Introduction

The complement system is a potent inflammatory system activated by three distinct pathways: the classical pathway (CP), the lectin pathway (LP), and the alternative pathway (AP). These combine to provide crucial defense against microbial infections and clearance of dying host cells; however, they also contribute to tissue damage and organ dysfunction in human diseases1,2,3. The number of diseases in which complement dysregulation has been identified as a pathology driver is increasing, as is the generation of new complement therapeutics4,5. Therapeutic targeting of complement can be challenging, as it may increase the host’s susceptibility to microbial infections. This could be addressed by specifically targeting key complement factors involved in pathologic activation, while leaving the rest of the system unaffected. For instance, in some antibody-mediated diseases and in ischemia-reperfusion conditions, specific targeting of the CP and LP while leaving the AP intact could be beneficial6,7,8. Both CP and LP contribute to the anti-microbial activities of complement, as they can be activated by antibodies bound to microbial antigens or by microbial surface glycans, respectively. However, therapeutic inhibition of only the CP and LP will compromise the complement-mediated anti-microbial defense less than that of a total blockade of the system, as the AP also exerts anti-microbial activity.

The complement cascade is activated by three independent pathways: the CP, the LP, and the AP (Fig. 1). The CP and LP rely upon the complement cascade components complement factor 2 (C2; cleaved into C2a, the smaller activation fragment, and C2b, the larger serine protease fragment) and C4. C4 and C2 do not participate in AP and are both positioned upstream of complement factors that mediate inflammation, in particular C3 and C5. Thus, by targeting C4 or C2, both the CP and LP can be inhibited while leaving the AP activity unaffected. C2 may be the preferred therapeutic target for inhibition of the CP and LP because (1) among genetic deficiencies of the CP, which are typically associated with an increased risk for autoimmune diseases, those of C2 have the lowest9,10, and (2) its plasma concentration is lower than C411,12.

Fig. 1: The complement cascade.
figure 1

The complement cascade is a major mechanism for host cells to mount an effective immune response against invading pathogens and is activated by three independent pathways: the classical pathway (CP), the lectin pathway (LP), and the alternative pathway (AP). Activation of the CP, by pathogen or antibody binding (1), or the LP, by the binding of pattern-recognition molecules to pathogen-associated molecular patterns (2), results in the cleavage of C4 to C4a and C4b and C2 to C2a (smaller activation fragment) and C2b (larger fragment containing the serine protease domain)44 via the C1s protease or MBL-associated serine proteases, respectively (3). C4b and C2b join to form an active C3 convertase. Activation of the AP by pathogens and injured tissue causes low-grade hydrolysis of C3, with the resultant C3b binding to FB (4). Cleavage of FB by FD produces the alternative C3 convertase (C3bBb), at which point the three pathways merge (5). The C3 convertases cleave C3 into C3a and C3b. C3a binds the C3a receptor (C3aR) to induce inflammation (6), while C3b joins the C3 convertases to create the C5 convertases (C4b2a3b and C3bBbC3b) (7). The C5 convertases cleave C5 to C5a and C5b, which lead to inflammation via the C5a receptor, and cell lysis and death via MACs in the cell membrane formed from C5b, C6, C7, C8, and C9 (8). C2 participates in both the CP and LP, making it a target for intervention therapy for disorders affecting these pathways. Binding of empasiprubart to C2 (9) prevents formation of the C4bC2 complex upstream of C3 and the subsequent activation of the CP and LP, while leaving the AP intact. Parts of this figure were adapted from Burgelman M, Dujardin P, Vandendriessche C, and Vandenbroucke RE (2023). Free complement and complement-containing extracellular vesicles as potential biomarkers for neuroinflammatory and neurodegenerative disorders. Front. Immunol. 13:1055050. doi: 10.3389/fimmu.2022.1055050 licensed under CC BY 4.0. Ab antibody, Ag antigen, C complement factor, C3aR C3a receptor, C5aR C5a receptor, FB factor B, FD factor D, MAC membrane attack complex, MBL-MASP mannose-binding lectin–associated serine protease.

Full size image

Empasiprubart (ARGX-117) is a humanized immunoglobulin (Ig)G1 monoclonal antibody that binds C2, thereby preventing the formation of the C4bC2 complex, the C3 proconvertase of the CP and LP (Fig. 1)13. Empasiprubart exhibits pH- and Ca2+-dependent target binding and contains mutations in the Fc region (H433K and N434F), which increase its affinity for antibodies for the neonatal Fc receptor (FcRn) at an acidic pH in endosomes14. These features facilitate the release and degradation of bound C2 in endothelial endosomes and the recycling of empasiprubart into the plasma compartment. Thus, empasiprubart is a recycling antibody with the potential to induce sustained C2 blockade in vivo, as previously confirmed in non-human primates13.

Here, we reveal the structure of empasiprubart in complex with C2 and report the outcome of a first-in-human study investigating its safety, tolerability, pharmacokinetics (PK), pharmacodynamics (PD), and immunogenicity after intravenous (IV) administration. Empasiprubart is well tolerated, and the rapid, sustained, and dose-dependent reduction in circulating free C2 levels may provide a valuable treatment option for patients with complement-mediated disorders.

