Interactions of the anti-FcRn monoclonal antibody, rozanolixizumab, with Fcγ receptors and functional impact on immune cells in vitro

ABSTRACT

Rozanolixizumab is a humanized anti-neonatal Fc receptor (FcRn) monoclonal antibody (mAb) of the immunoglobulin G4 (IgG4) sub-class, currently in clinical development for the treatment of IgG autoantibody-driven diseases. This format is frequently used for therapeutic mAbs due to its intrinsic lower affinity for Fc gamma receptors (FcγR) and lack of C1q engagement. However, with growing evidence suggesting that no Fc-containing agent is truly “silent” in this respect, we explored the engagement of FcγRs and potential functional consequences with rozanolixizumab. In the study presented here, rozanolixizumab was shown to bind to FcγRs in both protein-protein and cell-based assays, and the kinetic data were broadly as expected based on published data for an IgG4 mAb. Rozanolixizumab was also able to mediate antibody bipolar bridging (ABB), a phenomenon that led to a reduction of labeled FcγRI from the surface of human macrophages in an FcRn-dependent manner. However, the presence of exogenous human IgG, even at low concentrations, was able to prevent both binding and ABB events. Furthermore, data from in vitro experiments using relevant human cell types that express both FcRn and FcγRI indicated no evidence for functional sequelae in relation to cellular activation events (e.g., intracellular signaling, cytokine production) upon either FcRn or FcγR binding of rozanolixizumab. These data raise important questions about whether therapeutic antagonistic mAbs like rozanolixizumab would necessarily engage FcγRs at doses typically administered to patients in the clinic, and hence challenge the relevance and interpretation of in vitro assays performed in the absence of competing IgG.

KEYWORDS:

• FcRn
• neonatal Fc receptor
• rozanolixizumab
• Fcγ receptor
• antibody bipolar bridging

Introduction

The Fc gamma receptor (FcγR) family of receptors mediates a diverse range of functions following their engagement with the crystallizable fragment (Fc) portion of immunoglobulin G (IgG) antibodies.^1,2 The human FcγR family includes both activating receptors (FcγRI, FcγRIIa, FcγRIIIa, FcγRIIIb) and an inhibitory receptor (FcγRIIb) that, together, carefully control the activation status of immune cells. The neonatal Fc receptor (FcRn), expressed broadly on phagocytic leukocytes as well as on non-hematopoietic cells, is also a member of the FcγR family but, uniquely, its primary role is to prolong the circulating half-life of IgG and albumin.^3,4 It does this by specifically binding to these two proteins in the acidic environment of intracellular endosomes following their pinocytosis, which enables their recycling back to the cell surface where they are released into the circulation at neutral pH. This pathway is responsible for the very long half-life of IgG and albumin relative to other plasma proteins and is also the mechanism for preserving the half-life of pathogenic IgG autoantibodies. FcRn is also recognized as a receptor that can mediate transcytosis of IgG across epithelial barriers and transfer of IgG across the placenta.³ More recent data suggest FcRn is a receptor for fibrinogen and members of the echovirus family and can modulate immune complex processing and presentation/cross-presentation by antigen-presenting cells to T cells.^4,5

There has been considerable interest in targeting FcRn as a therapeutic approach in autoimmune diseases driven by pathogenic IgG autoantibodies.^3,4 For example, clinical efficacy has been demonstrated in patients with generalized myasthenia gravis, a prototypic autoimmune disease driven by IgG autoantibodies, with the high-affinity blocking monoclonal antibody (mAb), rozanolixizumab,⁶ and efgartigimod,⁷ a mutated IgG1 Fc (also known as ‘MST-HN’ IgG1 Fc) with enhanced affinity for FcRn over wild type (WT) Fc.⁸ The binding epitope for rozanolixizumab is on the FcRn α chain and overlaps with many of the residues known to be important for the binding of IgG (Fc) to FcRn.⁹ These agents have therefore been engineered to block the Fc binding site, but not the albumin binding site on FcRn, thus minimizing the impact on albumin levels in humans.^6,10,11

Rozanolixizumab was engineered as an IgG4 mAb, a format frequently selected for therapeutic mAbs, due to the intrinsic lower affinity of IgG4 for FcγRs and because it does not engage C1q and therefore inadvertently activate the complement pathway.^12–14 Nevertheless, published data continue to suggest that no mAb (or other Fc-containing construct) is truly “silent” with respect to FcγR engagement, even for aglycosylated mAbs or some variants of other mutated molecules specifically designed to eliminate FcγR binding.^12,15 This study therefore explored the nature and functional consequences of direct FcγR engagement by rozanolixizumab. Overall, the data aim to challenge the relevance and interpretation of in vitro FcγR binding assays performed in the absence of competing IgG.

Materials and methods

Preparation of anti-FcRn antibodies and Fc fragments

Rozanolixizumab was designed as a so-called IgG4P format, the ‘P’ denoting that a serine to proline change at position 241 of IgG4 was introduced to prevent ‘Fab arm exchange’ that can occur with IgG4 mAbs and to allow retention of a stable and functionally bispecific mAb in the circulation. Several batches of rozanolixizumab were used in the studies described herein, but all were prepared in UCB Pharma laboratories and shown to possess >97% monomeric antibody, <3% higher molecular weight species, endotoxin levels <1.0 EU/mg of protein and all material generated was formulated in phosphate-buffered saline (PBS), pH7.5. The following were also prepared at UCB Pharma: ‘rozanolixizumab’ in IgG1 and Fab formats, soluble WT IgG1 Fc, and MST-HN IgG1 Fc from the IMGT database.

