Development and characterization of agonistic antibodies targeting the Ig-like 1 domain of MuSK – Scientific Reports

Antibody isolation, production and selection

Two types of monoclonal MuSK Ig-like 1 domain antibodies were used in this study. The first set was isolated from a MuSK MG patient as described previously²⁴. In short, patients with MuSK MG were recruited in our MG outpatient clinic at the Leiden University Medical Centre (LUMC) and were selected based on the presence of a positive MuSK antibody test (RSR Ltd). The study was conducted in accordance with the Declaration of Helsinki and was approved by the local medical ethics committee (The Medical Ethics Committee Leiden The Hague Delft) and patients signed informed consent. To avoid immunogenicity in future preclinical development, the antibodies were germlined and derisked (effector functions knocked down and sequence optimized)²⁶, resulting in 20 variants of the original five clones. An overview of the clones, the variants introduced and their functional characteristics is given in Supplementary Table S1. The second set of antibodies resulted from libraries generated from two llamas that were immunized with recombinant human MuSK-Fc chimera protein (R&D Systems, cat. 9810-MK). RNA from peripheral blood lymphocytes was reverse transcribed to cDNA and the Fabs then cloned into a phagemid vector as described in de Haard et al.²⁷ and used for phage display selection against recombinant different domains of MuSK to isolate anti-domain MuSK specific Fabs (SIMPLE antibody™ platform of argenx). The variable domains of both types of antibodies (patient and llama-derived) were reformatted as a full human immunoglobulin G1 (IgG1) equipped with the Leu234Ala/Leu235Ala (LALA) mutations to knock down the antibody effector functions²⁶ and deletion of the C-terminal lysine (delK) to avoid heterogeneity. All antibodies were formulated in PBS with 0.02% Tween-80. Endotoxin levels were determined at Eurofins (cut-off value of < 0.1 EU/mL). The llama-derived antibodies were further selected for their binding features on mouse and human recombinant MuSK. All antibodies were produced by transient transfection in HEK293-E 253 cells (obtained from ImmunoPrecise Antibodies) and purified by MabSelect SuRe™ LX beads and size exclusion chromatography to remove potential aggregate of antibodies.

Patient polyclonal IgG(4) was derived from MuSK MG patients undergoing therapeutic plasmapheresis. For the use of the plasmapheresis material, patients signed informed consent (no permission was required from an ethical committee for the use of this “waste” material). The different plasmapheresis batches per patient were transported to Immunoprecise Antibodies Ltd, pooled and purified for IgG4 using a 35 mL CaptureSelect-hIgG4 XK26/20 column (Pharmacia Biotech). The concentrated IgG4 was shipped back to the LUMC for storage at − 20 °C and further use. The patient plasmapheresis batches that were selected for IgG4 purification had highly MuSK reactive IgG4 (tested in a MuSK enzyme-linked immunosorbent assay [ELISA], data not shown) and were of a large volume. For polyclonal patient total IgGs, plasmapheresis material was purified on a 200 mL CaptureSelect IgG total XK20/50 column (Pharmacia Biotech) at the LUMC and stored at − 20 °C for further use.

Affinity measurements

Anti-human Fc antibody (Jackson ImmunoResearch, cat. 109-005-098) was diluted in 10 mM sodium acetate pH 4.5 (Cytiva cat. BR100350) and immobilized on flow channels Fc1 and Fc2 on a CM5 chip series S (Cytiva cat. 29149603) using amine coupling chemistry (amine coupling kit Cytiva, cat. BR100050) on a Biacore™ T200 SPR device (Cytiva cat. 28975001). HBS-EP + pH 7.4 (diluted in MilliQ H₂O from a 10 × stock HBS-EP + pH 7.4 [Cytiva, BR100826]) was sterile-filtered and used as running buffer at a flow rate of 30 µL/min. Antibodies of interest were captured at around 150 RU in flow channel Fc2 with a contact time of 30 s at pH 7.4 and a flow rate of 10 µL/min, followed by a stabilization time of 30 s. Next, the antigen was injected as a 3 point 1 over 4 dilution series over flow channels Fc1 and 2 starting from 10 mM applying an association time of 5 min, a dissociation time of 10 min and a flow rate of 30 µL/min. Regeneration was done by injection of 10 mM glycine–HCl pH 1.5 (Cytiva cat. BR100354). Double referenced sensorgrams were fitted using a 1:1 binding model to obtain the on-rate ka (1/Ms), off-rate kd (1/s) and the affinity KD (M).

