Fcγ RIIIa chromatography to enrich a-fucosylated glycoforms and assess the potency of glycoengineered therapeutic antibodies

Therapeutic antibodies can elicit an immune response through different mechanisms, either cell independent via complement activation (CDC) or through activation of immune-effector cells (such as macrophages and NK cells).

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/chroma

Anne Freimoser–Grundschobera , ∗, Petra Ruegerb , Felix Fingasb , 1 , Peter Sondermanna , 2 ,

Sylvia Hertera , Tilman Schlothauerb , Pablo Umanaa , Christiane Neumanna

a Roche Pharma Research & Early Development, Roche Innovation Center Zurich, Roche Glycart AG, Wagistrasse 10, CH-8952 Schlieren, Switzerland

b Roche Pharma Research & Early Development, Roche Innovation Center Munich, Nonnenwald 2, 82377 Penzberg, Germany

a r t i c l e i n f o

Article history:

Received 20 June 2019

Revised 12 September 2019

Accepted 17 September 2019

Available online 18 September 2019

Keywords:

ADCC

Affinity column

Fc γRIIIa

Fucose

Glycoengineering

Monoclonal antibody

a b s t r a c t

Therapeutic antibodies can elicit an immune response through different mechanisms, either cell in- dependent via complement activation (CDC) or through activation of immune-effector cells (such as macrophages and NK cells) After target binding, the Fc part of the antibody will interact with Fc re- ceptors on the surface of effector cells, leading to activation and lysis of the target cells by a mechanism called antibody-dependent cell-mediated cytotoxicity (ADCC) The ADCC of an antibody can be increased

by modifying the carbohydrates on the Fc part If the fucose on the first N -acetylglucosamine is absent, the affinity for the Fc γRIIIa is increased and the ADCC enhanced We describe the development of a chromatography method that is based on the differential affinity of the Fc receptor Fc γRIIIa (high affin- ity V158 variant) for fucosylated and a-fucosylated antibodies Immobilized Fc γRIIIa can be used for the separation of immunoglobulins carrying these glycosylation variants for both, analytical and preparative purposes The biological activity and fucose content of three pools enriched for fully fucosylated, mono- fucosylated or a-fucosylated carbohydrates could be characterized Mono-fucosylated and a-fucosylated immunoglobulins have the same enhanced biological activity compared to fully fucosylated IgGs A di- rect, label- and modification-free analytical method for screening of IgGs from culture supernatant was developed and was amenable to high-throughput screening Clones producing antibodies with a high content of a-fucosylated oligosaccharides could be successfully selected

© 2019 The Authors Published by Elsevier B.V This is an open access article under the CC BY-NC-ND license

( http://creativecommons.org/licenses/by-nc-nd/4.0/)

The number of antibody-based therapeutics, either approved or

in clinical trials, is increasing constantly and the majority is based

on the human IgG1 isotype [1,2] The two Fab (fragment anti-

gen binding)-parts of these IgGs are responsible for target bind-

ing, whereas the constant Fc (fragment crystallizable) domain in-

teracts with components of the immune system, leading to me-

diation of immune effector functions such as antibody-dependent

cellular cytotoxicity (ADCC) and complement dependent cytotoxi-

city (CDC) The carbohydrate structures attached to the conserved

N-glycosylation site at Asparagine 297 (Asn297, N297) within the

∗ Corresponding author

E-mail address: anne.freimoser-grundschober@roche.com (A Freimoser–

Grundschober)

1 Present Address: GVG Diagnostics, GmbH, Deutscher Platz 5b, 04103 Leipzig,

Germany

2 Present Address: Tacalyx GmbH, Müllerstr 178, 13353 Berlin, Germany

CH2-domain of the constant Fc part are mandatory for mediating these effector functions These oligosaccharides consist predomi- nantly of a bi-antennary core pentasaccharide, comprised of N

acetylglucosamine and mannose, and complex type structures with

a variable content of bisecting GlcNAc ( N-acetyl-glucosamine), ter- minal galactoses, core fucose and sialic acids ( Fig 1 ) In addition, these oligosaccharides may be differently composed on each of the two heavy chains of the same IgG molecule

Numerous studies have shown that the carbohydrates play

an important role in maintaining the structure and stability of the IgGs and that the carbohydrate composition strongly affects the antibody-mediated immune effector functions Indeed, the Fc- attached oligosaccharides influence the affinity of the antibody for individual Fc γRs and regulate binding to different Fc γR classes: activating or inhibitory [1,3–8] To improve the biological functions

of therapeutic antibodies, several approaches have been developed

to modulate their glycosylation profile [2–4,9–11] A promising approach is the reduction or abolishment of the core fucosyla- tion, because the absence of the core fucose results in a 50-fold https://doi.org/10.1016/j.chroma.2019.460554

0021-9673/© 2019 The Authors Published by Elsevier B.V This is an open access article under the CC BY-NC-ND license ( http://creativecommons.org/licenses/by-nc-nd/4.0/ )

Fig 1 Carbohydrate moiety attached to Asn-297 of human IgG1-Fc The sugars in bold define the pentasaccharide core; the addition of the other sugar residues is variable

GlcNAc, N -acetylglucosamine; Fuc, fucose; Man, mannose; Gal, galactose; NeuAc, N-acetylneuraminic acid

increased antibody affinity to Fc γRIIIa and Fc γRIIIb, leading to

enhanced ADCC activity [12,13] At least three antibodies in

clinical trials have carbohydrates with reduced fucosylation on

their Fc part: obinutuzumab [14] , MEDI-551 [15] and GSK2831781

(ClinicalTrials.gov identifier: NCT02195349)