Surface plasmon resonance

The binding affinities for the interactions of different antibody formats for FcγRs were determined by surface plasmon resonance (SPR) on a Biacore T200 system using Series S CM5 sensor chips (Cytiva). HBS-EP⁺ buffer (10 mM HEPES pH7.4, 0.15 M NaCl, 3 mM ethylenediaminetetraacetic acid (EDTA), 0.05% Surfactant P20) was used as the running buffer (Cytiva). All the experiments were performed at 25°C. The FcγRs (FcγRI, FcγRIIa, and FcγRIIIa [R&D Systems]) were captured using Tetra-His antibody (Qiagen). Covalent immobilization of the capture antibody (10 mg/mL in 10 mM sodium acetate, pH4.5) was achieved by amine coupling chemistry using AIM for immobilized level to obtain approximately 3000 response units.

WT IgG1 Fc, MST-HN IgG1 Fc, rozanolixizumab IgG4, and rozanolixizumab IgG1 were titrated over the captured FcγRI from 25 nM, whereas WT IgG1 Fc, MST-HN IgG1 Fc, and rozanolixizumab IgG4 were titrated over the captured FcγRIIa and FcγRIIIa from 8 µM (WT IgG Fc) or 16 µM (MST-HN IgG1 Fc and rozanolixizumab IgG4). Initially, 20 start-up cycles consisting of 1 min of 50 mM HCl, 30 s of 5 mM NaOH, and 1 min of 50 mM HCl at a flow rate of 10 µL/min were performed to stabilize the immobilized surface. Each assay cycle consisted of first capturing the FcγR (10 µg/mL) sample using a 1-min injection at a flow rate of 10 µL/min, followed by an association phase consisting of a 2-min injection of the antibody/Fc (FcγRI) or 1-min injection (FcγRIIa and FcγRIIIa) at a flow rate of 30 µL/min, after which the dissociation was monitored. After each cycle, the capture surface was regenerated at a flow rate of 10 µL/min with a 1-min injection of 50 mM HCl, a 30-s injection of 5 mM NaOH, followed by a 1-min injection of 50 mM HCl. A blank flow cell was used for reference subtraction, and buffer blank injections were included to subtract instrument noise and drift.

For FcγRI, kinetic parameters were determined by simultaneous global fitting of the resulting sensorgrams to a standard 1:1 binding model with local Rmax using Biacore T200 Evaluation Software V3.0. For FcγRIIa and FcγRIIIa, affinity values were determined using the steady-state affinity model.

Intravenous immunoglobulin competition of binding to FcγRI-expressing cells

Cell surface FcγR binding studies were performed using HEK293 cells stably transfected with the extracellular domain of the high-affinity receptor, FcγRI. In all cases, background binding was measured on WT HEK293 cells. To allow background binding to be performed in the same well as FcγRI transfected cells, parental cells were pre-stained with calcein violet AM dye (Invitrogen) according to the manufacturer’s instructions. Violet-stained cells were gated as a distinct population using a flow cytometry gating process. Test reagents were pre-labeled with Alexa-647 (ThermoFisher Scientific), and it was confirmed that these were still able to bind FcγRI. To perform the competition binding assay, intravenous immunoglobulin (IVIg) (Gamunex, buffer-exchanged into PBS) and test antibodies were mixed, at the concentrations shown, with cells for 2 h at 4°C. Cells were then washed, and the binding of Alexa-647-labeled antibodies and Fcs was quantified using an iQUE3 (Sartorius) flow cytometer. Data were analyzed using Forecyt software (Sartorius) and mean fluorescence intensity (MFI) was used to represent binding of the test antibodies and Fcs. Binding to untransfected cells was subtracted from binding to FcγRI cells to measure specific binding, and 90% inhibition concentration (IC₉₀) values were obtained using Prism (GraphPad).

Antibody bipolar bridging studies

Peripheral blood mononuclear cells (PBMCs) from healthy donors were isolated through Ficoll-Paque PLUS (GE Healthcare; 1178538) density centrifugation and monocytes enriched from PBMCs using a Percoll (Sigma-Aldrich; 17-0981-02) density gradient and an adherence step of 2 h at 37°C. The adherent monocytes were differentiated into macrophages using 50 ng/mL recombinant human M-CSF (ImmunoTools; 11343112) for 5 d. After 5 d, macrophages were detached and seeded in 96-well imaging plates (Sigma-Aldrich; M0562), stimulated with interferon-gamma (IFNγ; ImmunoTools; 11343536; 2.0 ng/mL) at 37°C and allowed to adhere overnight. Macrophages were then treated with test molecules for the indicated periods in RPMI 1640 medium containing 10% heat-inactivated fetal calf serum (FCS). Following compound incubation, macrophages were surface labeled at 4°C with FITC-conjugated anti-CD32 (BD Biosciences; 555448) and APC-conjugated anti-CD64 (ThermoFisher; 17-0649-42), washed, fixed using 4% paraformaldehyde (PFA) (ThermoFisher; 28908) for 10 min at room temperature, washed, and nuclei labeled with Hoechst 33342 (ThermoFisher; H21492). Labeled macrophages were visualized by confocal microscopy using a Yokogawa CQ1 microscope or Molecular Devices ImageXpress. Images were quantified by automated analysis that segmented cells and quantified the intensity of fluorescence labeling.

