CSL305: A Dual Functional Therapeutic Antibody Targeting Complement C2 and FcRn

Abstract

Complement and pathogenic antibodies act independently and together to mediate the pathology of many autoimmune diseases. To address these drivers of disease, we generated a monoclonal antibody (mAb), CSL305, that binds and inhibits both complement and the neonatal Fc (fragment crystallizable) receptor FcRn. The fragment antigen binding (Fab) portion of CSL305 was engineered to bind both human C2 (huC2) zymogen and the active fragment huC2b to inhibit the classical and lectin complement pathways in vitro, and C3b deposition on primary lung endothelial cells using a 3-dimensional microvascular model system. Engineering of a triple amino acid mutation (“YPY” motif) into the Fc region of CSL305 increased its affinity to FcRn at both acidic and neutral pH, allowing it to also act as a potent FcRn antagonist. Intracellular trafficking experiments demonstrated that CSL305, but not the wild-type (WT) mAb lacking the YPY motif, was able to block immunoglobulin G (IgG) recycling in vitro. The generation of a high resolution 2.6Å crystal structure of CSL305 Fab region bound to huC2b showed that the epitope lies directly over the huC2b catalytic triad, providing evidence of its complement mechanism of action as a neutralising mAb. Early pharmacokinetic (PK)/pharmacodynamic (PD) studies using CSL305 in cynomolgus monkeys demonstrated both complement inhibition and FcRn antagonism in vivo, with reductions in complement classical pathway activity and endogenous IgG observed following single intravenous (IV) administration. CSL305 thus represents a dual-functional mAb as a potential therapeutic candidate.

1. Introduction

Autoimmunity is a complex phenomenon broadly defined as a condition where one’s own immune system recognizes proteins, known as autoantigens, on and within cells and tissues as foreign and elicits an immune response against them [1,2]. The autoimmune response involves a number of common mechanisms, such as the production of pathogenic autoantibodies, immune complex formation and complement activation, and T cell-mediated damage, and can be either organ/tissue-specific or systemic [1,3,4,5]. The autoimmune diseases that arise from this aberrant immune response often utilize multiple mechanisms, two of which we will focus on: complement activation (or dysregulation) and autoantibody production.

A review of the literature reveals several autoimmune indications that show evidence of complement and autoantibody involvement. These include, but are not limited to, myasthenia gravis [6,7]; neuromyelitis optica spectrum disorder [8,9]; lupus nephritis [10,11]; warm autoimmune hemolytic anemia [12,13]; dermatomyositis and other myositis indications [14,15]; and chronic inflammatory demyelinating polyneuropathy [16,17]. Antibody-mediated rejection could also be added to this list, although the pathogenic donor-specific antibodies are alloantibodies, rather than autoantibodies [18,19,20].

Targeting either complement or autoantibodies as a therapeutic approach requires us to understand the details of each biological mechanism. Complement is an important arm of the innate immune system and defines a system of multiple proteins that are sequentially activated and regulated as a cascade, and which lead to specific effector functions such as opsonization, anaphylatoxin release, and lysis [21]. Complement activation can be broadly divided into three pathways—classical, lectin, and alternative—depending on the initiating factors; the classical and lectin pathways converge at the level of complement C2, and all pathways at the level of C3 [21,22,23]. In addition to its well-known effects in the fluid phase, intracellular complement and its involvement in physiological processes has become a topic of recent interest [24].

IgG (which includes autoantibodies) and albumin homeostasis is regulated by the neonatal Fc receptor FcRn, a membrane-bound receptor composed of a unique major histocompatibility complex class I-like alpha chain, bound as a heterodimer to beta-2-microglobulin (B2M) [25]. Following internalization via fluid phase endocytosis, FcRn engages the most abundant immunoglobulin subtype IgG (but not other Ig subtypes) and albumin under acidic endosomal conditions. These complexes are then redirected back to the cell surface where the shift back to neutral pH promotes the dissociation of the complex and release of IgG and human serum albumin (HSA) back into the circulation. Non-FcRn-bound molecules and immune complexes alternatively move into the lysosomal compartment and are degraded. As a major contributor to IgG’s long (~21-day) serum half-life, FcRn is thus also recognized as an attractive target in autoimmune and other conditions where the pathology is mediated by IgG [25,26].

Targeting complement or FcRn are proven therapeutic strategies for the treatment of autoimmune disease. Therapeutics developed to effectively target each mechanism individually have been described. For complement [27], these include the terminal complement pathway inhibitor Eculizumab [28,29]; the pan complement inhibitors Pegcetacoplan [30] and the soluble complement receptor 1 fragment CSL040 [31,32]; and those that are pathway specific, such as the anti-C1s mAb Riliprubart/SAR445088 [33] and anti-C2 mAb Empasiprubart (ARGX-117) [34]. Recombinant FcRn inhibitors such as the Fc fragment Efgartigimod [35] and the anti-FcRn mAbs Nipocalimab (M281) [36] and Batoclimab (HBM9161) [37] have been developed as potent antagonists of FcRn to mediate the reduction of IgG in diseases such as generalised myasthenia gravis, where autoantibodies are the major drivers of pathology [25]. Efgartigimod, for example, reduced endogenous IgG levels by approximately 50% and 85% following single and multiple doses to healthy volunteers, respectively, with no effect on IgA, IgD, IgE, or IgM [38,39]. The effects of targeting both complement and FcRn together, rather than separately, are unclear. However, combination therapy is potentially challenging from a mechanistic perspective, as elimination of an IgG-based complement inhibitor would be expected to be accelerated by the co-administered FcRn antagonist.

In this study, we describe the generation and characterization of CSL305, a dual-functional mAb able to both potently antagonize FcRn via selective mutations within the Fc region, and to selectively inhibit the classical and lectin complement pathways via binding to complement C2, whilst leaving the alternative pathway intact.

2. Results

2.1. Generation of CSL305

A CSL proprietary naïve Fab phage-display library was screened for clones that bound both human C2 (huC2) zymogen and the active fragment huC2b as per recommended nomenclature [40], and 25 unique clones were identified. These unique clones were reformatted and expressed as whole human IgG4 mAbs with a hinge-stabilizing mutation (S228P; EU numbering) and assayed for potency. The most potent mAb in terms of inhibition of the complement classical and lectin pathways in vitro, hu4D8, was selected for back-mutations in the heavy and light chain framework regions to the corresponding germline sequences based on IgBLAST analysis [41]. Affinity maturation of this mAb was then undertaken, focusing on the six complementarity determining region (CDR) loops as defined by the Kabat scheme, and a mAb referred to as aghu4D8 or “WT” was identified. To generate the dual-functional CSL305 mAb, a triple mutation in the IgG4 Fc region (M252Y, V308P, N434Y, “YPY”; EU numbering) was introduced; in all other respects, the WT and CSL305 mAbs are identical.

The affinity of the WT and CSL305 mAbs to recombinant huC2 and its huC2a and huC2b fragments was determined by surface plasmon resonance (SPR) and the results shown in

Table 1. Binding of human C2, C2a, and C2b to CSL305. Shown are the kinetic rate constants and affinities of CSL305 and WT (aghu4D8) mAbs to huC2, huC2a, and huC2b. k_a, association rate constant; k_d, dissociation rate constant; K_D, equilibrium dissociation constant. K_D indicated as mean ± SD (in nM) from multiple experiments as indicated. Values were calculated from sensorgram data fit to a 1:1 binding model. C2a is designated as the smaller fragment following cleavage of C2 zymogen; C2b is the larger, catalytically active fragment.