During clone screening of a stable production cell line for a

classical monoclonal antibody, the major selection criterion is the

titer of the secreted antibody (usually determined by ProteinA

binding) In the case of antibodies with a-fucosylated carbohy-

drates, it is also critical to monitor the oligosaccharide composi-

tion in addition to the titer The carbohydrate analysis can be per-

formed with different methods that usually involve digestion of

the oligosaccharides from the IgGs and analysis via reverse-phase

UPLC with previous labeling of the carbohydrates or without la-

beling by mass spectrometry These methods are tedious and only

partially adaptable to high-throughput analysis The conventional

affinity chromatography matrices applied for IgG purification can-

not discriminate between different glycosylation patterns within

the IgG, since the immobilized capture proteins specifically bind

the protein backbone of an antibody For example, protein A and

protein G bind to the interface between the CH2 and CH3 do-

main of the Fc-part, whereas protein L interacts with the con-

stant part of the kappa light chain A recently published report us-

ing an immobilized non-glycosylated Fc γRIIIa could not differenti-

ate between glycoforms that were fucosylated or not [16] Another

approach takes advantages of carbohydrate binding lectins Lectin

chromatography, applied to enrich antibodies carrying defined car-

bohydrates [17] , as well as a lectin ELISA developed in house (data

not shown), are, however, limited by their inability to bind glyco-

proteins lacking a specific monosaccharide unit, which is the case

for a-fucosylated antibodies

Since the higher Fc γRIIIa affinity of a-fucosylated antibod-

ies results in an enhanced ADCC, screening for Fc γRIIIa bind-

ing has the potential to identify antibodies with improved bio-

logical activity Here, we describe a chromatography method that

is based on the differential affinity of the Fc receptor Fc γRIIIa

(high affinity V158 variant) for fucosylated and non-fucosylated

antibodies Immobilized Fc γRIIIa can be used for the separa-

tion of immunoglobulins carrying these glycosylation variants for

both, analytical and preparative purposes Preparative Fc γRIIIa-

chromatography enables the enrichment of non-fucosylated or fu-

cosylated immunoglobulins for the detailed characterization of

their biological activity Analytical Fc γRIIIa-chromatography can

be employed for characterizing the carbohydrate composition of

an antibody pool, as well as for the high-throughput screen-

ing of clones producing antibodies with a high content of non-

fucosylated oligosaccharides, the read-out correlating directly to

cell-mediated killing (e.g enhanced ADCC)

2 Results

2.1 Preparative separation of IgG with different a-fucosylation degree

An antibody pool comprises antibodies with different degrees

of fucosylation (i.e fully fucosylated antibody species, when both

carbohydrates of the Fc carry a fucose residue, mono- and a- fucosylated species if only one or no fucose is present, respec- tively) For the preparative separation of IgGs with different fu- cose content, a column with a volume of 2.7 ml and with the

Fc γRIIIa(V158)K6H6 receptor coupled to NHS-sepharose beads was used to isolate these different species from such an antibody pool The IgGs were loaded at pH 8.0 and eluted in three steps at pH 4.6, pH 4.2 and pH 3

First, the GEmAb1 (glycoengineered monoclonal antibody 1) was subjected to preparative Fc γRIIIa chromatography analysis The chromatogram showed three peaks corresponding to the flow through, and the elution steps with pH 4.6 and pH 4.2 No ad- ditional material eluted in the last wash at pH 3 The respective fractions of the peaks were collected for further analysis of their carbohydrate content, their binding to the Fc γRIIIa by surface plas- mon resonance (SPR) and their ability to induce ADCC The same procedure was subsequently performed for GEmAb2 ( Fig 2 )

2.1.1 Analysis of the carbohydrate composition

The carbohydrate composition of the antibody samples sep- arated by Fc γRIIIa affinity column was analyzed by matrix- assisted laser desorption/ionization time-of-flight mass spectrom- etry (MALDI TOF MS) following a PNGaseF treatment to release the carbohydrates from the IgG For the two glycoengineered IgGs that were analyzed here, the flow through peak had the lowest a- fucose content, followed by peaks two and three ( Tables 1 and 2 ) However, this approach delivered only the overall amount of non- fucosylated oligosaccharides in the antibody preparation

To determine the distribution of fucose residues on the two an- tibody heavy chains in the Fc domain, the samples were digested either with plasmin, EndoH and EndoS (GEmAb1), or with EndoH and EndoS (GEmAb2), to obtain Fc fragments (GEmAb1) or whole IgGs (GEmAb2) carrying only the first N-acetylglucosamine residue (with or without fucose) of the oligosaccharide core (both proce- dures allow equivalent quantification of fucose distribution per Fc) These Fc fragments or IgGs were subsequently subjected to elec- trospray ionization mass spectrometry (ESI-MS) analysis to deter- mine the distribution of the fucose per Fc fragment ( Tables 1 and

2 ) The ESI-MS analysis of antibodies 1 and 2 revealed 98% and 88% fucosylated antibodies, respectively, in the flow through frac- tion (peak 1) The fraction obtained by elution with pH 4.6 (peak 2) consisted of a mixture of all fucosylation types, but antibodies with mono-fucosylated oligosaccharides prevailed (64% and 68.5%, respectively) The fraction of GEmAb1 eluted with pH 4.2 (peak 3) consisted of 61% a-fucosylated and 39% mono-fucosylated IgGs The corresponding fraction of GEmAb2 demonstrated a higher pro- portion of antibodies with a-fucosylated carbohydrates (76.5%) and the mono-fucosylated and fucosylated species were represented by 18.5% and 5%, respectively In summary, the results demonstrate that the preparative Fc γRIIIa column can be successfully employed

to enrich a-fucosylated antibodies For both, GEmAb1 and GEmAb2, the first peak was strongly enriched in antibodies with fucosy- lated sugars on both Fc parts, the peak two comprised mostly Fcs with one fucosylated and one a-fucosylated carbohydrate, and the

Fig 2 Preparative Fc γRIIIa chromatography Chromatogram A280 for GEmAb2 GEmAb2 elutes in three peaks: peak one is the flow-through of the column, peak 2 and 3 elute with two pH steps (pH 4.6 and pH 4.2) Fractions pooled for peak 1, 2 and 3 are indicated Black solid line: A280, black dashed line: pH gradient, grey solid line: pH-value

Table 1

Content of carbohydrates with and without fucose for antibody pools of GEmAb1 separated in three peaks

by Fc γRIIIa chromatography A-fucosylation level was determined globally by MALDI TOF MS after PNGase F treatment (average from 7 runs) or the fucose distribution per Fc was determined by ESI-MS after Plasmin/Endo S/Endo H digest (pool of 3 runs)

Fractions

overall % a-fuc (MALDI) ± std dev

% of Fc without fucose with one fucose with two fucoses

Table 2

Content of carbohydrates with and without fucose for antibody pools of GEmAb2 separated in three peaks

by Fc γRIIIa chromatography A-fucosylation level was determined globally by MALDI-TOF MS after PNGase F treatment (average from 2 runs) or the fucose distribution per IgG was determined by ESI-MS after Endo S/Endo

H digest (pool from 2 runs)

