Fc-engineered antibodies with immune effector functions completely abolished

Ian Wilkinson; Stephen Anderson; Jeremy Fry; Louis Alex Julien; David Neville; Omar Qureshi; Gary Watts; Geoff Hale

doi:10.1371/journal.pone.0260954

Abstract

Elimination of the binding of immunoglobulin Fc to Fc gamma receptors (FcγR) is highly desirable for the avoidance of unwanted inflammatory responses to therapeutic antibodies and fusion proteins. Many different approaches have been described in the literature but none of them completely eliminates binding to all of the Fcγ receptors. Here we describe a set of novel variants having specific amino acid substitutions in the Fc region at L234 and L235 combined with the substitution G236R. They show no detectable binding to Fcγ receptors or to C1q, are inactive in functional cell-based assays and do not elicit inflammatory cytokine responses. Meanwhile, binding to FcRn, manufacturability, stability and potential for immunogenicity are unaffected. These variants have the potential to improve the safety and efficacy of therapeutic antibodies and Fc fusion proteins.

Citation: Wilkinson I, Anderson S, Fry J, Julien LA, Neville D, Qureshi O, et al. (2021) Fc-engineered antibodies with immune effector functions completely abolished. PLoS ONE 16(12): e0260954. https://doi.org/10.1371/journal.pone.0260954

Editor: Sophia N. Karagiannis, King’s College London, UNITED KINGDOM

Received: June 1, 2021; Accepted: November 20, 2021; Published: December 21, 2021

Copyright: © 2021 Wilkinson et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: All relevant data are within the paper and its Supporting information files.

Funding: This study was primarily funded by mAbsolve Limited https://mabsolve.com/. Some of the Biacore experiments were funded by Antibody Analytics Ltd https://www.antibodyanalytics.com/. IW was and SA is and employee of Absolute Antibody Ltd, JF is an employee of ProImmune Limited, LAJ is an employee of Antibody Analytics Limited, DN is an employee of Reading Scientific Services Limited, OQ is an employee of Celentyx Limited, GW is an employee of Abzena Limited and IW and GH are employees of mAbsolve Limited. mAbsolve Limited played an indirect role in the study design, data collection and analysis, decision to publish and preparation of the manuscript to the extent that IW and GH are founding shareholders and directors of mAbsolve Limited. Antibody Analytics Ltd provided support in the form of salary for LAJ, but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of this author are articulated in the ‘author contributions’ section.

Competing interests: I have read the journal’s policy and the authors of this manuscript have the following competing interests: The subject matter of this publication is also disclosed in patent applications GB2007613.9 and PCT/GB2021/051233 (‘Modified immunoglobulin Fc regions’) which are assigned to mAbsolve Limited. mAbsolve has issued licences to biotechnology and pharma companies to develop antibodies and Fc fusion proteins described in the patent applications. IW and GH are founders and directors of mAbsolve Limited. Each of them owns more than 5% and less than 20% of the shares of mAbsolve. SA owns fewer than 5% of the shares of mAbsolve. This does not alter our adherence to PLOS ONE policies on sharing data and materials. No other authors declare a competing interest.

Introduction

Pharmacologic properties of immunoglobulins depend very much on their Fc region. Interaction with C1q initiates activation of complement. Binding to various Fc receptors on leucocytes induces antibody-dependent cell-mediated cytotoxicity (ADCC) or antibody-dependent cell-mediated phagocytosis (ADCP). Binding to the FcRn receptor is responsible for the comparatively long half-life of IgG. For many years, scientists have investigated the particular residues involved in binding to these various ligands, with the aim of modifying the natural properties of antibodies—either to enhance (e.g. to improve killing of tumour cells, or to extend half-life) or to reduce (e.g. to avoid unwanted side effects) [1–4]. Many therapeutic IgG4 antibodies have been developed as it was once believed that this isotype was devoid of effector function. However, that is not the case, as was tragically shown by a disastrous trial of the IgG4 CD28 antibody TGN1412 [5]. Alterations to the Fc region are necessary if binding to Fcγ receptors is to be eliminated. The Fc region is glycosylated and the carbohydrate is easily eliminated by mutation of N297, resulting in a reduction in binding to Fc receptors and C1q [6]. At least three therapeutic antibodies using the N297A mutation entered clinical trials (e.g. atezolizumab NCT02409355, clazakizumab NCT04343989, otelixizumab NCT01123083). However, removal of carbohydrate is a drastic change which significantly destabilises the Fc structure, leading to potential problems with manufacturability, stability and pharmacokinetics [7, 8].

