcγRIIIa binding and activity of afucosylated and bisected IgG1.^24,26

In this manuscript, we created 3 FcαRI BsAb IgG-Fc glycovariants, that is, low fucose, high galactose, and a combination of high galactose with low fucose and investigated its effect on CDC as well as tumor cell killing mediated via neutrophils, macrophages and NK cells in vitro.

Methods

Generation of glycoengineered FcαRI BsAbs

Generation of FcαRI BsAb was previously described.²¹ Glycoengineering was achieved by producing both parental IgG1 anti-EGFR and parental IgG1 anti-FcαRI monoclonal antibodies (mAbs) in specific conditions. Low-fucosylated or high galactose FcαRI x EGFR BsAb were generated by producing parental mAbs in human embryonic kidney cells (HEK) Freestyle^TM 293 System in the presence of the fucose analogue 2-deoxy-2-fluoro-fucose (2FF) (Carbosynth, Berkshire, United Kingdom) or the glycosyltransferase B-1,4-galactosyltransferase 1 (B4GALT1), co-transfected with D-galactose to increase galactose, respectively, as previously described.²⁴ In addition, low fucosylated with high galactosylated FcαRI × EGFR BsAbs were generated in the presence of 2FF and D-galactose (Sigma Aldrich, Saint Louis, MO) and co-transfected with the galactosyltransferase B4GALT1. Culture supernatants, containing up to 200 mg/ml recombinant IgG with either (glycomodified) anti-EGFR or anti-FcαRI mAbs were collected 5 days after transfection, purified on protein G HiTrap HP columns using ActaPrime Plus and dialyzed against PBS overnight. To determine the IgG Fc N-glycosylation, antibodies were subjected to tryptic cleavage followed by Liquid-chromatography mass-spectrometry (MS) as described previously.³⁷ LC-MS spectra were analyzed with LaCyTools.³⁸ The IgG Fc N-glycosylation traits Fucosylation, Galactosylation, Bisection and Sialylation were calculated by normalizing the relative intensity of each glycopeptide to the sum of their total areas as previously described.³⁷

FcαRI × EGFR BsAb glycovariants were generated via controlled Fab-arm exchange (FAE) with DuoBody technology.^21,39 In short, equimolar amounts of the two different non- and glycoengineered IgG1 variants anti-EGFR G1m(f) allotype containing the point mutation F405L and anti-FcαRI G1m(a) allotype containing the K409R mutation) were mixed with 2-Mercapto-ethyl-amine hydrochloride (2-MEA) and incubated to enable the exchange. After incubation, 2-MEA was removed using various desalting methods and samples were stored overnight at 4 °C to allow reoxidation of the disulfide bonds. This process allowed the formation of BsAbs by swapping the heavy chains between the antibody variants.

Allotype specific IgG ELISA

FAE exchange was verified using a sandwich ELISA utilizing the allotypic epitope differences between the non- and glycoengineered parentals and glycoengineered FcαRI BsAbs, as described previously.²¹ 96 well plates were coated with 2 µg/mL antibodies against anti-EGFR IgG1_F405L (G1m(f) allotype)) or antibodies against anti-FcαRI (anti-allotype G1m(a)) (100 µl/well in PBS) overnight at 4 °C. The following day, plates were washed thrice with PBS-Tween (0.05% (v/v) Tween 20 (Sigma) in PBS) (200 µl/well) followed by a blocking step with 0.05% (v/v) milk (Skim milk powder) in PBS-Tween (200 µl/well) for 1 h at RT. After PBS-Tween washing, non- and glycoengineered anti-EGFR IgG1_F405L (G1m(f) allotype), anti-FcαRI IgG1_K409R (G1m(a) allotype) or the mixed glycoengineered FcαRI BsAbs was added at a serial dilution of 2-fold from 10ug in PBS-Tween (100 µl/well) and incubated for 1 h at RT. Plates were then washed thrice with PBS-Tween and 1 µg/ml of biotinylated antibodies against anti-EGFR (anti-allotype G1m(f)) or antibodies against anti-FcαRI IgG1_K409R (G1m(a) allotype)) were added in PBS-Tween (100 µl/well) and incubated for 1 h at RT. Anti-G1m(a) was used for coat and anti-G1m(f) for detection to detect bispecific antibodies. Streptavidin HRP (Strep-HRP, Sigma) diluted 1:2000 in PBS-Tween was added (100 µl/well) after three washes with PBS-Tween, and incubated for 1 hour at RT. Plates were washed thrice with PBS-Tween. After that, TMB substrate solution (containing 0.1 mg/ml 3,3′,5,5′ tetramethylbenzidine, 0.1 M NaAc and 0.003% (v/v) H₂O₂ in MilliQ set to pH5.5) was added. After sufficient color change, 2M H₂SO₄ was added to stop the reaction.