Results

Crystal structure of empasiprubart in complex with C2

Previous work has revealed that empasiprubart binds to an epitope located in the second complement control protein domain (CCP2) of human C213. To investigate the pH- and Ca2+-dependent interaction between empasiprubart and C2, X-ray crystal structure analysis of empasiprubart fragment antigen binding (Fab) with an N-terminal fragment of C2 (C2 Nt) containing the three complement control protein domains (CCP1–3) was performed after initial attempts to make crystals of complexes of recombinant C2 and the Fab of empasiprubart had failed. Upon adding a κ chain–binding variable domain of a heavy-chain antibody as a chaperone for crystal formation and growth15, crystals of complexes of C2 Nt and the Fab region of empasiprubart were obtained that diffracted X-rays to beyond 1.9 Å resolution and enabled structure determination (Supplementary Table 1). The structure confirmed binding of empasiprubart to the CCP2 domain of C2 and revealed that empasiprubart Fab almost exclusively contacts this domain by leveraging five of its six complementarity-determining regions (CDRs) (Fig. 2a; PDB Entry ID 8ACI). Only a few van der Waals interactions take place between the Fab and residues Gly159 and Ser179 in the C2 CPP3 domain. The interaction of empasiprubart with CCP2 of C2 involves both electrostatic and hydrophobic interactions, as well as a pivotal role for a Ca2+ ion. At the heart of the intermolecular interface of empasiprubart and CCP2, and wedged between the heavy and light chains, five of the six coordination sites of this Ca2+ ion are occupied by heavy-chain CDR3 residues (HCDR3) Asp99, HCDR3 Glu95, light-chain CDR3 (LCDR3) Tyr96, and by two water molecules kept in position by HCDR1 Asp35 and HCDR2 Asp50 (Fig. 2b). The crystal structure of empasiprubart obtained at a low pH (PDB Entry ID 8ACF and Supplementary Table 1) was practically indistinguishable from that obtained at a neutral pH.

Fig. 2: Crystal structure of empasiprubart.
figure 2

a Illustrative representation of the Fab-C2 Nt complex in the presence of a calcium ion (pink sphere). The anti-κ nanobody (kNb) is shown in yellow, empasiprubart VH in orange, empasiprubart VL in green, and C2 Nt (C2a) in gray. b Detailed view of interactions between the centrally positioned Ca2+ ion with both empasiprubart VH and C2 Nt main chain. Side chains of indicated residues are presented in sticks. Small red spheres are water molecules. Hydrogen bonds are shown as dotted lines. c View of the interaction between HCDR2 Tyr54 (stick representation) and the hydrophobic cavity of C2 (surface representation, red is more hydrophobic) (PDB Entry ID: 8ACI; https://doi.org/10.2210/pdb8ACI/pdb). C2 complement factor 2, CCP complement control protein, CL conserved light, HCDR2 heavy complementarity-determining region, Nt N-terminal, VH variable heavy, VL variable light.

Full size image

Completion of the Ca2+ coordination sphere is entrusted to the main chain of C2 Arg103. The side chain of Arg103 simultaneously forms a double hydrogen bond with HCDR2 Asp50, thereby excluding any bulk solvent accessibility to Ca2+ and securing a tight binding of the ion between C2 and the Fab. This predominant binding pocket centered around Ca2+ is further enforced by additional Fab-C2 interactions mediated by the heavy chain, with a prominent role for HCDR2 Tyr54 protruding into a hydrophobic pocket of C2 and, to a lesser extent, interactions mediated by residues of LCDR1 (Fig. 1c). Thus, the crystal structure demonstrates how CDR residues of both heavy and light chains of empasiprubart collectively form a Ca2+-mediated interface with CCP2 of C2. Interestingly, the crystal structure also provides a rationale for the selectivity of empasiprubart for C2, without cross reactivity for the CCP2 domain of factor B (FB). Despite a sequence identity of 53.3% between both CCP2 domains (Supplementary Fig. 1), steric hindrance between FB Arg105, Pro119, and Tyr120 with empasiprubart’s heavy chain arguably prohibits binding of FB, thereby supporting the previously determined lack of binding13.

Clinical evaluation of empasiprubart

The first step in the clinical development of empasiprubart was a phase 1 study to evaluate the PK, PD, and safety of single ascending doses (SADs) and multiple ascending doses (MADs) of empasiprubart in healthy participants. Demographics and baseline characteristics of 54 healthy participants receiving a single dose of empasiprubart or placebo (cohorts A1–A6 and A8; Fig. 3a) are shown in Supplementary Table 2. In addition, 24 participants were randomized to receive multiple doses of empasiprubart IV or placebo (cohorts B1‒B3; Fig. 3b and Supplementary Table 3). Demographics and baseline characteristics of participants were well balanced across the various dose groups, with the exception of age.

Fig. 3: Schematic presentation of the first-in-human study of empasiprubart in healthy participants.
figure 3

Investigation of a SADs and b MADs is shown. *One participant discontinued the study (lost to follow-up). †One participant did not receive the third dose on day 22 due to SARS-CoV-2 positivity, and one participant finished all study treatments but did not complete all the follow-ups (withdrawal by participant). ‡For two participants, empasiprubart treatment was withdrawn after the first administration, one because of SARS-CoV-2 infection, the other because of urticaria. One participant finished all study treatments but did not complete all the follow-ups (withdrawal by participant). IV intravenous, MAD multiple ascending dose, n numbers of participants, SAD single ascending dose.

Full size image

Of the 78 participants randomized to receive an IV dose of empasiprubart or placebo, 75 (96.1%) completed the study. One participant in cohort A2 receiving a single dose of 0.5 mg/kg empasiprubart was lost on follow-up and considered discontinued from the study (Fig. 3a). Two participants in the MAD part of the trial discontinued the study (withdrawal by participant), and three participants discontinued study treatment as per protocol, but they completed all follow-up visits. Of these participants, two discontinued because of a positive SARS-CoV-2 test and one due to a grade 1 adverse event (AE) of urticaria (Fig. 3b).