Epitope mapping

The respective human Fabs of the monoclonal antibodies of interest were produced in HEK293 cells (obtained from ImmunoPrecise Antibodies) and were coated on a Microlon half-area ELISA plate (Greiner cat. 675061) diluted at a concentration of 1 µg/mL in 1 × PBS and incubated overnight at 4 °C. The next day the ELISA plate was washed 3 times with 1 × PBS-0.05% Tween20 pH 7.4 and blocked with 100 µL/well 1% casein in 1 × PBS, followed by an incubation of 1 h on an ELISA plate shaking platform at room temperature. Recombinant Human MuSK His-tag Protein (R&D Systems, 10189-MK-MTO) diluted to 1 µg/mL in 1 × PBS 0.1% casein was added to each well and was incubated 1 h at room temperature on an ELISA shaking platform. After washing each well with 1 × PBS-0.05% Tween20 pH 7.4 mAbs of interest were diluted in 1 × PBS 0.1% casein at a concentration of 1 µg/mL for 1 h at room temperature while shaking. After 1 h the plates were washed with 1 × PBS-0.05% Tween20 pH 7.4 and detection antibody/well was added (Peroxidase affinipure goat anti-human IgG Fc fragment specific, Jackson ImmunoResearch 109-035-098, diluted 1 over 5000 in 1 × PBS 0.1% casein) and was incubated at room temperature while shaking. After 1 h the plates were washed with 1 × PBS-0.05% Tween20 pH 7.4 and TMB (Merck cat. CL07-1000ML) was added and color development was monitored. Reaction was terminated by adding 0.5N H₂SO₄ (ChemLab cat. CL052615) and OD was measured at 450 nm (reference 620 nm).

C2C12 culturing

C2C12 mouse myoblast cells (Cell Lines Service) were cultured in DMEM GlutaMAX (31966 Gibco) with added 10% fetal bovine serum and 1% penicillin/streptomycin (proliferation medium) until 60–70% confluency. For differentiation into myotubes the cells were plated on 10 cm dishes (Greiner Bio-One, 664160, 58 cm², MuSK phosphorylation assay) or on 96-wells plates (Greiner µclear, 655090, 0.32 cm²/well, AChR clustering assay), at 1.25 × 10⁴ cells/cm² in proliferation medium. Myoblast fusion into myotubes was induced after reaching 90–95% confluency (~ 2 days), by culturing in differentiation medium consisting of 2% horse serum, 1% penicillin/streptomycin and in DMEM GlutaMAX™. Medium was changed every 2 days. After 5 days the myoblasts were fused into myotubes. Cells were kept at 37 °C with 5% CO₂ during culturing.

MuSK tyrosine phosphorylation assay

C2C12 myotubes on 10 cm dishes were stimulated in different experimental conditions using 10 ng/mL agrin (Rat c-term agrin 3,4,8 R&D550AG) and/or 1.1 µg/mL mAbs (mono or bispecific) and/or 2.25 mg/mL patient-purified IgG-total (Patient (Pt)#1) for 30 min at 37 °C, unless otherwise specified. Due to a batch concentration problem, 13-3D10wt was not included in the MuSK tyrosine phosphorylation (MuSK-P) assay. To obtain quantitative MuSK-P data, myotubes were extracted in lysis buffer (30 mM triethanolamine, 1% NP-40, 50 mM NaF, 2 mM Na-orthovanadate, 1 mM Na-tetrathionate, 5 mM EDTA, 5 mM EGTA, 1 mM N-ethylmaleimide, 50 mM NaCl, 1 × protease inhibitor cocktail, 1 × phosphatase inhibitor cocktail) followed by centrifugation (5000 g). To precipitate MuSK, the lysate normalized for protein content was incubated using 5 µL/sample rabbit anti-MuSK polyclonal serum (ab94276 or ab94277, a kind gift by Prof. Steve Burden, NYU Medical School) or using 1 µg/sample 11-3F6 IgG1 anti-MuSK antibody at 4 °C overnight. Using protein A agarose beads (11134515001, Roche), the bound antigen–antibody complexes were captured. After 5 min incubation at 95 °C with sample buffer containing reducing agent dithiothreitol (DTT), the bead-precipitated proteins were run on SDS-PAGE gel electrophoresis and transferred to a polyvinylidene difluoride membrane. Goat anti-rat MuSK (R&D AF 562) and mouse anti-pTyr clone 4G10 (Millipore 05-321) were used as primary antibodies for the detection of MuSK protein and phosphorylated MuSK. As secondary antibodies for the fluorescent signal donkey anti-mouse-680RD (926-68072, Licor) and donkey anti-goat-800CW (926-32214, Licor) were used. Fluorescence intensity of bound antibodies was detected using the Odyssey CLx (Licor). Tyrosine phosphorylation experiments were repeated three times and normalized for the effect of agrin, unless otherwise stated.