One of the most widely used IgG1 variants is L234A/L235A (LALA) [9]. These substitutions reduce binding to the IgG Fc receptors FcγRI, FcγRII and FcγRIII as well as to complement component C1q. Such antibodies are useful where binding and activation of Fc receptors is undesirable, for example when the product is being used as an antagonist of a cytokine or similar. Numerous therapeutic antibodies using the LALA mutations have entered clinical trials (e.g. bimagrumab NCT01925209, cemiplimab NCT02383212, galcanezumab NCT03559257, progolimab NCT03912389, risankizumab NCT02684370, spesolimab NCT03482635, teplizumab NCT00385697). However, it is now known that LALA variants still have substantial binding to Fc receptors and so numerous others have been developed in the quest to completely eliminate binding to the Fcγ receptors [10–18]. Throughout the subsequent discussion, unless otherwise stated, we will refer to variants of human IgG1.

L234F/L235E/P331S (FES) and L234F/L235Q/K322Q (FQQ) are among the candidates [10, 11]. At least three antibodies with FES substitutions have entered clinical trials (e.g. anifrolumab NCT02446899, durvalumab NCT02369874, olendalizumab NCT01883544). However, in our tests, both variants still show significant levels of binding as well as being active in cell-based assays for the activation of receptors involved in ADCC and ADCP. An alternative set of variants is based on the substitutions A330S/P331S [9]. Serine is naturally found at 330 and 331 in IgG4, but when incorporated into IgG2, results in reduced FcR binding. This was used in a number of antibodies (e.g.bococizumab NCT01968967, fremanezumab NCT02621931, ponezumab NCT00945672, tanezumab NCT00733902). However, concerns emerged over the heterogeneity of disulphide bonding in IgG2 and reduced binding to FcRn leading to a shorter half-life [13]. More recently, L234A/L235A/P329G (LALAPG) was said to ‘completely abolish’ immune effector functions [14]. Several antibodies with this set of mutations have been taken into clinical trials (e.g. cergutuzumab NCT02350673, cibisatamab NCT03866239, faricimab NCT03038880, RG7386 NCT02558140). Other variants reported to reduce binding to Fcγ receptors include: L234A/G237A, L234A/L235A/G237A, L234A/L235A/G237A/P238S/H268A/A330S/P330S, L234A/L235E, G236R/L328R, and L234A/L235A/K322A [15–18].

In the above reports, different methods were used to measure binding to Fc receptors and functional activity of the variant Fc regions, so it is impossible to objectively compare the different mutations. We have now prepared a collection of antibodies having identical Fab regions but altered Fc regions as previously described. We compare them with a large number of novel Fc variants and find that all of the previous Fc variants still have a measurable level of FcγR binding activity. In contrast, some of our novel variants show no detectable activity at all. We go on to show that the new variants are not significantly different from wild-type with regard to other important properties, including binding to FcRn, stability and manufacturability or risk of unwanted immunogenicity. These new variants may find many applications in research, diagnosis and therapy.

Materials and methods

Human tissues

Human material for cytokine release studies was obtained by Celentyx from participants with written consent and approval from London-South East Research Ethics Committee REC reference 16/LO/0601. Cell samples for in vitro immunogenicity studies were obtained by ProImmune from NHS Blood & Transplant donors with written consent. They were processed and stored under Human Tissue Authority Licence 1258.

