Synthetic genes (VH and VL of empasiprubart) were generated by PCR-based assembly of overlapping oligonucleotides with random mutations on selected positions in the CDR regions based on empasiprubart’s crystal structure (12). After gene assembly via PCR, amplicons were digested and ligated into the phagemid vector as previously described (13). Phages were produced as described by van der Woning et al. (14) and screened for rat C2 binding.

All antibodies for in vitro analysis were produced as hIgG1 with L234A/L235A (LALA) mutations following protocols described elsewhere (15). Briefly, HEK293 cells were transfected with 1:1 heavy/light chain plasmid ratio using 1:3 DNA/PEI ratio for transient production followed by protein A affinity purification.

HEK293E-253 cells were transfected with endotoxin-free plasmid DNA using the rPEx^® Technology at ImmunoPrecise Antibodies Ltd (the Netherlands) to produce antibodies for in vivo analysis. Six days post-transfection conditioned medium containing recombinant antibody was harvested using SatroClear Dynamics, followed by MabSelect PrismA purification. Obtained antibodies were sterilized by filtration over a 0.22 µm syringe filter.

Microtiter wells (Greiner Bio-one, Cat N°: 675061) were coated with immunoglobulin (Ig)G (CP) or mannan (LP) and incubated overnight at 4 °C or at room temperature, respectively. After blocking with 1% w/v casein (Bio-Rad Laboratories, Cat N°: 1610783) and washing, wells were incubated with rat serum samples diluted in veronal buffered saline at 37 °C. As controls, MgEGTA (inhibits CP and LP) or EDTA (inhibits all pathways) were added to the buffer. After washing, wells were incubated with biotinylated anti-rat C3 antibody (Abcam, Cat N°: ab17456), followed by streptavidin-HRP (Pharmingen, Cat N°: 554066). Afterwards, wells were washed and incubated with TMB solution. The substrate reaction was stopped with sulfuric acid and OD450/620 nm was measured with a spectrophotometer (Infinite M Nano, Tecan).

Rat AP activation levels were assessed using a commercially available ELISA kit (Hycult Biotech, HIT412, Uden, the Netherlands). Serum was diluted to 2% v/v with kit provided sample dilution buffer. The manufacturer’s instructions were followed, and absorbance was measured at 450 nm. In addition to the kit’s positive and negative control, healthy pooled rat serum was added as positive control (InnoSer, Diepenbeek, Belgium).

Briefly, wells of a MaxiSorp 96-well plate (ThermoScientific, Cat N°: 442404) were coated with human, cynomolgus monkey, guinea pig, rabbit, mouse, hamster, or rat recombinant C2 protein (produced in HEK293E-253 cells, ImmunoPrecise Antibodies Ltd, the Netherlands) overnight at 4 °C, blocked with 1% w/v casein and incubated with 5 µg/mL of the anti-rat C2 antibody or empasiprubart. After washing, wells were incubated with HRP-conjugated goat anti-human IgG antibodies (Bethyl Laboratories, Cat N°: A80-319P). Afterwards, wells were washed and incubated with TMB solution. The substrate reaction was stopped with sulfuric acid and OD450/620 nm was measured with a spectrophotometer.

Multi-array 96-well plates (Meso Scale Discovery, Cat N°: L15XA-3) were coated overnight at 4 °C with recombinant rat C2. After washing the plates with wash buffer A (Supplementary Table 1A), wells were blocked with 1% w/v Bovine Serum Albumin (BSA) (Sigma-Aldrich, Cat N°: 05482). Subsequently, plates were washed with appropriate wash buffer (A-D, Supplementary Table 1A) and incubated with serial dilutions of antibody in the corresponding dilution buffer (A-D, Supplementary Table 1A). After washing the plates with appropriate wash buffer (A-D, Supplementary Table 1A), a final wash step with buffer A was performed. Next, Meso Scale Discovery-Sulfo tag-labeled goat anti-human IgG antibody (Meso Scale Discovery, Cat N°: R32AJ-1) was applied. Finally, wells were washed, MSD reading buffer (Meso Scale Discovery, Cat N°: R92TC-2, diluted ½ in MQ water before use) was added to the wells and signal was measured immediately by Meso™ QuickPlex SQ 120 (Meso Scale Discovery).

