Transgenic mice (BALB/c background) expressing human CD89 (FcαR1) (32) and knock-in for human IgA (33) were bred and housed at the animal facility of the VU University (Amsterdam, The Netherlands) under standard conditions. Experimental animals were littermates and cohoused. All animal experiments were performed according to the institutional guidelines and procedures. The animal ethical committee of the VUmc approved all animal experiments.

PMNs were isolated from heparinized peripheral healthy donor blood as described before (31). Briefly, cells were separated using lymphoprep (Alere Technologies) density gradient centrifugation after which erythrocytes were lysed using ammonium chloride buffer (155 mM NH₄Cl (Merck), 10 mM KHCO₃ (S.T. Baker) and 0.11 mM EDTA (Sigma). PMNs were washed with phosphate buffered saline (PBS; Fresenius Kabi) and resuspended in RPMI1640 (Gibco) supplemented with glutamine (Gibco), penicillin-streptomycin (Gibco) and fetal calf serum (FCS; Biowest). For NET-release assays, 1% heat-inactivated FCS was added (1% RPMI++), and for the other assays, 10% heat-inactivated FCS was added (10% RPMI++). Isolated PMNs were allowed to rest for 30-60 minutes at 37°C before starting experiments. All donors gave informed consent for collecting blood according to the guidelines of the Medical Ethical Committee of the VU University Medical Center (The Netherlands).

Hybridomas were established from splenocytes obtained from BALB/c mice immunized with recombinant extracellular human CD89 (Sino Biological) and membrane-bound CD89 in HEK293F cell lysates (see Domain and epitope mapping below). Hybridomas producing anti-CD89 mAbs were screened using conventional methods, like ELISA using solid-phase recombinant human CD89 and flow cytometry using human CD89 expressing HEK293F cells as targets, respectively. The blocking potency of anti-CD89 mAbs was evaluated using an in-house ELISA in which serum-derived human IgA (Bethyl Laboratories) binding to solid-phase recombinant human CD89 was measured. Seven hybridomas, which inhibited the binding of human IgA to human CD89 (data not shown), were expanded, and anti-CD89 mAbs (mouse IgG₁) were purified from collected supernatants with protein A, and subsequently used for experiments.

10⁵ cells were added per well in 96-wells round bottom plates (Greiner bio-one, Cellstar) and stained with 10 µg/ml mouse anti-CD89 mAbs or a mouse IgG₁ negative isotype control (Biolegend). Subsequently, PE-labeled goat-anti-mouse IgG antibody (Jackson) was added. After final washings, cells were fixed in PBS/0.1% bovine serum albumin (BSA; Fitzgerald)/2%formaldehyde (37%; Sigma). Binding of anti-CD89 antibodies was determined with flow cytometry (FACS Cyan, Beckman Coulter), and expressed as signal-to-noise (S/N) ratios, ie, dividing measured geometric fluorescence intensity of anti-CD89 antibody by measured geometric fluorescence intensity of mouse IgG1 isotype control.

Latex beads (carboxylate-modified polystyrene, non-fluorescent (0.9 μM) or green fluorescent (1.0 μM); Sigma-Aldrich) were coated with BSA (Akron) or purified serum IgA (MP Biomedicals) as described previously (6). Shortly, beads were resuspended in 2-(N-morpholino) ethanesulfonic (Sigma) buffer with 2 mg/ml BSA (Akron) or serum IgA (MP Biomedicals) in the presence of N-(3-Dimethylaminopropyl)-N’-acid ethylcarbodimide hydrochlorid (Sigma-Aldrich) and incubated O/N at room temperature (RT) (overhead shaker). After washing, beads were resuspended in PBS/0.1%BSA.