Anti-C2 mAb	Description	ka (1/Ms)	kd (1/s)	huC2 KD (nM)	N
WT	Affinity-matured wild-type	5.12 × 104	5.47 × 10−4	10.8 ± 1.2	57
CSL305	Affinity matured with Fc-YPY	5.05 × 104	5.49 × 10−4	11.0 ± 0.9	26
		ka (1/Ms)	kd (1/s)	huC2a KD (nM)
WT	Affinity-matured wild-type	No Binding			3
CSL305	Affinity matured with Fc-YPY	No binding			3
		ka (1/Ms)	kd (1/s)	huC2b KD (nM)
WT	Affinity-matured wild-type	3.61 × 104	7.79 × 10−4	21.7 ± 2.1	52
CSL305	Affinity matured with Fc-YPY	3.38 × 104	7.89 × 10−4	23.0 ± 2.1	24

and Figure S1. The affinity of the WT mAb to huC2 was 10.8 ± 1.2 nM and to huC2b 21.7 ± 2.1 nM; no binding to huC2a was detected. The YPY mutations within CSL305 do not affect the affinity of CSL305 to huC2 and huC2b compared to WT mAb, where values of 11.0 ± 0.9 nM and 23.0 ± 2.1 nM were measured for huC2 and huC2b, respectively (Table 1, Figure S1). CSL305 and WT mAbs are cross-reactive with cynomolgus monkey (cyno) C2 and C2b, with similar affinities compared to each other and to huC2/C2b measured (Table S1). However, no binding to additional surrogate species (mouse, rat, rabbit, dog, pig, and sheep) could be determined (Table S1). The ability of CSL305 to show pH- or calcium-dependent binding to huC2 was also assessed. As shown in

Table 2. Calcium- and pH-dependent binding of CSL305 to human C2. SPR was used to assess the affinity of CSL305 to huC2 in the presence or absence of calcium, at both acidic and neutral pH. A control anti-C2 mAb was also included. k_a, association rate constant; k_d, dissociation rate constant; K_D, equilibrium dissociation constant. K_D indicated as Mean ± SD (in nM) from N = 4 measurements. Values were calculated from sensorgram data fit to a 1:1 binding model.

Anti-C2 mAb	Experimental Conditions	ka (1/Ms)	kd (1/s)	huC2 KD (nM)	N
CSL305	pH 7.3/2 mM Ca2+	5.30 × 104	6.14 × 10−4	11.6 ± 0.21	4
CSL305	pH 7.3/0 mM Ca2+	5.79 × 104	6.24 × 10−4	10.8 ± 0.23	4
CSL305	pH 6.0/2 mM Ca2+	1.20 × 105	9.57 × 10−4	8.0 ± 0.12	4
CSL305	pH 6.0/0 mM Ca2+	1.04 × 105	8.63 × 10−4	8.4 ± 0.54	4
Control	pH 7.3/2 mM Ca2+	4.48 × 105	1.12 × 10−4	2.5 ± 0.10	4
Control	pH 7.3/0 mM Ca2+	-	-	No Binding	4
Control	pH 6.0/2 mM Ca2+	2.01 × 105	4.37 × 10−3	21.7 ± 0.29	4
Control	pH 6.0/0 mM Ca2+	-	-	No Binding	4

and in contrast to a control anti-C2 mAb, CSL305 showed only minor differences in affinity to huC2 at pH 6.0 compared to pH 7.3 and no difference in affinity in the presence or absence of 2 mM calcium.

The introduction of a triple mutation (M252Y, V308P, N434Y; the YPY motif) to the Fc region of mAbs has previously been shown to significantly increase their affinity to FcRn [42]. When the affinity of CSL305 to a complex of huFcRn and B2M was assessed and compared to the WT mAb under both acidic and neutral pH, a large increase in affinity for CSL305 to huFcRn/B2M at pH 6.0 was measured compared to WT (4.7 ± 0.1 nM and 1850 ± 14 nM, respectively;

Table 3. Binding of human FcRn/B2M to CSL305 under acidic and neutral conditions. Shown are the kinetic rate constants and affinities of CSL305 and WT (aghu4D8) mAbs to huFcRn/B2M. k_a, association rate constant; k_d, dissociation rate constant; K_D, equilibrium dissociation constant. K_D indicated as Mean ± SD (in nM) from N = 4 measurements. “Weak binding” refers to affinities over 10 μM. Acidic pH: 6.0; Neutral pH: 7.3.

Anti-C2 mAb	Analyte (pH 6.0)			Analyte (pH 7.3)
	ka (1/Ms)	kd (1/s)	KD (nM)	ka (1/Ms)	kd (1/s)	KD (nM)
WT	-	-	1850 ± 14	-	-	Weak binding
CSL305	4.25 × 105	2.00 × 10−3	4.7 ± 0.1	-	-	2424 ± 70

, Figure S2), attributable to the introduction of YPY. At (neutral) pH 7.3, only weak binding was observed for the WT mAb, but an affinity of 2424 ± 70 nM for CSL305 to huFcRn/B2M was measured (Table 3, Figure S2). The affinities of CSL305 and WT mAbs toward cynoFcRn/B2M were similar to those measured to huFcRn/B2M (Table S2). Comparative binding of CSL305 and WT mAbs to the human Fc gamma receptors cluster of differentiation (CD) 16 (158F), CD32A, CD32B, and CD64 by SPR was also conducted, with less than 2-fold differences in affinity measured (Table S3).

2.2. CSL305 Inhibits Complement Activity In Vitro

To assess the ability of CSL305 to block complement activity in vitro, we employed Wieslab^® commercial assays specific for the classical, lectin, and alternative pathways. As shown in Figure 1A, CSL305 was able to completely block classical pathway activity in a dose-dependent manner, with a calculated IC₅₀ value of 6.4 ± 0.7 nM. This activity was specific, given that an isotype control IgG4 mAb showed no inhibitory activity. The WT mAb showed similar activity compared to CSL305, with an IC₅₀ value of 5.2 ± 2.4 nM, indicating that the presence of the YPY motif on CSL305 did not deleteriously affect its inhibitory potential. In comparison, the CSL040 control was more potent for the classical pathway (IC₅₀: 0.5 ± 0.3 nM; Figure 1A). A similar profile was observed for the lectin pathway: CSL305 and WT mAbs were equipotent, with determined IC₅₀ values of 1.2 ± 0.9 nM and 1.3 ± 0.3 nM, respectively. CSL040 was more potent, with an IC₅₀ of 0.2 ± 0.04 nM, and the isotype IgG4 control mAb showed no inhibitory activity (Figure 1B). A classical pathway hemolytic (CH50) assay was also used to assess CSL305 potency in vitro. Complete and dose-dependent inhibition of CH50 activity was observed (Figure 1C) with CSL305 (IC₅₀: 12.7 ± 6.2 nM). No significant difference in WT mAb CH50 inhibitory activity was observed compared to CSL305 (IC₅₀: 14.2 ± 9.5 nM), confirming lack of an effect of the YPY mutation on CSL305 complement inhibitory activity; the isotype control showed no activity (Figure 1C). In contrast, neither CSL305 nor WT mAbs showed any inhibition of the complement alternative pathway as measured by the Wieslab^® assay (Figure 1D), or by the hemolytic (AH50) assay (Figure 1E), with the CSL040 positive control showing potent inhibition in both assays (IC₅₀: 0.81 ± 0.06 nM and 4.12 ± 1.13 nM, respectively). Potencies of CSL305 and the WT mAbs were also comparatively assessed using sera of several animal species. As shown in Figure S3, and confirming the binding data above, both mAbs were found to be similarly cross-reactive with only human and non-human primate species (Rhesus and Cynomolgus monkey). No inhibition of the classical pathway of mice, rats, rabbits, pigs, minipigs, or dogs was observed for either CSL305 or WT mAbs.