The surface plasmon resonance (SPR) measurements were performed on a Biacore T200 instrument (Cytiva). Recombinant rat C2 was diluted to 5 µg/mL in immobilization buffer (10 mM sodium acetate pH 4.5, Cytiva, Cat N°: BR100350), injected and immobilized to 150 response units on the surface of a carboxymethylated dextran sensor chip (Sensor Chip CM5 series S, Cytiva, Cat N°: 29104988) at a flow rate of 30 µL/min. Binding measurements of empasiprubart and the anti-rat C2 antibody were performed in 20 mM HEPES, 100 mM NaCl₂, 1.25 mM CaCl₂, 0.05% Tween20, pH 7.5, at 30 µL/min flow rate. After each measurement, the surface was regenerated by two injections of 10 mM glycine-HCl pH 1.5 (Cytiva, Cat N°: BR100354) at a flow rate of 30 µL/min without loss of C2 binding capacity. Sensorgrams were generated at concentrations of 33.33, 16.67, 8.33, 4.17, 2.08, 1.04, and 0.52 nM antibody. Kinetics of the antibodies were determined with the BIAevaluation software (version 5.0.18) using simultaneous k_a/k_d fitting with 1:1 mass transfer.

MaxiSorp microtiter plates were coated overnight with the anti-rat C2 antibody and blocked with 3% w/v BSA. After blocking, samples and calibration curve were pipetted into the ELISA plates and incubated on ice. Serum samples were diluted in assay buffer (TBS: Tris Buffered Saline, ThermoFisher, Cat N°: BP24711) supplemented with 2 mM CaCl₂ (Sigma, Cat N°: 21115). As a calibration curve, dilutions of recombinant rat C2 in assay buffer were used. After incubation and washing, free C2 was detected with biotinylated human anti-rat C2 antibody 3H07 (argenx). The epitope targeted by 3H07 is distinct from that recognized by the newly developed anti-rat C2 antibody. After incubation on ice, plates were developed with streptavidin-HRP and s(HS)TMB (SDT reagents, Cat N°: sTMB) ½ diluted in MilliQ. OD450/620 nm was measured with a spectrophotometer after the reaction was stopped with sulfuric acid.

High-binding half-area plates were coated overnight with mouse anti-rat C2 antibody 3H07. The following day, plates were blocked using 1% v/v casein in PBS and washed. Next, samples and calibration curve diluted in TBS supplemented with 2 mM CaCl₂, 10 µg/mL of the new anti-rat C2 antibody and 10%, v/v rabbit serum (Cederlane, Cat N°: CL3441S50), were incubated. The calibration curve consisted of dilutions of recombinant rat C2 in assay buffer. After incubation and washing, plates were developed with goat anti-human IgG Fc HRP (Abcam, Cat N°: ab97225) and subsequently TMB.

In brief, Maxisorp ELISA plates were coated with mouse anti-hIgG (CH2) (Bio-Rad Laboratories, Cat N°: 18570060) on which serum samples containing the anti-rat C2 antibody, 1/100 diluted in 1x PBS/0.1% v/v casein, were incubated. After a wash step, plates were incubated with HRP-conjugated goat anti-human IgG antibodies (Bethyl Laboratories, Cat N°: A80-319P). After a final wash step, plates were developed using TMB. Each run contained a calibrator curve (11 non-blank known anti-rat C2 antibody concentrations spiked in rat serum) and at least two sets of quality controls. Concentrations of the anti-rat C2 antibody in the study samples and QC samples were calculated from the calibrator curve using a five-parameter logistic (5PL) fit without additional weighting factors applied.

