Edited by: Dimitrios C. Mastellos, National Centre of Scientific Research Demokritos, Greece
Reviewed by: Trent M. Woodruff, The University of Queensland, Australia; Maartje G. Huijbers, Leiden University Medical Center, Netherlands
*Correspondence: Eva-Maria Nichols,
This article was submitted to Molecular Innate Immunity, a section of the journal Frontiers in Immunology
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Myasthenia Gravis (MG) is mediated by autoantibodies against acetylcholine receptors that cause loss of the receptors in the neuromuscular junction. Eculizumab, a C5-inhibitor, is the only approved treatment for MG that mechanistically addresses complement-mediated loss of nicotinic acetylcholine receptors. It is an expensive drug and was approved despite missing the primary efficacy endpoint in the Phase 3 REGAIN study. There are two observations to highlight. Firstly, further C5 inhibitors are in clinical development, but other terminal pathway proteins, such as C7, have been relatively understudied as therapeutic targets, despite the potential for lower and less frequent dosing. Secondly, given the known heterogenous mechanisms of action of autoantibodies in MG, effective patient stratification in the REGAIN trial may have provided more favorable efficacy readouts. We investigated C7 as a target and assessed the
The complement system plays an important role in innate and adaptive immune response and tissue homeostasis and is broadly involved in both common and rare diseases (
The 2007 FDA approval of Eculizumab (Soliris, Alexion Pharmaceuticals), an anti-C5 monoclonal antibody, for the treatment of Paroxysmal Nocturnal Hemoglobinuria (PNH), presented a significant milestone in the field of complement therapeutics (
In 2017, Eculizumab was also approved for the treatment of refractory Myasthenia Gravis (MG), an autoantibody-mediated disease characterized by disrupted cholinergic transmission due to decreased numbers of acetylcholine receptor (AChR) at the neuromuscular junction (NMJ), resulting in weakness and lack of muscular control (
Here, we describe the
All animal studies were ethically reviewed and carried out in accordance with Animals (Scientific Procedures) Act 1986 and the GSK Policy on the Care, Welfare and Treatment of Animals. Transgenic mice expressing a human V-gene repertoire, were immunized with human C7 protein purified from normal human serum (Complement Technology, Inc). B-cells enriched from the spleen and lymph node tissues were used for hybridoma fusion and direct single B-cell sorting. For the sorting cells were stained with a combination of fluorescently labelled antibodies against B cell markers. Memory and plasma blast B cells were labelled with B220-PECy7, IgM-BV605 and CD43-FITC, plasma cells with B220-PECy7 and CD138-PE. Contaminating cells were excluded by gating out cells positive for CD3, CD93, CD11c, Ter-119 and Gr1, all conjugated to PerCPCy5.5. Antigen specific B-cells and CD138+ plasma cells were single cell sorted using the BD FACS Aria III (Becton Dickinson). To identify C7 binding memory or plasma blast cells the cells were incubated with a biotinylated version of the human protein isolated from human plasma. C7 binding was visualized using streptavidin-PE and streptavidin-APC. cDNA was synthesized from the sorted B-cells and used for V-gene amplification by PCR. Cognate VH and VL chains were then cloned into the Adimab yeast-based platform (Adimab, LLC) and clonal yeast populations with concomitant HC and LC expression and human C7 binding were isolated by FACS, expressed and purified. Antibody clones derived from both the hybridoma and B-cell sorting methods were characterized and selected based on their binding affinity, inhibitory potency in the classical pathway hemolysis assay and epitope diversity.
Affinity maturation libraries were built for the selected, functional anti-C7 antibodies by diversifying each of the complementary determining regions (CDRs) 1, 2, and 3 of the heavy- and light-chain variable region (VH and VL) genes. Random mutations restricted to the CDRs were introduced by splice-overlap-extension (SOE) PCR using degenerate oligonucleotides synthesized with mixtures of nucleotide bases with a bias towards the wild-type nucleotide. Antibodies were selected from the affinity maturation libraries using the Adimab platform according to the protocols developed by Adimab, LLC. Human C7 protein was biotinylated
The monoclonal antibodies were obtained by transient expression in HEK293-6E cells. Supernatants were collected after 10 days, sterile filtered and the antibodies affinity purified using MabSelect SuRe columns on the Akta Xpress system. Purity and integrity of the purified mAbs was confirmed by analytical size exclusion chromatography and SDS-PAGE. Endotoxin levels were confirmed to be below 0.75 EU/mL using an EndoSafe endotoxin reader (Charles River Laboratories).