PK of empasiprubart

Serum levels of empasiprubart were measured with a specific enzyme-linked immunosorbent assay (ELISA) at different timepoints before and after administration and used to calculate PK parameters using non-compartmental methods. After a single IV infusion of empasiprubart 0.1 to 80 mg/kg, a dose-dependent increase of mean maximum observed serum concentration (Cmax) was seen (range 2‒2127 μg/mL), with a median time to reach Cmax (tmax) of 2 to 4 h post-dose (Fig. 4a and Table 1). The mean elimination half-life of empasiprubart ranged from 42 to 88 days for all dose groups (Table 1). The shortest half-life was observed after administration of 0.1 and 0.5 mg/kg (mean half-life values 43 and 42 days, respectively, see Table 1). However, these values should be interpreted with caution because the elimination phase could not be fully characterized due to the limited duration of sampling in these cohorts. Upon exclusion of the dose cohorts 0.1, 0.5, and 2.5 mg/kg, post-hoc analysis showed Cmax, the area under the serum concentration-time curve (AUC) from time 0 to 168 h (AUC0–168h), and time 0 to infinity (AUC0‒inf) all increased dose proportionally over the 10 to 80 mg/kg dose range, and the elimination half-life ranged between 70 and 88 days. The mean clearance (CL) values of empasiprubart ranged from 0.00245 to 0.00422 L/h, and the mean volume of distribution (Vz) values ranged from 5.66 to 8.57 L for doses 0.1 to 80 mg/kg.

Fig. 4: Time-concentration profiles of empasiprubart serum concentrations.
figure 4

a SAD and b MAD parts of the study. a Healthy participants were dosed with 0.1, 0.5, 2.5, 10, 30, 60, and 80 mg/kg empasiprubart or placebo (i.e., cohorts A1–A6, A8) IV in a 2-h infusion. Cohorts A1 and A6: n = 6; Cohort A2: n = 6 except weeks 5, 7, and 13 (n = 5); Cohort A3: n = 6 except weeks 3, 13, 21, 37, and 39 (n = 5) and week 17 (n = 4); Cohort A4: n = 5, except week 2 (n = 4); Cohort A5: n = 5, except week 7 (n = 4); Cohort A8: n = 6 except weeks 4 and 9 (n = 5). b Healthy participants were dosed with 10 mg/kg empasiprubart or placebo on days 1, 8, 15, and 22 (cohort B1), or 60 mg/kg empasiprubart or placebo on day 1 followed by 10 mg/kg on days 8 and 22 (cohort B2), or 10 mg/kg empasiprubart or placebo on day 1 followed by 50 mg/kg on day 8 and 20 mg/kg on day 22 (cohort B3), all doses administered as 2-h IV infusions. Cohort B1: n = 6 to week 33 and n = 5 after week 33; Cohort B2: n = 6 to week 3 and n = 5 for all timepoints after week 3; n = 6 to day 3, n = 5 to week 2, n = 4 for all remaining timepoints. Data were plotted as mean values ± SD of technical duplicates of samples collected from all participants within each cohort. IV intravenous, MAD multiple ascending dose, SAD single ascending dose, SD standard deviation.

Full size image
Table 1 Summary of empasiprubart PK parameters in the SAD part of the study
Full size table

To assess its potential to induce long-term inhibition of the CP and LP, the effects of MADs of empasiprubart were evaluated. Doses and intervals were based on the initial PK and PD results of the SAD part of this study and results previously obtained in cynomolgus monkeys16. Cmax after the first administration of empasiprubart during the MAD part of the study (Fig. 4b) was consistent with the results observed after single doses of 10 or 60 mg/kg. The elimination half-life of empasiprubart after multiple administrations ranged from 60 to 81 days (Table 2), which is in line with the observations after single dose administration (Table 1). In cohort B1, Cmax of empasiprubart increased from 212 to 505 µg/mL after four once-weekly doses of 10 mg/kg. In cohorts B2 and B3, empasiprubart Cmax reached 1060 and 1111 µg/mL after the last dose was administered.

Table 2 Summary of empasiprubart PK parameters after final administration (day 22) in the MAD part of the study
Full size table

PD of empasiprubart

PD analysis of empasiprubart included measurement of circulating levels of free and total (free and empasiprubart-bound) C2, of the activity of the CP, LP, and AP, and circulation levels of C1q, C3, C4, and C5. In the SAD part of the study, rapid, sustained, and dose-dependent reductions of free C2 levels occurred after administration of single doses of 2.5 to 80 mg/kg empasiprubart, whereas placebo had no significant effect on these levels (Fig. 5a). Immediately after the 2-h IV infusion of 60 and 80 mg/kg empasiprubart, free C2 levels were undetectable for 24 h, and reduced by up to 99% compared to baseline values for 1 week. A single IV administration of empasiprubart at 10, 30, 60, or 80 mg/kg induced a long-lasting reduction of free C2 levels by >90% compared to baseline for up to 63, 119, 203, and 259 days, respectively. IV administration of empasiprubart 2.5 mg/kg induced a gradual increase of total C2 concentration by 2.5-fold at a maximum, whereas the higher doses induced maximally three- to four-fold increases (Fig. 5b).