SPRi

Surface Plasmon Resonance imaging (SPRi) measurements were carried out on an IBIS MX96 device (IBIS technologies) as described previously.²⁴ For the EGFR affinity, the antibodies were spotted using a Continuous Flow Microspotter onto a single SensEye G Easy2Spot sensor in three-fold dilutions, ranging from 30 nM to 1 nM, using 10 mM acetate buffer supplemented with 0.075% Tween-80 (VWR) pH 4.5 as activation buffer. Deactivation of the sensor was done by flowing 100 mM ethanol amine, pH8.8 for 7 minutes. Binding affinity of EGFR toward the different antibodies was measured by injecting soluble EGFR over the sensor at 8 times dilution series starting at 0.78 nM until 100 nM in PBS + 0.075% Tween-80 pH7,4 (PBST). Regeneration was carried out after every EGFR injection with 100 mM H3PO4, pH 1.7. For the FcγRs and FcαRI affinities, biotinylated FcγRs and biotinylated FcαRI were spotted onto a single SensEye G-streptavadin sensor in the following concentrations: 100, 33, 11 and 3.7 nM for FcγRIIIa 158 V and 30, 10, 3.3, and 1.1 nM for the rest in PBST. The antibodies were injected over the sensor at 8 times dilutions series starting at 7.8 nM until 1,000 nM in PBST and regenerated after every injection with 10 mM Glycine-HCl, pH2.0. Calculation of the dissociation constant (KD) was performed by equilibrium fitting to Rmax= 300 for the EGFR affinity and Rmax= 500 for the FcγRs and FcαRI affinities. Analysis and calculation of all binding data was carried out with Scrubber software version 2 and Excel.

Cell cultures

The murine colon adenocarcinoma cell line MC38 transfected with a modified murine EGFR containing binding site for anti-human EGFR mAb CetuximAb, hereafter referred as MC38-chimeric-EGFR (MC38-cEGFR) (kindly provided by D.Y. Gout, Amsterdam UMC, The Netherlands) and human epidermoid colorectal cell line A431 (ATCC, CRL-2592) were cultured in Dulbecco’s Modified Eagle Medium (DMEM, Gibco, cat. no. 41966029) containing 10% (v/v) heat-inactivated fetal calf serum (Biowest, cat. no. S1810-500), 2 mM L-glutamine (Gibco, cat. no. A2916801), 100 U/ml⁻¹ Penicillin-Streptomycin (Gibco, cat. no. 15140122) and 5 ug/ml puromycin (for MC38cEGFR), hereafter referred to as complete DMEM. Cells were cultured at 37 °C and 5% CO2.

Effector cell isolation

Polymorphonuclear leukocytes (PMN) were isolated from buffy coats of healthy donors (Sanquin, Amsterdam, The Netherlands). Donors gave informed consent. Buffy coats were diluted 1:1 with 1% citrate supplemented PBS and carefully loaded on Lymphoprep (StemCell, 07851). Blood was centrifuged for 30 minutes at 800 g with acceleration setting on 4 and no brake. The PBMC fraction was washed with 10% plasma supplemented PBS (v/v). Monocytes and NK cells were subsequently isolated from the PBMCs using MACS isolation kit for CD14 positive selection (Miltenyi Biotec, 130-050-201) and NK cell negative selection (Miltenyi Biotec, 130-092-657), respectively, according to the manufacturer’s protocols. Isolated cells were collected in 100 U/ml Penicillin-streptomycin (Gibco-BrI Life Technologies), 2 mM L-glutamine (Gibco, A2916801) and 10% (v/v) heat-inactivated fetal bovine serum (FBS) (Biowest, S1810-500) supplemented RPMI 1640 (Gibco-BrI life Technologies), hereafter referred to as complete RPMI. PMNs were isolated by lysing erythrocytes for 10 min in lysis buffer (155 mM NH₄Cl, 10 mM KHCO₃, 0.1 mM EDTA, pH 7.4 in milli-Q (Millipore)). PMN were washed thrice with PBS with subsequent centrifugation for 7 min at 200 g before collection in complete RPMI.