For migration experiments cyanogen Bromide-Activated Sepharose beads (4B from Amersham; GE healthcare) were coated with BSA or IgA as described (7, 16). Briefly, 100 mg beads were washed in 1 mM HCl and then resuspended in 0.1M NaHCO3 (Merck)/0.5M NaCl (Fluka) (pH 8.3). 300 µg BSA (Akron) or IgA (MP Biomedicals) was added to the beads and incubated O/N at 4°C. Beads were washed with 0.1 M NaHCO3/0.5M NaCl (pH 8.3) and incubated with 0.5 ml 0.1M Tris (Merck)/HCl (Sigma)/0.5M NaCl (pH 8.0) for 2 hours (h) at RT. Subsequently beads were washed with alternating 0.1 M NaAc (Sigma)/0.5M NaCl (pH 4.0) and 0.1 M Tris/HCl/0.5M NaCl (pH 8.0) for 3 cycli and resuspended in PBS/20% etOH (for storage) or 1/10%RPMI++ (for use).

Isolated PMNs were resuspended in 10%RPMI++. 2x10⁵ cells were seeded per well in 96-wells round bottom plates (Greiner bio-one, Cellstar) and incubated with 20 µg/ml anti-CD89 mAbs or a mouse IgG₁ isotype antibody as negative control (Biolegend) for 20 minutes (min) at 4°C. Green fluorescent latex beads coated with BSA or IgA were incubated with the cells for 30 min at 37°C a cell-to-bead ratio of 1:60. After washing, fluorescence intensity was measured with flow cytometry (BD FACS Calibur, BD bioscience) and phagocytic index was calculated as the percentage of cells that phagocytized, multiplied by the mean fluorescent intensity of bead-positive cells (6).

The release of neutrophil extracellular traps (NETs) was analyzed as previous described (6). 10⁵ isolated PMN cells were seeded per well in 96-wells round bottom plates (Greiner bio-one, Cellstar) and incubated with 20 µg/ml anti-CD89 mAbs or a mouse IgG₁ isotype antibody as negative control (Biolegend) for 20 min at 4°C. Cells were incubated with non-fluorescent latex beads coated with BSA or IgA for 30 min at 37°C at a cell-to-bead ratio of 1:300. After washing, PMNs were resuspended in 1%RPMI++ and transferred to a black 96-wells plate (Greiner). 10 nM Phorbol myristate acetate (PMA; Sigma) was added to the positive control wells and 100 µg/ml DNAse 1 (Roche) to the negative control wells. PMNs were incubated for 3 h at 37°C after which extracellular DNA was detected by adding 2.5 µM SYTOX Green (Invitrogen Life Technologies). Optical density was measured with a fluorimeter (FLUOstar/POLARstar BMG Labtech GmbH) at 480 nm excitation, 520 nm emission.

For 2-D migration isolated PMNs were resuspended in 10%RPMI++ medium in a concentration of 4x10⁶ cells/ml (16). Cells were labeled with 1 µM Calcein AM (green fluorescence; ThermoFisher Scientific) for 30 min at 37°C, after which 2.5x10⁵ cells were seeded per well in 96-wells flat bottom plates (Greiner bio-one, Cellstar) to form a monolayer, and incubated with 20 µg/ml anti-CD89 mAbs or a mouse IgG₁ isotype antibody as negative control (Biolegend) for 20 min at 4°C. IgA and BSA coated sepharose beads were carefully added on top of the monolayer PMNs, and incubated for 40 min at 37°C after which supernatant was removed and beads were washed with PBS to remove unbound cells. PMNs bound to the beads were lysed using HTAB (Hexadecyltrimethylammonium bromide, containing 1 g/L Tween-20 (Sigma), 2 g/L CTAB (Sigma), 2 g/L BSA (Fitzgerald), 7.44 g/L Na-EDTA (Sigma)). Optical density was measured with a fluorimeter (FLUOstar/POLARstar BMG Labtech GmbH) at 485 nm excitation, 520nm emission. The amount of LTB₄ in the supernatant was measured by ELISA (see below).