For Western blotting experiments, following overnight incubation of human macrophages with rozanolixizumab and other agents, cells were washed in ice-cold PBS and lysed in RIPA Lysis and Extraction Buffer (Pierce 89900) supplemented with Halt Protease and Phosphatase Inhibitor Cocktail (ThermoFisher 78446) at 1 × 10⁷ cells/mL for 15 min on ice with gentle agitation. Cell lysates were cleared by centrifugation at 14,000 rpm for 15 min before being mixed with NuPAGE LDS Sample Buffer (ThermoFisher NP0007) and NuPAGE Reducing Agent (ThermoFisher NP0004) then heated at 70°C for 10 min. Protein was separated by SDS-PAGE using NuPAGE Bis-Tris 4–12% gels (ThermoFisher NP0321) and MOPS SDS running buffer (ThermoFisher NP0001). Separated protein was transferred to a nitrocellulose membrane using an iBlot dry blotting system (ThermoFisher IB301032). The membrane was rinsed in Tris Buffered Saline pH8.0 with 0.1% Tween 20 (TBST) then blocked for 1 h in blocking buffer (5% Nonfat Dry Milk in TBST) and washed three times for 5 min each in TBST before overnight incubation at 4°C with primary antibodies, anti-CD64 (Cell Signaling Technology #90694) and anti-COX IV loading control (Cell Signaling Technology #4850), diluted 1:1000 in TBST containing 5% bovine serum albumin. Membranes were washed three times for 10 min each in TBST and then incubated for 1 h with anti-rabbit-horseradish peroxidase (HRP) secondary antibody (Cell Signaling Technology #7074) diluted 1:1000 in TBST containing 5% Nonfat Dry Milk. Membranes were washed three times for 10 min each in TBST then incubated for 5 min in Luminata Crescendo Western HRP substrate (Millipore). The membrane was imaged using an ImageQuant LAS-4000 Imager (GE Healthcare).

Cytometry time of flight expression and signaling assays

To explore leukocyte signaling potential, the cytometry time of flight (CyTOF) technology was applied, a method that involves interrogation of single cells using antibodies that are conjugated to heavy metal isotopes which are detected using time-of-flight mass spectrometry.

For CyTOF profiling of FcRn expression on different human immune cells, PBMCs were isolated from fresh whole blood donated by healthy individuals using standard density gradient isolation (Leucosep Greiner #227288) and suspended in Maxpar Cell Stain buffer (Standard Bio Tools Inc., 201068). Cells were stained with an optimized panel of antibodies to cell lineage markers (listed in Supplementary Methods Table S1). To detect FcRn expression, an in-house anti-FcRn antibody (UCB1417; prepared in UCB Pharma laboratories) was conjugated as per manufacturer instructions (Standard Bio Tools Inc.). Cells were incubated with the antibody for 1 h on ice, washed twice in Maxpar buffer then resuspended in a DNA intercalator (Standard Bio Tools Inc., #S00093) diluted 1:5000 in 4% PFA. Samples were incubated at 4°C overnight, washed with cell acquisition solution (Standard Bio Tools Inc. #201239) then analyzed on CyTOF2 with Helios upgrade (Standard Bio Tools Inc.).

For CyTOF signaling experiments, human PBMCs were suspended at 1 × 10⁶ cells/mL in serum-free RPMI media (GIBCO) and allowed to rest for 30 min. Stimulation reagents were prepared at 10× working concentration in PBS and 100 µL diluted into 1 mL of cells. Final concentrations of 10 µg/mL used for rozanolixizumab and IgG4 isotype control (BioLegend, #403702) or 1 µg/mL ionomycin (Sigma-Aldrich, #I0634) plus 10 ng/mL phorbol 12-myristate 13-acetate (PMA; Sigma-Aldrich, #P8139). Cells were incubated with treatment for 5, 10, or 30 min at 37°C, alongside unstimulated controls.

After stimulation, cells were fixed through the addition of 16% PFA (ThermoFisher, #11586711) directly into cell suspension. Samples were subsequently washed with Maxpar PBS (Standard Bio Tools Inc., #291058) substituted with 1 mM EDTA (Sigma-Aldrich, #E7889) and then stained with palladium barcodes as per manufacturer instructions (Standard Bio Tools Inc., #S0014) to allow samples to be combined before antibody staining.

The full panel of heavy metal-conjugated antibodies used is shown in Supplementary Table S1. For extracellular staining, the panel of antibodies was diluted into Maxpar buffer and incubated with cells for 1 h at room temperature. Cells were washed twice in Maxpar buffer and then permeabilized for intracellular staining by incubation with ice-cold methanol for 20 min on ice. Intracellular antibodies were diluted into Maxpar buffer and then incubated with cells for 1 h at room temperature. Samples were washed twice in Maxpar buffer and then resuspended in a DNA intercalator diluted 1:5000 in 4% PFA. Samples were incubated at 4°C overnight, washed with cell acquisition solution and then analyzed on CyTOF2 with Helios upgrade (Standard Bio Tools Inc.).

After the acquisition, data were normalized to EQ beads using CyTOF Helios software (v6.7.1014), and cells were assigned back to the original sample based on the Pd barcode using the Standard Bio Tools Inc. Debarcoder App. Individual FCS files were then analyzed using FlowJo (v10.6.0) and R (v4.1.0).

Cytokine production assays

Cytokine release was investigated in multiple in vitro assay formats in several cell types to assess the potential for rozanolixizumab to induce cytokine release; five cell combinations were utilized with four rozanolixizumab delivery formats. The five cell combinations were as follows: 1) Human whole blood; 2) Human PBMCs; 3) Human umbilical vein endothelial cells (HUVECs); 4) Whole blood with HUVECs; 5) PBMCs with HUVECs. The four rozanolixizumab delivery formats were as follows: 1) In solution; 2) Plate immobilized; 3) Cross-linked with an F(ab’)₂ fragment anti-human Fc; 4) Bead-immobilized.