We then assessed the inhibition of complement pathway activation by CSL305 in a 3D in vitro micro-vessel system to mimic a human inflammatory setting. Primary human lung microvascular endothelial cells (HLMVECs) were grown in circular microchannels and cultivated under pulsatile flow conditions. Once confluent, HLMVECs were opsonized with an antibody against the surface protein CD105 to model an antibody-driven immune response. Microchannels were perfused with normal human serum (NHS, 20% in culture medium) and increasing concentrations of CSL305 (25, 50, 100, 500 nM), WT mAb (500 nM), or the CSL040 positive control (100 nM). Complement component C3b, HLMVEC cytoskeleton, and cell nuclei were immuno-fluorescently labeled and visualized using confocal microscopy (Figure 2A). The fluorescence intensity of C3b as a measure of deposition was then quantified as shown in Figure 2B. Mean C3b fluorescence intensity (arbitrary units) in un-opsonized HLMVECs was measured at 0.75 ± 0.54, which increased to 3.6 ± 1.13 upon opsonization, and was inhibited by the CSL040 positive control (1.13 ± 0.40; Figure 2B). CSL305 treatment reduced C3b deposition on HLMVECs in a concentration-dependent manner, with mean C3b fluorescence intensity values of 2.46 ± 1.18, 1.52 ± 0.73, 1.45 ± 0.65, and 0.72 ± 0.29 at doses of 25, 50, 100, and 500 nM, respectively (Figure 2B). All reductions in C3b deposition, except for the lowest concentration of CSL305, were statistically significant compared to untreated (otherwise opsonized and perfused) cells. 500 nM of WT mAb also significantly inhibited C3b deposition compared to untreated HLMVECs (1.14 ± 0.56); this level of inhibition was not significantly different from an equimolar dose of CSL305 (Figure 2B).

2.3. CSL305 Inhibits IgG Recycling In Vitro

We next assessed the ability of CSL305 to antagonize recycling of IgG and HSA using an in vitro intracellular trafficking assay. The results of this experiment are shown in Figure 3A and Figure S4A, with the quantification of fluorescent intensity also shown in Figure 3B–E and Figure S4B–E. Uptake of fluorescently labelled IgG (“No antagonist”) by macropinocytosis was readily observed (Figure 3A) and could be measured in the presence or absence of protease inhibitors (Figure 3B,C) at the 0 min chase time point. After a 15 min chase, labelled IgG1 was no longer detectable (Figure 3A,D), and signal was not rescued by protease inhibitors (Figure 3A,E), suggesting IgG was recycled by the cell rather than transported to the lysosomal compartment for degradation. No impact on this activity was observed when recombinant Fc-G1 (the Fc domain of wild-type human IgG1) and the WT anti-C2 mAb were added (Figure 3A–E). In contrast, a reduced signal for IgG uptake at time 0 min post pulse was observed in the presence of the FcRn antagonists CSL305 and control FcYTEKF protein (Figure 3A–C). This suggests a reduced interaction of labelled IgG1 with membrane-associated FcRn within the macropinosomes and the loss of free labelled IgG1 from the core of the macropinosome following the fixation step, rather than reduced internalization. Indeed, with live cell imaging, without fixation, uptake of labelled IgG by macropinocytosis could be demonstrated even in the presence of CSL305 (Figure S4F). Following a 15 min chase prior to fixation, in cells treated with CSL305 or FcYTEKF, labelled IgG1 was absent without protease inhibitors (Figure 3A,D), but detectable with the addition of protease inhibitors (Figure 3A,E), indicating lysosomal delivery and antagonism of FcRn-mediated IgG recycling. In a similar manner, we also examined whether CSL305 affected human serum albumin (HSA) recycling; the same experiment was conducted with fluorescently labelled HSA. In contrast to the IgG1 data, all molecules, including CSL305, did not prevent the uptake of labelled HSA (Figure S4A–C). After 15 min chase, all uptake was lost, which was not prevented with the addition of protease inhibitors (Figure S4A,D–E), indicating recycling of HSA. For all experiments, a Hoechst 33342 nuclear stain was used as a cellular control to verify nuclear integrity in the fixed cells used (Figure 3A and Figure S4A). The combined data suggest that IgG, but not albumin recycling, is inhibited by CSL305.

2.4. Structure–Function Analysis of the CSL305-C2 Mechanism of Action

The structural basis of complement inhibition by CSL305 was investigated using X-ray crystallography. A complex of C2b with an Fab fragment of CSL305 (identical to that of the WT/aghu4D8 Fab) was prepared by mixing a 1.2-fold molar excess of purified CSL305 Fab with C2b, followed by gel filtration to obtain a homogeneous C2b–CSL305 Fab complex (Figure S5). The crystal structure of the C2b–CSL305 Fab complex was determined at 2.6 Å resolution in space group P1 (Figure 4A;

Table 4. Diffraction data statistics for CSL305-C2b complex.

Data Collection Space Group	CSL305-C2b P1
Cell dimensions (a, b, c) Å (α, β, γ)°	62.24, 67.79, 155.37 98.32, 90.68, 101.2
Resolution (Å) a	48.63 − 2.6 (2.66 − 2.6)
Completeness (%) a	94.1 (95.8)
I/σI a	8.2 (1.1)
Rmerge b	0.053 (0.557)
R p. i. m. (%) c	5.3 (55.7)
CC 1/2 d	0.995 (0.627)
Number of reflections e	138,176 (71,245)
Redundancy	1.9 (2.0)
Wilson B factor (Å2)	48.6
Refinement
Resolution (Å)	48.63 − 2.6
Rwork (%) f/Rfree (%) g	20.76/25.61
Non-hydrogen atoms
Protein	14,236
Water	143
Ligands	182
r.m.s.d. from ideal geometry
Bond lengths (Å)	0.03
Bond angles (°)	0.684
Ramachandran plot
Most favoured and allowed region (%)	99.9
PDB code	pdb_00009zcj

^a Values in parentheses are for the highest resolution shell. ^b R_merge = å |(I_i − <I>)|/å |I|, where I_i is the intensity of an individual reflection and <I> is the average intensity of that reflection. ^c R_pim, precision-indicating (multiplicity-weighted) R_merge. ^d CC1/2 is the correlation coefficient of the mean intensities between two random half-sets of data. ^e Values in parentheses represent the number of unique reflections. ^f R_work = ∑||Fo| − |Fc||/∑|Fo|, where Fo and Fc are the observed and calculated structure factors for reflections, respectively. ^g R_free was calculated as R_work using the 5% of reflections that were selected randomly and omitted from refinement.