AChR clustering assay

AChR clustering in myotubes was induced for 24 h with varying experimental conditions using 10 ng/mL agrin and/or 1.1 µg/mL mAbs (mono or bispecific) and/or 2.25 mg/mL Pt#1 MuSK-reactive total IgG. In therapeutic experiments, the inhibiting bispecific IgG4 or patient IgG was incubated for 30 min before adding the agonist antibody and finalizing the remaining 23.5 h incubation. The experimental investigator was blinded for the conditions. Twenty-four hour incubation was followed by incubation with Alexa Fluor 488-conjugated α-bungarotoxin (B13422, Thermo Fisher) to stain AChRs and Hoechst 33342 (H1399, Thermo Fisher) to stain nuclei, for 30 min at 37 °C. The cells were fixed in 4% w/v paraformaldehyde for 10 min at room temperature and then used for imaging. Microscopy was performed with 20 × magnification on a Leica AF6000 fluorescence microscope. Four visual fields per well and five wells per condition were selected, based on evenly spread and clearly visible mature myotubes in the brightfield channel. Analysis of AChR-positive cluster count and area size were performed using ImageJ (1.52n) and MATLAB (R2018b). The number of mature clusters (≥ 15 µm²) per condition and its distribution of all cluster sizes were quantified. Only mature clusters were included for the cluster count analysis, as they represent the most functionally relevant clusters^28,29,30. AChR clustering distribution graphs were produced by deriving a fitted line from a histogram containing all quantified cluster area sizes in MATLAB. AChR clustering experiments were repeated 3 times and normalized for the effect of agrin, unless otherwise stated.

Passive transfer studies in mice

Mice

The five different passive transfer tolerability or MuSK MG mouse studies described in this paper were conducted at three different independent institutions. For clarity, an overview of the study design and outcomes can be found in Supplementary Table S2 and Supplementary Figure S1. Mouse studies in C57BL/6J mice were conducted at the laboratory of Christopher Borg (Besançon, France) (Supplementary Fig. S1a), at the Jackson (JAX) Laboratory (Bar Harbor, ME, USA) (Supplementary Fig. S1b) and at the Leiden University Medical Center (LUMC; Leiden, The Netherlands) (Supplementary Fig. S1c). The tolerability (Supplementary Fig. S1d) and MuSK MG (Supplementary Fig. S1e) studies in NOD/SCID mice were also conducted at the LUMC. C57BL/6JIco mice used in the LUMC study were purchased from Charles River (Wilmington, MA, USA) and those used in the Besançon study from Charles River (Domaine des Oncins, L’arbresle Cedex, France). Mice used in the JAX study were bred locally at the Jackson Laboratory. Original breeders of immunodeficient NOD.CB17-Prkdcscid/J (NOD/SCID) mice were also purchased from Jackson laboratory. All mice were approximately two months old at the start of the studies. Sterilized food and drinking water were provided ad libitum. NOD/SCID mice were housed in sterile individually ventilated cages. All experiments were carried out according to Dutch law and Leiden University guidelines, including approval by the local Animal Experiments Committee (Instantie voor Dierenwelzijn). The mouse experiments at the Besançon and Jackson Laboratory were performed based on French (The Animal Experimentation Ethics Committee) and American (Institutional Animal Care and Use Committee) law and ethics guidelines.