Surface plasmon resonance experiments were performed on a Biacore 8K instrument (Cytiva) using HBS-EP+ (Teknova) at pH 7.4 as a running buffer. For the SPR chip, Protein A (Pierce) was immobilized on a CM5 chip (Cytiva) using a Biacore amine coupling kit (Cytiva) according to the manufacturer’s instructions. Multi-cycle kinetics experiments were run as follows. Antibodies were diluted in running buffer to 0.5μg/ml and captured on the sample flow cells of all channels at a flow rate of 10 μL/min for 60s. Antigen was flowed over both flow cells of all channels at a flow rate of 30μl/min for 180s followed by a 600s dissociation time. The chip surface was regenerated between cycles using 50mM NaOH for 30s at a flow rate of 30μL/min. Antigen concentrations used were 0, 0.39, 1.56, 6.25 and 25nM. The data was analyzed in the Biacore analysis software to a 1:1 binding model using local Rmax and double referencing. Off-rates fitted as slower than 1x10-5 1/s and too slow to be fitted reliably were manually adjusted to 1x10-5 1/s. Antigens used were human complement C7 (Complement Tech) and cyno complement C7 purified from cynomolgus serum (Seralab).
Sheep erythrocytes (TCS Biosciences) were prepared for assay with gentle washing in CFD (Complement Fixation Diluent, Oxoid Ltd) followed by sensitization with a complement fixation antibody (Amboceptor, Siemens Healthcare) for 30min at 37°C. Control antibodies included were an in-house anti-C5 and disabled anti-C5 antibody (both isotype matched with test mAbs, human Fc-disabled IgG1κ) for human serum and a mouse anti-C7 mAb (Quidel), mouse IgG1κ was used for the rat serum, in addition to the isotype control for the test antibodies. There was no cyno cross-reactive positive control available at the time of the experiments, therefore only the test antibody isotype control was included. Serial dilutions of test antibodies, along with positive and negative control antibodies as above, were prepared in CFD and 50μL added to the wells of a 96-well U-bottomed assay plate (Greiner Bio-One). Pooled normal human serum (or rat serum/cyno serum) was diluted in CFD to a concentration previously determined to elicit 80% lysis of sheep erythrocytes and 50μL were added to the wells of the assay plate containing the serially diluted test molecules. 50μL of diluted serum was also added to wells containing an equal volume of CFD to determine maximum complement induced lysis. The plate was then incubated on ice for 30min. Following incubation, 50μL of sensitized sheep erythrocytes were added to all wells of the assay plate and also to wells containing 100μL of water only (to give 100% lysis control) and wells containing 100μL of CFD only (to give 0% lysis control). The plate was then incubated at 37°C for 30 min to allow complement mediated lysis of the cells to take place. Following lysis, the plate was centrifuged at 500g for 3 min and supernatants transferred to a Maxisorp 96-Well Flat-Bottom Assay Plate (Nunc). Absorbance was measured at 405nm on a Molecular Devices SpectraMax M5 plate reader. The percentage hemolysis was calculated as = 100 x (Test sample at 405nm - A405nm 0% lysis control)/(A405nm 100% lysis control - A405nm 0% lysis control). Dose response curves for the test and control antibodies were derived in GraphPad Prism (v 5.01) using a four-parameter logistic curve fit and EC50 values determined.
Hydrogen Deuterium Exchange Mass Spectometry (HDX-MS) was used to determine the C7 binding epitopes of TPP1657, TPP1653 and in the case of TPP1820, the parental molecule TPP1651 was included as TPP1820 was derived later through affinity maturation. Based on our previous experience the epitope in daughter clones does not change as a result of our affinity maturation process, protection patterns tend to be more pronounced. Full experimental details, method and supporting experimental data are located in
BLI experiments were performed on an OctetRed384 instrument (Fortebio) using phosphate buffer saline IgG free (PBSF) buffer. Anti-mouse capture (AMC) biosensors (Fortebio) were equilibrated in PBSF (10 min shaking 200xg) then 133nM anti-C6 monoclonal antibody (Quidel) or 133nM mouse IgG1,κ isotype control (GlaxoSmithKline) were loaded onto the AMC biosensors (600s). Following a PBSF wash step (30 seconds), 100nM of C5b6, C7, C8 and C9 (Complement Technologies) were added to the AMC biosensors sequentially (time varied between 300-800s depending on the protein being assessed). PBSF only controls were included for each protein addition step of forming the complex and represent the negative controls. To determine whether the anti-C7 mAbs prevent C5b6:C7 or C7:C8 interactions binding of C7 in the presence or absence of 2nM or 200nM anti-C7 antibody to previously captured C5b6 biosensors was evaluated (800 seconds). Subsequent steps then sequentially assessed the ability of C7, C8 or C9 complement proteins to bind to the AMC biosensors (800s). In all experiments a PBSF buffer only control, representing a reference biosensor, was included. To subtract background noise, values for the reference biosensor was subtracted from the sample biosensor values. When elucidating anti-C7 antibody method of action, all traces were aligned to the beginning of the addition of C7 step.