Nomenclature

The EU numbering system [19] is used throughout this article. Amino acid alterations are described thus: XnnnY, where X is the single letter code for the residue in the native amino acid sequence, nnn is the EU index position and Y is the single letter code for the replacement amino acid residue. The symbol Δ refers to a residue which is deleted. Some variants are referred to in short form: LALA (L234A/L235A), LALAPG (L234A/L235A/P329G), FES (L234F/L235E/P331S), FQQ (L234F/L235Q/K322Q), aglycosyl (N297A),

Antibody design, expression and purification

Codon-optimised synthetic genes encoding the wild-type human IgG1 heavy chain constant region or variants thereof were provided by Genewiz in pUC57-Kan vectors. The purified DNA was digested with NheI/NotI restriction enzymes and the gene extracted for ligation into a pUV mammalian expression vector. Similarly, a codon-optimised synthetic gene for human kappa light chain constant region was ligated into a pUV vector. Synthetic genes encoding the VH and VL variable regions of anti-CD20 rituximab, anti-CD3 muromonab or anti-CD52 alemtuzumab were provided by Genewiz with NheI and AvaI restriction sites at the 5’ and 3’ ends for cloning. VH genes were digested, excised and ligated into the wild-type human IgG1 heavy chain pUV vector. VL genes were digested, excised and ligated into the human kappa light chain pUV vector.

To create variants, synthetic genes containing variant Fc regions were synthesised with KasI and SacII restriction sites at the 5’ and 3’ ends and digested, excised and ligated into the existing vector encoding the immunoglobulin heavy chain of anti-CD20. Variants of the anti-CD3 heavy chain or anti-CD52 heavy chain were generated by digesting the appropriate VH variable regions with NheI and AvaI followed by excision and ligation into anti-CD20 variant heavy chain vectors similarly digested with NheI and AvaI to remove the canti-CD20 VH sequence for replacement with anti-CD3 or anti-CD52.

Intact antibodies were produced by transient transfection of HEK293 cells with a mixture of the appropriate heavy chain expression vector with the corresponding light chain expression vector encoding the cognate light chain. HEK293 cells were grown in suspension in serum-free culture medium (ThermoFisher Cat. No. 12338026) and transfected using linear polyethyleneimine. The cells were cultured for 6 days at 37°C with shaking at 140 rpm in 5% CO₂. Culture supernatants were analysed by non-reducing SDS-polyacrylamide gel electrophoresis (SDS-PAGE) followed by staining with Coomassie Blue. All samples gave a single band corresponding to a molecular weight of approximately 150 kDa. Antibodies were purified by affinity chromatography on Protein A (MabSelect SuRe, GE Life Sciences). Eluted antibodies were buffer exchanged into phosphate buffered saline (PBS) pH 7.2 and concentrated as required. The protein concentration was measured by absorbance at 280 nm using a Nanodrop ND-1000 spectrophotometer and adjusted to 1 mg/mL by dilution with PBS. The binding of CD3 and CD52 antibodies to recombinant antigens was verified by ELISA. Several separate transfection experiments were carried out. A selected list of samples is shown in Table 2.

Binding to immobilised Fc receptors

Binding analysis was carried out by surface plasmon resonance at 25°C using a Biacore T200 instrument with HEPES-buffered saline containing EDTA and P20 detergent (HBS-EP+) running buffer. The instrument was controlled using BiaCore T200 software version 1.0 and data were analysed using BiEvaluation software version 4.1 and Microsoft Excel 2016 32-bit. Mouse monoclonal IgG1 anti-histidine antibody (Qiagen Cat. No. 34670) was immobilised to a CM5 chip by carbodiimide chemistry giving a response in the range 6455 to 7834 RU. A reference cell was activated and blocked with ethanolamine. Histidine-tagged recombinant Fc receptors (Table 1) were diluted to 5 μg/mL in HBS-EP+ and injected for 2 min at 10 μL/min.

Download:

Table 1. Recombinant histidine-tagged Fc receptors.

https://doi.org/10.1371/journal.pone.0260954.t001

Test samples were diluted with running buffer to 100 μg/mL and injected for 1 min at 30 μL/min. Between each cycle, the chip was regenerated by injection of 10 mM glycine pH 3.0 for 0.5 min at 30 μL/min and recharged with a fresh injection of Fc receptor. At the beginning and end of each experiment, blank cycles of buffer alone were run. The mean SPR signal (relative to the reference cell) was measured approximately 5 seconds before the end of injection (response). The blank (buffer) response was subtracted to give a corrected response for each test sample. Statistical analysis was carried out using a two-tailed t-test, assuming unequal variances.