For ligand binding assays, 96-wells Maxisorp™ plates (Nunc Maxisorp) were coated with 10 µg/ml human IgA (MP Biomedicals) or BSA (Akron). Isolated PMNs were labeled with 1 µM Calcein AM (green fluorescence; ThermoFisher Scientific). After that, 2x10⁵ cells were seeded per well to 96-wells flat bottom plates (Greiner bio-one, Cellstar) and incubated with 20 µg/ml anti-CD89 mAbs or a mouse IgG₁ isotype antibody as negative control (Biolegend) for 20 min at 4°C. Subsequently, cells were transferred to coated Maxisorp™ plates and incubated for 30 min at 37°C, after which supernatant was collected and stored at -20°C. Plate bound cells were washed with PBS and lysed using HTAB. Optical density was measured with a fluorimeter (FLUOstar/POLARstar BMG Labtech GmbH) at 485 nm excitation, 520 nm emission. The amount of lactoferrin in the supernatant was measured with ELISA (see below).

Standard curves with known numbers of lysed calcein AM–labeled neutrophils (0 – 3 x 10⁵ cells/well) were used to quantify to numbers 2-D migrated or IgA (ligand)-bound neutrophils.

The amount of LTB₄ in supernatants from 2-D migration assays was measured with a commercial LTB₄ ELISA kit (Enzo life sciences) according to manufacturer’s protocol.

Lactoferrin in the supernatants of ligand binding assays was measured with an in-house developed ELISA (31). Briefly, Maxisorp™ plates (Nunc Maxisorp) were coated with 50 µg/ml polyclonal rabbit-anti-human-lactoferrin antibodies (Sigma, L-3262) O/N at 4°C and blocked with PBS/0.005% Tween-20/0.5% BSA 1 h at 37°C. Two times diluted supernatants were added to plates and incubated for 1h at 37°C followed by incubation with alkaline phosphatase-labeled rabbit-anti-human-lactoferrin antibodies (1:2500, MP Biomedicals) for 1h at 37°C. After adding the chromogenic substrate P-nitrophenyl phosphate (Sigma), optical density was measured with a microplate reader (Biorad) at 405 nm. Purified human lactoferrin (Sigma) was used as a standard.

The following human CD89 constructs were generated for domain mapping and transiently expressed ( Table 1 ) (1): wild type human CD89 construct, which contained both Ig-like extracellular 1 (EC1) and Ig-like EC2 domains of human CD89 (SEQ 1), and therefore denoted as ‘human EC1-EC2-CD89’ (2). membrane chimeric Ig-like EC1 domain of human CD89 combined with Ig-like EC2 domain of bovine Fcγ2R construct (SEQ 3 and SEQ 4, i.e., combined with bovine transmembrane-intracellular region or with human transmembrane-intracellular region, respectively), and therefore denoted as ‘human EC1-CD89’ (3). membrane chimeric Ig-like EC1 domain of bovine Fcγ2R combined with Ig-like EC2 domain of human CD89 construct (SEQ 7), and therefore denoted as ‘human EC2-CD89’ (4). membrane full-length bovine Fcγ2R construct was also generated, which contained both Ig-like EC1 and Ig-like EC2 domains of bovine Fcγ2R (SEQ 9), and therefore denoted as ‘bovine Fcγ2R’. cDNAs encoding for above-described ‘human EC1-EC2-D89’, ‘human EC1-CD89’, ‘human EC2-CD89’, and ‘bovine Fcγ2R’ constructs were optimized for mammalian expression and synthesized by GENEART, Regensburg, Germany (SEQs 2, 5, 6, 8, and 10, respectively). These cDNAs were subcloned in pcDNA3.1-derived expression plasmids. Detailed information about SEQs is shown in Supplement Figure S1 .