The protocol for studies incorporating HUVECs and whole blood exposed to soluble and cross-linked rozanolixizumab are detailed as follows; the HUVECs were confirmed to express FcRn and several FcγRs, notably FcγRI and FcγRIII but negligible FcγRII (data not shown). HUVECs (Lonza, CC-2517) grown to confluence were displaced with Accutase (Merck, SCR005), washed, and plated at 1 × 10⁴/well in flat-bottom tissue culture plates and incubated at 37°C for 24 h to allow cells to adhere and form a monolayer. 12.5 µL of various antibodies at appropriate concentrations was added to each well followed by 237.5 µL heparinized whole blood obtained from healthy human donors. Where required, rozanolixizumab and control antibodies were cross-linked in solution using an F(ab’)₂ fragment donkey anti-human Fc-specific antibody (Jackson ImmunoResearch, 709-006-098) before the experiment by mixing at a 2:1 Molar ratio and incubating for 20 min at room temperature with gentle, continuous movement to avoid precipitation. For non-cross-linked controls, individual antibodies were mixed with an equivalent volume of PBS. In addition to rozanolixizumab prepared in both IgG4 and IgG1 formats, Campath (alemtuzumab; Genzyme; EU/1/01/193/002 Art: 3231032) and A33 (a tumor-associated antigen) IgG1 and IgG4 isotype controls (prepared at UCB Pharma, stocks PB1205 and PB1204, respectively) were included. Dilutions of the mixtures were eventually made in assays to give final concentrations of 50 µg/mL antibody/16.6 µg/mL anti-human Fc F(ab’)₂ or 5 µg/mL antibody/1.66 µg/mL anti-human Fc F(ab’)₂. Plates were incubated at 37°C for 24 h and then centrifuged at 750g for 5 min and supernatants were frozen at −80°C. Cytokines in the supernatant were measured in a 7-plex pro-inflammatory kit (MSD), measuring the cytokines IFNγ, interleukin (IL)-12p70, IL-1β, IL-6, IL-8, IL-10, and tumor necrosis factor (TNF). The method was as described in the manufacturer’s protocol.

Results

SPR affinity of different antibody formats and Fc fragments for FcγRs are consistent with literature data

The affinity values derived from SPR studies of various Fc-containing molecules for purified FcγRs captured on the Biacore sensor chip are summarized in Table 1. The affinity (KD) values are higher (in the low nM range) for the high-affinity receptor, FcγRI compared to data for binding to FcγRIIa and FcγRIIIa which were in the low µM range. Rozanolixizumab expressed as an IgG1 bound with a higher affinity compared to the therapeutic IgG4. The binding of the MST-HN IgG1 Fc mutant was weaker than WT IgG1 Fc, albeit to a lesser extent for FcγRI. All these data are in agreement with published data.^12,13,16 Note that data for other FcγRs (FcγRIIb and FcγRIIIb) revealed very weak binding, >10–20 µM, for all agents tested.

Table 1. The affinity (KD) of various Fc-containing molecules for FcγRs measured by SPR.

	WT IgG1 Fc	MST-HN IgG1 Fc	Rozanolixizumab (IgG4)	Rozanolixizumab (IgG1)
FcγRI	0.80 nM	1.35 nM	2.44 nM	0.29 nM
FcγRIIa	3.54 µM	>9 µM	5.84 µM	ND
FcγRIIIa	1.30 µM	2.09 µM	>20 µM	ND

FcγR proteins were captured on the Biacore sensor chip and Fc-containing molecules were titrated over to allow the determination of affinity (KD) values. ND, not determined; WT, wild type.

The presence of exogenous IgG (IVIg) abolishes the capacity for all antibody formats to bind to the high-affinity receptor, FcγRI

Many of the published data exploring interactions of IgG entities with FcγRs use culture media lacking exogenous human serum or plasma or purified human IgG. Under physiological conditions, FcγRs, and FcγRI in particular, are often described as being ‘occupied’ by normal, circulating levels of IgG. This means that an Fc-containing therapeutic agent would need to compete for binding to FcγRs. This fact has led to the exploration and discovery of solutions to enhance binding to specific FcγRs and hence enhance effect or function in situations where this is considered to be advantageous for therapeutic effect. However, for many therapeutic antagonistic mAbs, engagement of FcγRs is not deemed to be either required or even advantageous.

Cell binding experiments were conducted on stably transfected HEK-293 cells expressing the extracellular domain of the high-affinity receptor, FcγRI. Competition was assessed in the presence of a titration of human IgG using a therapeutic IVIg preparation (Figure 1). IVIg is prepared from human plasma and typically contains >90% IgG; the major constituent (60%) is human IgG1 with lower levels of IgG2, IgG3, and IgG4. Therefore, these competition binding studies are in effect against IgG1, which accurately reflects the competition faced by rozanolixizumab in humans. IVIg showed a concentration-dependent inhibition of the binding of all agents tested. Based on the data, we were able to calculate the IC₉₀ values (the capacity for IVIg to inhibit the binding to FcγRI by 90%) for each test agent (Table 2). Even at the highest concentration of rozanolixizumab tested, the IC₉₀ was ~20 µg/mL IgG (~130 nM), and Figure 1 indicates that binding was almost completely inhibited at ~100 µg/mL IgG (~660 nM). Data for the other agents tested were broadly similar, although IC₉₀ values tended to be higher for the IgG1 Fc monomer control, which is not surprising given the tighter binding affinity of IgG1 for FcγRI.