Fig. 5: Time-concentration profiles of PD parameters in the SAD part of the study.
figure 5

Healthy participants were dosed with 0.1, 0.5, 2.5, 10, 30, 60, and 80 mg/kg empasiprubart or placebo (i.e., cohorts A1–A6, A8), administered IV in a 2-h infusion. Cohorts A1 and A6: n = 6; Cohort A2: n = 6 except weeks 5, 7, and 13 (n = 5); Cohort A3: n = 6 except weeks 3, 13, 21, 37, and 39 (n = 5) and week 17 (n = 4); Cohort A4: n = 5, except week 2 (n = 4); Cohort A5: n = 5, except week 7 (n = 4); Cohort A8: n = 6 except weeks 4 and 9 (n = 5). a Serum concentrations of free C2 and b total C2 levels over time. Data were plotted as mean values ± SD. c Activity of the CP, d LP, and e AP over time. Data of AP activity represent levels at various timepoints of the placebo, 30 mg/kg, or 60 mg/kg cohorts (A5 and A6, respectively). LP and AP values are normalized against a human serum pool, which is set as 100% activity, data were plotted as mean values ± SD of technical duplicates of samples collected from all participants within each cohort. Note that individuals deficient in LP activity are not included in this representation of the LP data. AP alternative pathway, C2 complement factor 2, CP classical pathway, IV intravenous, LP lectin pathway, PD pharmacodynamic, RR reference range, SAD single ascending dose, SD standard deviation.

Full size image

To evaluate the impact of empasiprubart on CP activity, total hemolytic complement activity (CH50) was measured. As expected, all participants had normal CP activity at baseline. A dose-dependent reduction of CH50 occurred upon IV infusion of empasiprubart at doses of 2.5 to 80 mg/kg (Fig. 5c). In the 2.5 mg/kg cohort, this reduction was followed within days by an apparent increase of CH50 activity. Notably, levels varied widely between participants in this dose group at that time (Fig. 5c), with levels increasing in only three of the six participants. Such an increase was not observed in any other dose group and may reflect variation due to the semi-quantitative nature of the assay. At higher doses of empasiprubart, the reduction of CH50 activity following empasiprubart administration was more pronounced and sustained, gradually returning to baseline in all cohorts. Following infusion of empasiprubart at 80 mg/kg, CH50 levels immediately decreased by 95% and remained low until 21 weeks, when levels returned to normal (Fig. 5c). Of note, at empasiprubart doses of 10 mg/kg or higher, after a sharp initial decrease, CH50 activity was partially regained within the first 1 to 3 weeks, after which activity was more gradually restored (Fig. 5c). Free C2 levels followed a comparable pattern, that is, a partial restoration after an initial sharp decrease (Fig. 5a). It is tempting to speculate that this partial restoration of free C2 levels, and hence CH50 activity, may have reflected the mobilization of C2 from extravascular sources.

Similarly, empasiprubart induced an immediate dose-dependent sharp decrease of LP activity (Fig. 5d), which was followed by an early partial restoration and gradual increase thereafter, reaching near-baseline values by the end of the study (day 260). The partial restoration of LP activity during the first weeks likely reflected the same phenomenon as described above for CH50 activity. As expected, empasiprubart had no effect on AP activity, as is shown in Fig. 5e. Finally, empasiprubart had no significant effect on circulating levels of C1q, C4, C3, or C5 (Supplementary Fig. 2).

Multiple IV administrations of empasiprubart induced sustained reductions of free C2 levels in all cohorts (Fig. 6a), with a >99% reduction in free C2 levels from baseline after the first dose and a >98% reduction after the last dose. These reductions were consistent during the early observation period; although levels of free C2 were more variable toward the end of the study, reductions of >50% were still observed in two of six participants in cohort B1, five of six participants in cohort B2, and three of six in cohort B3 at week 41. Levels returned to baseline in all other participants, with free C2 values coinciding with normal complement activity as measured by the CH50 assay. These variations in free C2 levels at the end of the observation period were unrelated to other study parameters or anti-drug antibody (ADA) formation and probably reflected inter-individual biological differences.

Fig. 6: Time-concentration profiles of PD parameters in the MAD part of the study.
figure 6

Healthy participants were dosed with 10 mg/kg empasiprubart or placebo on days 1, 8, 15, and 22 (cohort B1), or 60 mg/kg empasiprubart or placebo on day 1 followed by 10 mg/kg on days 8 and 22 (cohort B2), or 10 mg/kg empasiprubart or placebo on day 1 followed by 50 mg/kg on day 8 and 20 mg/kg on day 22 (cohort B3), all doses administered as 2-h IV infusions. Serum concentrations of a free C2, b total C2 over time, and c CP activity are given. Data were plotted as mean values ± SD of technical duplicates of samples collected from all participants within each cohort. Cohort B1: n = 6 to week 33 and n = 5 after week 33; Cohort B2: n = 6 to week 3 and n = 5 for all timepoints after week 3; n = 6 to day 3, n = 5 to week 2, n = 4 for all remaining timepoints. C2 complement factor 2, CP classical pathway, IV intravenous, MAD multiple ascending doses, PD pharmacodynamic, RR reference range, SD standard deviation.

Full size image

As observed in the SAD cohorts, total C2 increased after multiple administrations of empasiprubart, up to a maximum three- to four-fold increase compared to baseline levels (Fig. 6b). This increase of total C2 lasted until the end of the study. As expected, multiple administrations of empasiprubart markedly reduced CP activity as measured with the CH50 assay (Fig. 6c). This activity was reduced by up to 76.1% (mean CH50 value = 25 U Eq/mL) after the last dose in cohort B1 and by up to 91.3% (mean CH50 value = 12 U Eq/mL) and 91.2% (mean CH50 value = 10.5 U Eq/mL) in cohorts B2 and B3, respectively. These marked decreases in CH50 levels were followed by a gradual return to baseline in all cohorts near the end of the study period, with the exception of one participant who had a CH50 activity level of 60% at week 41. As other parameters in this participant were unremarkable, this somewhat lower CH50 activity likely reflects biological variation. Finally, multiple IV doses of empasiprubart had no significant effects on circulating levels of C1q, C3, C4, and C5 (Supplementary Fig. 2).