Binding to FcRn was measured by indirect capture of histidine-labelled FcRn of various species as described above, except that the running buffer was adjusted to pH 6.0 by titration with acetic acid. Binding to human FcRn was also measured by direct capture. Human FcRn was immobilised on a CM5 chip by EDC/NHS chemistry to give a response of 645.8 RU. Samples were diluted with HBS-EP+ pH 6.0, injected for 2 min at 30 μL/min and allowed to disassociate for 1 min. Between each cycle, the chip was regenerated by injection of HBS-EP+ pH 7.4 for 1 min. A baseline correction was calculated by interpolating the response of blank samples at the beginning and end of the run. To correct for any loss of binding capacity during the run, a normalization factor was calculated by interpolating the response of the wild-type reference sample of anti-CD20 antibody at the beginning and end of the run. The baseline-corrected responses were divided by the normalisation factor to give corrected and normalised responses which were expressed as a percentage of the response of the corresponding wild-type reference.

Binding to immobilised antibodies

Human, cynomolgus monkey and mouse Fcγ receptors and FcRn were produced in HEK293 cells. FcRn was expressed as a heterodimer with β2 microglobulin. The extracellular domains had a C-terminal hexa-His-tag and were purified by immobilised Ni^2+-affinity chromatography using a 1 mL HisTrap FF column (Cytiva Cat. No. 17531901) followed by size exclusion chromatography using a 120 mL HiLoad 16/600 Superdex 200 pg column (Cytvia Cat. No. 28989335). Samples were eluted in HBS-EP+ at pH 7.4 and the monomer peak were pooled and analysed by SDS-PAGE.

Binding analysis was carried out in HBS-EP+ pH 7.4 (for Fcγ receptors) or PBS-Tween pH 5.8 (for FcRn) at 25°C using a Biacore 8K. The instrument was controlled using Biacore 8K Control Software version 3.0.12.15655. Data were collected and analysed using Biacore Insight Evaluation Software version 3.0.12.15655. Streptavidin was immobilised on flow cells 1 and 2 of a Series S CM5 Sensor Chip by standard amine coupling chemistry. Biotin-labelled anti-human κ light chain (CaptureSelect, Life Technologies Cat. No. 7103302100) was injected until an acceptable level (798 to 1392 RU) was achieved. Test samples, either Remicade (rituximab, wild-type human IgG1) or variant 2–52, were diluted with running buffer to 5 μg/mL and injected for 90 s at 10 μL/min on flow cell 2. Human, cynomolgus monkey, and murine Fcγ receptors were diluted to an optimal concentration for maximal binding to wild-type human IgG1, ranging from 12.5 nM (human and cynomolgus FcγRI) to 18 μM (human FcγRIIb). Human, cynomolgus and mouse FcRn were diluted to 1,500 nM, 500 nM, and 250 nM, respectively. The binding of FcRn was assessed over a 6-point, 1 in 2-fold dilution series (except for the murine FcRn: 5-point dilution series), in duplicate. Association and dissociation were set at 50 s and 60 s. Between each cycle, the chip was regenerated by injection of 10 mM glycine HCl pH 1.5 and 10 mM NaOH for 1 min at 50 μL/min. At the beginning and end of each experiment, blank cycles of PBS-Tween (pH 5.8) alone were run. After reference and blank subtractions, affinity constants were calculated using a non-linear curve fit equilibrium model (human and cynomolgus monkey FcRn) or a Langmuir 1:1 kinetic model (mouse FcRn).

Fc effector cell bioassays

Antibodies were assessed for their ability to engage in antibody-dependent cell-mediated phagocytosis (ADCP) or antibody-dependent cell-mediated cytotoxicity (ADCC) using Promega Fc effector bioassay systems. The assay kit contained CD20+ Raji target cells (Promega cat. no. G7016) and engineered Jurkat effector cells which stably express the desired Fc receptor and an NFAT response element to drive expression of firefly luciferase. Experiments were carried out according to the manufacturer’s instructions using effector cells as listed in Table 2

Download:

Table 2. Effector cells.

https://doi.org/10.1371/journal.pone.0260954.t002

Target cells, effector cells and sample dilutions were all prepared in RPMI1640 culture medium containing 4% low IgG bovine serum (assay buffer). 75 μL of assay buffer was added to the edge wells of a white, flat-bottomed microplate and 25 μL of target cell suspension was added to the central wells. Sample dilutions were prepared in an off-line plate; 25 μL was transferred to the central wells of the assay plate and mixed on a plate shaker. 25 μL of effector cell suspension was added and the plate was incubated at 37°C in 5% CO₂ for 6 hours, then equilibrated at room temperature for 15 min. 75 μL of luciferase assay substrate was added and luminescence was measured using a Glomax 96 luminometer (Promega).

