A panel of glycoengineered IgG1 (mAb SO57) (48, 49) with different N-glycoforms using stably glycoengineered CHO cells was previously generated (45). In brief, we first established a CHO wildtype clone stably expressing IgG1 (IgG1-CHO^WT) and used IgG1-CHO^WT to generate a library of N-glycoengineered clones by knockout (KO) using CRISPR/Cas9 and stable knock-in (KI) of key glycosyltransferases in the N-glycosylation pathway using Zinc finger nucleases and the ObLiGaRe Strategy (50) ( Figure 1 , Supplementary Table S1 ). To obtain afucosylated IgG1, a Fut8 and B4galt1 KO was created, resulting in the IgG1-G0 clone. An Mgat1 KO was created (IgG1-Oligomannose) to obtain oligomannosylated IgG1 and KO of Mgat2 and Man2a1/2 resulted in hybrid (IgG1-Hybrid) and monoantennary (IgG1-Monoantennae) IgG1-expressing clone, respectively. To increase galactosylation compared to IgG1-CHO^WT, the IgG1-G2F clone was created by KI of B4GALT1. To further increase linkage-specific sialylated IgG1, KI of ST3GAL4 or ST6GAL1 was introduced on top of the B4GALT1 KI clone, resulting in α2-3 linked (IgG1-G2FS(2-3)) or α2-6 linked Sia (IgG1-G2FS(2-6)), respectively, as previously described (45) ( Figure 1 , Supplementary Table S1 ).

The isolated glycoengineered IgG1 glycoforms were validated by SDS-PAGE for correct protein expression and purity, and to confirm concentration determination. All IgG1 glycoforms resulted in a heavy chain (HC) and low chain (LC) band of 50 and 25 kDa, respectively ( Figure 2A ). Further validation of glycosylation patterns was done by MALDI-TOF-MS employing linkage-specific sialic acid derivatization to differentiate between α2-3 and α2-6 linked Sia ( Supplementary Figure S1 ) ( 52, 53). IgG1-CHO^WT showed a heterogeneous profile including dominant G0F glycans with small amounts of G1F ( Supplementary Figure S1A ) (45). Homogenous afucosylated, agalactosylated IgG1 was produced by IgG1-G0 ( Supplementary Figure S1B ) and homogenous Man5 IgG1 by IgG1-Oligomannose ( Supplementary Figure S1C ). The dominant glycoform for IgG1-Monoantennae was the galactosylated monoantennary IgG1 with small amounts being agalactosylated and minor amounts being galactosylated and α2-3 sialylated ( Supplementary Figure S1D ). The same galactosylation and sialylation trend was seen for IgG1-Hybrid ( Supplementary Figure S1E ). IgG1-G2F produced homogenous G2F, which is a significant galactosylation increase in comparison to IgG1-CHO^WT ( Supplementary Figure S1F ). For both sialyltransferase KI clones (IgG1-G2FS(2-3) and IgG1-G2FS(2-6)), the dominant glycoforms remained G2F, with a minority of species being sialylated ( Supplementary Figures S1G, H ). Sialylation in IgG1-G2FS(2-6) was lower compared to our previous study (45), and this may be due to differences in cell density and viability during cell culturing. However, sialylation was increased compared to IgG1-CHO^WT and IgG1-G2F ( Supplementary Figures S1A, F–H ).