To measure
All animal studies were ethically reviewed and carried out in accordance with Animals (Scientific Procedures) Act 1986 and the GSK Policy on the Care, Welfare and Treatment of Animals.
Maxisorp plates (Nunc, Loughborough, UK) were coated with rabbit anti-rat C9 IgG (in house; 10µg/ml in bicarbonate buffer, pH 9.6) at 4°C overnight; wells were blocked (1h at 37°C with 3% BSA in PBS), washed once in PBS containing 0.05% Tween 20 (PBS-T). Standard curves of zymosan-activated normal rat serum (serial dilution series staring at 1 in 5) and rat serum samples diluted 1 in 50 in 1% BSA-PBS, 20mM EDTA were added in duplicate and incubated overnight at 4°C. Wells were washed with PBS-T 3X, then incubated (2h, room temperature (RT)) with monoclonal 12C3 anti- rat/mouse C5b-9 (in house; 5µg/ml in PBS-T). After washing, wells were incubated (1h, RT) with biotinylated donkey anti-Mouse IgG (1:2000 in PBS-T; Jackson ImmunoResearch #715-005-150). After washing, Streptavidin-HRP (1:2000 in PBS-T; R&D Systems, # 890803) was added and incubated 30 minutes at RT. After washing, plates were developed using o-phenylenediamine dihydrochloride (SIGMAFAST; Sigma-Aldrich, St Louis, MO) and absorbance (492 nm) was measured. Protein concentrations (Units/ml) of serum samples were automatically calculated by reference to the standard curve using GraphPad Prism (GraphPad, La Jolla, CA, USA).
The assay system comprised optimally antibody-sensitized sheep erythrocytes (ShEA) incubated with rat serum (NRS) dilutions in HEPES-based buffer supplemented with Ca2+ and Mg2+ cations (assay buffer) (SOP in (
Hemolytic activity in IgG-depleted NHS was tested as described above but using complement fixation diluent (CFD) (Oxoid Ltd.) rather than HBS buffer. The percentage hemolysis was calculated as = 100 x (A405 nm test sample - A405 nm 0% lysis control)/(A405 nm 100% lysis control - A405 nm 0% lysis control). The NHS hemolysis dose-response curves were plotted in GraphPad Prism v7 and analyzed using a four-parameter logistic curve fit to calculate the EC50 concentrations.
AChRαβ & AChRδϵ plasmids: cDNAs for the α1/β1 and δ/ε subunits of human nAChR were cloned into the dual expression vectors pBiCIH and pBiCIN respectively. pBiCIH and pBiCIN are hygromycin and neomycin resistant derivatives of the bi-cistronic vector pCIN5 (
Rapsyn plasmid: The pCIP4 plasmid backbone (made internally) contains an IRES element upstream of the puromycin resistance cassette, allowing stable selection of cell lines. The Rapsyn gene (NM_005055) was synthesized at GenScript, then subcloned into pCIP4 using NotI and BamHI restriction sites.
ARPE-19 CD46, CD55, CD59 KO cells were available as a suitable tool cell line as they are susceptible to complement attack, without the need of blocking antibodies to the above complement regulatory proteins. The cells were cultured in DMEM/F12 + GlutaMAX medium (Invitrogen), supplemented with 10% HI-FBS (Invitrogen), 15mM HEPES (Invitrogen), and 1X MEM-NEAA (Invitrogen) (complete culture medium). Generation of these cells is described in
The cells were blocked following transfection with 200μL/well of complete culture medium + 1% BSA (staining buffer) for 15 – 30 min at 37°C, 5% CO2, high humidity, then stained with 50μL/well Alexa Fluor™ 647 conjugated α-Bungarotoxin (AF647 α-Bungarotoxin; Invitrogen) at 2.5μg/ml in staining buffer and incubated for 30 min at 37°C, 5% CO2, high humidity. The cells were washed twice with 200μL/well of PBS and fixed for 10 min at RT with 50μL/well of 10% Formalin (Sigma), washed twice with 200μL/well of PBS and stained with a solution, containing 10 Units/ml Alexa Fluor™ 488 Phalloidin (actin stain, Invitrogen) and 1μg/ml Hoechst 33342 (nuclear stain, Invitrogen). The cells were incubated for 30min at 37°C, 5% CO2, high humidity, and washed 2x with 200μL/well PBS. PBS was added to the wells at 100μL/well and images were captured on an InCell 2200 or InCell 6000 imager (GE Healthcare), using a 10X objective. Images were analyzed using Columbus v2.8 (Perkin Elmer). The following image analysis algorithm was used. The fluorescence from the DAPI channel (Hoechst 33342 stain) was used to find the nuclei. The fluorescence from the FITC channel (Alexa Fluor™ 488 Phalloidin stain) was then used to identify the cytoplasm and draw the cell perimeter. Spots of clustered surface AChR were identified within each cell, based on the AF647 α-Bungarotoxin (Cy5) fluorescence, and various spot parameters were calculated. The sum of the spot fluorescence per cell was used to set the threshold for identifying “AChR Pos Cells”. The % positive cells was calculated by dividing the number of positive cells by the total number of cells, multiplied by 100. The data was plotted in GraphPad Prism v7.05.