Binding to C1q

Antibody binding to human C1q was measured by ELISA. Human C1q (Sigma Cat. No. 20476) was coupled to horse-radish peroxidase (HRP) using Lightning-Link^® HRP Conjugation Kit (Innova Biosciences Cat No 701–0003). ELISA was carried out with shaking at room temperature (18 to 22°C). Samples of purified antibodies comprising variant Fc regions were diluted to 10 μg/mL in PBS. Negative controls consisted of PBS alone. Replicates of 100 μL each were added to 96-well clear Maxisorp microplates (Sigma Cat. No. M9410-ICS) and incubated for 60 min. The plates were rinsed with PBS containing 0.05% (v/v) Tween 20 (wash buffer) and 200 μL PBS containing 1% casein (block buffer) was added. The plates were incubated for 60 min, then rinsed with wash buffer. 100 μL HRP-labelled C1q at 100 ng/mL in block buffer was added. The plates were incubated for 90 min then washed four times with wash buffer and twice with water. 100 μL of 3,3’,5,5’-tetramethylbenzidine liquid substrate, supersensitive (TMB) (Sigma Cat. No. T4444) was added and the plates were incubated for 10 min. 50 μL of 1M sulphuric acid was added to stop the reaction and the absorbance was read at 450 nm with 620 nm subtraction using a microplate spectrophotometer (Anthos Labtec HT). At least 6 replicates of each sample were tested in each experiment and the mean absorbance for each sample was compared with the mean absorbance obtained with buffer alone using a one-tailed t-test, assuming unequal variances and the resulting probability (P) values were calculated. Results were flagged if they were significantly higher (p ≤ 0.05) than the response for buffer alone.

Cytokine release

Human peripheral blood mononuclear cells (PBMC) were isolated from healthy donors by centrifugation over Ficoll-Paque PLUS (GE Healthcare Cat. No. 11778538) and washed with RPMI medium containing 1% heat-inactivated foetal bovine serum, depleted of bovine IgG (FBS) and 1% penicillin/streptomycin (PS). The PBMC were resuspended at 2×10⁶ per mL in RPMI containing 10% FBS and 1% PS (cRPMI). Anti-CD3 antibodies at 1 mg/mL in PBS were diluted with cRPMI to give concentrations of 20, 2, or 0.2 or 0.02 μg/mL. A negative control consisted of PBS diluted five-fold with cRPMI. Each sample was tested in triplicate. 100 μL of test sample was mixed with 100 μL of cell suspension in a 96-well round-bottom microplate and cultured at 37°C and 5% CO₂ for 24 h. The plate was centrifuged at 400 × g for 5 min and the culture supernatants were collected. Cytokines (GM-CSF, IFNγ, IL-2, IL-4, IL-10 and TNFα) were quantified by Luminex assay (Bio-Rad Cat. No. M50000007A) on a Bio-Plex^® 200 system with Bio-Plex Manager software according to manufacturer’s instructions. For each donor, the median value of the triplicate response was calculated. For each test sample, the median responses for the five donors were compared with the median responses for the negative control using a single-tailed, paired Student’s t-test to determine the significance of any differences. The distribution of each set of responses for the five donors was tested using the Shapiro-Wilk test and in the majority of cases gave a p-value >0.05 indicating the assumption that the data are normally distributed can be retained. The exceptions were limited to samples with responses very close to baseline. The complete dataset (i.e. 15 samples consisting of three replicates with each of five donors) was also analysed using the non-parametric single-tailed, paired Wilcoxon signed-rank test to confirm significant differences between test samples and the negative controls.