Safety of empasiprubart

An overview of treatment-emergent adverse events (TEAEs), defined as AEs starting or worsening at the time of or after first dosing, across cohorts in both the SAD and MAD treatment arms are summarized in Table 3. Across the study, the number of reported TEAEs among participants receiving empasiprubart was comparable to those receiving placebo (46/58 [79.3%] vs 18/20 [90%] participants, respectively). This trend was also observed in individual treatment arms (75.0% of empasiprubart-treated participants vs 85.7% of placebo-treated participants in the SAD treatment arm, and 88.9% vs 100% of empasiprubart- vs placebo-treated participants in the MAD treatment arm). Across the whole study, related (empasiprubart, coronavirus disease 2019 [COVID-19] vaccination, or study procedure related) TEAEs occurred in 59.0% of participants and were less frequent in the SAD cohort (47.5% vs 57.1% of empasiprubart- or placebo-treated participants, respectively) than the MAD cohort (77.8% vs 83.3% of empasiprubart- or placebo-treated participants, respectively). All TEAEs were mild or moderate (grade 1 or 2) in severity. Severe or life-threatening TEAEs (grade 3 or 4) or deaths were not reported.

Table 3 Overview of TEAEs observed in the study
Full size table

The most frequently reported TEAEs in the total study population are listed in Table 4 and Supplementary Data 1. Among empasiprubart-treated participants, the most frequently reported AEs were headache (15.0%), diarrhea (12.5%), nausea (10%), and fatigue (7.5%) for participants in the SAD cohorts, and catheter site hematoma (33.3%) and headache (33.3%) in the MAD cohort (Table 4). The following empasiprubart-related AEs were reported by more than one study participant: fatigue (one participant each in cohorts B2 and B3 of the MAD treatment arm), feeling cold or hot (one participant who received 80 mg/kg empasiprubart reported feeling cold in the SAD part of the study; one participant reported feeling cold in cohort B2 and two participants reported feeling hot in cohort B3 in the MAD cohort), and headache (one participant in the 80 mg/kg group in the SAD cohort; two participants in cohort B2 and one in cohort B3 of the MAD group) (Supplementary Data 2).

Table 4 Overview of TEAEs (reported in ≥2 participants in the overall study population)
Full size table

One participant in cohort B3, with a history of asthma, hay fever, and house dust and mite allergy, reported an urticarial rash on day 4 after administration of empasiprubart (10 mg/kg). Except for this rash, this person had no other allergic symptoms. Without medication, the urticarial rash had resolved the next day, and this AE (grade 1) was considered to be probably related to the study drug. This participant was discontinued from further study treatment. Only one serious AE was reported, for a participant who developed an abscess of the lower back (by Klebsiella oxytoca bacteria) 39 days after a single IV dose of 2.5 mg/kg empasiprubart, the third lowest dose evaluated in the SAD part of the study. This event was considered unrelated to the study drug. Notably, 39 days after the administration of empasiprubart, free C2 levels and CH50 activity in this participant had returned to normal (see 2.5 mg/kg dose in Fig. 5a, c), which further supports that this AE was unrelated to the administration of empasiprubart. Infection events experienced by another 13 participants during the trial were all considered unrelated to empasiprubart administration. There were five infection events reported by five participants in the SAD part of the trial: three cases of nasopharyngitis, including one in a placebo-treated participant, one reported abscess in a participant receiving empasiprubart, and one case of COVID-19 in a participant receiving placebo. In the MAD cohort, no infection events were reported in the placebo group, whereas 12 AEs of infections were reported in eight empasiprubart-treated participants: folliculitis (n = 4), nasopharyngitis (n = 4), COVID-19 (n = 2), gastroenteritis (n = 1), and gingivitis (n = 1), none of which are typically associated with a genetic C2 deficiency10. As inherited deficiencies of the CP may predispose individuals to the development of systemic lupus, antinuclear and anti-dsDNA autoantibodies were measured in serum samples collected before administration of empasiprubart and afterwards at several timepoints (days 8, 29, 50, 92, 148, and 204), including at study end. No clinically meaningful changes in levels or presence of these autoantibodies were noted in any of the healthy participants receiving empasiprubart, and no AEs of lupus or autoimmune disorders were reported.

Overall, no clinically significant changes from baseline in laboratory values, vital signs, or electrocardiogram were observed in any of the study participants.

Immunogenicity of empasiprubart

In the SAD cohorts, ADA-positive baseline samples (i.e., pre-existing antibodies)17 were reported in two (5.0%) empasiprubart-treated participants. Thirty-seven (92.5%) empasiprubart-treated participants were ADA negative, and three (7.5%) were positive for treatment-induced ADAs in at least one sample during the study. In the MAD cohorts, two (11.1%) empasiprubart-treated participants had ADA-positive samples at baseline. Fourteen (77.8%) empasiprubart-treated participants were ADA negative, and four (22.2%) were ADA positive in at least one sample during the study and classified as treatment-induced ADAs. ADA levels were not boosted in any participant from either the SAD or MAD parts of the study following empasiprubart treatment. Overall, anti-empasiprubart ADA titers were low, and there was no clear correlation between incidence or prevalence of anti-empasiprubart ADA and the dose of empasiprubart received. There was no apparent impact of ADA on safety or on PK or PD of empasiprubart, identified upon visual inspection of individual PK/PD curves.