Dosing and assessments

To test the tolerability of the Ig-like 1 candidate antibodies in mice, we used 25 NOD/SCID mice at the LUMC. After measuring body weight and establishing baseline values for the in vivo neuromuscular tests during 6 days, the mice were injected intraperitoneally every 3 days with 20 mg/kg of Ig-like 1 binding MuSK agonist antibodies 13-3D10a, 13-4D3a, 11-3F6c or 11-3D9b (all antibodies generated contained the same IgG1-LALA-delK backbone), or were left untreated for 25 days. As positive control we included a monospecific version of the Ig-like 1 binding antibody 13-3B5 (produced with a hIgG4S228P Fc tail in CHO cells), known to be able to induce myasthenia in NOD/SCID mice²⁵. Each treatment group consisted of 2 male and 2 female mice, except for the untreated group which included 2 male and 3 female mice. The clinical signs test consisted of daily measurements of body weight and neuromuscular performance in grip strength and hanging time on an inverted mesh (maximum 180 s, 3 attempts)¹¹.

At Besançon, a total of 36 C57BL/6 mice (18 male, 18 female) were injected intraperitoneally every 3 days for 8 weeks in a blinded fashion with 20 mg/kg or 5 mg/kg of antibodies 13-4D3, 11-3F6c, or 11-3D9b or 20 mg/kg of 13-3D10a, MuSK Fz domain binding antibody mAb13 (benchmark, Genentech, known MuSK agonistic antibody, positive control) or Mota (RSV binding antibody, irrelevant binding domain, negative isotype control) in the hIgG1-LALA-delK backbone (C-terminal lysine was removed genetically to avoid heterogeneity). Each treatment group included 2 male and 2 female mice.

At the Jackson laboratory, C57BL/6 mice (N = 20) received intraperitoneal injections with 0.03, 0.1, 0.3, 1 or 5 mg/kg 11-3F6c or Mota (2 males, 2 females per treatment group) every 3 days for 5 weeks.

At the LUMC, a total of 24 C57BL/6 mice (12 males, 12 females) were injected every 3 days in an investigator-blinded fashion with either 11-3F6c (n = 18), or Mota (n = 6). Six mice (3 males and 3 females) were euthanized after 5, 8 and 11 weeks of 11-3F6c injections consecutively, whilst 6 Mota mice were euthanized after 12 weeks of injections, following endpoint analyses.

For passive transfer of patient purified IgG4 in the MuSK MG mouse model, the minimal dose inducing a maximal myasthenic phenotype (40 mg/kg patient 2017-004 polyclonal IgG4) was administered to 2-month-old female NOD/SCID mice daily. One day prior to injection of IgG4, mice were injected every 3 days with either 2.5 mg/kg 11-3F6c (n = 6), or Mota (n = 6) in a blinded fashion. The daily IgG4 dose was dissolved in either sterile PBS with Mota or sterile PBS with 11-3F6c on the days of planned agonist administration. Before the start of injections, (daily) body weight and baseline values for neuromuscular performance were established of each mouse, and continued until humane endpoint. If body weight loss of ≥ 20% occurred, compared with the starting weight, mice were euthanized by CO₂ inhalation (i.e. humane endpoint). Blood samples for analyzing serum titers were taken through tail vein puncture shortly before each injection of 11-3F6c (and IgG4).