Myasthenia Gravis patient plasma samples were purchased from Tissue Solutions (Glasgow, UK). These patients were confirmed as anti-nicotonic AChR positive based on a autoanitobdy titre determined by immunoassay (see
Pooled control plasma was prepared from blood obtained from healthy volunteers at GSKs internal blood donation unit. 100mL of EDTA anti-coagulated blood was obtained per donor from six donors in total. The blood was centrifuged for 10 min at 2000xg, 4°C and the plasma supernatants were transferred to clean 50mL falcon tubes which were centrifuged again for 5 min at 2000 x g, 4°C. All plasma supernatants from all donors were pooled into a sterile reservoir, aliquoted into sterile Eppendorf tubes, and stored at -80°C.
Pooled normal human serum (NHS) was prepared from blood obtained from healthy volunteers at GSKs internal blood donation unit. Blood was collected from 10 healthy volunteers in serum separator tubes (S-Monovette® 7.5ml Z-Gel, Sarstedt) and processed to preserve complement activity. The blood was left to clot at RT for 20 – 30min, then the tubes were immersed in ice water to contract the clots and centrifuged for 10min at 2000xg at 4°C. The serum supernatants from all tubes were pooled into a 50 ml Falcon tube. The tube was centrifuged for 5min at 2000xg, 4°C. The supernatant was transferred into a fresh reservoir, aliquoted and frozen immediately at -80°C.
The human biological samples described were sourced ethically and their research use was in accord with the terms of the informed consents under an IRB/EC approved protocol.
The Anti-Acetylcholine Receptor ELISA (IgG) kit from Euroimmun (catalogue number EA 1435-9601 G) was used for quantification of patient autoantibody titers in-house. The plasma test samples were thawed at RT and diluted 1:26 in sample buffer. The 1:26 sample dilutions were considered as neat samples on the standard curve, and controls were also diluted 1:26. Some of the samples which had higher titers were further diluted and dilution factor from this step was used when calculating the sample concentrations from the standard curve. The absorbance was read at 450 nm and 650 nm on a PHERAstar FSX. The data analysis was carried in the PHERAstar FSX MARS Software v3.20 R2: the blank and 650 nm values were subtracted from all wells and a standard curve was plotted and analyzed using a four-parameter logistic curve-fit. The unknown sample concentrations were calculated using the four-parameter logistic equation and multiplied by the dilution factor (if applicable). Each sample was tested in three independent experiments and the mean of the three experiments was calculated.
The pooled NHS was depleted of immunoglobulins using Protein A/G agarose (Thermo Scientific) in order to remove natural antibodies that may bind to the cells and cause non-specific MAC deposition. 15mL Protein A/G agarose resin was used per 3.5mL NHS. The resin was washed three times with 35mL PBS and placed on ice. The NHS was thawed in a water bath at 37°C, mixed with the resin and incubated for 30min in a 4°C cold cabinet on a roller mixer. The NHS/resin suspension was centrifuged for 2min at 1000xg, 0°C. The supernatant was aliquoted in Eppendorf tubes and immediately frozen at -80°C. Except for the assay validation data, all MG experiments reported used Ig-depleted NHS as source of complement.
To confirm whether immunoglobulin depletion was successful, NHS and Ig-depleted NHS were tested for IgG, IgA, and IgM levels using a commercial immunoassay kit (Human/NHP Isotyping Panel 1 kit, Meso Scale Discovery). NHS samples were assayed at dilutions ranging from 1/20 – 1/1,562,500 (1/5 serial dilution intervals) according to the manufacturer’s instructions. The data was analyzed using the MSD Discovery Workbench Software v4.0.12. The background (Blank values) was subtracted from all samples/standards (ECL Signal - Blank). A standard curve was generated by plotting the “ECL Signal – Blank” values against the concentration and analyzed using a four-parameter logistic curve-fit, including a 1/Y2 weighting function. The unknown sample concentrations were calculated using the four-parameter logistic equation and multiplied by the dilution factor where applicable.