Antibody-dependent cellular cytotoxicity (ADCC)

Tumor cells were plated in a concentration of 8000 cells/well in a flat bottom 96 well plate (MC38cEGFR were plated on fibronectin-coated wells (1 µg/cm2) (Sigma-Aldrich, S5171)). After 24 h, plates were washed with PBS and PMN (E: T 50:1) or NK cells (E: T 5:1) in the absence or presence of (glycomodified) BsAbs (0.1 and 1.0 μg/ml) were added and incubated for either 4 or 24 h at 37 °C and 5% CO2. Cell survival was determined using the Cell Titer-Blue Cell Viability Assay (Promega, cat. no. G8080). Plates were washed twice with PBS, blotted dry and incubated with Cell Titer Blue reagent (CTB stock concentration was diluted 1:6 in complete RPMI) for 1 h at 37 °C and 5% CO2. After incubation plates were shaken at 400 RPM for 5 min and fluorescence of formed resorufin, proportional to the amount of living cells in the wells, was recorded by a FLUOstar Galaxy Microplate Reader (MXT Lab systems). Resorufin was excited using a 560/10 nm excitation filter and emission was acquired using a 590/12 nm filter. Gain of the detector was set to 115 (gain 0-255). The percentage of cellular cytotoxicity was calculated in Excel using the equation: % cytotoxicity = (1 – (experimental value—medium control)/(no antibody—medium control)) × 100%. Data were plotted and analyzed using Graphpad Prism 9.0.

ADCP assay

Isolated monocytes were cultured for 6 d in the presence of 50 ng/ml M-CSF to generate macrophages. Macrophages were counted, labeled with DiO (Molecular Probes, no, D275) and seeded in 24 well plates (1.5 × 10⁵ cells/well). After 48 hours, culture medium was refreshed with complete RPMI 1640 in the absence or presence of BsAbs (0.1 and 1.0 μg/ml). Cell proliferation dye eFluor 670 (eBioscience, no, 65-0840-85) labeled A431 or MC38cEGFR were added as target cells (E: T ratio 5:1). Tumor cells were co-cultured with macrophages for 4 h at 37 °C and 5% CO2. Cells were washed with PBS and harvested by trypsin for 15 min at 37 °C and 5% CO2) followed by cell scraping and washing with 0.2% PBS-BSA and fixation with 1% formaldehyde in PBS. Samples were analyzed with flow cytometry (BD LSRFortessa) and analyzed with FlowJo software (FlowJo, LLC). Percentage phagocytic events was quantified using the equation: % phagocytosis = (1 – (% tumor cells in experiment)/(% tumor cells in no antibody)) × 100%.

CDC assay

CDC assay was previously described.⁴⁰ 1M 2,4,6-trinitrobenzenesulfonic acid (TNBS) (Sigma-Aldrich) was diluted to 4 mM in 0.15M Na₂HPO₄ (Merck, Milipore). TNP solution (1 mM) was added 1:1 with the cell lines (A431 and MC38cEGFR) to a concentration of 3 × 10⁶ cells and incubated for 15 minutes at 37 °C. To remove unbound TNP, cells were washed 3 times with PBS, and cells were diluted to a concentration of 1 × 10⁶ cells/ml. A431 and MC38cEGFR cell lines (without TNP) were also diluted to a concentration of 1 × 10⁶ cells/ml and 55.000 cells/well were plated in a U-bottom plate. (Glycoengineered) FcαRI BsAbs antibodies were added to a concentration of 0.1 μg/ml and 1.0 μg/ml. The plate was centrifuged for 1 min at 300 g and incubated for 15 min at RT. After incubation, 20 μL NHS (normal human serum), containing complement factors was added and incubated for 45 min at 37 °C on a shaking plate. Cells were washed with 100 μl PBS/0.1% FCS and centrifuged for 3 min at 300 g. After centrifugation, cells were incubated with 150 μl Propidium Iodide-stain (1:1000) and incubated for 20 min in the dark. After staining, cells were washed 3 times with 100 μl PBS/0.1%BSA and transferred to a V-bottom plate. Samples were measured with flow cytometry (BD LSRFortessa) and analyzed with the FlowJo Software (FLowJo, LLC). Percentage of dead cells was quantified and compared to untreated samples.