Repetitive nerve stimulation electromyography

Repetitive nerve stimulation electromyography (RNS-EMG) was conducted at the LUMC on C57BL/6 and NOD/SCID mice from the Besançon and LUMC studies as described. Mice were anaesthetized with a 1.5:1 (v/v) mixture of ketamine hydrochloride (Nimatek; 100 mg/mL, Eurovet) and medetomidine hydrochloride (Domitor; 1 mg/mL, Pfizer), at 1.25 μL/g mouse body weight, adjusted with Ringer solution to 200 μL volume and administered intraperitoneally. Mice were maintained at 37 °C on a heating pad. A grounding needle electrode was inserted subcutaneously in the right thigh. Stimulation needle electrodes were inserted near the sciatic nerve in the left leg thigh. Subcutaneous recording needle electrodes were inserted near the calf muscles of the left leg. Grounding and recording electrodes were coupled via an AI402 pre-amplifier to a Cyberamp-380 signal conditioner (Axon Instruments/Molecular Devices). The nerve was stimulated supramaximally from a computer-controlled programmable electrical stimulator (AMPI). Trains of 20 stimuli were applied at increasing frequencies of 40 Hz, with a 30 s pause between trains. Compound muscle action potentials (CMAPs) were digitized using a Digidata 1440 interface and Axoscope 10 (Axon Instruments/Molecular Devices). Peak–peak amplitudes were determined in Clampfit 11 (Axon Instruments/Molecular Devices). After completing the recordings, mice were euthanized by CO₂ inhalation without recovery from anesthesia and muscles were dissected for the studies described below.

C57BL/6 mice from the JAX study were analyzed for neuromuscular transmission using RNS-EMG at the Jackson laboratory using a similar method as described above, except mice were anesthetized with isoflurane for up to 5 min with 2–3% isoflurane in O₂. In addition, for RNS, stimulation intensity was increased to 120% of the intensity required to record the maximum CMAP. One train of 20 stimuli was administered at 40 Hz, while the CMAPs were recorded. Decrement upon RNS was expressed as the percentage of decrease of the CMAP from the 1st stimulus of each train.

Ex vivo muscle contraction studies

Muscle contraction was analyzed in dissected muscles from NOD/SCID and C57BL/6 mice (from the LUMC studies only). Contraction force of left phrenic nerve-hemidiaphragms was recorded in Ringer’s medium containing (in mM): NaCl 116, KCl 4.5, CaCl₂ 2, MgCl₂, NaH₂PO₄, NaHCO₃, glucose, pH 7.4) at room temperature (20–22 °C) with a force transducer (type K30, Harvard Apparatus, Hugo Sachs Elektronik GmbH), connected to an amplifier TAM-A 705/1 (Hugo Sachs Elektronik). The signal was digitized using a Digidata 1440 digitizer (Axon Instruments/Molecular Devices), connected to a PC running Axoscope 10 (Axon Instruments). The phrenic nerve was stimulated supramaximally once every 5 min with 280 stimuli of 100 µs duration at 40 Hz, i.e. for 7 s. The safety factor of neuromuscular transmission was assessed by measuring contraction force before and after equilibration with 125 nM d-tubocurarine (Sigma-Aldrich). The area under each contraction curve was determined in later off-line analyses, using Clampfit 11 (Axon Instruments).

Fluorescence microscopy of NMJs

Morphology of NMJs was analyzed in diaphragms of all surviving mice in all studies and in the epitrochleoanconeus (ETA) muscle for the two LUMC studies. Diaphragm strips (right hemidiaphragm, most dorsal area) and ETA muscles were fixed in 1% paraformaldehyde in PBS for 1 h, washed in PBS and incubated for 2.5 h with 1 µg/mL Alexa Fluor 488 conjugated α-bungarotoxin to stain AChRs, followed by PBS wash (1 h), all at room temperature. Muscles were whole-mounted and viewed under an epi-fluorescent microscope (Zeiss Axioskop), using identical settings for all muscles. Low-magnification overview pictures (5 × objective), containing about 150 NMJs, were taken using Axiovision software (Zeiss). The integrity of the AChR area at NMJs was semi-quantitively scored by the investigator (blinded to the treatment) either as ‘normal’ (i.e. a clear, continuous pretzel-like structure without much fragmentation of AChR clustering) or ‘abnormal’ (i.e. fragmented and/or faint AChR clusters). The proportion of abnormal NMJs was calculated by dividing the number of NMJs scored as abnormal by the total number of NMJs evaluated in the picture.

Statistical tests

One-way-ANOVA with Dunnett’s multiple comparison test, (multiple) unpaired t-tests with Holm-Šídák or Bonferroni corrections for multiple testing, or a Log-rank (Mantel-Cox test) were performed wherever appropriate. Differences with P values < 0.05 were considered statistically significant. GraphPad Prism 9.3.1 was used for statistical analyses and to test data sets for normality.

The studies reported in this paper were conducted according to the ARRIVE guidelines (https://arriveguidelines.org).