Potential for binding to MHC Class II

The potential for variant Fc regions to give rise to peptides with the theoretical capacity to bind to human MHC Class II was assessed using the MHC-II Binding Prediction tool provided by the Immune Epitope Database (IEDB), available at http://tools.iedb.org/mhcii/. The prediction method was IEDB recommended version 2.22 [20]. This used a Consensus method, version 2.22 combining NN-align, SMM-align, CombLib and Sturniolo if any corresponding predictor was available for the molecule, otherwise NetMHCIIpan. The input consisted of a list of 16 sequences (wild-type and variants) corresponding to the Fc region of human IgG1, i.e. residues E216 to K448 inclusive. A reference set of 27 MHC Class II alleles [21] was selected with the default peptide epitope length of 15 amino acid residues. The output was sorted by rank score. A peptide was considered “at risk” of binding to MHC Class II if it gave a rank score of 10% or less for binding to any of the selected alleles. The total number of ‘at risk’ peptide/allele combinations was calculated for each variant.

Potential for immunogenicity in vitro

The ProImmune ProMap^® T cell proliferation assay was used to assess the potential immunogenicity of peptides derived from variant Fc regions. 20-mer peptides were synthesized by Kendall Scientific Inc (Lincolnshire, IL, USA), purified by HPLC and analysed by HPLC and mass spectrometry to confirm purity (> 95%) and molecular weight (within 1Da of theoretical). The test samples were provided as freeze dried TFA salts and were dissolved in DMSO to give 1 mM. Two of the peptides were synthesized twice. The analyst was blinded to the peptide sequences and duplicates. PBMC were isolated from 20 healthy human donors and depleted of CD8+ T cells. Each sample was HLA-typed and stored in liquid nitrogen prior to use. They provided both T cells and antigen presenting cells (APC) in the assay. Cells were labelled with carboxyfluorescein succinimidyl ester (CFSE) prior to incubation with test peptides. Six replicates of each peptide were tested at a final assay concentration of 5μM. Controls included: (a) whole proteins Tuberculin Purified Protein Derivative (PPD), 5 μg/mL (b) Keyhole Limpet Hemocyanin (KLH), 0.25 mg/mL (c) ProMix^™ CEFT Peptide Pool, 5 μM (d) a pool of peptides from KLH, 5 μM (e) culture medium alone. After 7 days of culture, cells were stained with anti-CD4 antibody, washed and fixed for analysis by flow cytometry using FlowJo Software (TreeStar Inc.). Proliferation was determined by measuring a decrease in CFSE intensity.

The number of CD4+ CFSEdim (i.e. proliferating) cells was determined as a proportion of the total CD4+ population to determine percentage stimulation. The background percentage stimulation of between 0.05% and 3.78% for the individual donors was subtracted from the percentage stimulation of the test samples and a one-way analysis of variance (ANOVA) was calculated to determine (to a significance level of p ≤ 0.05) whether the stimulation obtained from each peptide differed significantly from the unstimulated control. A sample was flagged ‘positive’ if all the following were true: (a) Percentage Stimulation ≥ 0.5% above Background, (b) Percentage Stimulation ≥ 2 × SEM above Background, (c) test sample significantly different from background p ≤ 0.05. Conventionally, a positive response in two or more independent donors is considered indicative of a potential T cell epitope.

Aggregation by size exclusion chromatography

Size-exclusion chromatography (SEC) was carried out using an Agilent 1100 series HPLC with a Superdex 200 Increase 5/150 GL column equilibrated in PBS, pH 7.2 at a flow rate of 0.2 mL/min. Absorbance was measured at 280 nm and the areas of the peaks corresponding to IgG monomer and dimer were measured as a percentage of the total eluted protein. Samples were analysed shortly after preparation and again after incubation at 40°C for 7 days and 14 days.

Thermal stability using fluorimetry and static light scattering

Antibodies were assessed using the Uncle^® Biostability Platform (Unchained Labs, Pleasanton CA, USA) in accordance with the manufacturer’s instructions. Samples were subjected to thermal ramping from 25°C to 95°C at 0.3°C per minute. Intrinsic protein fluorescence and static light scattering (266 nm) were measured as indicators of protein unfolding and aggregation. Each sample was tested at 1 mg/mL in PBS and a sample volume of 9 μL. The whole experiment was carried out three times to obtain triplicate independent temperature profiles. The melting temperature (Tm) was defined as the major inflection point on the fluorescence versus temperature plot, i.e. the temperature as which the rate of change of fluorescence reached a maximum. The software algorithm was unable to reliably determine a Tm from some of the fluorescence traces (even though the profiles were very similar to their replicates) and these were dropped from the analysis. The aggregation temperature (Tagg) was defined as the temperature at which the light scattering started to increase.