The following human CD89 constructs were generated for epitope mapping and transiently expressed (1): membrane full-length human CD89 construct, which contained the full-length EC1 domain of human CD89 (SEQ 1), and therefore denoted as ‘human EC1-CD89’ (2). membrane chimeric human CD89/cynomolgus monkey CD89 construct (Gln22 – Lys46 from human EC1-CD89) (SEQ 13), and therefore denoted as ‘™Gln22 – Lys46 human EC1-CD89’ (3). membrane chimeric human CD89/cynomolgus monkey CD89 construct (Ile47 – Ile71 from human EC1-CD89) (SEQ 15), and therefore denoted as ‘™Ile47 – Ile71 human EC1-CD89’ (4). membrane chimeric human CD89/cynomolgus monkey CD89 construct (Gly72 – Gly96 from human EC-CD89) (SEQ 17), and therefore denoted as ‘™Gly72 – Gly96 human EC1-CD89’ (5). membrane chimeric human CD89/cynomolgus monkey CD89 construct (Arg97 – Gly121 from human EC1-CD89) (SEQ 19), and therefore denoted as ‘™Arg97 – Gly121 human EC1-CD89’ (6). membrane full-length cynomolgus monkey CD89 construct, which contained the full-length EC1 domain of human cynomolgus monkey CD89 (SEQ 11), and therefore denoted as ‘cynomolgus EC1-CD89’. cDNAs encoding for above-described ‘human EC1-CD89’, ‘™Gln22 – Lys46 human EC1-CD89’, ‘™Ile47 – Ile71 human EC1-CD89’, ‘™Gly72 – Gly96 human EC1-CD89’, and ‘cynomolgus EC1-CD89’ constructs were optimized for mammalian expression and synthesized by GENEART, Regensburg, Germany (SEQs 2, 14, 16, 18, 20, and 12, respectively). These cDNAs were subcloned in pcDNA3.1-derived expression plasmids as well. Detailed information about SEQs is shown in Supplement Figure S2 .

Using the FreeStyle™ 293 Expression System (Invitrogen), FreeStyle™ HEK293F cells (Invitrogen) were transiently transfected with the different constructs. After 48-72 hours, flow cytometry analyses were performed to determine binding of mouse anti-human CD89 antibodies to chimeric human/bovine Fc receptors (domain mapping) and chimeric human CD89/cynomolgus monkey Fc receptors (epitope mapping) on transfected cells.

Transient transfected HEK293F cells were put at 10 x 10⁶ cells/mL in ice-chilled phosphate-buffered saline containing 0.1% BSA (Sigma)/0.05% NaN₃ (PBS/BSA/NaN₃) supplemented with 50 µg/mL human IgGs (blocking Fcγ receptors; Sigma) for 10 minutes at 4°C. Then, 10 μL/tube (i.e., 0.1 x 10⁶ cells) of these cells were incubated with or without 100 μL purified mouse anti-human CD89 antibody at 10 μg/mL (in PBS/BSA/NaN₃) for 30 minutes at 4°C. In parallel, 100 μL purified mouse IgG1 isotype control (BD Biosciences) at 10 μg/mL (in PBS/BSA/NaN₃) was used as a negative control, and 100 μL at 10 μg/mL (in PBS/BSA/NaN₃) purified mouse anti-human CD89 antibody clone MIP8a (BioRad), clone A59 (BD Biosciences), and clone A3 (Santa Cruz Biotechnology) were used as positive controls. After extensive washing in PBS/BSA/NaN_3, cells were incubated with 1:200 diluted PE-conjugated goat anti-mouse IgG Fcγ-specific antibodies (Jackson ImmunoResearch) for 30 minutes at 4°C. After extensive washing in PBS/BSA/NaN₃, cells were fixed in 2% formaldehyde in PBS/BSA/NaN₃ for 30 minutes at 4°C. Binding of antibodies was measured with flow cytometry (BD FACSCalibur; BD Biosciences), and is expressed as signal-to-noise (S/N) ratios, i.e., dividing measured geometric fluorescence intensity of anti-CD89 antibody by measured geometric fluorescence intensity of mouse IgG1 isotype control.