Fc receptor expression

Macrophages, NK cells, and neutrophils were first stained with the Fixable Viability Dye (FvD) in PBS or no dye for 10 to 15 min one ice. After FvD, samples were washed with FACS buffer (PBS/1% FCS) and centrifuged at 1500 rpm for 5 min. Then cells were incubated with the following antibodies, CD16 (eBioscience no. 47-0168-42), CD32 (eBioscience no. 25-0329-41), CD64 (Biolegend no. 305010), CD89 (BD Pharmingen (no. 555686) and hIgG1 isotype control (eBioscience no. 53-4714-42) for 30 min on ice covered in aluminum foil. After incubation, cells were washed with FACS buffer at 1,500 rpm for 5 min. Cells were fixed with 1% PFA for 15 min at 4 °C. After fixation, cells were washed with FACS buffer and transferred to FACS tubes until analysis by flow cytometry.

Binding assay

A431 and MC38cEGFR were plated separately in a concentration of 1.5 × 10⁴ cells/well. Cells were incubated with FcαRI-BsAb at various concentrations (0.001, 0.01, 0.1, 1.0, and 10 μg/ml) for 1 h at 4 °C. After incubation cells were washed with 0.5% PBS-BSA and centrifuged for 5 min at 1,500 rpm. After centrifugation directly labeled goat anti-human IgG1 secondary (Jackson ImmunoResearch, no. 109-605-003) was added for 30 min on ice covered in aluminum foil. After incubation, cells were washed with 0.5% PBS-BSA for 5 min at 1,500 rpm. Cells were fixed with 4% PFA for 10 min at RT. After fixation, cells were washed with 0.5% PBS-BSA and transferred to FACS tubes until analysis by flow cytometry. Data were plotted and analyzed using Graphpad Prism 9.0.

SDS-page

Antibody stability was analyzed by SDS-page on a non- and reducing 10% acrylamide gel (Bio-Rad, no. 1610172). Antibodies (1ug) were diluted in milli-Q and Laemmli sample buffer (Bio-Rad, no. 1610737) with or without β-mercaptoethanol (no. 63689-100ml-F). Next, samples were denatured for 10 min at 95 °C, vortexed, and shortly centrifuged. Samples were loaded and ran for 1,5 h at 100 V, after which gels were stained with Instant Blue Coomassie Protein Stain (Abcam, no. ab119211) and imaged using the Azur Biosystems c200.

Statistics

Statistical comparison between experimental groups was performed using Two-Way Anova tests corrected for multiple comparisons by Tukey multiple comparisons test. A P-value of <0.05 was considered significant. Unless otherwise stated, representative experiments out of three independent experiments are shown.

Results

Generation and characterization of glycoengineered FcαRI BsAbs

To improve immune effector functions, glycoengineered IgG1 FcαRI BsAbs were produced (Fig. 1A). The IgG-Fc N-glycosylation of glycoengineered IgG1 anti-FcαRI and CetuximAb hIgG1 variants, including WT, low fucose, high galactose and the combinations of the latter were determined by LC-MS and are shown in Fig. 1B. Stability and correct assembly of glycoengineered FcαRI BsAbs were tested using SDS-PAGE showing that all antibodies were intact as indicated in the non-reduced gel. The reduced gel demonstrated heavy- and light chain showing that glycoengineered FcαRI BsAbs were corresponding to the parentals (Fig. 1C). Unmodified and modified parental IgG1 antibodies against antigens EGFR (CetuximAb) and parental IgG1 anti-FcαRI were mixed and subjected to controlled reducing conditions in vitro that separated the two parental antibodies into half-molecules, which subsequently allowed reassembly and reoxidation to form glycoengineered FcαRI BsAbs.²¹ FAE was verified using a sandwich ELISA utilizing the allotypic epitope differences between the non- and glycoengineered EGFR/FcαRI parentals (Fig. 2A–D and E–H) and glycoengineered FcαRI BsAbs (Fig. 2I–L). As expected, only the FAE antibodies gave strong signal in the allotype-mismatched sandwich ELISA.