Fractions

overall% a-fuc (MALDI) ± std dev

% of Fc without fucose with one fucose with two fucoses

population of peak three contained predominantly completely a-

fucosylated antibodies

2.1.2 Surface plasmon resonance

The three antibody pools isolated from the Fc γRIIIa column

were expected to possess different affinities for the receptor, since

they eluted more or less easily These affinities were determined by

SPR for GEmAb2, because the recombinant antigen was available

The antibody samples were captured to the antigen immobilized

on the chip surface and soluble Fc γRIIIa was injected The affini-

ties fell in two categories: the fucosylated fraction (peak 1) as well

as the WTmAb2 (not glycoengineered, carrying wild-type fucosy-

lated carbohydrates) had KDs around 70 nM, with rapid on and off

rates, whereas the peak two and peak three fractions, as well as the glycoengineered GEmAb2, had KDs of around 3 nM and a much slower off rate ( Table 3 ) The IgGs of the peak three, with the high- est proportion of antibodies with a-fucosylated carbohydrates, had the highest Fc γRIIIa binding affinity SPR thus confirmed the corre- lation between elution time, affinity to Fc γRIIIa and a-fucosylation level

2.1.3 Antibody-dependent cell-mediated cytotoxicity

To assess if the measured affinities correlate with biological ac- tivity, the ADCC activity of the separated antibody species of both test IgGs were examined in a cellular assay (as described in the experimental section) The analysis of the ADCC activity revealed

Table 3

Affinity between Fc γRIIIa and pools of GEmAb2 separated in three peaks by Fc γRIIIa chromatography KD obtained by surface plasmon resonance at 25 °C The three peaks of the antibody pool of GEmAb2 sep- arated by Fc γRIIIa chromatography, the starting material and the WTmAb2 with wild type carbohydrates were captured on immobilized antigen and the Fc γRIIIa was used as analyte Values represent the aver- age ± standard deviation of SPR measures from the fractions of two independent Fc γRIIIa chromatography runs (values without standard deviation are single measure) Fitting: kinetic (1:1 binding) or steady state (for interactions with too fast on and off-rates) Percentage of overall a-fucosylation determined by MALDI- TOF MS

Overall a-fucosylation level (MALDI) KD [nM] Model

Fig 3 Biological activity of antibody fractions collected from preparative Fc γRIIIa chromatography: ADCC assays were performed for the three peaks and the starting material for GEmAb1 (A) and GEmAb2 (B) Black squares: starting material, white diamonds: peak one, black triangles: peak two, black circles: peak three, white squares (only shown in B): WTmAb2 wild-type (not glycoengineered)

again a separation in two categories: the antibodies containing

mainly fucosylated carbohydrates on both heavy chains (peak one

and wild-type, not glycoengineered IgG) had a lower ability to

induce ADCC than antibody species from elution peaks two and

three, as well as glycoengineered IgG, which demonstrated a sim-

ilar and higher ability to induce ADCC ( Fig 3 ) For instance,

GEmAb1 peak two fraction contained ∼60% mono-fucosylated an-

tibodies, whereas peak three had the same amount of completely

a-fucosylated antibodies, but the ADCC assay demonstrated almost

identical results These cell-based activity measurements thus con-

firmed and agreed with the KD determined by SPR

Taken together, these results demonstrated that only one a-

fucosylated glycan per Fc was enough for a maximum affinity to

Fc γRIIIa and a corresponding ADCC activity of the antibody There-

fore, an overall a-fucosylation level of 50% is enough to obtain

the benefits of glycoengineering (i.e increased affinity for Fc γRIIIa

and increased ADCC) Based on this finding, we developed a high-

throughput screening method that allowed separating IgGs with double-fucosylated carbohydrates from IgGs with mono- and a- fucosylated carbohydrates and quantifying of the respective frac- tions

2.2 High-throughput screening for selection of clones with highly a-fucosylated carbohydrates

To identify clones producing a-fucosylated carbohydrates dur- ing cell line generation for glycoengineered antibodies, a high- throughput screening method was established For this method, a small Fc γRIIIa column (60 μl volume) with Fc γRIIIa coupled to POROS TM material via amine coupling was used and allowed a short run time of 7 min per sample The IgGs were loaded at pH 8.0 and eluted in a step at pH 3.0

First, wild type mAb1 and mAb2, as well as their respective gly- coengineered variants (carrying high proportions of a-fucosylated

Fc-oligosaccharides) were analyzed The chromatograms, obtained

by monitoring absorption at 280 nm, showed two peaks: the flow-

through peak and the elution peak Because the area of the second

peak quantifies the antibody fraction with enhanced ADCC activ-

ity as confirmed by respective analyses, we calculated the relative

biological activity (RBA) of an antibody as the percentage of the

second peak area over the total area

The RBA of glycoengineered GEmAb1 and 2 was determined to

be 66% and 75%, respectively The a-fucose content obtained by

MALDI analysis for these antibodies was 48% and 75% respectively

For the wild type antibodies WTmAb1 and 2, the RBA was of 26%

and 31%, whereas the a-fucose content determined by MALDI was

10% and 9%, respectively This test supported the use of Fc γRIIIa

affinity chromatography to identify antibodies highly enriched in

a-fucosylated carbohydrates

Next, wild-type antibodies mAb1 or mAb2 and their respec-

tive glycoengineered variants were mixed in different proportions

to obtain samples with different fucosylation levels These sam-

ples were subsequently analysed by Fc γRIIIa chromatography and

MALDI TOF MS The relationship between the percentage of a-

fucosylation assessed by MS analysis and the RBA measured by

Fc γRIIIa chromatography column was linear Five different mix-

tures were assessed in triplicates and a linear regression was fit-

ted through the obtained 15 data points For mAb1 the slope was

0.939 and the intercept 18.35 with a standard error of 0.024 for the

slope and of 0.772 for the intercept The correlation coefficient R2

was 0.991 For mAb2 the slope was 0.645 and the intercept 26.80

with a standard error of 0.007 for the slope and of 0.343 for the

intercept The correlation coefficient R2was 0.998

These results demonstrated that the fucosylation level deter-

mined by Fc γRIIIa affinity chromatography correlates to the actual

level of fucosylation measured by MS analysis

Next, we tested this screening method on 53 culture super-

natants from a clone screen instead of using purified antibodies

Cell culture supernatants of different clones expressing the gly-

coengineered GEmAb3 were analyzed in parallel by two different

methods: (1) Protein A chromatography followed by MALDI TOF

MS of the released carbohydrates and (2) Protein A chromatog-

raphy with subsequent Fc γRIIIa chromatography For this high-

throughput analysis, 100 μl of supernatant were injected on a

Protein A chromatography column The eluate was subsequently

neutralized and either digested with PNGase F for MALDI TOF

MS analysis of the carbohydrates or injected onto the Fc γRIIIa

affinity column For the latter approach, the injected sample vol-

ume was adjusted, to contain 10 μg of the analyte antibody

The Protein A purified samples were eluted into a 96-well plate,

which could directly be used for subsequent Fc γRIIIa chromatog-

raphy without any additional buffer exchange or pipetting step

The RBA (which corresponds to the a-fucose level of the sample)