Figure 1. Competitive inhibition of the binding of various Fc-containing agents to FcγRI-expressing cells by IVIg.

Competitive binding of Alexa-647-labeled rozanolixizumab (a), MST-HN IgG1 Fc (b), and their respective IgG4 IgG (c) and IgG1 Fc (d) isotype controls with IVIg. Three concentrations of each test agent were tested at 25 nM (red solid circles), 5 nM (green solid squares), and 1 nM (blue solid triangles), and inhibition of binding was compared in the absence of IVIg at 25 nM (red open circles), 5 nM (green open squares) and 1 nM (blue open triangles). The data show MFI and are the mean of n=3 experiments ± SD.

Table 2. IC₉₀ values (μg/mL) for competitive inhibition by IVIg for binding of various Fc-containing compounds to cells expressing FcγRI.

Compound	25 nM	5 nM	1 nM
Rozanolixizumab	23 (21–24)	15 (13–17)	28 (14–42)
MST-HN IgG1 Fc	82 (65–98)	21 (15–27)	7 (6.6–7.6)
IgG1 Fc monomer	280 (206–354)	39 (28–49)	18 (17–19)
IgG4 isotype control	33 (25–41)	12 (10–13)	4 (3.5–5.5)

IC90 values for inhibition by IVIg of binding to FcγRI-expressing cells were calculated at three different concentrations of each test agent (25 nM, 5 nM, and 1 nM). Data are the geometric mean (and minimum/maximum values) of three separate experiments and units are expressed in µg/mL.

Cell surface labeling of FcγRI is reduced on human macrophages by rozanolixizumab treatment in vitro, but this is abolished in the presence of exogenous IgG

Antibody bipolar bridging (ABB) is the capacity through which a mAb binds its membrane antigen through its Fab domain and concomitantly via its Fc region to a membrane Fc receptor.^17–19 Critically, the Fab arm-mediated binding effectively provides a valency boost to the affinity of the Fc for any receptors present. This phenomenon can be especially notable for the weaker interactions, such as IgG2 and IgG4 with low-affinity FcγRs, converting barely detectable binding into measurable and physiologically relevant binding. Given that macrophages express high levels of both FcRn and FcγRs, we explored the potential for rozanolixizumab to induce ABB in vitro. Following the culture of human monocyte-derived macrophages with rozanolixizumab, there was a reduction in surface labeling of the high-affinity FcγR, FcγRI, whereas those of FcγRII were unaffected (Figure 2a,b). Consistent with an ABB mechanism, the reduction in FcγRI appeared to be dependent on both the Fab and the Fc regions, since the Fab version of rozanolixizumab and the MST-HN Fc mutant did not mediate a reduction in labeling (Figure 2c). At higher concentrations of the IgG4 isotype control, some reduction of the FcγRI labeling was evident, potentially a result of a competitive monovalent interaction between the Fc region of the IgG4 and the labeling anti-FcγRI antibody. Nevertheless, in the denaturing immuno-blotting conditions where a competitive interaction with the detection antibody is not expected (Figure 2d), in the presence of the IgG4 isotype control no change in FcγRI levels was detected. Despite the ~10-fold enhanced affinity of the IgG1 version of ‘rozanolixizumab’ (Table 1), a comparison of IgG1 and IgG4 formats showed a very similar ABB profile (Supplementary Figure S1).

Figure 2. Rozanolixizumab specifically reduces the surface labeling of FcγRI on human macrophages, but only in the absence of exogenous human IgG.

(a) Surface expression of FcγRI (red; labeled with APC-conjugated anti-CD64) and FcγRII (green; labeled with FITC-conjugated anti-CD32) on IFNγ stimulated (2.0 ng/mL; overnight) monocyte-derived macrophages (differentiated with M-CSF [50 ng/mL] for 5 d), and treated with vehicle (PBS), rozanolixizumab or IgG4 isotype control (1.0 µg/mL), for 4 h. Cells were fixed and labeled with Hoechst (cyan) before image acquisition by confocal microscopy. Images are maximum intensity projections from consecutive confocal z-stacks of representative fields of view. (b) Quantification of FcγRI and FcγRII for macrophages as described in (A) treated with rozanolixizumab, IgG4 control, or rozanolixizumab Fab for 4 h. Points represent mean and error bars = SEM of normalized data from three independent experiments. (c) FcγRI expression on macrophages as described in (A) treated with rozanolixizumab (10 µg/mL), rozanolixizumab Fab (10 µg/mL), IVIg (100 µg/mL), ET904 (IgG4 isotype control; 10 µg/mL) or MST-HN IgG1 Fc mutant (10 µg/mL). Data show mean and individual data points from triplicate cultures. (d) SDS-PAGE/Western Blotting analysis of FcγRI in macrophages following culture with rozanolixizumab. (e) Macrophages were cultured with the indicated concentrations of IVIg and rozanolixizumab for 72 h followed by quantification of FcγRI surface expression. The heat map shows the staining intensity of FcγRI.