Eight-week-old male Sprague Dawley rats weighing 250–300 g were used for a PK and PD study of the anti-rat C2 antibody. Animals were housed under conventional conditions at InnoSer (Diepenbeek, Belgium). A single dose of 2, 5, 10, 25, or 50 mg/kg anti-rat C2 antibody (hIgG1 LALA) was administered intravenously (IV). Each dose group included three animals. As a control group, 3 rats received 50 mg/kg isotype control (Mota hlgG1 LALA). Baseline blood samples were collected 7 days and 1 day prior to the study’s start. Intermediate blood samples were taken at 5 minutes, 6 hours, 24 hours, 48 hours, 3 days, 5 days, 7 days, 14 days, and 21 days post-dosing without anesthesia. After the final tail vein blood collection on day 21, all animals were euthanized humanely via an intraperitoneal overdose of Euthasol vet (200 mg/kg). For free C2, the relative percentage change of concentration, normalized to the mean baseline value within each group, was determined. A non-compartmental analysis was used for parameter estimation using Phoenix WinNonlin (Certara LP, USA). The IV bolus model (Calculation method: Linear Up, Log Down) was used on individual time-concentration curves, using nominal dose levels and nominal sampling times relative to the dose administration.

Population PK/PD modeling was performed using NONMEM version 7.5.1 (ICON Development Solutions, Ellicott City, MD, USA), employing the stochastic approximation expectation-maximization algorithm for parameter estimation. Due to a high proportion of PD data below the limit of quantification (BLQ), the M3 method was applied to incorporate BLQ data into the likelihood calculation and ensure robust parameter estimation. Diagnostic plots and visual predictive checks were used to evaluate model performance. Simulations based on the final population PK/PD model were conducted in a virtual rat population (n = 1000) to determine an optimal dosing regimen of the anti-rat C2 antibody that maintained complete suppression of free C2 levels over a 7-day period.

The study was performed in the Charles River facility in Laval (Quebec, Canada) according to internal standard operating procedures and approved by the local animal ethical committee (Study 20441571, IACUC CRL Montreal ULC-Laval). Syngeneic inbred, male, adult Lewis rats (CRL, North Carolina, USA) weighing 300–400 g served as donors and recipients to avoid graft rejection. The animals were housed under conventional conditions. Kidney transplantation was performed as adapted from Yu et al. (11). The left donor kidney was flushed with heparinized saline (80 U/mL) followed by a flush with cold preservation solution (Belzer UW cold storage solution) before removal. The kidney was then placed in a sterile container with UW cold storage solution for 32 hours at 4 °C and then kept on ice pending transplantation. After left nephrectomy of the recipient (10–11 weeks), the donor kidney was transplanted by connecting the arterial and venal cuff to the recipient abdominal aorta and vena cava, respectively. Next, the ureter was connected by end-to-end anastomosis, and the kidney was reperfused. After the left kidney transplant, the recipient right kidney was removed and discarded. Isoflurane at 2–3% was used to anesthetize the rats during the procedure. Animals were sacrificed via exsanguination from the abdominal aorta after isoflurane anesthesia, unless deemed inappropriate by the Study Director and/or the clinical veterinarian, when humane endpoints were reached following CRL SOPs. Surgery day was defined as day 1.

Recipient animals were dosed IV in the tail vein immediately prior to reperfusion with a bolus of 50 mg/kg anti-rat C2 antibody (18 animals) or a hIgG1 LALA isotype control mAb (9 animals), followed by a second intravenous bolus injection of 25 mg/kg 72 hours post-transplant. Animal mortality within 24 hours post-surgery was considered related to surgical procedure, and these animals were replaced. Blood samples were collected on day 3, 5, 7, and 14 for assessment of renal function (serum creatinine and blood urea nitrogen [BUN] levels) and biomarker (Neutrophil Gelatinase-Associated Lipocalin (NGAL)) analysis. On day 14, animals were sacrificed for histological examination and weighing of the transplanted kidney. Due to a few unscheduled deaths at earlier time points (n = 3 per group), and 7 prospective scheduled deaths for early kidney histology (data not shown), final sample sizes were: 18, 11, 9, and 7 for the anti-rat C2 mAb group and 9, 8, 7, and 6 for the isotype control mAb group on days 3, 5, 7, and 14 respectively.