was compared to the percentage of a-fucosylation determined by

MALDI TOF MS A similar ranking was obtained with both meth-

ods ( Fig 4 ), demonstrating that Fc γRIIIa affinity chromatography,

combined with Protein A purification, is a powerful tool for the

high throughput screening of cell culture supernatants and for

the ranking of clones according to their titer and a-fucosylation

grade

2.3 FcγRIIIa affinity chromatography with Fc-dimerized

FcγRIIIaV158

Soluble, his-tagged Fc γRIIIa is difficult to produce, results in

low yields after purification (up to 14 mg/L), and needs to be

chemically coupled to the chromatographic material, which might

lead to decreased receptor binding To circumvent these prob-

lems, a new construct was designed The C-terminus of the ex-

Fig 4 Comparison of Protein A chromatography followed by MALDI TOF MS and

Protein A chromatography with subsequent Fc γRIIIa chromatography as two differ- ent methods to analyze the a-fucosylation degree of antibodies purified from cell culture supernatant: Similar ranking obtained by RBA from Fc γRIIIa chromatogra- phy as by MALDI TOF MS of glycoengineered GEmAb3 53 clones of GEmAb3 were analyzed

tracellular domain of Fc γRIIIa was fused to an AviTag for site- specific biotinylation, followed by an IgA protease cleavage site and the Fc region of an IgG1, which included the P329G, L234A and L235A amino acid substitutions to avoid interactions of the

Fc part with the Fc γRIIIa [18] The Fc γRIIIa-Avi-Fc PG LALA was expressed and purified as described in material and methods sec- tion and the resulting expression yields were increased by a fac- tor of 5.5 (up to 78 mg/L) compared to the construct without the

Fc fusion The engineered Fc γRIIIa-Fc-fusion protein was subse- quently biotinylated via the AviTag, and 3 mg were coupled to Streptavidin Sepharose and packed into a 1 ml XK column Fifty micrograms IgG were loaded at pH 6.0 and eluted with a gradi- ent to pH 3.0 Upon optimization of the elution gradient, it was possible to shorten the time of analysis to 10 min while retaining resolution

Chromatography with this column also resulted in two peaks,

as with the Fc γRIIIa(V158)K6H6-tagged construct ( Fig 5 ): the first one containing the fully fucosylated IgG species and the second the mono- and a-fucosylated IgGs In contrast to the his-tagged con- struct, the first peak was not the flow-through, but eluted within the pH gradient The binding of the IgG to the immobilized, enzy- matically biotinylated Fc γRIIIa-Fc fusion was stronger than for the chemically coupled his-tagged receptor Nevertheless, we observed also here a linear relationship between the RBA and the fucosyla- tion level determined by reversed phase (RP)-UPLC as observed for the Fc γRIIIa his construct Five mixtures of GEmAb4 and WTmAb4 were analyzed on this column and compared to the a-fucosylation level determined by RP-UPLC The slope was 0.901 and the inter- cept -5.11 The correlation coefficient R2was 0.995

For comparison the Fc γRIIIa-Avi-Fc PG LALA construct was treated with IgA protease to remove the Fc-polypeptide, biotiny- lated and coupled to SA-sepharose before packing into a column Comparative analysis of the retention profiles obtained by the two affinity columns demonstrated similar chromatography per- formance in separating antibody samples by their fucose content, confirming that the Fc fusion had no influence on the chromatog- raphy result (data not shown)

Finally, we examined the stability and longevity of the Fc γRIIIa-

Fc affinity column The chromatography results were highly repro- ducible for up to 500 runs and no loss of receptor activity could

be detected ( Table 4 )

Fig 5 Elution profile of two differently fucosylated antibodies from cell line development obtained by analytical Fc γRIIIa-Avi-Fc chromatography Peak1 fully fucosylated IgG species, peak2 at least one arm a-fucosylated; black dashed line: medium a-fucosylation level, black solid line: high a-fucosylation level

Table 4

Stability of the Fc γRIIIa chromatography column The same probe (50 μg of

WTmAb5) was injected 500 times on the same column The total peak area

as well as the percentage of peak 1 and peak 2 were recorded The area and

percentages remained stable over 500 runs

Run number Total area (mAU) Area peak 1 [%] Area peak 2 [%]

High levels of a-fucosylated oligosaccharides result in strong

Fc γRIIIa binding and increased biological activity in ADCC assay

for glycoengineered therapeutic antibodies [6,19] In this study, we

present a Fc γRIIIa based chromatography to separate antibodies

according to the degree of fucosylation of their Fc-attached carbo-

hydrates

The method presented here allowed analyzing the fucosylation

level of the carbohydrates without prior labeling (with 2-AB for ex-

ample) and without the need to cleave the carbohydrates from the

polypeptide chain (with PNGaseF) and correlated directly with the

biological activity The analysis of fucosylation by Fc γRIIIa chro-

matography is thus a direct, label- and modification-free method

that avoids artefacts resulting from processing and modification

steps and therefore results in higher and more reliable quality data

The only preparation step is the removal of cell culture medium by

Protein A purification

We demonstrated, for the first time, that the Fc γRIIIa affin-

ity column can be used to enrich antibody species with differ-

ent fucosylation levels Three antibody fractions were isolated by

this procedure: antibodies enriched in fully fucosylated (1), mono-

fucosylated (2) or a-fucosylated (3) carbohydrates Each antibody

population was subjected to further analysis of its biological activ-

ity The mono- and a-fucosylated antibodies revealed a similar be-

havior in ADCC assay and SPR-assessed Fc γRIIIa interaction These

results demonstrated that one a-fucosylated glycan on the anti-

body is sufficient for increased affinity to Fc γRIIIa, which is in line

with the crystal structure of Fc with Fc γRIIIa [12] showing that

only one side of the Fc is binding to the Fc γRIIIa As expected,

the higher affinity of the a-fucosylated glycan dominates and de-

termines the overall affinity of the antibody Consequently, 100% a-fucose content is not required to achieve enhanced effector func- tions, but 50% a-fucosylation content per IgG molecule is, theoret- ically, enough Among the different methods of glycoengineering that were developed, deletion of the fucosyltransferase [20] would allow reaching higher a-fucosylation levels than overexpressing recombinant wild-type β-1,4- N-acetyl-glucosaminyltransferase III and wild-type Golgi α-mannosidase II [19] However, according

to the results shown here, this difference in a-fucosylation levels would not be reflected in ADCC activity and both methods can therefore be considered equally suitable to obtain highly active, glycoengineered antibodies