FcγRIIIa has five consensus N-glycosites occupied by heterogeneous N-glycans. These display site-specific N-glycan structures and variation hereof in different cell types and individuals (54). We generated a library of glycoengineered HEK293 cells with different N-glycosylation capacities by combinatorial KO and KI of glycosyltransferase genes as previously reported (47). For afucosylated FcγRIIIa, FUT8 was knocked out. To obtain oligomannose, monoantennae, and hybrid FcγRIIIa-158F/V we used KO of MGAT1, MGAT2/3, or MAN2A1/2, respectively. KO of B4GALN3/4 and MGAT3/4A/4B/5 in combination with ST3GAL3/4/6 or ST6GAL1, resulted in clones expressing complex-type biantennary α2-3 or α2-6 linked sialylated N-glycans without LacNAc repeats, respectively. Recombinant FcγRIIIa-158F and 158V were transiently expressed in CHO^WT and HEK293^WT, resulting in FcγRIIIa-CHO^WT and FcγRIIIa- HEK^WT, respectively ( Figure 1 , Supplementary Figure S2 , Supplementary Table S1 ). Transfection of the above-mentioned glycoengineered HEK cells resulted in FcγRIIIa-Afucosylated, FcγRIIIa-Hybrid, FcγRIIIa-Oligomannose, FcγRIIIa-Monoantennae, FcγRIIIa-G2FS(2-3), and FcγRIIIa-G2FS(2-6) ( Figure 1 , Supplementary Figure S2 , Supplementary Table S1 ).

The produced FcγRIIIa-158F/V were validated by SDS-PAGE and mass spectrometry to confirm N-glycan structures and evaluate glycan heterogeneity ( Figures 2 , 3 , Supplementary Figures S3, 4 ). Glycoengineered FcγRIIIa resulted in a heterogeneous glycoprofile migrating as a broad band with an apparent molecular weight ranging from 40 to 50 kDa depending on clone, highlighting the N-glycan heterogeneity and mass contribution of the five N-glycans per FcγRIIIa molecule.

For glycoprofiling, PNGaseF-released esterified N-glycans of FcγRIIIa-158F/V were analyzed by MALDI-TOF-MS ( Supplementary Figures S3, 4 ) ( 52). Since the Asn162 glycosite of FcγRIIIa has shown to influence IgG1 binding (29), site-specific analysis of Asn162 by LC-MS/MS was performed. For this, FcγRIIIa was digested by chymotrypsin and Glu-C before subjection to LC-MS/MS and the top 90% most abundant glycan structures for the Asn162 glycosite are depicted ( Figure 3 ). In general, the Asn162 N-glycan structures for both FcγRIIIa-158F/V confirm gene editing signatures and generally followed MALDI-TOF-MS data. However, underprocessed glycans were seen for Asn162 in comparison to the total N-glycan profiling, where more complex N-glycans were found for FcγRIIIa-HEK^WT, FcγRIIIa-CHO^WT, and FcγRIIIa-afucosylated (33, 35). The Asn162 of FcγRIIIa-CHO^WT was predominantly of the G2S type with overall high levels of galactosylation (G2S>G2S2>G2) and variable sialylation levels (G2S>G2S2) ( Figure 3 , Supplementary Figures S3A, 4A ). The FcγRIIIa-HEK^WT Asn162 site had mainly biantennary N-glycans with GalNAc extensions and no sialylation. These were also the major glycan species found for the total FcγRIIIa-HEK^WT glycoprofiling data ( Figure 3B ). N-glycan profiling of FcγRIIIa-HEK^WT also showed minor sialylated, elongated, and branched glycospecies ( Supplementary Figures S3A, 4A ). Most complex N-glycans were seen for FcγRIIIa-afucosylated, supporting the higher molecular weight as seen on SDS-PAGE ( Figures 2B , 3C , Supplementary Figures S3B, 4B ). Confirmed by total protein glycoprofiling and Asn162 site-specific glycoanalysis, homogeneous Man5 was found for FcγRIIIa-Oligomannose ( Figure 3D , Supplementary Figures S3C, 4C ). The major Asn162 glycan structure for FcγRIIIa-Monoantennae is a sialylated monoantennary species ( Figure 3E , Supplementary Figures S3D, 4D ). However, a wider variety of other glycan structures were seen for both Asn162 and total FcγRIIIa glycoprofiling such as hybrid species and biantennary N-glycans and total N-glycoprofiling of FcγRIIIa-hybrid clone showed more complex N-glycans than expected. For both FcγRIIIa-G2FS(2-3) and FcγRIIIa-G2FS(2-6), biantennary N-glycans with variable sialylation levels were seen ( Figures 3G, H , Supplementary Figures S3F, G, 4F, G ). Furthermore, branching fucose was found in FcγRIIIa-G2FS(2-3) and -G2FS(2-6).