Affinity of parental and glycoengineered derivatives was determined by Surface Plasmon Resonance imaging (SPRi), using an array of c-terminally site-specifically biotinylated FcγR and FcαRI at various concentrations. Binding of antibody variants was monitoring in real time by flowing the antibody variants over the chip (Fig. S1). Glycoengineering showed no marked changes to any of the Fc-receptors except for FcγRIII-family members, as expected (Fig. 3). Removal of low fucose resulted in enhanced affinity to FcγRIIIa 158F and FcγRIIIa 158 V (Fig. 3E, F), but also FcγRIIIb (Fig. S1), although binding to FcγRIIIb was generally too low for actual affinity determination. Limit of detection/determination for affinity calculation for NA1/2 was marked as angle brackets. Combination of galactose with afucosylation did not have a marked effect on top of afucosylation alone (Fig. S1/Fig. S3). Taken together, glycan changes in the IgG-Fc only affected binding to FcγRIIIa and FcγRIIIb with a major effect of afucosylation increasing affinity to FcγRIIIa/b.

No increased neutrophil-mediated tumor cell killing in the presence of glycoengineered FcαRI BsAb

Next, killing ability of neutrophils in the presence of glycoengineered FcαRI BsAb was investigated. Neutrophils expressed FcαRI (CD89), FcγRII (CD32) and FcγRIII (CD16) (Fig. S2A), of which the latter one is known to be exclusively FcγRIIIb.^41,42 Neutrophils of the donor showed expression of FcγRI, which is found on activated neutrophils.¹³ Moderate killing of A431 cells by neutrophils was observed in the presence of 0.1 µg/ml BsAb, which was increased in the presence of 1.0 μg/ml (Fig. 4A). However, no significant differences in neutrophil-mediated tumor cell killing were observed in the presence of different glycovariants of FcαRI BsAbs. Similar results were found when the MC38cEGFR cell line was used as target. Dose-dependent elimination of tumor cells by neutrophils was observed with all FcαRI BsAb, but no differences were observed between glycoengineered and WT FcαRI BsAbs (Fig. 4B). These results are in line with our previous results indicating that FcαRI bispecifics mainly mediate their effector functions on neutrophils through the FcαRI.^15,18,21

Glycoengineered FcαRI BsAbs do not enhance ADCP by macrophages

To investigate if glycoengineered FcαRI BsAbs enhanced phagocytosis by macrophages, ADCP assays were performed with A431 and MC38cEGFR cells. Fc receptors of macrophages was as expected with FcαRI, FcγRI, FcγRII, and FcγRIII, the latter of which is known to be FcγRIIIa⁴¹ (Fig. S2B). Increased ADCP of A431 was observed compared to ADCP of MC38cEGFR (Fig. 5A, B). No differences were observed between the differently glycoengineered FcαRI BsAbs, suggesting that FcαRI contributes to ADCP.

Glycoengineered low fucose FcαRI BsAb boost NK cell-mediated cytotoxicity

Next, we tested the capacity of NK cells which only express FcγRIII as expected⁹ (Fig. S2C), to mediate ADCC of A341 and MC38cEGFR. At neither timepoint tested did any of the glycoengineered BsAbs show differential NK cell-mediated tumor cell killing at either concentration tested (Fig. 6A, B).

However, ADCC of MC38cEGFR by NK cells was much improved using afucosylated FcαRI BsAbs (Fig. 6C, D). In the presence of afucosylated FcαRI BsAbs increased ADCC of MC38cEGFR by NK cells was observed at both timepoints and concentration BsAb tested (Fig. 6C, D). This could be due to the EGFR expression on MC38cEGFR which was found to have lower EGFR expression compared to A431 (Fig. S4). In conclusion, NK cell-mediated afucosylation of bispecific FcαRI-EGFR antibodies with intact FcγR-binding region can be enhanced by afucosylation, but the effect magnitude may depend on the target.