Glycosylation

N-glycans were released with rapid PNGase F (NEB Cat. No. P07010) essentially according to the manufacturer’s instructions. The released glycans were fluorescently labelled by reductive amination with 2-aminobenzoic acid (2AA) (42.5 mg/ml) and sodium cyanoborohydride (0.33M), final concentrations, in a sodium acetate/boric acid/tetrahydrofuran buffer at 80°C for 1 hour. The labelled oligosaccharides were purified by HILIC using DPA-6S solid phase extraction cartridges [22]. The purified labelled oligosaccharides were analysed using an Agilent 1260 Quaternary HPLC system with fluorescence detection using a BEH Amide Column, 3.5 μm, 2.1 mm × 150 mm (Waters Cat. No. 186004861) at 30°C. The column was eluted at 0.15 mL/min with acetonitrile:0.525M ammonium acetate pH 3.75:water from 85:10:5 to 50:15:35. A dextran hydrolysate labelled with 2-aminobenzoic acid was used as an external calibrant to calculate glucose unit values for each peak.

Enzyme digests

Test samples (0.5 mg/mL final concentration) at were incubated with proteases for 24 h at 37°C as shown in Table 3. The final volume of each digest was 10 μL.

Download:

Table 3. Enzyme digestion conditions.

https://doi.org/10.1371/journal.pone.0260954.t003

Digestion was stopped by addition of 3.5 μL of E-PAGE loading buffer 1 and the samples were incubated at 70°C for 10 min. 10 μL (3.7 μg protein) was applied to an 8% E-PAGE 48 polyacrylamide gel (Invitrogen Cat. No. EP04808). SeeBlue Plus2 pre-stained molecular weight standards were applied to the reference lanes. The samples were electrophoresed for 25 min on an E-Base electrophoresis device (Invitrogen Cat. No. EB-M03). The gel was stained with 0.03% Coomassie Brilliant Blue R in 10% acetic acid/40% methanol and destained with 8% acetic acid. Images were captured using an E-gel camera (Invitrogen).

The experimental work described in this paper was carried out between Nov 2019 and Jun 2021.

Results

Design and initial screening of immunoglobulin variants

Previous work identified at least 39 different residues in human IgG1 which are relevant for binding to Fcγ receptors [4, 23–28]. Residues between 232 and 239 received particular attention and many of the variants in the clinic include alterations in this area [9–11, 14–18]. With advances in gene synthesis and rapid transient expression, it is now possible to explore a wider range of potential changes. For the initial screens we focused on the high affinity receptor FcγRI because this provided the most sensitive measure of any residual binding activity. Preliminary experiments showed that certain substitutions at 234 and 235 combined with G236K or G236E gave low levels of binding to FcγRI, comparable with LALAPG. When the same substitutions were combined with G236R, binding to FcγRI became undetectable (S7 Table). In subsequent experiments the permissible substitutions at positions 234 and 235 were systematically tested to find a variant that gave the lowest levels of binding to Fcγ receptors without compromising any desirable properties. We avoided residues with structural liabilities (Asn, Cys, Met, Pro) or those which increased the in-silico immunogenicity score (Phe, Trp, Tyr). A panel of culture supernatants from 165 variants was screened for binding to FcγRI S1 Table). From these, 63 variants giving the lowest responses were purified for analysis in more detail in parallel with variants representing previously described mutations. Results from three separate experiments are shown in Table 4.

Download:

Table 4. Binding responses of purified CD20 antibodies binding to human FcγRI.

https://doi.org/10.1371/journal.pone.0260954.t004

The data were compared with those for a LALAPG reference antibody, chosen because it gave the lowest level of binding of the previously available variants. Fifty-six of the new variants gave significantly lower binding than LALAPG and many of them were indistinguishable from the buffer control. In contrast, all of the previously published variants (Samples 2–65, 2–67 to 2–78) and controls (Samples 2–79 to 2–83) gave higher or indistinguishable binding compared with LALAPG. Sample 2–72, which contains mutations designed to enhance binding to FcγRIII [29], gave binding to FcγRI which was comparable to the wild-type reference.

Selected sensorgrams are shown in Fig 1 to illustrate the range of responses seen with previously published variants. The mean responses for those variants are charted in Fig 2.