Rat blood samples were collected in normal Eppendorf tubes and clotted for 30 min at room temperature followed by centrifugation (4 °C, 1300 - 2000 × g, 10 min). Serum samples were aliquoted and snap-frozen within 2 hours after sampling. Samples were stored at -70 °C and were not subjected to freeze-thaw cycles prior to analysis of CP, free C2, total C2, and PK. Unfrozen serum samples were analyzed for BUN by in vitro kinetic testing with urease and glutamate dehydrogenase, and for creatinine by the Jaffé alkaline picrate method, both on a Roche/Hitachi Cobas c analyzer according to the packaging inserts (Ref ID 04460715 190 and 4810716 190, respectively). Transplanted kidneys were dissected, weighted, and then stored in 10% neutral buffered formalin, followed by paraffin embedding, sectioning, mounting, and staining with hematoxylin and eosin for microscopic analysis.

Log-transformed serum creatinine and BUN concentrations at day 3 were analyzed with a linear model including treatment group as a fixed effect and a heterogeneous residual variance across treatment groups. A longitudinal linear mixed model was fitted to the log-transformed concentrations of serum creatinine, BUN and normalized NGAL (= fold increases to healthy serum) obtained at days 3, 5, 7, and 14. Treatment group, day and their interaction were included as categorical fixed effects. A flexible correlation structure heterogeneous across treatment groups was incorporated to correlate the outcomes between consecutive time points within one animal: heterogeneous compound symmetry for serum creatinine, and heterogeneous first-order autoregressive structures were applied for BUN and normalized NGAL. The linear models were fitted using restricted maximum likelihood and the Kenward-Roger approximation for the degrees of freedom using SAS v9.4. Wald hypothesis tests were used to compare the (time profiles of) log-transformed concentrations of serum creatinine, BUN, and normalized NGAL between treatments. Results were visualized using GraphPad Prism v10. p-values <0.05 were considered statistically significant.

Superimposition of a predicted model of rat C2a (AF-A0A0U1RRP9-F1) on top of our previously published crystal structure of empasiprubart bound to CCP2 of human C2 (PDB 8ACI, manuscript submitted) suggested that the weak affinity to rat C2 was possibly due to the presence of a threonine in rat C2 at the equivalent position of V93 in human C2 (10). The more polar side chain of threonine likely stabilizes the position of the side chain of HCDR3 Y54 which interacts less efficiently with the main chain of C2 A91 (Figure 2A).