No complement activation via glycoengineered FcαRI BsAbs

Finally, we investigated if CDC can be increased of these antibodies by enhanced galactosylation. While these cells were sensitive to CDC using control anti-TNP antibodies on TNP-lated cells (Fig. S3), neither CDC was induced by (glycoengineered) FcαRI bispecifics, nor by control WT CetuximAb (Fig. 7).

Discussion

Here, we generated various IgG-Fc glycoengineered FcαRI BsAbs including low fucose, high galactose, and combination of both. These variants were investigated for their functional capacity to engage and activate immune cells and complement.

Previously, binding of WT FcαRI-BsAb to FcαRI and EGFR was tested,²¹ showing successful binding to FcαRI. For binding of FcαRI-BsAbs to EGFR multiple cell lines expressing EGFR. Here, it was shown that FcαRI-BsAbs were strongly bound to A431 compared to other cell lines.²¹ Glycoengineering of these FcαRI-BsAbs shown no abnormalities by SDS-page, nor did it affect their affinities to the respective antigens, as expected. Removal of fucose increased affinity of BsAb for FcγRIIIa and FcγRIIIb, but not for the other FcγRIIa/b nor affected FcαRI binding. Both increased phagocytosis of CLL cells and decreased ADCC of BJAB cells by neutrophils have been reported in the presence of low fucose antibodies.^12,36 Unstimulated human neutrophils express FcγRIIa and high levels of FcγRIIIb. Previously, it has been shown that IgG1-mediated ADCC by neutrophils occurs through FcγRIIa.^43–46 In contrast, FcγRIIIb is generally suggested to function as a decoy receptor by competing for IgG binding with FcγRIIa. In accordance with this, blocking FcγRIIIb enhances neutrophil-mediated tumor cell killing, supporting it plays as a decoy receptor, shifting the focus of activity from FcγRI/FcγRIIa.¹³

Here, we found that afucosylation of a IgG1-based BsAb targeting both FcαRI and a tumor target did not affect neutrophil-mediated tumor cell killing. As we and others have demonstrated that FcαRI is a much more potent activator of neutrophils than the FcγR-mediated activities,^{15,16,21,41,47,48} it likely that FcγRIIIb-decoy activity is less affected as this BsAb-induced killing of tumor cells is mostly through the FcαRI arm of the BsAb. In essence, this suggest that BsAb strategy targeting FcαRI in this way, may overcome the potential negative effect of introducing afucosylated IgG in tumor therapies.^12,13

ADCP by macrophages was also mainly effectively induced by FcαRI BsAb. However, no differences in phagocytosis were observed in the presence of different (glyco)engineered BsAb, likely due to the low FcγRIIIa expression of the effector cells used. Other groups have shown that sensitivity to afucosylation of mono/macrophage cell lines such as THP-1 is restored after inducing FcγRIIIa expression, even in the presence of FcγRI and FcγRIIa.³⁵ Removing fucose increased affinity to FcγRIIIa, but not to FcγRIIa and FcγRIIb. It was shown that in the presence of afucosylated therapeutic antibodies, binding to FcγRIIIa did not lead to significantly enhanced phagocytosis by human macrophages in vitro.^11,49,50 In mice, Kupffer cells express FcγRI and FcγRIV, which are involved in inducing phagocytosis.⁵¹ This suggests that divergent receptor expression patterns on different macrophage subtypes determine whether afucosylated antibodies increases ADCP.⁵² Presumably, ADCP by human monocyte-derived macrophages is mostly mediated through FcγRI and FcγRIIa, and not via FcγRIIIa, which has lower expression, which explain why no differences in ADCP was observed in the presence of (glyco)engineered BsAb.^53–55