The transfected cells were blocked with 200μL/well of complete culture medium + 1% BSA and a 1/40 dilution of TruStain FcR block (Biolegend) for 30-60min at 37°C. MG and control plasma samples were thawed and incubated in a heat-block for 30-45min at 56°C to heat-inactivate complement. The heat-inactivated samples were diluted 1/10 in complete culture medium and added to the blocked wells at 50μL/well. The plates were incubated for 30min at 37°C, 5% CO2, 95% humidity. The wells were washed twice with 200μL/well of PBS and stained with 2.5μg/ml AF647 α- Bungarotoxin, combined with 20μg/ml of either Cy3 Donkey Anti-human IgG, Fcγ specific, or Cy3 Donkey Anti-human IgM, Fc5μ specific, or Cy3 Donkey Anti-human IgA, α chain specific (all Jackson ImmunoResearch). The plates were incubated for 30min at 37°C, 5% CO2, high humidity and washed twice with 200μL/well of PBS. The cells were fixed for 10min at RT with 50μL/well of 10% Formalin (Sigma), then washed and stained with 1μg/ml Hoechst 33342 (nuclear stain) + 10 Units/ml AF488 Phalloidin (actin stain) for 30min at 37°C, 5% CO2, high humidity. The wells were then washed, filled with 100μL/well PBS and imaged on an InCell 6000 imager (GE Healthcare), using a 10X objective. Images were analyzed using Columbus v2.8 (Perkin Elmer). The following image analysis algorithm was used. The fluorescence from the DAPI channel (Hoechst 33342 stain) was used to find the nuclei. The fluorescence from the FITC channel (Alexa Fluor™ 488 Phalloidin stain) was then used to identify the cytoplasm and draw the cell perimeter. Spots of clustered surface AChR were identified within each cell, based on the AF647 α- Bungarotoxin (Cy5) fluorescence, and various spot parameters were calculated. The sum of the spot fluorescence per cell was used to set the threshold for identifying “AChR Pos Cells” (Cy5) and “Anti-human Immunoglobulin Pos Cells” (dsRed). The % positive cells was calculated by dividing the number of positive cells by the total number of cells, multiplied by 100. The same calculation was used for the % negative cells but using the number of negative cells. To correct for non-AChR specific immunoglobulin binding, the % Immunoglobulin positive cells within the AChR negative cell population were subtracted (background corrected). The average background-corrected values from three independent experiments (except for MG patient 1 where there was only sufficient sample for one experiment) were plotted in GraphPad Prism v7.03, where samples were classed into the following four categories, based on the average percentage of positive cells: between 0% - 15% (-), between 15% - 40% (+), between 40% - 65% (++), between 65% - 100% (+++). Statistical analysis was carried out in GraphPad Prism v7.05, using a repeated measures 1way ANOVA without correction, with Dunnet’s multiple comparison test, comparing the means (n=3 experiments) of each plasma sample to the mean of the pooled control plasma sample. For MG Patient 1 (n=1 experiment, 2 replicates) the mean of the two replicates was compared to the mean of the two pooled control plasma replicates from the same experiment using an unpaired t-test.
The transfected cells were blocked with 200μL/well of complete culture medium + 1% BSA and a 1/40 dilution of TruStain FcR block for 30-60 min at 37°C. MG and control plasma samples were thawed and incubated in a heat-block for 30-45 min at 56°C to inactivate complement. The heat-inactivated samples were diluted 1/10 in complete culture medium. A 10μg/ml mAb35 (Rat anti-AChR tool Ab, Genetex) solution was also prepared in culture medium. The blocking solution from the plates was aspirated and the diluted MG plasma samples or controls were added to the cells and incubated for 30min at 37°C, 5% CO2, high humidity. The wells were washed twice with 180μL/well of PBS and the anti-C7 blocking antibody or the corresponding isotype control antibody (both at 20μg/ml in culture medium) added to the appropriate wells at 50μL/well. Culture medium alone was added to control wells. Ig-depleted NHS aliquots were quickly thawed in a water bath at 37°C and placed on ice prior to dilution to 30% in culture medium immediately before addition to the plates (50μL/well). Culture medium alone was added to control wells. The plates were incubated for 2.5-3h at 37°C, 5% CO2, high humidity, washed twice with 180μL/well of PBS and stained with 50μL/well of 2.5μg/ml AF647 α-Bungarotoxin + 10μg/ml polyclonal Rabbit anti-C5b9 antibody (Abcam) in staining buffer (culture medium + 1% BSA) for 30 min at 37°C, 5% CO2, high humidity. The wells were washed twice with 180μL/well of PBS and fixed with 50μL/well of 10% Formalin for 10min at RT, the washed again as above. A staining cocktail was prepared, containing 1μg/ml Hoechst 33342 (nuclear stain) + 10 Units/ml AF488 Phalloidin (actin stain) + 1/200 dilution