) and shows a 1:1 stoichiometry. Structural comparison of CSL305-bound C2b (protein data bank (PDB) ID: 9ZCJ) with the apo conformation C2b structure (PDB ID: 2ODP [43]) revealed minimal conformational changes in the von Willebrand factor type A (VWA) and serine protease (SP) domains upon Fab binding (Figure S6). The superposition yielded a root mean square deviation of 1.0 Å, and the catalytic triad residues (His487, Asp541, Ser659) retained their orientation. Notable differences were confined to loop regions that directly contact the complementarity-determining region (CDR)-H3 of CSL305, such as Gln624–Asp634. Interestingly, the Leu686–Asn692 loop, unresolved in previous structures, was stabilized upon Fab binding through interactions with CDR-H2.

Detailed inspection indicates that CSL305 inhibits complement activity by active-site occlusion, achieved through penetration of CDR-H3 into the catalytic cleft of C2b (Figure 4B). This interaction buries approximately 1054 Ų of surface area, with ~700 Ų contributed by the heavy chain and ~355 Ų by the light chain. The paratope primarily comprises CDR-H3, H2, L1, and L2 loops, which form a network of salt bridges, hydrogen bonds, and π-stacking interactions (Figure 5). The extended CDR-H3 loop engages deeply within the catalytic cleft, where Leu105, Trp106, and Phe107 establish coordinated contacts with C2b residues, including Glu536, Tyr538, Lys656, Tyr682, and Trp679. Critically, Trp106 lies adjacent to Ser659 of the catalytic triad (Figure 5B), potentially forming π-stacking interactions and sterically blocking substrate access. Trp106 also appears to directly interact with Ser659 through hydrogen bonds. Charged residues from CDR-H2—Asp57, Asp59, and Arg61—create key salt bridges with Asp690 and Lys695 (Figure 5A). While light-chain residues form fewer direct contacts, they contribute affinity-enhancing interactions; for example, CDR-L1 Asp33 forms an ionic interaction with Gln469, and Arg26 and Val31 form hydrogen bonds with Lys467 and Gln469, respectively (Figure 5C). Additional stabilization is provided by a salt bridge between CDR-L2 Arg54 and Glu536 of C2b. Notably, Lys67, positioned outside canonical IMGT/Kabat CDRs, forms hydrogen bonds with Lys467 and Glu469 (Figure 5D).

Overall, the structure of CSL305 Fab bound to C2b reveals its mechanism of complement inhibition. CSL305 binding directly disrupts C2 protease function by extending CDR-H3 into the catalytic cleft, sterically occluding the active site and preventing substrate access. The proximity of Trp106 to Ser659 and its potential π-stacking interaction further compromises catalytic triad chemistry by interfering with the proton relay required for cleavage (Figure 5B). CSL305 would therefore block C3 convertase formation and downstream complement activation events, including C3 cleavage, opsonization, and membrane attack complex assembly.

Despite the common epitope of WT and CSL305 mAbs lying over the catalytic triad of huC2/C2b and the mechanistic implications thereof, we wished to confirm whether huC1s-mediated cleavage of huC2 might also be impacted. We used the parental hu4D8 mAb, which was shown to have an almost identical epitope to CSL305—as determined by exhaustive alanine mutagenesis (Table S4)—to inhibit huC2 cleavage by huC1s using a purified system. Compared to a control anti-C2 mAb that was able to block huC1s-mediated huC2 cleavage, hu4D8 showed no effect on huC2 cleavage in the presence or absence of huC1s (Figure S6A), even at a 10-fold molar excess over huC2 (Figure S6B).

2.5. Comparative Assessment of CSL305 and WT mAb PK/PD in Mice and Cynomolgus Monkeys

Having demonstrated CSL305 acting as a complement inhibitor and FcRn antagonist in separate experiments in vitro, we made a comparative in vivo assessment against the WT anti-C2 mAb. Despite the lack of cross-reactivity of both CSL305 and WT mAbs to mouse C2, their PK and effect on endogenous IgG levels could still be tested. A single dose of 10 mg/kg of WT mAb or 10 and 100 mg/kg of CSL305 was administered IV to mice, and PK and mouse IgG levels were measured over 7 days post dose. To account for differences in affinity between human IgG binding to human or mouse FcRn, huFcRn transgenic mice (32HOM) were also tested alongside wild-type mice. A single dose of each mAb was well tolerated. As shown in Figure 6A and

Table 5. Clearance and half-life of CSL305 and WT mAb administered IV at a dose of 10 or 100 mg/kg in mice and cynomolgus monkeys.

Species	Dose (mg/kg)	Clearance (mL/h/kg)		Half-Life (h)
		CSL305	WT mAb	CSL305	WT mAb
Wild-type mice	10	22.7	0.4 a	2.0 a	113 b,c
	100	9.0	nt	4.0 a	nt
32HOM mice	10	5.1	0.4 a	41 a	250 b,c
	100	6.0	nt	49a	nt
Cynomolgus monkey	10	nc	4.3	nc	84 b

^a Related to distribution phase; ^b Related to terminal phase; ^c Reduced reliability related to AUCextrap >20%; nc: not calculated due to data unreliability associated with high LLOQ); nt: not tested in this study.

, the plasma clearance of CSL305 in either wild-type or 32HOM mice was much higher (>10-fold) compared to the clearance of the WT mAb, although the terminal half-life in wild-type mice could not be calculated since CSL305 exposure fell below the lower limit of quantitation within 24 h post-administration. Dose-dependence of clearance of CSL305 in wild-type mice suggests binding is limited by cell-surface-expressed FcRn numbers; this was not assessed with the WT mAb. Consequently, PK of WT mAb is not impacted by mouse strain, suggesting that plasma FcRn binding affinity is too weak in both mouse strains to impact PK significantly. The efficient and strong binding of CSL305 to FcRn is supported by CSL305 showing a stronger initial distribution phase than WT mAb in both mouse strains, and the lower exposure of CSL305 as compared to WT mAb following drug distribution. Differences in the effect on endogenous mouse IgG by CSL305 and WT mAbs were also observed (Figure 6B); despite high inter-sample variability, the administration of the WT mAb to mice was shown to have no effect on endogenous mouse IgG levels. This is consistent with the PK data, which suggest very limited binding of WT mAb to FcRn. In contrast, single-dose CSL305 administration to wild-type mice resulted in dose-dependent reductions in endogenous mouse IgG levels up to 74% (no samples were taken after 48 h) and up to 46% reduction in 32HOM mice (Figure 6B).

Due to the limitations of using mice to assess both mechanisms of action of CSL305, 10 mg/kg of both CSL305 and WT mAbs were administered IV to cynomolgus monkeys and PK/PD outputs compared. As shown in Figure 6C, the initial plasma clearance of CSL305 was again found to be much higher than for the WT mAb, likely related to drug distribution. Detectable CSL305 levels fell below the lower limit of quantitation after 72 h, while WT mAb was measurable until 360 h. Endogenous IgG levels in WT mAb-treated animals were unaffected, but CSL305-treated monkeys showed a maximum reduction of ~30% by day 6 post-administration, returning to baseline by the end of the study at 720 h (Figure 6D), confirming that CSL305 was able to act as an FcRn antagonist in vivo via its engineered Fc YPY motif. A CH50 assay was used to assess ex vivo complement classical pathway activity. As shown in Figure 6E,F, both CSL305 and WT mAbs were able to achieve an initial reduction of >70–80% in CH50 activity. For CSL305, this was followed by a plateau phase with >50% reduction in CH50 until study day 10, while the WT mAb CH50 values returned to baseline at 8 days post administration. Taken together, these data demonstrate that CSL305 acts as a dual complement and FcRn antagonist in vivo, whereas the control WT mAb inhibits complement activity only.