Download:

Fig 1. Sensorgrams showing binding of variant CD20 antibodies to FcγRI (CD64).

Binding of a fixed concentration (100 μg/mL) of variant antibodies to immobilised receptor was measured by surface plasmon resonance. Sensorgrams were ‘double subtracted’, i.e. the response of a reference cell was subtracted from the response of a test cell to give the sample response and then the response of a buffer control was subtracted. For visualisation of high and low-level responses, the same data are shown at two different scales (Fig 1A and 1B). “New variants” include: L234G/L235S/G236R, L234S/L235T/G236R, L234S/L235V/G236R, L234T/L235Q/G236R and L234T/L235T/G236R, all of which gave baseline responses.

https://doi.org/10.1371/journal.pone.0260954.g001

Download:

Fig 2. Binding of variant CD20 antibodies to FcγRI (CD64).

The binding responses (measured at 283 s) for the selected samples shown in Fig 1 are expressed as a percentage of the binding of wild-type IgG1. Results are the mean ± standard deviation of two independent experiments (Expt 2 and Expt 3 in Table 4).

https://doi.org/10.1371/journal.pone.0260954.g002

Binding to Fc gamma receptors

Five of the new variants giving the lowest binding responses to FcγRI were selected for more detailed analysis on a range of receptors. They were re-expressed in combination with three different Fab regions, corresponding to rituximab (CD20), muromonab (CD3) and alemtuzumab (CD52). As controls, wild-type IgG1, LALA, LALAPG and aglycosyl variants were also expressed with each of the three variable regions. Each antibody was tested in duplicate for binding to each of the Fc receptors. The results are shown in Table 5. Significant binding of LALA, LALAPG and aglycosyl variants to human FcγRI was confirmed for all three antibody specificities. However, only aglycosyl antibodies gave significant binding to mouse and rat FcγRI. LALA variants showed significant binding to both alleles of FcγRIIIa and equivocal weak binding to FcγRIIa, FcγRIIb and FcγRIIIb. In contrast, all of the new variants gave very low binding responses with all of the receptors and were not significantly different from buffer alone.

Download:

Table 5. Binding responses of CD20, CD3 and CD52 antibodies binding to Fc gamma receptors.

https://doi.org/10.1371/journal.pone.0260954.t005

Additional experiments were carried out using an alternative Biacore assay format in which either wild-type IgG1 (Remicade) or variant 2–52 (L234S/L235T/G236R) was captured onto the Biacore chip and soluble Fc receptors were injected. An independently-made panel of receptors comprising: human FcγRI, FcγRIIa H131, FcγRIIa R131, FcγRIIIa V158, FcγRIIIa F158, FcγRIIIb; cynomolgus monkey FcγRI, FcγRIIa, FcγRIIb, FcγRIII; and mouse FcγRI, FcγRIIb, FcγRIII, FcγRIV, was tested to measure the binding of the 2–52 variant. All of the receptors gave positive binding responses of between 20 and 110 RU with the wild-type antibody whereas no significant binding (< 1 RU) was detected with the 2–52 variant.

Fc effector cell activation assays

The ability of anti-CD20 antibodies to engage with cellular Fcγ receptors was measured using the Promega Fc effector cell bioassay reporter system to indicate the potential for antibody-dependent cell-mediated phagocytosis (ADCP) or antibody-dependent cell-mediated cytotoxicity (ADCC). Target cells were Raji, a CD20+ human B lymphocyte cell line and effector cells were Jurkat, a human T cell line with a luciferase gene under the control of a nuclear factor of activated T-cells (NFAT) promoter and transfected with the required human Fc receptor. In each assay, a dose response curve was established using wild-type IgG1 CD20 antibody. Samples of variant antibodies were tested at a final concentration (10 μg/mL) corresponding to maximal response of the control. The results are shown in Table 6. Consistent with the previous binding responses, LALA, LALAPG and aglycosyl variants all gave cellular responses via FcγRI significantly above the negative control and the LALA variant gave significantly elevated responses with all of the receptors. However, all of the new variants gave responses which were indistinguishable from buffer alone.

Download:

Table 6. Activity of CD20 antibodies in Fc receptor effector cell bioassays.

https://doi.org/10.1371/journal.pone.0260954.t006