Using CDR randomization and phage display techniques, several candidate Fabs derived from empasiprubart’s sequence showed increased binding to rat C2 and were produced as human IgG1 antibody for further characterization (data not shown). In vitro analysis of these mAbs yielded one particular mAb that stood out as an inhibitor of both CP (IC₅₀ of 7.45 µg/mL, Figure 2B) and LP (IC₅₀ of 16.10 µg/mL, Supplementary Figure 1A) activity without interfering with AP activity of rat complement (Supplementary Figure 1B). This anti-rat C2 antibody demonstrated increased binding to rat C2 as compared to empasiprubart (maximum MSD signal of 101.58×10³ and 12.04×10³ at pH 7.4, respectively) (Figure 2C). Moreover, binding of the antibody to rat C2 was also calcium-dependent like empasiprubart to human C2, as illustrated by the increase and decrease in maximum MSD signal at neutral pH with addition of calcium or EDTA, respectively, to the buffer (117.85×10³ vs 24.66×10³ and 45.04×10³ vs 1.07×10³, Figure 2C). Lowering the pH from 7.4 to 5.5 impaired the binding capacity of the newly developed antibody to rat C2 (maximum MSD signal 3.79×10³ vs 0.32×10³). However, once bound to C2 at neutral pH (with calcium), a lower pH did not disrupt the binding as indicated by the overlapping curves. Consequently, the anti-rat C2 antibody, in contrast to empasiprubart, lacks pH-dependent target release properties for rat C2. Finally, affinity data confirmed low cross-reactivity of empasiprubart to recombinant rat C2 (Supplementary Figure 1C) whereas the anti-rat C2 antibody binds to both recombinant rat C2 (Figure 2D) and recombinant human C2 with high affinity at pH 7.4 (Supplementary Figure 1C). No k_d values could be calculated due to the low off-rate of the anti-rat C2 antibody. Finally, cross-reactivity of the anti-rat C2 antibody to other species was assessed (Figure 2E). Next to strong C2 binding for rat, the antibody illustrated cross-reactivity to human, cynomolgus monkey, and hamster C2. To reduce unwanted stimulation of effector functions in rats, the anti-rat C2 antibody was formatted as a hIgG1 with L234A/L235A (LALA) mutations, diminishing its interaction with Fc gamma receptors and C1q (16).

Thus, an anti-rat C2 antibody was developed that exhibits a Ca²⁺-dependent high affinity binding to rat C2 at neutral pH but lacking pH-dependent target release.

To evaluate the PK and PD characteristics of the anti-rat C2 antibody, single ascending doses of the mAb were injected IV in eight-week-old male Sprague Dawley rats. Serum concentrations of the antibody were measured at multiple time points and used to estimate PK parameters using a non-compartmental model (Supplementary Table 1B). The C_max was anticipated at the initial bleeding time point following the intravenous injection. Overall, anti-rat C2 antibody exposure (C_max, AUC) increased in a dose-proportional manner over the 2–50 mg/kg dose range tested (Figure 3A). The apparent terminal half-life was estimated at ~12.5 days, however the PK study duration was insufficient to accurately determine this parameter.

In addition to the PK, different PD read outs were assessed to capture the characteristics of the anti-rat C2 antibody in vivo. The time course of free C2 (i.e., C2 unbound to mAb) and total C2 values in serum with their fitted curves after a single IV dose in male Sprague Dawley rats are depicted in Figures 3B, C, respectively. A dose-dependent reduction of free C2 was observed for the different anti-rat C2 antibody-treated groups (Figure 3B). The highest dose (50 mg/kg) resulted in reductions of ≥95% compared to isotype-treated animals for up to 21 days.

Figure 3C shows that total C2 levels dose-dependently increased in the rats upon administration of anti-rat C2 antibody. Saturation of C2 build-up was reached when dosing ≥25 mg/kg. Both the 25 mg/kg and 50 mg/kg dosed animals exhibited a 10-fold increase in total C2.

Next, the course of CP activity in serum was measured in vitro as another PD measure for complement blocking activity. Administration of the anti-rat C2 antibody reduced classical pathway activity in a dose-dependent manner (Figure 3D). Complete inhibition, which is a similar inhibition as that by MgEGTA, an established inhibitor of CP activity, was observed upon administration of 50 mg/kg mAb for up to 2 days. Complement activity gradually restored over time but had not reached baseline levels after 21 days. All other dose groups returned to their baseline CP activity values after 14 days or even earlier, depending on the dose.

A two-compartment model with linear elimination and zero-order infusion could adequately describe PK data. The PD component of the model captured the synthesis, degradation, and interaction with free C2 using a target binding model. A schematic representation of the final PK/PD model is presented in Supplementary Figure 2A. The final model included inter-individual variability (IIV) on all parameters and used a proportional residual error model. Model-based simulations revealed that a loading dose of 50 mg/kg followed by a maintenance dose of 25 mg/kg administered 3 days later, would be able to sustain reductions of free C2 concentrations above a threshold of ≥95% for at least 7 days (Supplementary Figure 2B).