With respect to the screening of clones for antibody production, our experiments imply that quantifying the ratio between double fucosylated and the mixture of mono- and a-fucosylated antibodies

is sufficient As demonstrated here, the mono- and a-fucosylated antibodies both have similar biological activity and can therefore

be equally selected To identify clones producing antibodies with low fucosylation levels, both steps of Protein A purification and

Fc γRIIIa chromatography were amenable to high-throughput anal- ysis and were performed in this study in 96-well plates on an HPLC system The proportion of a-fucosylated antibodies in the analyzed samples, assessed by the presented chromatography method, cor- related with the one obtained by MS analysis The Fc γRIIIa chro- matography analysis, combined with preceding protein A purifica- tion, enabled fast and efficient ranking of the antibody expressing clones for their titer and fucosylation level The duration of the chromatography run was optimized to 7 min, which allowed au- tomated screening in a 96-well format of large number of samples

in a short time, representing a clear advantage of this approach over MS-analysis

The Fc γRIIIa construct was further optimized by dimerization through the introduction of an inert Fc-tag to increase produc- tion yields In addition, this construct contained an Avi-tag for spe- cific, directed biotinylation and coupling of the receptor to Strep- tavidin sepharose This ensured that the receptor bioactivity was not affected in any way due to unspecific crosslinking In our experiments, only the mono- and a-fucosylated antibody species bound to the chemically coupled Fc γRIIIa, which was likely due

to impaired receptor accessibility In contrast, Fc γRIIIa-Avi cou- pled via biotin efficiently interacted with both, fucosylated and a- fucosylated antibodies, which however differ in their affinity, as reflected in the elution pH The Fc γRIIIa-Avi tagged, as well as the Fc γRIIIa-his, allowed separating fully fucosylated antibodies

Table 5

Amino acid sequence of the two Fc γRIIIa constructs immobilized on the chromatographic matrices The biotinylated Fc fusion is the recommended construct for matrix preparation

Human Fc γRIIIa (V158) with C-terminal (Lysine)6- and (Histidine)6-fusion

GMRTEDLPKAVVFLEPQWYRVLEKDSVTLKCQGAYSPEDNSTQWFHNESLISSQASSYFIDAATVDDSGEYRCQTNLSTLSDPVQLEVHIGWLLLQAPRWVFKEEDPIHLRCHSW KNTALHKVTYLQNGKGRKYFHHNSDVYIPKATLKDSGSYFCRGLVGSKNVSSETVNITITQGLAVSTISSFFPPGYQGKKKKKKGHHHHHH

Human Fc γRIIIa (V158) with C-terminal AviTag, IgA Protease cleavage site and Fc fusion

GMRTEDLPKAVVFLEPQWYRVLEKDSVTLKCQGAYSPEDNSTQWFHNESLISSQASSYFIDAATVDDSGEYRCQTNLSTLSDPVQLEVHIGWLLLQAPRWVFKEEDPIHLRCHSW KNTALHKVTYLQNGKGRKYFHHNSDFYIPKATLKDSGSYFCRGLVGSKNVSSETVNITITQGLAVSTISSFFPPGYQGLNDIFEAQKIEWHELVVAPPAPEDKTHTCPPCPAPEAAGG PSVFLFPPKPKDTLMISRTPEVTCVVVDVSHEDPEVKFNWYVDGVEVHNAKTKPREEQYNSTYRVVSVLTVLHQDWLNGKEYKCKVSNKALGAPIEKTISKAKGQPREPQVYTLP PSRDELTKNQVSLTCLVKGFYPSDIAVEWESNGQPENNYKTTPPVLDSDGSFFLYSKLTVDKSRWQQGNVFSCSVMHEALHNHYTQKSL SL SPGK

from the mono- and a-fucosylated species and to precisely deter-

mine the percentage of these two antibody groups Scoring these

percentages allowed selecting the clones with strongest Fc γRIIIa

binding, which in turn correlated with higher biological activity

in ADCC assays This analysis was performed successfully for 222

clones during the clone screening of a production cell line [21]

Preparative Fc γRIIIa chromatography allowed to isolate three

different fractions with antibodies enriched in fully fucosylated,

mono-fucosylated or a-fucosylated carbohydrates The antibodies

carrying mono- and a-fucosylated carbohydrates had equivalent in-

creased biological activity, compared to the fully fucosylated ones

Consequently, screening for at least one-fold a-fucosylated glyco-

engineered antibodies is sufficient during clone selection The rec-

ommended design for matrix preparation is the fusion of the extra-

cellular domain of the Fc γRIIIa V158 to an Fc with the P329G LALA

mutation and an Avi-tag and immobilization of the biotinylated

receptor to Streptavidin sepharose The resulting Fc γRIIIa affinity

chromatography enabled the separation of antibodies based on the

fucosylation of their carbohydrates and/or the assessment of the

fucosylation level of an antibody sample during clone screening It

is a new, fast, stable, label- and modification-free efficient tool for

both, preparative and analytical purposes, the read-out of which

correlates directly with the biological activity of the analyzed anti-

bodies

5.1 Expression of proteins

5.1.1 Expression and purification of soluble human

FcγRIIIa(V158)K6H6

Human Fc γRIIIa (V158) with C-terminal (Lysine)6- and

(Histidine)6-fusion ( Table 5 ) [22] was produced by calcium

phosphate-transfection of HEK293-EBNA cells and harvested after

7 days The secreted protein was purified via immobilized metal

chelate chromatography (IMAC, NiNTA Superflow cartridge, Qiagen,

Germany) using the manufacturer recommendations and polishing

by size exclusion chromatography (HiLoad 16/60 Superdex 75; GE

Healthcare, Sweden) with a mobile phase of 2 mM MOPS, 150 mM

NaCl, 0.02% (w/v) NaN 3, pH 7.4

5.1.2 Expression and purification of soluble human

FcγRIIIa(V158)-Avi-IgA Protease-Fc fusion

Human Fc γRIIIa (V158) with C-terminal AviTag, IgA Protease

cleavage site and Fc fusion ( Table 5 ) was produced by transient

transfection of HEK293 cells with 293-Free transfection reagent

(Novagen) and harvested after 7 days The secreted protein was pu-

rified via affinity chromatography using HiTrap MabSelectSuRe (GE

Healthcare) and polishing by size exclusion chromatography on Su-

perdex 200 (GE Healthcare) with a mobile phase of 2 mM MOPS,

125 mM NaCl pH 7.2

5.1.3 Expression and purification of GEmAb and WTmAb

GEmAb1, WTmAb1, GEmAb2 and GEmAb4 were produced in CHO cells in serum free medium (CD-CHO, Gibco) WTmAb4 and WTmAb5 were produced in HEK293F cells in serum free Opti-MEM