To determine the fate of FcγRI, we performed SDS-PAGE/Western blotting of macrophages following incubation with rozanolixizumab, and a reduction in FcγRI immunoreactivity was observed, but not following an incubation with rozanolixizumab Fab, which is consistent with internalization and degradation of FcγRI induced by ABB (Figure 2d). While, hypothetically, the binding of rozanolixizumab might have been enhanced by ABB through an avidity effect (albeit with a monovalent interaction via the Fc with FcγRI), similar to flow cytometric analysis of rozanolixizumab binding to FcγRI described above, competition by IgG (using a therapeutic IVIg preparation) prevented the reduction in FcγRI labeling evoked by rozanolixizumab. Furthermore, the reduction was completely inhibited at the sub-physiological IgG concentration of ~100 µg/mL (~660 nM) (Figure 2e). Taken together with the flow cytometric analysis of rozanolixizumab binding to FcγRI, we present evidence supportive of the idea that, under physiological conditions, rozanolixizumab does not affect ABB of FcγRI and hence, by logical inference, of the lower affinity FcγRs in the blood of rozanolixizumab-treated subjects.

We have not performed a detailed analysis of the re-expression kinetics of FcγRI on monocyte-derived macrophages, although we did not observe recovery within a 2-h incubation period (Supplementary Figure S2). However, this may reflect a need to re-apply either differentiation cytokines or an additional stimulus. Finally, seeding macrophages more sparsely did not reduce the impact of rozanolixizumab on FcγRI (Supplementary Figure S2) suggesting trans binding of rozanolixizumab was unlikely to account for the FcγRI down-regulation.

Leukocyte signaling events are unaffected by the engagement of rozanolixizumab

FcRn expression in blood generally appears to be more prominent on leukocyte subsets capable of phagocytosis,^4,20 and our data confirmed this to be the case using the CyTOF methodology (Supplementary Figure S3). Expression was highest on monocytes and myeloid dendritic cells (mDCs), followed by neutrophils and basophils, whereas FcRn expression on B-cell and T-cell subsets and natural killer (NK) cells was low. These cells express other members of the FcγR family to a variable degree.

CyTOF phosphoflow analyses were performed on human PBMCs, investigating a broad panel of signaling components on several leukocyte subsets, including monocytes, NK cells, and DCs. The key data from monocytes are shown in Figure 3, quantifying pAKT, p38, pERK, and pMAPKAPK2 signaling at different time points and incorporating PMA/ionomycin as a positive control. These cells were capable of being stimulated, as evidenced by increased signal using PMA/ionomycin, but rozanolixizumab did not exert an impact on signaling through any component at any time point. Similar conclusions were reached based on the data from NK cells and DCs (data not shown).

Figure 3. Rozanolixizumab does not induce signaling in human monocytes.

CyTOF analysis showing (a) representative histogram plots from a single donor showing metal intensity of pAKT signal in monocytes after stimulation and (b) line plots of median metal intensities in monocytes for four phosphoproteins downstream of FcγR signaling (p38, pAKT, pERK, and pMAPKAPK2) from three donors after stimulation.

Rozanolixizumab does not induce cytokine release from leukocytes

Data published previously,⁹ using human PBMCs exposed to soluble or cross-linked rozanolixizumab, showed no propensity to induce the release of the pro-inflammatory cytokine, IFNγ. These data have now been extended to a broader range of assay formats using five human cell combinations (whole blood, PBMCs, HUVECs, whole blood with HUVECs, PBMCs with HUVECs) and four rozanolixizumab delivery formats (in solution, plate immobilized, cross-linked with a F[ab’]₂ anti-human Fc antibody, bead-immobilized rozanolixizumab). The data from studies incorporating HUVECs and whole blood exposed to soluble and cross-linked rozanolixizumab are shown in Figure 4, where the release of the pro-inflammatory cytokines, IL-6 and TNF, were assessed. The anti-CD52 mAb, Campath (alemtuzumab), was used as a positive control in these experiments and was shown to induce cytokine production both in soluble and cross-linked formats, but rozanolixizumab did not induce such a response, even after cross-linking. There was also no cytokine release using an IgG1 format of rozanolixizumab, which might be expected to bind to FcγRs with a higher propensity than an IgG4. The same conclusions were drawn from experiments conducted under the other conditions described above, where a broader range of cytokines were measured (IL-6, IL-8, IFNγ, IL-1β, IL-12p70, IL-10, and TNFα) and where other suitable positive controls (anti-CD3 and Fc multimers) were shown to be capable of inducing cytokine release (data not shown).

Figure 4. Rozanolixizumab does not induce the release of IL-6 (a) or TNFα(b) from whole blood/HUVEC cultures.

Human whole blood was added to confluent HUVEC cultures (as described in Methods and Materials) in the presence of soluble or anti-human Fc cross-linked antibodies (rozanolixizumab [RLZ] in IgG4 and IgG1 formats, Campath or appropriate isotype control mAbs) for 24 h at 37°C before supernatants were harvested for subsequent cytokine quantification. Data are mean ± SD from two determinations each from two separate donors.

Discussion

IgG4 is known to have negligible binding to the first complement component, C1q, and accordingly does not activate the complement pathway. However, although WT human IgG4 is widely considered to be “silent” in relation to the engagement of FcγRs on immune cells, some natural, albeit generally weak, binding is retained. Specifically, IgG4 can bind to FcγRI moderately well, but very weakly to FcγRIIa, FcγRIIb, and FcγRIIIa, and not at all to FcγRIIIb.^12,13,21 IgG4 binding to FcγRIIb is thought to be physiologically relevant, as IgG4 is seen to be the class-switching sub-class achieved in chronic inflammatory situations. We have therefore established the extent of rozanolixizumab binding to FcγRs in this study.