3. Discussion

The generation of complement inhibitors and FcRn antagonists to separately treat a variety of human diseases has been the subject of significant activity in recent years [25,27]. However, combining these two therapeutic approaches, or any strategy involving FcRn antagonism, is challenging because monoclonal antibodies—a preferred modality—are inherently degraded in vivo when FcRn antagonists are co-administered. Here, we describe a novel therapeutic mAb candidate—CSL305—that neatly circumvents this issue by containing both functions within the same molecule.

CSL305 was engineered to bind via its Fab arms to both huC2 zymogen and the active huC2b fragment and therefore to potently inhibit the classical and lectin complement pathways whilst leaving the alternative pathway intact (Table 1, Figure 1). An affinity maturation step was included in the engineering process to address the likelihood of insufficient inhibitory potential for complement classical and lectin pathway activity by the parental hu4D8 mAb; this was conducted experimentally, rather than computationally. The nomenclature of huC2a and huC2b can be confusing, with both being used interchangeably throughout the scientific literature. We have decided to adhere to the recommendations of complement field [40] and adopt huC2a as defining the N-terminal smaller fragment of C2, and huC2b as the C-terminal and larger, proteolytically active fragment. CSL305 binding to huC2 zymogen and cleaved huC2b was found to differ by approximately 2-fold (Table 1), which may reflect conformational changes introduced by the generation of their respective recombinant forms used in the binding studies. The introduction of a triple amino acid (YPY) mutation has previously been shown to increase the binding of mAbs to FcRn at both acidic and neutral pH [42], and we hypothesized that the introduction of this mutation into CSL305 would result in its ability to act as an FcRn antagonist, despite the lack of data in that study. Our binding studies confirmed the significant improvements in binding to FcRn conferred by the introduction of the YPY mutation to CSL305 (Table 2), as well as the ability to inhibit IgG but not HSA recycling in vitro (Figure 3 and Figure S4). Importantly, the introduction of YPY did not deleteriously affect the complement inhibitory potential of CSL305, as the results of several in vitro assays comparing CSL305 to the unmodified WT mAb showed (Figure 1 and Figure 2).

There are several points at which complement C2 can be targeted therapeutically to impact downstream classical and lectin pathway activity. These occur in the order of C2 activation: C2 zymogen binding to activated C4b, which then leads to cleavage of C2 by C1s (classical pathway) or MASP2 (lectin pathway) into C2a and C2b, and then association of C2b with C4b to form the C3 convertase [44,45]. The anti-C2 mAb ARGX-117 binds to the S2 domain of the small huC2a fragment (noted as huC2b in the publication) and its mechanism of action is proposed to be via interference of the C4b-C2 interaction [34]. However, the in vivo mechanism of action of ARGX-117 also appears to be due to its sweeping activity, which is mediated by its pH- and calcium-dependent binding to huC2. We had originally conducted extensive alanine mutagenesis studies with parental hu4D8, which showed that its epitope lay directly over the catalytic triad of huC2b (Table S4). Concurrent studies also demonstrated no inhibition of C1s-mediated cleavage of huC2 (Figure S6). The crystal structure of huC2 bound to the shared Fab region of both CSL305 and WT mAbs (Figure 4 and Figure 5) confirmed that their epitope is largely unchanged compared to the parental hu4D8 mAb. Importantly, the structure supports a complement-specific mechanism of action for CSL305 that is consistent with direct inhibition of C3 convertase formation. The structure clearly demonstrates the ability of CSL305 to block substrate binding in the serine protease domain of huC2 (Figure 4 and Figure 5). This interaction would then prevent C3 association, thereby inhibiting C3 convertase formation. In addition, CSL305 makes several contacts to C2 adjacent to the catalytic triad, which could lock the serine protease domain in a zymogen-like state, rendering C2 inactive. Consistent with this, the parental version of CSL305 had no effect on C2 cleavage by C1s (Figure S7), suggesting that CSL305 binding to C2 does not induce any major structural rearrangements. Instead, the data support more local conformational changes upon binding, particularly within the serine protease domain (Figure S6).

Demonstration of the dual function of CSL305 in vivo was the critical next step, and so we conducted comparative experiments with WT and CSL305 mAbs in mice. Despite the lack of C2 cross-reactivity of both mAbs to mice and other species (Table S1, Figure S2), comparative PK and endogenous IgG could still be assessed. An asymmetry in the interaction of IgG and FcRn has previously been noted between mice and humans, with weak to no binding of mouse IgG to human FcRn compared to the similar or even stronger affinity of human IgG to mouse FcRn (Table S2) [26,46]. To overcome this, both wild-type and 32HOM mice were used. The data in Figure 6A,B show that regardless of mouse strain employed, CSL305 exhibited a much faster plasma clearance compared to the WT anti-C2 mAb. A reduction in endogenous IgG was also noted following CSL305 administration to wild-type mice, peaking one to two days after administration. A subsequent comparative PK/PD study in cynomolgus monkeys confirmed these findings and showed immediate ex vivo classical pathway inhibition (Figure 6C–F), although there was a lack of either free or total C2 assay availability to link effects on C2 with ex vivo complement activity.

4. Materials and Methods

4.1. Antibody Discovery

A CSL proprietary naïve Fab phage-display library was screened for clones that bound both huC2 zymogen and its activated fragment huC2b fused to the Fc region of human IgG1. Three rounds of phage display selection were performed using decreasing concentrations of huC2-Fc/huC2b-Fc (15 µg, 10 µg, and 5 µg). At the end of the panning process, 380 individual phage clones were screened for binding to huC2. Of these, 75 positive binders were identified, of which 25 were unique clones. The unique clones were reformatted as whole human mAbs with IgG4 heavy chain containing the hinge-stabilising mutation (S228P, EU numbering) and lambda or kappa light chains and analysed for potency. Six mAbs were found to bind both huC2 and activated huC2b and inhibit classical and lectin pathway activity. The most potent mAb, hu4D8, was chosen for further engineering. Variable regions of hu4D8 were analysed by sequence homology searches using IgBLAST analysis [41]. Seven amino acid residues within the light chain variable region and two residues within the heavy chain variable region within the frameworks were substituted for the germline sequence. The germline back-mutated mAb (ghu4D8; Table 1) showed a reduction in affinity and potency compared to the parental hu4D8 mAb. Affinity maturation of this antibody was then carried out to identify mAbs with a higher affinity for huC2 and huC2b and increased classical/lectin pathway inhibitory activity.