I medium All IgGs were purified by affinity chromatography us- ing HiTrap MabSelectSuRe (GE Healthcare) and size exclusion chro- matography (HiLoad 16/60 Superdex 200; GE Healthcare, Sweden) WTmAb2 was produced in HEK EBNA cells in D-MEM with 10% FBS ultra-low IgG (Gibco) and purified under the same conditions with

an additional wash at pH 5.5 to remove any bovine IgGs Super- natants from GEmAb3 were produced in CHO cells in serum free medium (CD-CHO, Gibco)

5.2 Matrix preparation and chromatography conditions 5.2.1 Preparation of FcγRIIIa(V158)-affinity matrix using NHS Sepharose 4 FF for preparative purposes

30 mg Fc γRIIIa(V158) were coupled to NHS activated Sepharose 4FF (GE Healthcare, Sweden) Shortly, Fc γRIIIa(V158) was trans- ferred into 0.2 M NaHCO 3, 0.5 M NaCl, pH 8.2 and incubated for

4 h at room temperature with 3 ml NHS activated beads, pre- washed with 1 mM cold HCl The reaction was quenched with 0.1 M Tris, pH 8.5 for 2 h at room temperature The beads were then packed into a Tricorn 5/150 column (GE Healthcare) by grav- ity flow, followed by packing at 1.2 ml/min using an Äkta Explorer

10 (GE Healthcare), to a final column volume of 2.7 ml, at a col- umn length of 14 cm 30 mg of human Fc γRIIIa(V158) were im- mobilized in total Sepharose particle diameter is of 45-165 μm, pressure drop constraint was 1 bar and loading capacity of final column was around 1 mg antibody/ml column volume

5.2.2 Preparation of the affinity matrix using POROS AL for analytical purposes

10 mg Fc γRIIIa(V158) in 0.1 M sodium phosphate, 0.05% (w/v) NaN 3, pH 7.0 were coupled to 0.14 g of dry POROS AL beads (Ap- plied Biosystems, USA) with 0.1 M NaCNBH 3 and overnight incu- bation at room temperature The reaction was quenched with 1 M Tris, pH 7.4 and 50 mM NaCNBH 3for 30 min at room temperature Finally, the beads were washed four times with 1M NaCl and three times with 2 mM MOPS, 150 mM NaCl, 0.02% (w/v) NaN 3, pH 7.3

14 mg Fc γRIIIa(V158) were coupled per gram of POROS AL beads POROS particle diameter is of 20 μm, pore diameter 50 0–10 0 0 0 ˚A, pressure drop constraint max 170 bar, loading capacity of final col- umn was 0.16 mg antibody/ml column volume

5.2.3 Preparation of the affinity matrix using streptavidin sepharose for analytical purposes

An affinity column with Fc γRIIIaV158 with avi-tag was pre- pared by in vitro biotinylation and subsequent coupling to Strepta- vidin sepharose This was done with the intact fusion polypeptide

as well as with the receptor after having cleaved off the Fc-region

5.2.3.1 Cleavage of Fc fusion protein by IgA protease (optional). The

Fc γRIIIa(V158) Avi-IgA-Fc construct was dyalized to 50 mM Tris

pH 8 and incubated with IgA Protease (Roche Diagnostics GmbH)

at a ratio of w(protease)/w(fusion polypeptide) 1:100 at 21 °C

overnight The Fc-tag was removed by HiTrap MabSelectSuRe (GE

Healthcare) and the IgA protease by size exclusion chromatography

on Superdex 75 (GE Healthcare)

5.2.3.2 Biotinylation of receptor. 3 mg Fc γRIIIaV158 or 6 mg Fc

tagged Fc γRIIIaV158 in 2 mM MOPS, 125 mM NaCl pH 7.2, 0.02%

Tween20, and 1 tablet Complete protease inhibitor (cOmplete UL-

TRA Tablets, Roche Diagnostics GmbH) in 3 ml PBS were biotiny-

lated using the biotinylation kit from Avidity according to the man-

ufacturer instructions (Bulk BIRA, Avidity LLC, Denver, CO, USA) Bi-

otinylation reaction was done at room temperature overnight The

modified protein was dialyzed against 20 mM sodium phosphate

buffer com prising 150 mM NaCl, pH 7.5 at 4 °C overnight to re-

move excess of biotin

5.2.3.3 Coupling to streptavidin sepharose. One gram Streptavidin

Sepharose High Performance (GE Healthcare) was added to the bi-

otinylated and dialyzed receptor and incubated for 2 h while shak-

ing and finally packed in a 1 ml XK column (GE Healthcare) Strep-

tavidin Sepharose particle diameter is of 34 μm with a loading ca-

pacity of biotinylated bovine serum albumin of 6 mg/ml medium

(according to the manufacturer) Loading capacity of the FcyRIIIa

column was about 4 mg antibody/ml column volume (with 3 mg

Fc dimerized FcyRIIIaV158 bound)

5.2.4 Conditions for preparative separation using FcγRIIIa(V158)

immobilized on NHS Sepharose 4 FF

The chromatography column was equilibrated with 10 column

volumes of 20 mM Tris, 20 mM MOPS, 20 mM sodium citrate,

100 mM NaCl, pH 8.0, followed by load of 3 mg of purified an-

tibody at a flow rate of 0.1 ml/min The column was washed with

5 CV of 20 mM Tris, 20 mM MOPS, 20 mM sodium citrate, 100 mM

NaCl, pH 8.0 The elution was carried out with a three step gradi-

ent of pH steps 4.6, 4.2 and 3.0 The peaks were collected, concen-

trated and purified by protein A

5.2.5 Conditions for analytical chromatography using FcγRIIIa(V158)

immobilized on POROS AL

POROS AL beads with Fc γRIIIa(V158) were packed in a

2 × 20 mm Upchurch Scientific column (column volume: 60 μl)