To study the role of N-glycoforms on the IgG1:FcγRIIIa interaction, we used the produced glycovariants for SPR analysis on the IBIS MX96 biosensor system, as described in Dekkers et al. (55). For this all FcγRIIIa-158F/V glycoforms were enzymatically biotinylated at the AviTag™ and immobilized on a sensor chip. Subsequently, all IgG1 variants were injected as analytes in a multi-cycle mode ( Figure 1B ). Humira, a recombinant human IgG1 monoclonal antibody with a G0F or G0NF glycan structure (56), and recombinant FcγRIIIa-158F/V purchased from SinoBiologicals, were used as controls ( Supplementary Table S2 ). Binding was observed for all IgG1:FcγRIIIa combinations, however, the FcγRIIIa-158F binding to IgG1-Monoantennae and IgG1-Hybrid was weak and therefore K_D values could not be calculated from these interactions ( Supplementary Figures S5, 2 ).

We could confirm that FcγRIIIa-158V exhibited on average a 3-6 fold increased affinity for IgG1 compared to FcγRIIIa-158F regardless of the IgG1 glycoform tested ( Figure 4 , Supplementary Table S2 ). The highest affinity for all IgG1 glycoforms was seen to FcγRIIIa-Oligomannose with a 2-3 fold increase compared to all other FcγRIIIa glycoforms ( Figure 4 , Supplementary Table S2 ). Surprisingly, FcγRIIIa-Monoantennae and FcγRIIIa-Hybrid, carrying immature N-glycan structures often found on Asn162 expressed by native immune cells (57), had a similar affinity to all IgG1 glycoforms when compared to FcγRIIIa-HEK^WT and FcγRIIIa-CHO^WT, with the exception of IgG1-hybrid ( Figure 4 , Supplementary Table S2 ). Sialylation of FcγRIIIa (FcγRIIIa-G2FS(2-3) and FcγRIIIa-G2FS(2-6)) influenced IgG1 binding to a certain degree, with FcγRIIIa-2FS(2-3) having approximately a 2-fold decrease in binding compared to FcγRIIIa-G2FS(2-6) and FcγRIIIa-HEK^WT. However, it needs to be taken into account that the sialylation levels for FcγRIIIa-G2FS(2-3) are much higher compared to G2FS(2-6), where G2F is the dominant glycoform on Asn162 ( Supplementary Figures S3F, G ). Surprisingly, FcγRIIIa-G2FS(2-3) has a lower affinity to most glycoengineered IgG1 compared to FcγRIIIa-CHO^WT, even though the Asn162 glycan pattern is similar.

As expected, the highest affinity of all FcγRIIIa receptors was observed to afucosylated IgG, both IgG1-G0 and the IgG1-Oligomannose ( Figure 4 , Supplementary Table S2 ) ( 29). Interestingly, the N-glycosylation state of the FcγRIIIa had minimal effect on the binding affinity when probed with afucosylated IgG, except for oligomannosylated FcγRIIIa where binding affinity is increased by a factor of two. All FcγRIIIa glycoforms bound equally to IgG1 capped by α2-3 or α2-6 Sia (IgG1-G2FS(2-3) and IgG1-G2FS(2-6)), with minor differences that appear to be negligible ( Figure 4 , Supplementary Table S2 ). The highest K_D, and thus lowest affinity, was seen for IgG1-Hybrid and IgG1-Monoantennae to all FcγRIIIa. In contrast to IgG1-F0, for these suboptimal glycoengineered IgG1, the FcγRIIIa glycosylation influences binding ( Figure 4 , Supplementary Table S2 ). This combinatorial setup with glycoengineered IgG1 and FcγRIIIa-158F/V confirms that afucosylated IgG1 has the highest binding affinity to oligomannosylated FcγRIIIa-158V (FcγRIIIa-Oligomannose).