Cryosections of human TMAs (TSP-TMA-001, Tissue Solutions, Glasgow, UK) were kept at −80°C. Tissue sections (5 μm) were fixed with acetone for 10 minutes at RT, air dried, encircled with a hydrophobic pen, washed in PBS and blocked with PBS containing 2% BSA and 10% normal goat serum for 30 min at RT. Tissues were stained with mouse anti-hCD89 IgG1 clone 10E7, 20B4, 30C7 (Polpharma Biologics, Utrecht, The Netherlands) and mouse IgG1 isotype control (Biolegend, 400166). Subsequently, neutrophils were stained with Alexa fluor 488-conjugated mouse IgM anti-CD66b (NovusBio, NB100-77808AF488) and Alexa fluor 488-conjugated mouse IgM isotype control (Biolegend, 401617 AF488). Primary antibodies were diluted to 10,0 µg/ml and incubated overnight at 4°C. Alexa fluor 647-conjugated goat anti-mouse-IgG1 (Invitrogen, A-21236) was used as secondary antibody at 10,0 µg/ml and incubated for 1 hour at RT. Slides were washed, and nuclei were counterstained with 4’,6-Diamidino-2-Phenylindole, Dihydrochloride (DAPI, 1:5000) for 5 minutes at RT and slides were embedded with fluorescence mounting medium.

Stained sections were microscopic scanned with the Vectra^® Polaris™ (Akoya Biosciences) at a 20x magnification. Images were manually analyzed with Phenochart software (Akoya Biosciences). Isotype control staining’s of the mouse IgG1 and mouse IgM were compared with the hCD89 co-staining with CD66b to investigate whether neutrophils were stained with the hCD89 antibodies of Polpharma Biologics.

Transgenic mice expressing human CD89 and knock-in for human IgA were subcutaneously (sc) injected with 10 µl (7mg/ml) human-anti-mouse Collagen XVII IgA (auto)antibodies (Amsterdam UMC and Polpharma Biologics Utrecht) in the right ear and 10 µl PBS in the left ear as control. Injections were performed on day 0, 2, 4, 6, 8, 10 and 12. Mice were treated with 100 µl (1.5 mg/ml) of anti-human CD89 mouse antibody clone 10E7 (n=8, m/f) or a mouse IgG₁ negative isotype control (Biolegend) (n=4, m/f). Antibodies were injected intraperitoneally at day 7 and 11. Mice were monitored daily for discomfort. At the end of the experiment, day 14, mice were sacrificed, ears were excised and subsequently snap-frozen to use for cryosectioning and immunofluorescence staining.

Mice ear cryosections (6 µm) were fixed in acetone for 10 min at RT and air-dried. Then, cryosections were incubated with 400x diluted Alexa Fluor 488-conjugated rat-anti-mouse Ly-6G (GR-1 staining, eBioscience) for 1 hour at RT. After washing in PBS, nuclei were stained using DAPI (Invitrogen) at 1 µg/ml for 5 minutes at RT and embedded with fluorescence mounting medium. Tile scanning to obtain an image of the whole ear was performed using the Vectra^® Polaris™ (Akoya Biosciences) microscope with the following settings: DAPI MSI 0.43ms, FITC 81.70ms and a 20 times magnification. GR-1 staining of cryosections was analyzed with ImageJ/Fiji software. The total area (µm²) of mice ears and the area of the specific GR-1 staining (µm²) was measured. Quantification was calculated as the GR-1 area (µm²) divided by the total area (µm²). The following formula was used to determine the GR-1/total area ratio:

Statistical analysis was performed using graph pad prism 9.1.0. In order to determine the differences in GR-1 area between treated and untreated mCOL17 ears, Mann-Whitney U test was performed. Result was considered significant if p< 0.05.

First, the ability of the developed blocking anti-CD89 mAbs clones to bind CD89 on human neutrophils was analyzed. Isotype control mouse IgG1 was run in parallel as a negative control. Although some variation between the clones was seen, all tested anti-CD89 mAbs had the ability to bind CD89 ( Figure 1A ). Since the developed anti-CD89 mAbs effectively bound CD89 on primary immune cells, next we investigated whether IgA binding on neutrophils was prevented, and IgA-induced neutrophil activation subsequently inhibited.