which was mounted in the Agilent 1200 HPLC system (Agilent

Technologies, USA) 10 mM Tris, 50 mM glycine, 100 mM NaCl, pH

8.0 was used to equilibrate and wash the column; the elution was

carried out with 10 mM Tris, 50 mM glycine, 100 mM NaCl, pH

3.0 (flow rate 0.5 ml/min) At time zero the antibody preparation

(10 μg) was injected and washed for 2 min, then eluted in a step

gradient of 0.66 min, before re-equilibration for 4.33 min

For high throughput analysis of IgGs from supernatants the

samples were first purified using Protein A on the Agilent 1200

HPLC system and collected in a 96-well plate The samples were

neutralized by adding 1:40 v/v 2 M Tris pH 8.0 and either re-

injected on the Fc γRIIIa(V158) chromatography column (injection

volumes adjusted to inject 10 μg) or were digested with PNGase F

for MALDI TOF MS analysis of the carbohydrates

5.2.6 Conditions for analytical chromatography using

FcγRIIIa(V158)-Fc immobilized on Streptavidin sepharose

Antibody samples containing 30 to 50 μg of protein were di-

luted at a volume ratio of 1:1 with equilibration buffer, 20 mM cit-

ric acid/150 mM NaCl pH 6.0, and applied to the Fc γRIIIa column

The column was washed with 2.5 column volumes of equilibration

buffer and elution of samples was done with a linear pH gradi- ent to 20 mM citrate, 150 mM NaCl pH 3.0 in 15 column volumes The experiments were carried out at room temperature The elu- tion profile was obtained by continuous measurement of the ab- sorbance at 280 nm

5.3 Analytical methods for analyses of carbohydrate composition 5.3.1 ESI-MS analysis on Fc fragments or whole IgG

Oligosaccharides were digested by EndoS and EndoH prior to ESI-MS analysis as described [23]

5.3.2 MALDI-TOF analysis on released carbohydrates

The N-linked oligosaccharides were cleaved of the purified IgGs

by incubation with 0.005 U of PNGase F (QAbio, USA) and EndoH (QAbio, USA) in 20 mM Tris pH 8.0 at 37 °C for 16 h This resulted

in free oligosaccharides that were analyzed by mass spectrometry (Autoflex, Bruker Daltonics GmbH) in positive ion mode according

to [24]

5.3.3 2-AB (aminobenzamide) labeling and RP-UPLC of carbohydrates for the determination of the relative oligosaccharide distribution

The content of a-fucosylated N-glycans was determined by glycan release with PNGaseF, followed by derivatization with 2- aminobenzamidine and separation by RP-UPLC Glycans were re- leased by PNGaseF (New England Biolabs) digestion overnight at

37 ° After addition of final 20 mM acetic acid and incubation for

15 min at 45 °C, 2-AB labeling of the released glycans was per- formed with the GlycoProfile 2-AB labeling kit (Sigma) according to the manufacturer’s instructions Fucosylated and non-fucosylated labeled glycans were separated by a linear gradient of 0.1% aque- ous formic acid to 0.1% formic acid in acetonitrile on an Acquity HSS T3 column (Thermo Scientific) at 80 ° and detected by fluores- cence

5.4 Surface plasmon resonance

SPR interaction analysis was performed on a Biacore T100 sys- tem (GE Healthcare) Human antigen for mAb2 was immobilized

by amine coupling on a CM5 chip using the manufacturer’s in- structions (GE Healthcare) The IgG fractions were captured for

90 at 100 nM and 10 μl/min The human Fc γRIIIa(V158) was in- jected as analyte at a concentration range from 1.95–500 nM and

a flow rate of 50 μl/min for 120 s The dissociation is monitored for 220 s The surface was regenerated by two injections of 10 mM glycine, pH 2.0 for 60 s Bulk refractive index differences were cor- rected by subtracting the response obtained on the reference flow cell Association rates (k on) and dissociation rates (k off) were calcu- lated using a one-to-one Langmuir binding model with RI = 0 and Rmax = local (BIACORE  R T100 Evaluation Software version 1.1.1)

by simultaneously fitting the association and dissociation sensor- grams The equilibrium dissociation constant (KD) was calculated

as the ratio k off/k on

5.5 ADCC

Raji (for mAb1 ADCC assay) or A549 (for mAb2 ADCC assay) cells were harvested (adherent cells with Trypsin/EDTA), washed and labeled for 30 min at 37 °C with Calcein (Invitrogen) After

30 min, cells were washed 3 times with AIM V and re-suspended

in AIM V medium Subsequently, they were plated in a round- bottom, 96-well plate at a concentration of 30,0 0 0 cells/well The respective antibody dilutions were added and incubated for 10 min before contact with human effector cells (NK92 1708 clone LC3 E11 = NK92 cells transfected with Fc γRIIIa(V158)) Effector and target cells were co-incubated at a ratio of 3:1 for 4 h at 37 °C

Lactate dehydrogenase (LDH) release was measured using the LDH

Cytotoxicity detection Kit (Roche Applied Science) The calcein re-

tention was measured by lysing the remaining cells with borate

buffer (5 mM borate + 0.1% v/v Triton X100) followed by mea-

surement of the calcein fluorescence For calculation of antibody-

dependent killing, spontaneous release (only target +effector cells

without antibody) was set to 0% killing and maximal release (tar-

get cells + 2% v/v Triton X-100) was set to 100% killing

None

Acknowledgments

We thank Hans Koll for the analysis of the repartition of carbo-

hydrates on whole Fc fragments

Funding

This research did not receive any specific grant from funding

agencies in the public, commercial, or not-for-profit sectors All au-

thors are (or were) Roche employees (at the time when the study

was conducted)

Supplementary material associated with this article can be

found, in the online version, at doi:10.1016/j.chroma.2019.460554

References

[1] R Jefferis , Isotype and glycoform selection for antibody therapeutics, Arch

Biochem Biophys 526 (2012) 159–166

[2] X.R Jiang , A Song , S Bergelson , T Arroll , B Parekh , K May , S Chung ,

R Strouse , A Mire-Sluis , M Schenerman , Advances in the assessment and con-

trol of the effector functions of therapeutic antibodies, Nat Rev Drug Discov

10 (2011) 101–111

[3] R.M Anthony , F Nimmerjahn , The role of differential IgG glycosylation in the

interaction of antibodies with FcgammaRs in vivo, Curr Opin Organ Trans-

plant 16 (2011) 7–14

[4] J.N Arnold , M.R Wormald , R.B Sim , P.M Rudd , R.A Dwek , The impact of gly-

cosylation on the biological function and structure of human immunoglobu-

lins, Annu Rev Immunol 25 (2007) 21–50

[5] T Dashivets , M Thomann , P Rueger , A Knaupp , J Buchner , T Schlothauer ,

Multi-angle effector function analysis of human monoclonal IgG glycovariants,

PLoS One 10 (2015) e0143520

[6] K Mori , S Iida , N Yamane-Ohnuki , Y Kanda , R Kuni-Kamochi , R Nakano ,

H Imai-Nishiya , A Okazaki , T Shinkawa , A Natsume , R Niwa , K Shitara ,

M Satoh , Non-fucosylated therapeutic antibodies: the next generation of ther-

apeutic antibodies, Cytotechnology 55 (2007) 109–114

[7] T.S Raju , Terminal sugars of Fc glycans influence antibody effector functions of