HEK293 were grown in suspension in serum-free F17 culture media (Invitrogen) supplemented with 0.1% Kolliphor P188 (SIGMA) and 4 mM GlutaMax as previously described (47). CHOZN GS-/- cells (Sigma) were maintained as suspension cultures in serum-free media (EX-CELL CHO CD Fusion). Cultures were grown at 37°C and 5% CO₂. HEK293 and CHO cultures were under constant agitation (120 rpm).

KO was performed using a validated gRNA library for all human glycosyltransferases (GlycoCRISPR) (72), as previously described (73). In brief, HEK293 cells were seeded in 6-well plates (NUNC), transfected with 1 μg of gRNA and 1 μg of GFP-tagged Cas9-PBKS using Lipofectamine 3000 (Invitrogen) according to manufacturer’s instructions. Cells were harvested after 24 hours, and bulk sorted for GFP expression by FACS (SONY SH800). After culturing, the sorted cell pool was further single-sorted into 96 well plates and screened by Indel Detection by Amplicon Analysis (IDAA) (74), and indels of selected clones were confirmed by Sanger sequencing.

One day prior to transfection, cells were seeded in T25 flasks (NUNC). Electroporation was conducted with 2 × 10⁶ cells with 1 μg of both endotoxin-free plasmid DNA of Cas9-GFP and gRNA in the U6GRNA plasmid (Addgene Plasmid #68370) using Amaxa Kit V and program U-24 with Amaxa Nucleofector 2B (Lonza). Electroporated cells were moved to a 6-well plate with 3 ml growth media. Forty-eight hours after nucleofection, the 10–15% highest GFP-labeled pool of cells was enriched by FACS, and after one week of culturing, cells were single-cell sorted by FACS (SONY SH800) into 96-well plates. KO clones were identified by IDAA, as mentioned above, as well as immunocytology with appropriate lectins or monoclonal antibodies, whenever possible. Selected clones were further verified by Sanger sequencing.

Site-specific CHO Safe-Harbor locus KI was based on ObLiGaRe strategy (50) and performed with 2 µg of each ZFN (Merck/Sigma-Aldrich) tagged with GFP/Crimson (74), and 5 µg donor plasmid with full coding human genes, as described before (46).

The anti-rabies human IgG1 SO57 (48) was used to establish stable expressing CHO clones as described by Schulz et al. (45). For IgG production, clones were seeded at 0.5 x 10⁶ cells/ml without L-glutamine and cultured for three days. IgG was purified by HiTrap™ protein HP (CE Healthcare) as previously described (45). Protein purity and concentration were evaluated by SDS-PAGE and Coomassie staining.

HEK293-6E cells were seeded at a cell density of 0.5 × 10⁶ cells/ml and transfected with FcγRIIIa-158F/V with 10X his-tag and AviTag™ the next day with 1:3 μg DNA : PEI per 1 × 10⁶ cells. Secreted protein was harvested after 72 hours and purified from culture medium by nickel affinity chromatography. For this, the culture medium was centrifuged, filtered (0.45 μm), mixed 3:1 (v/v) in 4X binding buffer (100 mM sodium phosphate, pH 7.4, 2 M NaCl) and applied to self-packed nickel-nitrilotriacetic acid (Ni-NTA) affinity resin column (Qiagen), which was pre-equilibrated in washing buffer (25 mM sodium phosphate, pH 7.4, 500 mM NaCl, 20 mM imidazole). After washing, bound protein was eluted with 250 mM imidazole in washing buffer. Purity and quantification were evaluated by SDS-PAGE and Coomassie staining. Purified protein was buffer-exchanged to approximately 1 mg/ml in 50mM AmBic buffer with 2 ml Zeba Spin Desalting Column 7K MWCO (ThermoFisher).