Several functional assays were performed including phagocytosis, NETs-release, 2-D migration, IgA ligand binding and concomitant lactoferrin release. All anti-CD89 mAbs reduced IgA (coated to latex beads)-mediated phagocytosis and IgA induced NETs-release by CD89 expressing neutrophils, although clone 16D6 was relatively less effective ( Figures 1B, C ). Similarly, 2-D migration of neutrophils towards IgA coated sepharose beads was diminished in the presence of all blocking anti-CD89 mAbs ( Figure 1D ). Furthermore, a comparable pattern was observed for LTB4 release, a neutrophil chemoattractant, which was measured in the supernatant of 2-D migration assays ( Figure 1E ). Lastly, we determined the blocking capacity in a IgA ligand binding assay. The developed anti-CD89 mAbs blocked the binding of neutrophils to plate bound IgA ( Figure 1F ) and concomitant lactoferrin release, a measurement of degranulation, in the supernatant was reduced ( Figure 1G ). Again, clone 16D6 performed poorly in comparison with other mAbs, but collectively, these data show that the developed anti-CD89 mAbs have the capacity to bind CD89 on neutrophils and inhibit IgA-mediated functions.

Based on the functional and epitope mapping studies, clones 10E7, 20B4 and 30C7 were selected for further evaluation, as these clones showed the best capacity to inhibit IgA-induced neutrophil activation and bound to a different epitope compared to MIP8a. Potential cross-reactivity of our lead anti-CD89 mAbs clone 10E7, 20B4 and 30C3 was determined on cryosections of normal human tissues. As expected, all clones bound to neutrophils in lymphoid tissues, which was exemplified by their double staining on CD66B⁺ cells (e.g. in spleen, Figure 3B and in tonsil tissue, see Supplement Figure S3 ). However, we observed unexpected cross-reactivity of 20B4 to e.g. prostate and skeletal muscle tissues, whereas 30C7 cross-reacted to e.g. prostate, cerebellum and skeletal muscle tissues ( Figures 3A, B ). Since this may lead to serious side effects, we concluded that it was unsafe to further develop clone 20B4 and 30C7. Some aspecific background staining of 10E7 was observed in the epidermis, but this was also seen for the isotype control. Similarly, some aspecific staining of both 10E7 and the isotype control was observed in the pituitary, which may have been debris as no co-localization with nuclei was present (see Supplement Figure S4 ). Thus clone 10E7 binding was restricted to CD66B⁺ neutrophils and some tissue macrophages. Therefore, clone 10E7 became our lead mAb for preclinical animal testing.

To investigate whether anti-CD89 mAb clone 10E7 blocks IgA induced neutrophil influx and tissue damage in vivo, we used a preclinical mice model for LABD (18) using in-house bred human IgA knock in x human CD89 transgenic mice (32, 33) ( Figure 4A ). We make use of this mice model since wild type mice do not express (an equivalent receptor for human) CD89 (FcαR1) on their immune cells, precluding functional studies of the role of CD89 in wild type mice. Previous characterization of CD89 transgenic mice showed that human CD89 is expressed on neutrophils, on a subpopulation of monocytes and on macrophages after activation (32). To further mimic the human situation, mice were crossbred with human IgA knock-in mice (32, 33). Injection of human IgA anti-mouse COLXVII (the autoantigen in LABD) in the ears of human IgA knock in x human CD89 transgenic mice resulted in neutrophil influx in the skin (for injection schedule, see Figure 4A ), which was significantly reduced after treatment with anti-CD89 mAb clone 10E7 on day 7 and 11 ( Figures 4B, C ).

These results demonstrated that anti-CD89 mAb clone 10E7 had the ability to reduce experimentally induced human IgA-mediated neutrophil influx in human IgA x human CD89 transgenic mice.