The six CDR loops of the ghu4D8 mAb were determined using the Kabat definition scheme. Ten libraries targeting 6 residues at a time were designed as follows: two libraries in CDR1 of the light chain; two libraries in CDR3 of the light chain; one library in CDR1 of the heavy chain; two libraries in CDR2 of the heavy chain; three libraries in CDR3 of the heavy chain. Several conservative residues within the CDRs were excluded from randomization. Each residue within each library was randomly diversified for all 19 possible amino acids (excluding cysteine) using a Kunkel mutagenesis-based method for the generation of Fab-on-phage libraries [47]. For each position, saturation mutagenesis was introduced using two degenerate codons: NWS and NSG, where N = A, T, G, or C; W = A or T; and S = G or C. Following five rounds of selection against decreasing concentrations of recombinant huC2b-8His, 24 randomly selected clones from each library were analyzed by Sanger sequencing. Forty unique variants from all selection strategies were converted to hinge-stabilized (S228P mutation, EU numbering) IgG4 mAbs with a lambda light chain [48]. Affinity-matured mAb aghu4D8 (WT; with an affinity-matured light chain with mutations in CDR-L1 and an unmodified heavy chain) was selected based on affinity and potency. The terminal lysine was also removed from the IgG4 heavy constant region to improve product uniformity. This antibody is referred to as agh4D8/WT. To generate CSL305, a triple mutation (YPY, M252Y, V308P, N434Y based on EU numbering) was introduced to the IgG4 constant region of aghu4D8/WT mAb. Plasmids encoding both CSL305 and WT mAbs (with codon-optimized sequences) suitable for mammalian cell expression were generated and sequence verified using standard techniques.

4.2. Generation of cDNA Expression Plasmids

Copy DNA (cDNA) encoding the following proteins were codon-optimized for human expression and synthesized by GeneArt^® (Thermo Fisher Scientific, Waltham, MA, USA), with in-frame C-terminal 8x histidine-tags and double stop codons, using the amino acid (aa) sequences retrieved from their respective GenBank accession numbers and based on Met+1: full-length huC2 (aa1-752, NP_000054); huC2a (aa 1-243); huC2b (aa 244-752); cynoC2 (aa 1-752, XP_00555358; identical to rhesus C2, XP_014991516); cynoC2a (aa 1-243); cynoC2b (aa 244-752); dog C2 (XP_013973817); rat C2 (NP_757376); mouse C2 (NP_038512); rabbit C2 (XP_002714335); pig C2 (NP_001095285); sheep C2 (XP_004018979); soluble huFcRn (aa 1-297, NP_004098); huB2M (untagged, NP_004039); soluble cyno FcRn (aa 1-297, AAL92101); cyno B2M (untagged, AAL92100); soluble mouse FcRn (aa 1-297, NP_034319); mouse B2M (untagged, NP_033865); soluble huCD16 (aa 1-206, NP_001121065); soluble huCD32a (aa 1-215, NP_067674); soluble huCD32b (aa 1-221, NP_003992); soluble huCD64 (aa 1-291, NP_000557); huC1s (NP_001725); aghu4D8 (WT) and CS305 mAbs. Alanine point mutants of huC2 were generated based on the unmodified full-length sequence above, and a construct encoding a Fab fragment of CSL305 ending N-terminal to the hinge region used CSL305 as a template. Also generated were cDNAs encoding recombinant wild-type human IgG1 Fc domain (Fc-G1) and an FcRn antagonist variant thereof with the mutations M252Y, S254T, T256E, H433K, N434F (FcYTEKF). Each cDNA was generated with a Kozak consensus sequence (GCCACC) immediately upstream of the start codon and cloned into pcDNA3.1 (Thermo Fisher Scientific, Waltham, MA, USA). CSL040 was generated as previously described [32]. Preparations of plasmid DNA were carried out using QIAGEN plasmid maxi or giga kits according to the manufacturer’s instructions (QIAGEN, Hilden, Germany). The nucleotide sequences of all plasmid constructs were verified by sequencing both strands using BigDye™ Terminator Version 3.1 Ready Reaction Cycle Sequencing (Thermo Fisher Scientific, Waltham, MA, USA) and an Applied Biosystems 3130xl Genetic Analyzer (Applied Biosystems, Carlsbad, CA, USA).

4.3. Cell Culture, Recombinant Protein Expression, and Purification

Cell culture and transient transfection of plasmids encoding all non-mAbs (including Fabs) into Expi293F™ and plasmids encoding mAbs into ExpiCHO™ cells were carried out according to the manufacturer’s recommendations (Thermo Fisher Scientific, Waltham, MA, USA) as previously described [32,49]. Expression of FcRn required the co-expression of B2M at a plasmid ratio of 1:1. Purification of 8His-tagged recombinant proteins was carried out as previously described [32], and CSL040 as previously described [49]. For purification of recombinant mAbs and Fc domains, culture supernatants were loaded directly onto MabSelect SuRe^TM affinity resin (Cytiva, Marlborough, MA, USA) pre-equilibrated with 10 mM Na₂HPO₄, 150 mM NaCl, pH 7.4. Resin was washed with this buffer, then again with 0.1 M sodium acetate, pH 5.0, and resin-bound mAbs block eluted with 0.1 M sodium acetate, pH 3.5. Eluates were neutralized with the addition of 15% v/v 2 M Tris pH 8.0, and fractions containing mAb loaded onto a Superdex^TM 200 column (Cytiva, Marlborough, MA, USA) pre-equilibrated in 10 mM Na₂HPO₄, 150 mM NaCl, pH 7.4 to separate out contaminating proteins. Purified mAbs were concentrated using Amicon^® ultra centrifugal filters (Merck Millipore, Carlsbad, CA, USA) with a 50-kilodalton molecular weight cut-off, sterile filtered, and stored at −80 °C. To purify the co-expressed FcRn/B2M complex, culture supernatants were adjusted to pH 5.8 and then loaded directly onto IgG Sepharose FF resin (Cytiva, Marlborough, MA, USA) pre-equilibrated with 10 mM Na₂HPO₄, 150 mM NaCl, pH 5.8. Resin-bound FcRn/B2M proteins were washed as above then block eluted with 10 mM Na₂HPO₄, 150 mM NaCl, pH 7.4, based on absorbance at 280 nm. Elution fractions were further purified on a Superdex^TM 200 column (Cytiva, Marlborough, MA, USA), concentrated, sterile filtered, and stored at −80 °C as above. The CSL305 used for the single-dose PK study was produced recombinantly in Chinese Hamster Ovary cells. The mean achieved concentration of this batch of CSL305 was within 3% of the nominal concentration as determined by measuring its absorbance at 280 nm, confirming the accuracy of the formulation. The difference from the mean remained within 3%, confirming precise analysis.