Discussion

Recently published clinical guidance estimates the risk of meningococcal disease to be elevated in patients treated with C3 or C5 inhibitors compared with healthy individuals18. Although reported data on the safety profiles of complement inhibitors remain conflicted, targeted inhibition of specific pathways may be preferred over total inhibition of the complement system. Previously, we described empasiprubart, a novel monoclonal antibody that inhibits C213 and has a long half-life. Empasiprubart binds to the CCP2 domain of C2 and prevents the formation of C4bC2 complexes, the C3 proconvertase of the CP and LP. Here, we provide the structural basis for its mechanism of action, and in addition show that administration of empasiprubart in humans is safe and induces sustained inhibition of both the CP and LP, but not of the AP of complement.

Empasiprubart binds C2 in a Ca2+- and pH-dependent manner13. This predicts that the antibody will behave in vivo as a so-called recycling antibody19,20, which, upon pinocytosis by endothelial cells, will release bound C2 to the lysosomal degradation pathway, while being recycled itself via FcRn into the circulation, resulting in a long clearance half-life. To unravel the molecular basis of these properties, crystals of empasiprubart in complex with a C2 Nt fragment were grown, allowing structure determination of the complex at 1.9 Å resolution. The high-resolution information provided by the crystal structure confirmed that the binding epitope of empasiprubart is located in the CCP2 domain of C2. This structure also provided the molecular basis of the calcium dependency of target binding by empasiprubart, by revealing a calcium ion in the heart of the paratope-epitope interface. Remarkably, LCDR3 Tyr96 seems to actively contribute to the calcium coordination, which could favor a deprotonated state of the side chain hydroxyl group. Given the predicted pKa value of 10.4621, deprotonation at a physiological pH would come with a significant energetic penalty. However, the apparent polarizability of the side chains of individual amino acids may vary significantly from that predicted by their pKa, dependent, among other factors, on their micro-environment21,22. The molecular mechanism of the pH dependency of empasiprubart is difficult to infer from the structure, as structures obtained at a neutral pH are practically indistinguishable from those obtained at a low pH. Often, the pH dependency of antigen-antibody binding is mediated by histidine residues functioning as a pH switch at the molecular interface. However, the empasiprubart:C2 interface residues include only one of three CDR histidines (HCDR3 His98), which does not seem to directly interact with C2. Dissection of the molecular mechanism underlying the pH dependency of empasiprubart is beyond the scope of this paper and will be reported elsewhere.

The primary objective of the first-in-human study was to characterize the safety, tolerability, and PK and PD profile of empasiprubart in healthy participants. To this end, single IV infusions of up to 80 mg/kg or multiple IV infusions of up to 60 mg/kg of empasiprubart were administered to 40 and 18 participants in the SAD and MAD parts of the study, respectively. The doses used in the MAD part of the study were selected because they were considered to yield maximal information about empasiprubart safety, its potential to induce a sustained inhibition of the CP in recipients, and to optimally build a PK/PD model. Notably, genetic deficiency of C2 is common in humans and often goes unnoticed23, though affected individuals have an increased risk of systemic lupus and of bacterial infections by encapsulated bacteria10,24. Typically, the infections occur at a young age and are often associated with decreased levels of one or more antibody class or subclass25,26. No such infections were observed during the trial reported here. Of the TEAEs reported by more than two participants in the overall study, numerically more empasiprubart-treated participants in the MAD part of the study reported infections than placebo-treated participants (Table 4). These were all grade 1 events, except one grade 2. COVID-19 and nasopharyngitis accounted for the majority of infections, and the remaining infections were folliculitis, a common skin condition frequently observed in healthy individuals. Two participants in the placebo group of the SAD part of the study developed similar infections (one report of nasopharyngitis and one report of COVID-19).

Hypersensitivity reactions are a recognized AE associated with the use of therapeutic antibodies16. No anaphylactic or severe/serious hypersensitivity reactions were reported in either the SAD or MAD parts of the study. One participant participating in the MAD part of the study developed a mild urticarial rash on day 4 after the first administration of empasiprubart, suggesting a type 1 delayed hypersensitivity reaction16. Typically, such reactions develop within minutes, or sometimes a few hours, after exposure to the drug, at least in individuals with pre-existing IgE antibodies. Occurrence on day 4 after exposure to empasiprubart does not definitively preclude the possibility that empasiprubart elicited the urticarial rash, but given this participant’s history of multiple allergies, it may have been caused by another allergy. Furthermore, none of the participants receiving empasiprubart developed an antinuclear or anti-DNA antibody response. Overall, empasiprubart was safe and well tolerated and did not raise any specific safety concerns.

A noticeable finding among the PD effects of empasiprubart was the rise of circulating total C2 levels up to three- to four-fold that of baseline levels at the highest doses administered. Notably, these increased levels of total C2 were associated with a marked reduction of free C2 levels, implying that the increase of total C2 levels was due to circulating levels of complexes of C2 and empasiprubart. As C2 bound to empasiprubart cannot interact with C4b, the risk of a rebound effect on complement activation due to increased total C2 levels seems minimal, if any.