IgGs, Curr Opin Immunol 20 (2008) 471–478

[8] M Thomann , K Reckermann , D Reusch , J Prasser , M.L Tejada , Fc-galactosy-

lation modulates antibody-dependent cellular cytotoxicity of therapeutic anti-

bodies, Mol Immunol 73 (2016) 69–75

[9] C Ferrara , P Brunker , T Suter , S Moser , U Puntener , P Umana , Modulation of

therapeutic antibody effector functions by glycosylation engineering: influence

of Golgi enzyme localization domain and co-expression of heterologous beta1,

4-N-acetylglucosaminyltransferase III and Golgi alpha-mannosidase II, Biotech- nol Bioeng 93 (2006) 851–861

[10] M Schuster , P Umana , C Ferrara , P Brunker , C Gerdes , G Waxenecker ,

S Wiederkum , C Schwager , H Loibner , G Himmler , G.C Mudde , Improved effector functions of a therapeutic monoclonal Lewis Y-specific antibody by glycoform engineering, Cancer Res 65 (2005) 7934–7941

[11] N Yamane-Ohnuki , M Satoh , Production of therapeutic antibodies with con- trolled fucosylation, MAbs 1 (2009) 230–236

[12] C Ferrara , S Grau , C Jager , P Sondermann , P Brunker , I Waldhauer , M Hen- nig , A Ruf , A.C Rufer , M Stihle , P Umana , J Benz , Unique carbohydrate-car- bohydrate interactions are required for high affinity binding between Fcgam- maRIII and antibodies lacking core fucose, Proc Natl Acad Sci USA 108 (2011) 12669–12674

[13] R.L Shields , J Lai , R Keck , L.Y O’Connell , K Hong , Y.G Meng , S.H Weikert , L.G Presta , Lack of fucose on human IgG1 N-linked oligosaccharide improves binding to human Fcgamma RIII and antibody-dependent cellular toxicity, J Biol Chem 277 (2002) 26733–26740

[14] V Goede , K Fischer , R Busch , A Engelke , B Eichhorst , C.M Wendtner ,

T Chagorova , J de la Serna , M.S Dilhuydy , T Illmer , S Opat , C.J Owen ,

O Samoylova , K.A Kreuzer , S Stilgenbauer , H Dohner , A.W Langerak , M Rit- gen , M Kneba , E Asikanius , K Humphrey , M Wenger , M Hallek , Obinu- tuzumab plus chlorambucil in patients with CLL and coexisting conditions, N Engl J Med 370 (2014) 1101–1110

[15] E Ward , N Mittereder , E Kuta , G.P Sims , M.A Bowen , W Dall’Acqua , T Tedder ,

P Kiener , A.J Coyle , H Wu , B Jallal , R Herbst , A glycoengineered anti-CD19 antibody with potent antibody-dependent cellular cytotoxicity activity in vitro and lymphoma growth inhibition in vivo, Br J Haematol 155 (2011) 426–

437 [16] M Kiyoshi , J.M.M Caaveiro , M Tada , H Tamura , T Tanaka , Y Terao , K Morante ,

A Harazono , N Hashii , H Shibata , D Kuroda , S Nagatoishi , S Oe , T Ide ,

K Tsumoto , A Ishii-Watabe , Assessing the heterogeneity of the Fc-Glycan of a therapeutic antibody using an engineered FcgammaReceptor IIIa-Immobilized Column, Scientific Rep 8 (2018) 3955

[17] P.M Drake , B Schilling , R.K Niles , M Braten , E Johansen , H Liu , M Lerch , D.J Sorensen , B Li , S Allen , S.C Hall , H.E Witkowska , F.E Regnier , B.W Gibson , S.J Fisher , A lectin affinity workflow targeting glycosite-specific, cancer-related carbohydrate structures in trypsin-digested human plasma, Anal Biochem 408 (2011) 71–85

[18] T Schlothauer , S Herter , C.F Koller , S Grau-Richards , V Steinhart , C Spick ,

M Kubbies , C Klein , P Umana , E Mossner , Novel human IgG1 and IgG4 Fc-engineered antibodies with completely abolished immune effector func- tions, Protein Eng Des Sel 29 (2016) 457–466

[19] E Mossner , P Brunker , S Moser , U Puntener , C Schmidt , S Herter , R Grau ,

C Gerdes , A Nopora , P.E van , C Ferrara , P Sondermann , C Jager , P Strein ,

G Fertig , T Friess , C Schull , S Bauer , P.J Dal , N.C Del , K Dabbagh , M.J Dyer ,

S Poppema , C Klein , P Umana , Increasing the efficacy of CD20 antibody ther- apy through the engineering of a new type II anti-CD20 antibody with en- hanced direct and immune effector cell-mediated B-cell cytotoxicity, Blood 115 (2010) 4393–4402

[20] N Yamane-Ohnuki , S Kinoshita , M Inoue-Urakubo , M Kusunoki , S Iida ,

R Nakano , M Wakitani , R Niwa , M Sakurada , K Uchida , K Shitara , M Satoh , Establishment of FUT8 knockout Chinese hamster ovary cells: an ideal host cell line for producing completely defucosylated antibodies with enhanced antibody-dependent cellular cytotoxicity, Biotechnol Bioeng 87 (2004) 614–

622 [21] O Popp , S Moser , J Zielonka , P Rüger , S Hansen , O Plöttner , Development

of a pre-glycoengineered CHO-K1 host cell line for the expression of anti- bodies with enhanced Fc mediated effector function, mAbs 10 (2018) 290–

303 [22] C Ferrara , F Stuart , P Sondermann , P Brunker , P Umana , The carbohydrate at FcgammaRIIIa Asn-162 An element required for high affinity binding to non– fucosylated IgG glycoforms, J Biol Chem 281 (2006) 5032–5036

[23] C Jager , C Ferrara , P Umana , A Zeck , J.T Regula , H Koll , Determination of antibody glycosylation by mass spectrometry, Methods Molec Biol 901 (2012) 195–208

[24] D.I Papac , A Wong , A.J Jones , Analysis of acidic oligosaccharides and gly- copeptides by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry, Anal Chem 68 (1996) 3215–3223