Perhaps surprisingly, we observed enhanced NK-cell mediated killing of MC38cEGFR. However, this was not the case using A431 target cells, which is presumably due to high EGFR expression.^56–58 Therefore, we tested EGFR expression on both cell lines, and found MC38cEGFR to have lower expression of EGFR compared to A431. This probably led to less saturation of antibodies on MC38cEGFR cells resulting in added effects that is observed between the glycoengineered FcαRI BsAbs. It is generally known that FcγRIIIa is the only activating IgG receptor on NK cells, which leads to subsequent ADCC. Furthermore, afucosylated antibodies have a higher affinity to FcγRIIIa.^24,25,32,59 Interestingly, Obinutuzumab (afucosylated anti-CD20) apparently boosts NK cell-mediated tumor cell killing and faster interaction with opsonized targets (CD20-expressing cell line) as well as elevating their capacity to induce serial killing.^60,61

Unlike our previous work, where we saw an increased NK cell-mediated ADCC by with galactosylated afucosylated antibodies compared to afucosylated antibodies alone,²⁴ here no such affect was seen. We also observed a slight increase of Fc-sialylation (7%) in the antibodies engineered with increased galactosylation. This is in alignment that sialylation has only marginally negative effect on FcγRIIIa activity when engineered at even higher levels.²⁴ Although fucosylated IgG apparently can also kill our MC38cEGFR cells, it may be an underlying FcαRI-component that is also active in the process according to a recent report.⁶² This makes it even more exciting to see that this BsAb also functions through FcγR, and an enticing effector molecule capable of affecting a vast number of different effector cells with different Fc-receptor profiles.

Furthermore, we tested if glycoengineering our BsAbs could lead to improved CDC. It was reported that complement-mediated immune responses could be induced by Cetuximab itself. This led to tumor growth inhibition in a xenograft mouse tumor model.⁶³ Previously work has shown that both expression level affects CDC, and oligo/polyclonality, even only with 2 monoclonals can induce CDC even more. Both factors affect the distance between antibodies as well as their stoichiometry, and thereby to bridge the distance and orientation to form IgG hexamers.^64,65 In line with this, other studies also reported that targeting A431 with individual EGFR antibodies do not trigger CDC, while combination therapies, such as CetuximAb and MatuzumAb, both IgG1 formats, led to complement activation and therefore increased A431 lysis.^57,66 As our glycoengineered FcαRI BsAbs did not induce CDC on A431 and MC38cEGFR cells, unlike our anti-hapten antibodies that mimic a polyclonal response, it seem that despite the high EGFR expression our antibodies are not exposed in a proper way to support efficient complement activation.⁶⁷

In conclusion, glycoengineered of FcαRI BsAbs seems not to have a major impact on macrophage mediated phagocytosis, but unlike classical afucosylated IgG anti-tumor antibodies, do not downregulate their killing capacity likely mediated mostly through FcαRI. In addition, afucosylated FcαRI BsAb recruit better NK cell-mediated effector functions. All in all, this show that afucosylated FcαRI-directed bispecifics with a classical IgG-format have an added effector function repertoir of NK cells, which might also translate into stronger therapeutic potential for tumor eradication.

Acknowledgments

We would like acknowledge the Microscopy and Cytometry Core Facility (MCCF) at the Amsterdam UMC (location VUmc) for providing assistance with flow cytometry experiments.

Author contributions

C.A.N.S. and L.t.K. performed ADCC and ADCP experiments. Complement experiments were performed by C.A.N.S in collaboration with T.D. D.Y.G. generated the MC38cEGFR cell line and helped with the binding assays. A.R.T., S.L.T., and A.E.H.B. produced glycoengineered FcαRI BsAbs and performed experiments verifying the format of the FcαRI BsAb glycovariants. J.V.C. provided the characterization of the glycovariants. C.W.T. tested the production of the glycovariants. R.V. helped with the production of glycoengineered FcαRI-BsAbs and SDS-page experiments. L.M.B. helped with the SDS-page experiments. C.A.N.S. wrote the manuscript. M.v.E. and G.V. supervised the study and co-wrote the manuscript.

Supplementary material

Supplementary material is available at The Journal of Immunology online.

Funding

This work was funded by the KWF Dutch Cancer Society (Grant no: 12749). Name of Grant Holder: Prof. Dr Marjolein van Egmond.

Conflicts of interest

The authors declare no potential conflicts of interest.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

References

Author notes