FcγRI is the sole ‘high-affinity’ receptor (KD in the nM range), whilst all the others are considered to have a low affinity (KD in the µM range) to IgG4. Hence, stimulation of FcγR-bearing immune cells typically requires receptor crosslinking or clustering, such as via an IgG immune complex, or by high levels of IgG decoration on target pathogens (‘opsonization’). The presented SPR data showed that rozanolixizumab, expressed in an IgG4 format, had a relatively low affinity for the high-affinity receptor, FcγRI, whilst the affinity of rozanolixizumab expressed in an IgG1 format is around 10 times stronger, aligning with previous studies comparing IgG1 and IgG4.^12,13,22 The KD of MST-HN, an IgG1 Fc fragment specifically mutated to enhance its affinity for FcRn,⁸ was somewhat weaker than WT IgG1 Fc, which is also in keeping with data from previous studies.¹⁶ The trends observed in this study were broadly similar for binding to FcγRIIa and FcγRIIIa, although KDs were in the low μM range.

Most assays used to investigate the interaction of different IgG isotypes with FcγRs are performed in the absence of exogenous human IgG, which competes for binding to FcγRs. We performed binding assays to FcγRI expressed on cells in the presence of a titration of human IgG (as an IVIg preparation) and showed that binding of rozanolixizumab is effectively abolished in the presence of IgG, even at the low concentrations that would be present in humans treated long term with rozanolixizumab. To clarify, optimal maintenance of anti-FcRn treatment in humans generally leads to a sustained reduction of IgG levels by around 60–80% from baseline levels,^10,11,23 such that circulating IgG levels would be in the region of 2–4 mg/mL under these conditions; these concentrations exceed those shown to prevent binding events in the current study. Previous studies have shown that low concentrations of exogenous IgG can effectively compete and abolish the binding of a defined IgG entity to FcγRs.^21,24 Additionally, IVIg can significantly suppress IgG-IC-driven functional responses in vitro,²⁵ questioning whether therapeutic IgG4 mAbs such as rozanolixizumab are likely to bind to FcγRs in vivo after their administration. As already mentioned, this consideration would not apply to mAbs that have been deliberately engineered for enhanced FcγR engagement and effector function, such as to induce depletion of target cells.²⁶

ABB (or The Kurlander effect) is a long-established phenomenon whereby an antibody binds to both a membrane antigen through its Fab domain and concomitantly via its Fc region to a membrane Fc receptor,^17,18 which can lead to a block in FcγR binding.¹⁹ In addition, antigen binding may increase the affinity of IgG1 or IgG2 mAbs for FcγRs.²⁷ Since FcRn is co-expressed with FcγRs on monocytes or macrophages,^20,28 we confirmed potential ABB in our study, as evidenced by a reduction in FcγRI labeling, following incubation with rozanolixizumab in the absence of exogenous human IgG. We presume that this occurs after FcRn engagement via its Fab domain and FcγRI binding through its Fc domain, leading to loss of cell surface FcγRI by an ‘innocent bystander’ effect following normal internalization of FcRn. This presumption is supported by the Western blotting data showing the degradation of FcγRI after rozanolixizumab (Figure 2d). In some regards, FcRn is an unusual antigen to mediate ABB since its cell surface expression is relatively low,²⁰ and early work suggested surface density of the Fab-binding antigen was an important factor in this phenomenon.¹⁹ However, the overall recycling pool of FcRn may be higher,^29,30 and consequently ABB could internalize FcγRI. It is also relevant to question the valency of binding between rozanolixizumab and FcRn, given that the mAb is bivalent and, in theory, could bind two FcRn molecules on the cell membrane. However, data in both the human FcRn-transgenic mouse and Cynomolgus monkeys comparing monovalent and multivalent anti-FcRn formats⁹ strongly suggest that rozanolixizumab binds in a monovalent manner in vivo, possibly due to the low receptor density already discussed. In the in vitro monocyte-derived macrophages, we do not believe there is significant ‘unmasking’, as we see degradation of FcγRI (Figure 2b) and no clear increase in FcγRI labeling following a wash and 2-h re-incubation (Supplementary Figure S2). The kinetics of FcγRI re-expression are likely to depend on the differentiation status of the cells, and exposure to subsequent stimuli that upregulate FcγRI. This could be via type I interferon signaling or toll-like receptors³¹ but, given this interaction is unlikely to occur in the presence of physiological concentrations of IgG, we consider that this down-regulation is unlikely to occur in vivo, and so the IgG4 backbone is unlikely to impact surface levels of FcγRI.

FcγRI is upregulated in numerous inflammatory disease states and often correlates with increased disease activity, and is capable of mediating various inflammatory functions in vivo and being particularly pertinent in the context of immune complex-driven disorders.^32–36 Blocking FcγRI has been proposed as a therapeutic approach,³² and down-modulating the function of the activating FcγRs using so-called Fc ‘multimers’ is receiving a lot of attention as a potential therapeutic approach in autoimmune and chronic inflammatory diseases.^21,37 Monovalent engagement or occupancy (i.e., in the absence of cross-linking) of FcγRI is expected to occur naturally with physiological concentrations of IgG, and so a monovalent interaction with FcγRI leading to a reduction of FcγRI expression with rozanolixizumab might therefore be expected to play an anti-inflammatory role. However, given that low levels of exogenous IgG prevent rozanolixizumab ABB in vitro, presumably by preventing the Fc portion of the mAb binding to FcγRI, it seems unlikely that this would occur under physiological conditions. Nevertheless, individual pharmacodynamic responses or tissue-specific differences in IgG reduction could impact the relevance of this mechanism to clinical response.