4.4. Affinity Measurements by Surface Plasmon Resonance (SPR)

For C2 binding analyses, experiments were conducted on a Biacore 8K instrument (Cytiva, Marlborough, MA, USA) and binding sensorgrams and were fit to a 1:1 Langmuir model using the accompanying Biacore Insight evaluation software version 6.0. Goat anti-human IgG (ThermoFisher Scientific, Waltham, MA, USA) was directly immobilized onto the carboxymethyl dextran surface of CM5 sensorchips to approximately 12,000 response units (RU) using standard NHS/EDC chemistry at pH 5. The immobilized antibody surface was pre-conditioned with ten injections of polyclonal human IgG prepared at 1 µg/mL. Ligands (CSL305, anti-C2 mAbs, 2 µg/mL) were captured at the beginning of each cycle. Purified C2 zymogen and its C2a and C2b fragments, plus any single-point mutants, from different species were tested at concentrations ranging from 3.9 nM to 500 nM. The concentration range was prepared from 2-fold serial dilutions from a 500 nM stock in HEPES-buffered saline (Cytiva, Marlborough, MA, USA) supplemented with 0.1% w/v bovine serum albumin (BSA, Merck Life Science, Melbourne, Australia). Each analyte concentration was tested in duplicate injections. Analyte association and dissociation were monitored for 120 and 600 s, respectively. Each antibody ligand was tested in at least four experiments. The anti-human IgG antibody surface was regenerated in 100 mM H₃PO₄, injected for 60 s at the end of each cycle. Experiments were performed at 37 °C at a flow rate of 30 µL/min. Sensorgram data were double referenced against a reference anti-IgG surface and blank buffer injections obtained in each experiment. Buffers and solutions were filtered (0.22 µm) prior to use. The effect of Ca²⁺ on human C2 binding to anti-C2 mAbs was evaluated by measuring each interaction in the presence of 2 mM CaCl₂. Ethylenediaminetetraacetic acid (EDTA; 3 mM) was added to remove traces of Ca²⁺ or other divalent ions in the solution. Experiments were performed in 0.1% BSA under neutral (pH 7.3) and acidic (pH 6.0) conditions.

For FcRn/B2M binding experiments, mAbs were prepared in acetate buffer pH 5 at 1.5 µg/mL and directly immobilized onto the carboxymethyl dextran surface of CM5 sensorchips to between 3000 RU using standard NHS/EDC chemistry. Purified recombinant human, cyno, or mouse FcRn/B2M (analyte) was tested under 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES)-2-(N-morpholino) ethanesulfonic acid (MES) buffered at either neutral pH 7.3 or acidic conditions (pH 6.0). FcRn/B2M was tested at pH 7.3 at concentrations ranging from 0.2 to 50 µM and at pH 6.0 at concentrations ranging from 0.02 to 5 µM. Analyte samples were prepared in 2-fold serial dilutions from either 50 µM or 5 µM stock prepared in running buffer. Each analyte concentration was tested in duplicate. Analyte association and dissociation were monitored for 120 and 600 s, respectively. Sensorgrams were double referenced using reference surface and blank buffer injection data. Rate constants and binding affinities were calculated using a 1:1 Langmuir model with local Rmax and null refraction index (RI = 0) whenever possible.

For other Fc receptor binding studies, analyses were conducted on a Biacore 8K instrument (Cytiva, Marlborough, MA, USA) and binding sensorgrams and were fit using the accompanying Biacore Insight evaluation software version 6.0. Briefly, protein A: G (Thermo Scientific Scientific, Waltham, MA, USA) was prepared in a 1:1 ratio in 10 mM sodium acetate, pH 5.0, and immobilized to approximately 3000 RU in the active channel of each flow cell. The immobilized protein A: G surface was pre-conditioned with ten injections (60 sec) of polyclonal IgG (Merck Life Science, Melbourne, Australia) at pH 7.3, followed by a regeneration cycle (60 sec) of 10 mM glycine, pH 1.5. Anti-C2 mAb ligands were prepared at 1 μg/mL in running buffer prior to capture (60 sec). Purified soluble recombinant Fcg receptors (FcgRIIIA [158F], FcgRIIA, FcgRIIB, and FcgRI) were prepared in HEPES-buffered saline (Cytiva, Marlborough, MA, USA) supplemented with 0.1% w/v BSA. Low-affinity receptors were tested at concentrations ranging from 0.1 to 40 μM, while FcYRI was tested at concentrations ranging from 0.05 to 40 nM. All FcR samples were prepared in running buffer at pH 7.3, and each concentration was tested in duplicate. In all experiments, analyte association and dissociation were monitored for 120 and 420 sec, respectively. For the above, when kinetics are too fast to accurately measure k_a or k_d, a steady-state affinity assay is used to determine the K_D. The report point is taken at the end of the association phase and plotted against the concentration of analyte—half of the vMax of that binding plot is the K_D.

4.5. Wieslab^® and Hemolytic Complement Inhibition Assays

Wieslab^® complement assays (Svar Life Sciences, Malmo, Sweden) specific for the classical, lectin, and alternative pathway were performed according to the manufacturer’s recommendations and as previously described [50]. Optimal serum dilutions for non-human species were experimentally defined. Hemolytic assays specific for the complement classical (CH50) and alternative (AH50) pathways were also performed as previously described [50]. The pre-optimized serum dilutions used to assess species cross-reactivity in CH50 assays are as follows: Human 1/80; Mouse 1/80; Rat 1/60; Rhesus 1/40; Cyno 1/40; Dog (beagle) 1/40; Pig 1/16; Rabbit 1/15; Minipig 1/8.

4.6. Microfluidic Channel Production and Surface Modification

Microfluidic polydimethylsiloxane (PDMS) chips containing channels with 550 μm cross sections were prepared according to [51]. The microfluidic chips contain 2 microfluidic channels of 2 cm length. PDMS chips were bonded to 17 μm thick high precision coverslips (24 × 60 mm; Chip Direct AG, Roth, Germany) with a plasma treatment system (HPT-200; Henniker Plasma, Runcorn, UK) at 8 standard cubic centimeters per min for 1 min. The hydrophobic PDMS surface was silanized and extracellular matrix proteins immobilized. Microchannels were then coated with human fibronectin (50 μg/mL in PBS, Merck, Hessen, Germany) for 1 h at 37 °C, followed by bovine collagen I (100 μg/mL in 0.2 mol/L acetic acid, Thermo Fisher Scientific, Basel, Switzerland) for 1.5 h at RT. Microchannels were rinsed and primed for 30 min at 37 °C with micro-vessel medium (MVM: Endothelial Cell Growth Medium-2 EGM™-2 (Lonza, Basel, Switzerland), 4% dextran w/v, 1% BSA w/v, both from Sigma-Aldrich, Saint Louis, MO, USA).

4.7. Cell Seeding, Culture, and Pulsatile Flow

HLMVEC were grown confluently in T75 cell culture flasks using complete medium for HLMVEC: EGM™-2 Endothelial Cell Growth Medium-2 BulletKit™ (Lonza, Basel, Switzerland). Cells were rinsed with Hank’s Balanced Salt Solution (without CaCl₂, without MgCl₂, Thermo Fisher Scientific, Basel, Switzerland) and detached with Accutase^® (Biowest, Nuaille, France) for 5 min at 37 °C. Cells were resuspended in MVM, seeded into the channels, and medium changed 3x/day. Polyvinyl chloride tubes (Maagtechnic AG, Dupendorf, Switzerland) and Pump head tubing (Gilson, Baar, Switzerland) were sterilized. Tubing was attached to the PDMS chips once cells were confluent within the microchannels (~20 h after seeding). Before attachment, tubes were rinsed using a peristaltic pump (Masterflex peristaltic pump Masterflex™ DRIVE L/S 100 RPM 115/230 with multichannel pump head Masterflex™ Stainless Steel Rotor Multichannel Pump Head; Fisher Scientific, Basel, Switzerland). Falcon tubes with 10 mL MVM were attached to the tubing start and end to create a closed system flow. Placeholder tubes were removed, and the chip was attached. Microchannels were then perfused with MVM at 0.6 mL/min in pulsatile mode, resulting in a laminar shear stress of approximately 12 dyn/cm² (viscosity of the medium μ = 2.1 mPa/s). The medium in the Falcon tube reservoirs was changed after 24 h.