of PE Donkey Anti-Rabbit secondary antibody (Jackson ImmunoResearch) in staining buffer and added to the wells at 50μL/well. The plates were incubated for 30min at 37°C, 5% CO2, high humidity (95%), washed twice with 180μL/well of PBS, 100μL/well PBS added, and the plates were imaged on an InCell 6000 imager using a 10X magnification. Images were analyzed using Columbus v2.8 (Perkin Elmer). The following image analysis algorithm was used. The fluorescence from the DAPI channel (Hoechst 33342 stain) was used to find the nuclei. The fluorescence from the FITC channel (Alexa Fluor™ 488 Phalloidin stain) was then used to identify the cytoplasm and draw the cell perimeter. Fluorescent spots of clustered surface AChR (Cy5 fluorescence) and deposited MAC (dsRed fluorescence) were identified within each cell, and various spot parameters were calculated. The sum of the spot fluorescence per cell was used to set the threshold for identifying “AChR Pos Cells” and “MAC Pos Cells”. The % positive cells was calculated by dividing the number of positive cells by the total number of cells, multiplied by 100. The “AChR Pos Cell Population: % MAC Pos Cells” was calculated by dividing the AChR/MAC double positive cells by the total number of AChR positive cells, multiplied by 100. To be able to class patients into different categories based on levels of AChR loss and MAC deposition, a “fold change relative to NHS treatment” ratio was calculated for all treatment conditions to normalize the data from different experiments, by dividing each treatment value by the NHS-treatment value for that plate. For the AChR readout the classification was as follows: 1 – 0.8 (-), 0.8 – 0.6 (+), 0.6 – 0.3 (++). For the MAC readout the classification was as follows: 1 – 1.3 (-), 1.3 – 1.8 (+), 1.8 – 2.3 (++), 2.3 – 3.5 (+++). The % positive cells values were plotted in GraphPad Prism v7.05 and statistical analysis was performed using a repeated measures 1way ANOVA without correction, using Tukey’s multiple comparisons test. For patient categories 1, 2, and 4, the average of all donors within the category was analyzed. For individual donors (except Patient 1), statistical analysis was performed on the average of n=3 experiments, each carried out in duplicate. For MG Patient 1 (category 3) (n=1 experiment, 2 replicates) an unpaired t-test was used to compare the mean of duplicates between conditions.
A two-tailed, non-parametric Spearman correlation with 95% confidence interval was calculated in GraphPad Prism v7.05 between the following pairs of sample sets: IgG cell binding to AChR vs. MAC deposition on the AChR positive cells, IgG cell binding to AChR vs. ELISA anti-AChR titres, and ELISA anti-AChR titres vs. MAC deposition on the AChR positive cells. The values used for IgG cell binding were the background corrected fluorescent intensity averages of 3 replicates for all samples, except for MG Patient 1, where the average of two replicates from one experiment was used as there was not enough sample for further repeats. The values used for MAC deposition on the AChR positive cells were the “MG Plasma + NHS” fluorescent intensity averages of 3 replicates for all samples, except for MG Patient 1, where the average of two replicates from one experiment was used as there was not enough sample for further repeats. The values used to calculate the correlations for the ELISA anti-AChR titres were the average of 3 replicates for all samples, except for MG Patient 1, where the titer provided in the supplier datasheet was used as there was not enough sample to re-test in-house.
Functional anti-C7 antibodies were discovered by immunization of transgenic mice, that have a diverse human V-gene repertoire, combined with standard hybridoma techniques and B-cell sorting methods. 397 antibodies were screened by Surface Plasmon Resonance (SPR) to measure human C7 binding affinity (Kd), functional blockers were subsequently identified in a single dose classical pathway hemolysis assay (>30). Selected antibodies were grouped according to epitope bins using a sandwich Surface Plasmon Resonance (SPR) based method. The top seven functional, epitope diverse antibodies, were optimized for affinity and potency. Affinity optimization was achieved using the Adimab yeast-based platform, improved binders were selected from mutagenesis libraries generated by introducing random diversity into the CDRs of the heavy- and light-chain-variable genes. A total of 649 sequence unique antibodies were screened by Surface Plasmon Resonance for C7 binding. The optimization campaign delivered a hit panel of high-affinity and high-potency antibodies (with single-digit nanomolar IC50 potencies in the hemolysis assay). The antibodies were purified, stability and biophysical properties evaluated (data not shown) and further characterized functionally as described below.