To further establish the functional consequences of rozanolixizumab binding, cell signaling experiments were performed on leukocytes known to have high FcRn expression (e.g., monocytes, DCs). These cells showed no propensity for rozanolixizumab to induce signaling; importantly, these studies were conducted in a serum-free medium to avoid any confounding potential for competition by exogenous IgG. This is not unexpected given that the C-terminal (intracellular) domain of FcRn, whilst containing motifs associated with endocytosis and trafficking, lacks immunoreceptor tyrosine-based activation or inhibitory motifs.³⁸ In a further series of experiments that explored the potential for induction of cytokine production in a range of human cellular assay formats, we established that rozanolixizumab was completely inert with respect to cytokine production, consistent with the results from signaling experiments.

Our data strongly suggest that interactions of the anti-FcRn mAb, rozanolixizumab, with FcγRs can occur under defined conditions in vitro, but that the functional consequences are likely to be limited, given the fact that even low levels of exogenous IgG inhibit such events. Nevertheless, the functional co-operation of FcRn with FcγRs, in particular FcγRIIa, within endosomal compartments in leukocytic cells has recently been proposed,⁵ enhancing our existing understanding of how FcRn may modulate key functions, such as immune complex processing, presentation and cross-presentation to T cells, and induction of cytokine production.^5,39,40 In the future, it will be highly relevant to explore the impact of FcRn blockade on such pathways.

We also considered how our findings relate to the situation with other therapeutic mAbs. The fact that doses and dosing intervals vary considerably between different therapeutic mAbs makes extrapolation difficult, although Cmax levels for a range of (non-FcRn) mAbs in humans were measured in the 10–60 µg/mL range after their subcutaneous administration⁴¹, which is very low compared to endogenous IgG levels (10–12 mg/mL).

There was only around a 10-fold reduction in FcγRI binding affinity and no tangible differences in the functional studies between IgG4 and IgG1 formats of rozanolixizumab, indicating that IgG1 could be a viable format for this agent, although C1q binding and function (a key feature of IgG1 mAbs) or impact on expression of other FcγRs were not assessed in this study. As already discussed, FcRn is highly endocytic, and cell surface levels of the receptor are low, such that any engagement of FcγRI appears to simply lead to the down-regulation of FcγRI in the absence of immune cell activation. However, for a highly expressed target that is stabilized or clustered at the cell surface, investigation of ‘completely’ silenced Fc formats may be warranted.¹⁵

In conclusion, the anti-FcRn mAb rozanolixizumab was shown to bind to FcγRs in both protein-protein and cell-based assays, and the kinetic data were broadly as expected based on published data for an IgG4 mAb. The presence of exogenous IgG was able to inhibit such binding events, suggesting that therapeutic mAbs do not necessarily engage FcγRs at doses typically administered to patients in the clinic. Furthermore, data from in vitro experiments using relevant cell types that express both FcRn and FcγRI indicated no evidence for functional sequelae in relation to cellular activation events (e.g., intracellular signaling, cytokine production) upon either FcRn or FcγR binding of rozanolixizumab.

Abbreviations

ABB	=	antibody bipolar bridging
CyTOF	=	cytometry time of flight
DC	=	dendritic cell
EDTA	=	ethylenediaminetetraacetic acid
FcγR	=	Fc gamma receptor
FcRn	=	neonatal Fc receptor
FCS	=	fetal calf serum
HUVEC(s)	=	human umbilical vein endothelial cell(s)
IFNγ	=	interferon-gamma
IgG	=	immunoglobulin G
IVIg	=	intravenous immunoglobulin
mAb(s)	=	monoclonal antibody(ies)
MFI	=	mean fluorescence intensity
NK	=	natural killer
PBMC(s)	=	peripheral blood mononuclear cell(s)
PBS	=	phosphate-buffered saline
PFA	=	paraformaldehyde
RPMI	=	Roswell Park Memorial Institute
SPR	=	surface plasmon resonance
TNFγ	=	tumor necrosis factor-alpha
WT	=	wild type

Author contributions

OQ wrote the manuscript, designed the experiments, performed the experiments, and analyzed, interpreted, and critically reviewed the data; ES, RB, SW, RC, GM, AM, and GC designed the experiments, performed the experiments, and analyzed and interpreted the data; NB, SR, and BS designed the experiments and analyzed and interpreted the data; AS wrote the manuscript and analyzed, interpreted, and critically reviewed the data; DH analyzed, interpreted, and critically reviewed the data; all authors reviewed and approved the manuscript.

Acknowledgments

The authors acknowledge the following scientists at UCB Pharma, Slough, UK, for the preparation of protein reagents used in the studies described: Geofrey Odede, Helen Brand, Sue Cross, and Shirley Peters. We also thank Veronica Porkess from UCB Pharma for editorial review and assistance. Editorial support was provided by James Pickford and Jo Cook, Ogilvy Health UK, and funded by UCB Pharma, in accordance with Good Publication Practice 2022 (GPP2022) guidelines (https://www.ismpp.org/gpp-2022).

Disclosure statement

All the studies reported here were funded by UCB Pharma. All authors except for NB & OQ were employees of UCB Pharma when all studies were carried out: OQ was an employee of Celentyx Ltd when some of the experiments were conducted and NB was an employee at Celentyx Ltd throughout the study; GM is now an employee of AstraZeneca UK; GC is now an employee of Dark Blue Therapeutics, and AM is now an employee of Exscientia.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/19420862.2023.2300155

Additional information

Funding

The work was supported by UCB Pharma.