One possible explanation for the increase in total C2 levels is that empasiprubart interfered with the clearance of C2 as described in the literature for interactions of monoclonal antibodies with soluble ligands27. For example, receptor-mediated clearance or cellular uptake of empasiprubart-C2 complexes may be slower compared to free C2. C2 is among the least studied complement factors, and the mechanisms involved in its clearance are unknown. Therefore, it cannot be ruled out that the increase of total C2 upon administration of empasiprubart is due to interference of the antibody with such mechanisms. Another possible explanation for the increase in total C2 is that empasiprubart inhibits a low-grade continuous activation of the CP and LP occurring under physiological conditions. Such low-grade activation is supported by the presence of detectable levels of C1s-C1-inhibitor and MASP-C1-inhibitor complexes in the circulation of healthy individuals, and very low C4 and C2 levels, independently of clinical symptoms, in patients with hereditary angioedema due to a deficiency of C1-inhibitor28,29,30, the main inhibitor of the CP and LP. Accordingly, increasing circulating levels of total C2 upon administration of empasiprubart may reflect increasingly effective inhibition of the CP and LP activation in vivo. Further studies should validate this concept and reveal to what extent free C2 levels need to be reduced to achieve a clinical benefit in complement-mediated diseases.

Several CP and LP inhibitors are in development. Among the first inhibitors clinically available are plasma-derived and recombinant C1-inhibitor products31. As with empasiprubart, these inhibitors target the CP and LP, and additionally inhibit factor XIIa and kallikrein of the contact system31. Compared to empasiprubart, C1-inhibitors are less potent CP/LP inhibitors, requiring supraphysiological levels for inhibition. This, combined with their relatively short clearance half-life (1–3 days for plasma-derived products and up to 4 h for recombinant C1-inhibitor), makes them unsuitable to treat chronic conditions. Therapeutic monoclonal antibodies in development or approved for clinical application include sutimlimab, a C1s inhibitor targeting the CP with a clearance half-life of up to 5.5 days approved for the treatment of cold-agglutinin disease32, and narsoplimab, a mannan-associated lectin-binding serine protease-2 (LP) inhibitor with a half-life of 8–9 days33 that is in development as a treatment for adult hematopoietic stem-cell transplantation–associated thrombotic microangiopathy34. As empasiprubart has broader specificity, inhibiting both the CP and LP, and a longer half-life, it may offer a preferred option for treating chronic clinical conditions associated with unwanted activation of both the CP and LP.

The pH and Ca2+ dependency of target binding in combination with the NHance mutations, which increase the affinity of empasiprubart for FcRn at a lower pH, is well known to contribute to the ability of antibodies to eliminate antigens35. Designed to be a recycling antibody capable of sustained inhibition of the CP, empasiprubart is equipped with these features13. Yet, its apparent elimination half-life of clearance from the circulation of 70 to 88 days after a single IV dose stands out even among those reported for other engineered antibodies36,37,38. Whether this half-life only results from its engineering, or whether other additional features contribute, is a topic of further study. Whatever the mechanism, the findings reported here collectively show that empasiprubart is a first-in-class, recycling anti-C2 agent that is generally well tolerated. With its dose-dependent long elimination half-life, and efficacy in reducing free C2 levels and CH50 activity, empasiprubart may offer a novel treatment option for patients with complement-mediated disorders requiring sustained inhibition of both the CP and LP of the complement system. This concept is currently being investigated in a phase 2 study (ARDA; NCT05225675) and extension trial (ARDA+; NCT05405361) in patients with multi-focal motor neuropathy, and in a second phase 2 study evaluating the safety, efficacy, and tolerability of empasiprubart in recipients of a renal allograft at risk for delayed graft function (VARVARA; NCT05907096). Dose regimens for these trials were based on modeling using the data obtained from the first-in-human study reported here.

Methods

Protein production, crystal structure determination, and crystal model refinement

The C2 Nt fragment of human C2 (residues 1–217, Uniprot entry: P06681) was transiently expressed in HEK293-F cells (Thermo Fisher Scientific, catalog number R79007), prepared from low-passage master cell bank derived from parental HEK293-F cells that were recloned by limiting dilution and maintained in serum-free conditions for 30–35 total passages. The secreted C2 Nt-containing residues C2 21–217 was purified on a 5 mL HisTrap column (GE Healthcare). The empasiprubart Fab was prepared by limited proteolysis with papain (from papaya latex, Sigma). The anti-κ nanobody, κ Nb, was expressed and purified as described previously39 and used to reduce the conformational mobility of the empasiprubart Fab.

The C2 Nt:ARGX117 Fab:κ Nb ternary complex (1:1.3:2 molar ratio) was purified by size exclusion chromatography, and crystals were obtained using hanging-drop vapor diffusion. Diffraction data were collected at 100 K at the MAX IV beamline BioMAX and processed with XDS40. The structures were determined with phenix.phaser41,42, manually rebuilt in Coot43, and further refined with phenix.refine in an iterative manner using positional refinement, individual B-factors, and a Translation-Liberation-Screw-rotation model of rigid-body harmonic displacements. Molecular graphics figures were prepared using PyMOL (The PyMOL Molecular Graphics system, version 4.6.0, Schrodinger, LLC). Throughout this manuscript, Kabat numbering has been used for empasiprubart. Crystallographic coordinates and data have been deposited in the PDB (www.rcsb.org) PDB entry ID 8ACI and 8ACF. Full details are provided in the Supplementary Information.

Study population and trial design

From July 16, 2020, to August 26, 2022, we performed a first-in-human, single-center, randomized, double-blind, placebo-controlled, phase 1 study evaluating SADs and MADs of empasiprubart administered IV as a 2-h infusion in healthy males and females of non-childbearing potential (NCT04532125). Sex as assigned to the participant at birth was self-reported by the patient with the available binary options of “male” or “female” and was collected in this study for demographics and baseline characteristics. Gender was not collected in this study. All patients provided written informed consent before the commencement of any study-related procedures. Study participants were selected based on the inclusion and exclusion criteria detailed in the Supplemental Information. This study was performed by ICON, previously PRA Health Sciences (PRAH) (The Netherlands). The primary objective of the study was to evaluate the safety and tolerability of empasiprubart. The secondary objectives were to determine PK, PD, and immunogenicity of empasiprubart. As empasiprubart is not anticipated to increase the risk of infection, participants were not vaccinated against encapsulated bacteria (Neisseria meningitidis, Streptococcus pneumoniae, or Haemophilus influenzae).