A diverse panel of antibodies, that represented four distinct epitope bins determined by SPR, was selected from the hits using binding and affinity data to human and cyno C7. Three representative clones (TPP1657, TPP1653 and TPP1820 or its parental clone TPP1651) were included in the present work. Species cross-reactivity of the three mAbs was assessed using classical pathway hemolysis assays (CP CH50) using normal human, rat and cyno sera. In hemolysis assays using normal human and cyno serum, TPP1657 and TPP1820 showed comparable function with IC50s below 2nM for both species, while TPP1653 had IC50s of 2.6nM and 2.4nM for human and cyno, respectively (
Functional characterization of an anti-C7 monoclonal antibody and HDX-MS epitope mapping. Classical pathway sheep haemolysis assays were used to test the function of TPP1653 (closed triangle), TPP1657 (open circle) and TPP1820 (open square) in normal human serum, comparing to isotype matched active (open triangle) and disabled anti-C5 (closed circle) control antibodies
The three mAbs were determined to be in separate epitope bins by competitive ELISA (data not shown). HDX-MS was applied to determine the C7 binding epitopes of TPP1657, TPP1653 and in the case of TPP1820, the parental molecule TTP1651 was included. For this, mAb/C7 complexes or free C7 were deuterium labelled, the reactions quenched, and samples subjected to mass spectrometry. Protected peptides, indicative of binding epitopes on C7, were determined by comparing peptide uptake plots for complexed and free C7 (
To further assess the mechanism of action of the mAbs, a bio-layer-interferometry assay (Octet system) was developed to model stepwise assembly of the MAC on AMC biosensors. For this C5b6 was captured on the sensors and dipped sequentially into C7 with an excess of anti-C7 mAb, isotype control or without antibody, followed by C8 and C9 (
Assembly of the MAC complex using BLI technology – determine anti-C7 antibody mechanism of action. The sensograms demonstrate the assembly of MAC in the presence of absence of anti-C7 antibodies TPP1653, TPP1657 and TPP1820 following C5b6 capture onto AMC sensor (previous capture of anti-C6 mAb and C5b6 binding are not shown). Sensograms for each anti-C7 antibody evaluated are aligned to the C7/anti-C7 antibody addition step and the data has been reference subtracted (reference = αC6 mAb and C5b6 only, lighter blue trace).
Rats treated with the isotype control mAb, either at the time of disease initiation or 16h post-initiation, developed progressive disease and lost weight as expected in the model. In contrast, rats treated with TPP1820 at the time of induction showed no clinical disease and steady weight gain over the course of the experiments (
Therapeutic effect of monoclonal antibody (mAb) TPP1820 in experimental autoimmune myasthenia gravis (EAMG). TPP1820 or an isotype control IgG was administered either at the time of EAMG induction (time 0; 3A) or 8 hours after disease induction (3B) with 8 rats in each group in each experiment. Clinical score, assessed as described below, and weight change were monitored (
In order to make the case for a complement-targeted therapy in MG and identify patients appropriate for such a therapy, an assay system was sought that enabled stratification of MG patients according to the complement-activating ability of their anti-nicotinic AChR autoantibodies, differentiating them from MG patients with only ligand-blocking or crosslinking anti-AChR antibodies that do not activate complement and would not be amenable to complement therapies (
We screened MG patient plasma samples for IgG binding to the AChR+Rapsyn transfected cells; bound IgG was detected using Cy3-labelled anti-human IgG; AF647-labelled α-Bungarotoxin was used to stain the AChR clusters.
MG patient autoantibody binding pattern to AChR+Rapsyn transfected cells.
Myasthenia Gravis patient categories according to how the patient Ig mediates AChR loss, MAC deposition and effect of C7 inhibition.
MG Patient | AChR Loss | AChR Loss blocked by anti-C7 Ab? | MAC Deposition | IgG Cell Binding | Anti-AChR Titre (nmol/L) | Category |
---|---|---|---|---|---|---|
2 |
|
|
|
|
5.43 | 1 |
3 |
|
|
|
|
143.49 | 1 |
14 |
|
|
|
|
13.65 | 1 |
15 |
|
|
|
|
12.09 | 1 |
16 |
|
|
|
|
23.60 | 1 |
17 |
|
|
|
|
67.52 | 1 |
MSDN04 |
|
|
|
|
130.32 | 1 |
MSDN09 |
|
|
|
|
1.78 | 1 |
MSDN19 |
|
|
|
|
53.84 | 1 |
4 |
|
|
|
|
4.80 | 2 |
12 |
|
|
|
|
10.13 | 2 |
13 |
|
|
|
|
3.80 | 2 |
1 |
|
|
|
|
18.40 | 3 |
5 |
|
|
|
|
16.28 | 4 |
8 |
|
|
|
|
2.18 | 4 |
9 |
|
|
|
|
3.21 | 4 |
11 |
|
|
|
|
4.37 | 4 |
18 |
|
|
|
|
41.89 | 4 |
MSDN12 |
|
|
|
|
<Detection Range | 4 |
Pooled Ctrl Plasma |
|
|
|
|
<Detection Range | N/A |
Max Scale |
|
|
|
N/A, not applicable.
The patient stratification assay was then validated with MG patient plasma as the source of anti-AChR autoantibodies (
Example of MG patient plasma tested in the AChR loss and MAC deposition assay.