Both FcγRIIIa-158F/V with 10X his-tag and AviTag™ constructs were both subcloned into a modified pCGS3 vector (Merck/Sigma-Aldrich) for glutamine selection in CHOZN GS−/− cells (Sigma). CHO cells were seeded at 0.5 X 10⁶ cells/ml in T25 flasks (NUNC) one day prior to transfection. Two million cells were electroporated with 5 μg plasmids using Amaxa kit V and program U24 with Amaxa Nucleofector 2B (Lonza) and plated in 6-wells with 3 ml growth media. Three days after transfection, cells were plated in 96-wells at 1000 cells/well in 200 μl Minipool Plating Medium containing 80% EX-CELL^® CHO Cloning Medium and EX-CELL CHO CD Fusion serum-free media without glutamine (Sigmaaldrich). High-expressing clones were selected by testing the medium using anti-HIS antibodies, and selected clones were scaled up in serum-free media without L-glutamine TPP TubeSpin^® shaking Bioreactors (180 rpm, 37 °C, and 5% CO₂) for protein production. FcγRIIIa was purified as described above.

N-glycan protein profiling employing linkage-specific sialic acid esterification was obtained as described by Reiding et al. (52). In brief, 10 μg IgG1 and FcγRIIIa were denatured by incubating the samples 10 min at 60°C in 2% SDS. N-glycans were released by adding a release mixture containing 2% NP-40 and 0.5 mU PNGaseF at 37°C over night. Released N-glycans were esterified to obtain sialic acid linkage specificity. In brief, esterification reagent containing 0.5 M EDC and 0.5 M HOBt in ethanol were added to 1 μl glycan mixture and incubated for one hour at 37°C. After incubation, 25% NH₄OH and subsequently 100% ACN was added (52). Glycan enrichment was performed by cotton hydophilic interaction liquid chromatography (HILIC) (75). Briefly, pipet tips were packed with cotton thread which was conditioned by 85% ACN. The sample was loaded by pipetting 20 times into the reaction mixture. Tips were washed in 85% ACN 1% TFA, followed by 85% ACN and eluted in 10 μl MQ. For MALDI-TOF-MS analysis, 1μl HILIC purified glycan sample was spotted on an MTP AnchorChip™ 384 BC mixed on plate with 1μl sDHB (5mg/ml in 50% ACN) and left to air dry. The sample was recrystallized using 0.2μl ethanol. Measurement was performed on the Bruker Autoflex (Bruker Daltonik GmbH, Bremen, Germany) using the Bruker Flex Control 3.4 software. For each spectrum, 10 000 laser shots were accumulated at a laser frequency of 100 Hz. Spectra were recorded in the positive reflector mode (900-4500 Da) and the raw spectra were processed by the Flexanalysis 5.1.

The purified protein was dissolved in 50mM ammonium bicarbonate buffer, reduced in 10 mM dithiothreitol (DTT), alkylated with 20 mM iodoacetamide (IAA), reduced again with 10 mM DTT, and digested with 1:20 chymotrypsin:protein followed by 1:20 GluC:protein (Roche). The proteolytic digest was desalted by in-house produced modified StageTip columns containing 3 layers of C18 membrane (3M Empore disks, Sigma Aldrich) (76). Samples were eluted with 50% methanol in 0.1% formic acid (FA), dried down, and re-solubilized in 0.1% FA for LC-MS/MS analysis.