The rat kidney transplantation model is summarized in Figure 4A. During the entire experiment, free C2 serum concentrations of anti-rat C2 antibody-treated rats were reduced by ≥95% as compared to reference levels in pre-surgery donor animals (Figure 4B). Rats receiving isotype control mAb developed severe kidney failure, as shown by a drastic increase in serum creatinine and BUN levels 2 days post-transplant (day 3, Figure 4C). Creatinine and BUN levels reduced over time post-transplant but were still higher on the day of sacrifice (day 14) compared to pre-transplant donor reference values. In contrast, in anti-rat C2 antibody-treated rats, serum creatinine and BUN concentrations were significantly lower than in isotype control rats 2 days post-transplant and remained lower up to the day of sacrifice. Creatinine levels in anti-rat C2 antibody-treated rats were on average 40% lower on day 3 (95% CI [14%; 59%], p = 0.008) and 49% lower across time points (95% CI [15%; 70%], p = 0.014). Additionally, these animals also had significantly reduced BUN concentrations at day 3 (33%, 95% CI [13%; 49%], p = 0.005, Figure 4C) and across time points (49%, 95% CI [13%; 70%], p = 0.017) compared to isotype-treated animals. Interestingly, NGAL release in serum was lower in rats treated with the anti-rat C2 mAb, suggesting that the improved renal clearance was related to less severe acute kidney injury (Supplementary Figure 3).

Notably, three rats in each treatment group died. All these mortality cases were associated with high serum creatinine post-transplantation, but only two deaths per treatment arm were considered related to IRI, based on microscopic findings, associated secondary changes in distant organs, and renal impairment (data not shown). The other two deaths were considered to be related to the transplantation/surgery procedure. Mortality rates were too low for meaningful conclusions. As planned sacrifice and unplanned mortality/sacrifice numbers were similar across treatment groups, missing values due to death were considered randomly distributed and were not taken into account in statistical analyses of creatinine, BUN, and NGAL levels. Furthermore, treatment did not impact incidence or severity of post-operative clinical signs, body weights, or body weight gains (data not shown).

Transplanted kidneys from all surviving animals were histologically examined on day 14, at which time the acute injury is expected to have subsided and changes are more reflective of (sub)chronic changes. Indeed, in isotype-treated control rats, IRI had resulted in gross and microscopic findings consistent with regeneration and subacute/chronic inflammation (Table 1). Specifically, kidneys appeared enlarged, discolored, and with abnormal consistency. Microscopically, IRI was evidenced by findings of tubular regeneration, hyaline/granular casts (within necrotic and regenerating tubules in the medulla, cortex and/or papilla), mineralization (mineralized sloughed necrotic tubular epithelial cells), and infiltration of interstitial lymphocytes and plasma cells accompanied by fibroplasia (Table 1). No notable histological changes were observed in glomeruli. Importantly, these tubular IRI-related findings were less in severity and incidence in animals treated with anti-rat C2 antibody than in isotype-treated control rats (Table 1). This reduction of IRI-related histological findings correlated with a decrease of the kidney weights of 23% as compared to the isotype-treated control group (Table 1), thereby suggesting a protective effect of complement blockade on kidney health at 2 weeks post-transplant.

A select number of animals were examined by histology at earlier time points (Supplementary Table 2). Although sample sizes were too small to assess treatment response during the peak of acute injury, histological assessment of these cohorts confirmed that the chronic inflammatory and regenerative responses, observed at day 14, were preceded by (sub)acute inflammation and tubular cell injury. Specifically, tubular necrosis, characterized by injured, markedly attenuated or hypereosinophilic epithelia, was observed in all four animals that prematurely died from IRI and in 3 out of 7 animals from the scheduled cohorts. Furthermore, interstitial and luminal neutrophil infiltration was observed in 5 animals, confirming activation of innate immunity.