MG patient plasmas were grouped into four categories based on degree of IgG cell binding, ability to cause AChR loss and MAC deposition, and whether these can be blocked by the anti-C7 antibody (
Patient Ig-mediated loss of AChR and MAC deposition - plotted means of Myasthenia Gravis patient categories, showing % AChR positive cells
Patient samples from category 1 showed high (++) AChR loss that was fully or partially blocked by the anti-C7 antibody, low (+) to high (+++) MAC deposition and moderate (++) to high (+++) IgG cell binding (
Patient samples from category 2 showed low (+) AChR loss that was fully blocked by the anti-C7 antibody, no detectable (-) MAC deposition and low (+) to moderate (++) IgG cell binding (
Only one sample fitted into category 3; because of limited sample it was tested in a single experiment in duplicate. The profile comprised moderate (++) IgG binding to the AChR on the cells (
Patient samples from category 4 showed no detectable (-) AChR loss, no MAC deposition and no (-) or low (+) IgG cell binding (
Statistically significant correlation was observed when comparing IgG cell binding vs. MAC deposition (r = 0.8, p < 0.0001
The role of complement in human disease has been well described and many therapeutic concepts developed to target the system and harness the clear potential for therapeutic benefit in diverse diseases with significant unmet patient need (
Therapies in the terminal pathway space have been focused on C5 (Eculizumab/Ravalizumab, Alexion Pharmaceuticals; Crovalimab, Roche; Zilucoplan, UCB) and, apart from a few recent examples, including C6 (Regenemab, CP010), other terminal pathway targets have received little attention (
In the present work, we report the characterization of three molecules from a discovery campaign that selected C7 inhibiting mAbs. In CP CH50 using normal human, cyno and rat sera, all three mAbs showed inhibition of human and cyno complement mediated lysis (
We used HDX-MS to map the C7 epitopes of the mAbs (
TPP1820 was selected for an
MG is a heterogenous disease, both at the level of auto-antigen (AChR, muscle-specific kinase (MuSK), LDL receptor related protein 4 (LRP4/agrin)) and within each serological subtype (
However, the complexity of the cell-based assay we developed presents challenges for its use in a clinical setting in the present form. A flow cytometry assay using stably transfected cells and detecting C3/MAC deposition may be more practical. Here, the assay system developed by Obaid et al, is more suited. The complement regulator deficient ARPE-19 were chosen as tool to interrogate the MAC-dependent mechanisms and the ability to perform reproducible transfection, similar to the choice of HEK cells by Obaid et al, 2022 (
In conclusion, we characterized a set of novel anti-C7 monoclonal antibodies, and provided novel insights into tractable, functionally relevant epitopes on C7 and further validation of C7 as target for MG. With view to improved clinical trial design, we report a proof of concept patient stratification assay, developed to assess the heterogeneity of complement-dependence in MG. Taken together, these findings are relevant to future development and testing of new complement therapies in MG and other terminal pathway mediated pathologies, to facilitate bringing the right drug to the right patient with the associated benefits of faster and better treatment outcomes for patients and lowered burden on healthcare systems for society.
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation. Anti-C7 mAbs can be made available under MTA to repeat the in vitro experiments. The CRISPR edited APRE-19 cell line cannot not be shared due to licensing limitations. A protocol to replicate the effect of the regulator k/o is included as supplementary file.
Ethical review and approval was not required for the study on human participants in accordance with the local legislation and institutional requirements. The patients/participants provided their written informed consent to participate in this study. All animal experiments were approved by the Committee for Animal Care, Welfare and Use committee. All animal studies were ethically reviewed and carried out in accordance with European Directive 2010/63/EEC and the GSK Policy on the Care, Welfare and Treatment of Animals.
E-MN, SD, BM, MF, TW, and SK contributed to design of the study and supervised. EL, WZ, DGow, CS, IO, LS, DGorm, AS, AB, EW, RP, MB, and SP-F performed experiments, analysis, and statistical analysis. MF and E-MN performed review of data integrity. EL and E-MN wrote the first draft of the study. All authors contributed to the article and approved the submitted version.
BM and WZ would like acknowledge UK Dementia Research Institute and Alzheimer's Research Race Against Dementia Fellowship.
BM is supported by the UK Dementia Research Institute (UK-DRI), funded in part by the Medical Research Council. WZ is a Race Against Dementia Fellow and UK-DRI Future Leader Fellow. The authors would like to acknowledge the expert contributions of Joselli Silva O’Hare, Victoria Martin and Irene Sanjuan-Nandin in the
EL, DGow, CS, IO, LS, DGorm, AS, AB, EW, MB, TW, RP, SK, MF, SD, E-MN are employees and shareholders of GSK. SP-F is an employee of GSK.
The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
The Supplementary Material for this article can be found online at:
Containing supplementary data and methods for the HDX-MS and BLI experiments:
Generation of complement regulator deficient ARPE-19 cells and protocol for inhibition of complement regulators in lieu of the cell line.
Containing supplementary data for the MG patient stratification assay.
Plasmid maps of pBiCIH-hnAChRα1β1 and pBiCIN-hnAChRδϵ, created in SnapGene 5.3.2.
Plasmid map of PCIP4_Rapsyn, created in SnapGene 5.3.2.