Participants were allocated to receive a placebo or empasiprubart according to a randomization list prepared by ICON using SAS software (SAS Institute Inc., USA). An unblinded pharmacist or appropriately qualified member of the study staff prepared the study drug that corresponded to the assigned participant randomization number. The participants, clinical study staff, and sponsor were blinded to treatment. The placebo matched empasiprubart in appearance and contained the same formulation, but without the active ingredient.

Of the 320 healthy participants screened, 78 were randomized to either the SAD (n = 54) or MAD cohorts (n = 24). Participants participating in the SAD part of the study received either placebo or 0.1, 0.5, 2.5, 10, 30, 60, or 80 mg/kg empasiprubart (cohorts A1–A6 and A8; Fig. 3a). In cohorts A1 and A2, two sentinel participants were randomized to empasiprubart and one to placebo. The remaining participants in these cohorts were randomized in a 4:1 ratio to empasiprubart or placebo, respectively. In cohorts A3‒A6 and A8, one sentinel participant was randomized to empasiprubart and one other to placebo; the remaining participants were randomized to empasiprubart or placebo in a 5:1 ratio. Dose escalation to the next SAD cohort as well as initiation of MAD cohorts was dependent upon satisfactory review of the blinded safety, PK, and PD data by the safety committee, consisting at a minimum of the sponsor’s medical monitor, safety physician, and clinical pharmacologist, the principal investigator, and an independent immunologist.

Participants of the MAD part of the study received IV doses of (1) 10 mg/kg empasiprubart or placebo on days 1, 8, 15, and 22 (cohort B1); or (2) 60, 10, and 10 mg/kg empasiprubart or placebo on days 1, 8, and 22, respectively (cohort B2); or (3) 10, 50, and 20 mg/kg empasiprubart or placebo on days 1, 8, and 22, respectively (cohort B3; see Fig. 3b). Participants of each of these cohorts were randomized to empasiprubart or placebo in a 6:2 ratio. The procedure for dose escalation was identical to that of the SAD part.

Serum collection schedules for the SAD and MAD parts of the study are detailed in the Supplementary Information.

PK of empasiprubart

A validated quantitative ELISA was used to determine total empasiprubart concentrations in human serum samples. The following PK parameters were estimated and calculated using a non-compartmental method: AUC0–t, AUC0‒168 h, or AUC0–inf, Cmax, tmax, apparent terminal half-life (t½) of clearance from the circulation, CL (SAD part only), and Vz (SAD part only). Cmax, AUC0‒168h, and AUC0‒inf for empasiprubart were compared across each dose level in the SAD part to assess dose proportionality.

PD of empasiprubart

To evaluate the PD of empasiprubart, serum concentrations of total (free and empasiprubart-bound) and free C2, of C1q, C3, C4, and C5, and of the activities of the CP, LP, and AP were measured. Total C2 concentrations were determined using an automated quantitative ELISA. Free C2 in serum, not bound by empasiprubart, was measured using a validated quantitative method on the Gyros® platform. Activities of CP and AP were measured using the MicroVue CH50 Eq EIA (A018; Quidel) and Wieslab AP330 assay, respectively, according to the manufacturers’ instructions. LP activity was measured using an in-house assay developed by argenx. Serum levels of C1q, C3, C4, and C5 were measured using routine diagnostic assays (immune nephelometric assays performed by Sanquin, Amsterdam, The Netherlands). Full details are provided in the Supplementary Information.

Serum concentrations of complement proteins were measured over time following administration of empasiprubart and used to assess the maximum percent reduction value, time to reach maximum serum concentration (tmax), and area under the percent reduction curve from pre-dose to the last sampling timepoint. AP activity was measured at multiple timepoints and values were pooled per treatment group.

Statistics

All statistical calculations for PK and PD analyses were performed using SAS software (version 9.4; SAS Inc., USA) and Phoenix WinNonlin software (version 8.1; Certara, USA). No formal sample size calculations were performed. Descriptive statistics such as number of participants, arithmetic mean, standard deviation, coefficient of variation, median, minimum, maximum, range, and 95% confidence interval were used to summarize continuous data, including serum levels, and derived absolute and relative changes compared to baseline. Categorical data were summarized using counts and percentages. Dose proportionality was assessed only in the SAD part using a power model.

Safety

Empasiprubart safety data included AE monitoring and vital sign measurements, electrocardiogram recordings, physical examination, and clinical laboratory test results (clinical chemistry, international normalized ratio, urinalysis, hematology, and systemic lupus erythematosus panel test). Safety was assessed based on AE reporting. The number of events, frequency, and percentage of TEAEs (coded using the Medical Dictionary for Regulatory Activities [version 23.1]) were calculated, and summarized overall and by dose level for the SAD and MAD parts of the study. The severity (mild, moderate, or severe) and causality (unrelated, unlikely, possibly, probably, or certainly related) of each AE were assessed by the investigator. TEAEs were considered related to the study drug when classified as possibly, probably, or certainly related.

Immunogenicity

Anti-empasiprubart antibodies were detected using a multi-tiered electrochemiluminescence immunoassay method developed and validated with a pre-treatment step to meet the drug tolerance requirements. Full details are provided in the Supplementary Information.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.