LC-MS/MS analysis was performed on EASY-nLC 1200 UHPLC (Thermo Scientific) interfaced via nanoSpray Flex ion source to an Orbitrap Fusion Lumos MS (Thermo Scientific). Briefly, the nLC was operated in a single analytical column set up using PicoFrit Emitters (New Objectives, 75 μm inner diameter) packed in-house with Reprosil-Pure-AQ C18 phase (Dr. Maisch, 1.9-μm particle size, 19-21 cm column length). Each sample was injected onto the column and eluted in gradients from 3 to 32% B in 75 min, and from 32% to 100% B in 10 min, and 100% B for 10 min at 200 nL/min (Solvent A, 100% H₂O; Solvent B, 80% acetonitrile; both containing 0.1% (v/v) formic acid). A precursor MS1 scan (m/z 350–2 000) of intact peptides was acquired in the Orbitrap at the nominal resolution setting of 120 000, followed by Orbitrap HCD-MS2 and at the nominal resolution setting of 60 000 of the five most abundant multiply charged precursors in the MS1 spectrum; a minimum MS1 signal threshold of 50 000 was used for triggering data-dependent fragmentation events. Targeted MS/MS analysis was performed by setting up a targeted MSⁿ (tMSⁿ) Scan Properties pane. A target list was composed of the top 30 most abundant glycans or glycopeptides from the proposed compositional list. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD035846.

Glycan and glycopeptide compositional analysis was performed from m/z features extracted from LC-MS data using in-house written SysBioWare software, as previously described (46, 77). Briefly, For m/z feature recognition from full MS scans LFQ Profiler Node of the Proteome discoverer 2.2 (Thermo Scientific) was used. The list of precursor ions (m/z, charge, peak area) was imported as ASCII data into SysBioWare and compositional assignment within 3 ppm mass tolerance was performed. The main building blocks used for the compositional analysis were: NeuAc, Hex, HexNAc, dHex and the theoretical mass increment of the most prominent peptide corresponding to each potential glycosites. The most prominent peptide sequence related to the N-glycosite of interest was determined experimentally by comparing the yield of deamidated peptides before and after PNGase F treatment. One or two phosphate groups were added as building blocks for assignment. To generate the potential glycopeptide list, all the glycoforms with an abundance higher than 10% of the most abundant glycoform were used for glycan feature analysis.

Prior to SPR measurements, glycoengineered FcγRIIIa was site-specifically biotinylated on the BirA tag using BirA enzyme as described by Rodenko et al. (78). For biotinylation of 1 µM FcγRIIIa protein, 3.3 nM BirA ligase was used. After biotinylation overnight at 25°C, the biotinylated FcγRIIIa mixture was buffer-exchanged and subsequently concentrated in PBS pH 7.4 using Amicon Ultra centrifugal filter units (MWCO 3 kDa) (Merck, Millipore).

The biotinylated, recombinant, human FcγRIIIa-158F and 158V from SinoBiological (10389-H27H-B and 10389-H27H1-B, respectively) were used as an control for SPR. SPR measurements were performed on an IBIS MX96 (IBIS technologies) device as described previously by Dekkers et al. (8). All biotinylated FcγRIIIa were spotted using a Continuous Flow Microspotter (Wasatch Microfluidics) onto a SensEye G-streptavidin sensor (Senss, 1–08–04–008) allowing for binding affinity measurements of each glycoengineered antibody, and Humira^®, to all glycoengineered FcγRIIIa simultaneously on the IBIS MX96.

The biotinylated FcγRIIIa were spotted in 4 concentrations with a 3-fold dilution ranging from 0.3 to 100nM, depending on the FcγRIIIa in PBS supplemented with 0.075% Tween-80 (VWR, M126–100ml), pH 7.4. Glycoengineered IgG1 was then injected over the IBIS at 8 times dilution series starting at 15.6 nM until 2000 nM in PBS + 0.075% Tween-80. Regeneration after every sample was carried out with 10 nM Gly-HCl, pH 2.4. Calculation of the dissociation constant (K_D) was performed by equilibrium fitting to R_max= 500. Analysis and calculation of all binding data were carried out with Scrubber software version 2 (Biologic Software) and Excel. Three independent experiments for each receptor/Fc pair were carried out on at least two different days and representative data are reported.