4.8. Complement Deposition Assay in a Human Inflammatory Setting

After ~40 h of pulsatile flow, HLMVEC were opsonized using a murine mAb against CD105/Endoglin (mouse IgG1; 1 μg/mL in MVM, Abcam, Waltham, MA, USA) for 30 min on ice. During this time, CSL040 (100, 50, 25, 12.5, and 6.25 nM) was incubated together with 20% NHS derived from pooled donors (Dunn Labortechnik GmbH, Asbach, Germany) in MVM at 37 °C. After opsonization, tubes were exchanged for MVM containing NHS with or without CSL040 or MVM alone as control and perfused for 2 h at 37 °C, 0.6 mL/min. After treatment, microchannels were washed with phosphate-buffered saline (PBS), fixed with 4% paraformaldehyde (PFA) for 20 min at RT, and washed again with PBS. Treatment with 20% NHS on non-opsonized HLMVEC served as a negative control, and 20% NHS on opsonized HLMVEC was used as a positive control.

4.9. Immunofluorescence Staining and Evaluation

HLMVECs in microchannels were blocked with 3% BSA/PBS for 1 h at RT or overnight at 4 °C. A fluorescein isothiocyanate (FITC)-coupled antibody against C3b/c (Rabbit Anti-Human C3c-FITC, 1:100, Agilent Technologies, Lausanne, Switzerland), F-actin (Acti-Stain 555 Phalloidin, 1:200, Cytoskeleton, Inc., Denver, CO, USA), and DAPI stain (1:1000, Sigma Aldrich, Saint Louis, MO, USA) were diluted in staining solution (PBS, 1% bovine serum albumin, 0.05% Tween, Sigma Aldrich, Saint Louis, MO, USA) and incubated in the microchannels for 3 h at RT or overnight at 4 °C. Channels were washed with PBS and filled with fluorescence mounting medium (Sigma Aldrich, Saint Louis, MO, USA) before confocal imaging. Images were taken at 10× magnification with a confocal laser-scanning microscope (LSM800, Zeiss, Feldbach, Switzerland) and analyzed by ImageJ version 1.54 (National Institutes of Health, Bethesda, MD, USA). Three to four images per channel were analyzed for fluorescence intensity of C3b staining. The region of interest was outlined using the F-actin staining and transferred to the C3b staining image to obtain the mean fluorescence intensity. An area outside of the region of interest (ROI) was measured as background and subtracted from the ROI intensity. All data are displayed as violin plots, with lines indicating the median and quartiles. Statistical analysis was performed with Graphpad Prism 9 software (GraphPad, Boston, MA, USA). Treatment group differences were evaluated by the Kruskal–Wallis test with Dunn’s multiple comparisons against the positive control.

4.10. FcRn Recycling of IgG and HSA by Bone Marrow Derived Macrophages (BMDMs)

Ten-week-old 32HOM mice were euthanized by CO₂ asphyxiation and BMDMs were generated essentially as previously described [52]. For each experiment, 6 ×10⁶ BMDMs were seeded in 12-well removable chamber slides (Ibidi GmbH, Gräfelfing, Germany) at 2.5 ×10⁵ cells/well. Cells were differentiated in 200 µL of BMDM media (RPMI; Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 15% heat inactivated fetal calf serum (FCS; Sigma Aldrich, Saint Louis, MO, USA), 20% L cell media (Cell Biologics, Chicago, IL, USA), 100 Units/mL penicillin and 0.1% streptomycin (Sigma Aldrich, Saint Louis, MO, USA); 2mM Glutamax (Thermo Fisher Scientific, Waltham, MA, USA) for 3 days before another 200 µL of BMDM media was added for 48 h to allow further differentiation of cells to macrophage-like cells. Cells were then starved for 16 h by media removal, washing twice with PBS, and replacing with 200 µL C-RPMI (RPMI, heat-inactivated FCS (10%), 100 Units/mL penicillin/streptomycin, 2 mM Glutamax) per well. Medium was then aspirated from wells and cells pre-incubated in C-RPMI in the presence of dimethylsulphoxide (1/100 dilution; Sigma Aldrich, Saint Louis, MO, USA) or protease inhibitor (P1860, Sigma Aldrich; 1/100 dilution) for 4.5 h. At 30 min before the end of pre-incubation, mouse serum (Thermo Fisher, Waltham, MA, USA; 1/100 dilution) was added to block surface Fc gamma receptors and incubated on ice for 30 min. A 30 μM concentration of each of the following treatment samples: CSL305, WT mAb, Fc-G1, and FcYTEKF, as well as 1 μM HSA-AF488 or 1 μM IgG-AF594, and protease inhibitors, were prepared and pre-warmed at 37 °C. Following incubation with mouse serum, cells were washed twice with phosphate-buffered saline (PBS), pre-mixed pre-warmed experimental treatments (CSL305, WT, Fc-G1, FcYTEKF, or no treatment) were added with either IgG-AF594 or HSA-AF488, and cells pulsed for 15 min at 37 °C (pH 7.3). Cells were then washed twice in PBS, and either fixed immediately (“0 min chase”) in C-RPMI ± protease inhibitor addition or incubated for a “15 min chase” period before washing and fixation ± protease inhibitors. Cells were fixed in 4% PFA (Millipore Sigma, Burlington, MA, USA) for 10 min at RT, washed once in PBS containing 50 mM NH₄Cl to quench excess PFA, and incubated for a further 10 min at RT. They were then washed once more in PBS and blocked in 5% FCS/PBS for 30 min. Samples were then treated with Hoechst 33342 (Thermo Fisher Scientific, Waltham, MA, USA) to stain the nuclei, and the slides mounted in Mowiol^® mounting reagent (Millipore Sigma, Burlington, MA, USA). Imaging was conducted by confocal microscopy using a Leica SP8 confocal microscope. Fluorescence intensity of acquired images (5 images per condition and timepoint, each containing an average of 20–30 cells) was analyzed by the “analyze particle” plugin in Image J, and the data were analyzed using GraphPad Prism 10.5.0. Data are expressed as box plots, where the boxes show the minimum and maximum values and the line represents the data mean. Each data point represents the average fluorescence intensity of 20–30 cells from 3 independent experiments.

4.11. Complex Formation

Purified huC2b-8His at 10.8 mg/mL and CSL305 Fab (identical to WT/aghu4D8 Fab) at 24.9 mg/mL were incubated at a 1 to 1.25 molar ratio, respectively, at RT for 20 min. The CSL305-C2b complex was then purified in 10 mM HEPES (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid), 100 mM NaCl, 5% glycerol, pH 7.4, by size exclusion chromatography to remove unincorporated components and aggregates. Fractions were then pooled, concentrated to 10.3 mg/mL using Amicon Ultra-15 concentrators with a 50-kilodalton molecular weight cut-off (MilliporeSigma, Burligton, MA, USA), sterile filtered, and stored at −80 °C.

4.12. Crystallization

The human C2b/aghu4D8-Fab (this Fab is common to both WT and CSL305 mAbs) complexes were screened against commercial crystallisation screens using an NT8 crystallisation robot (Formulatrix, Bedford, MA, USA) in MRC 96-Well Triple Drop Plates (Molecular Dimensions, Rotherham, UK). Sitting drops containing 0.2 μL protein + 0.2 μL reservoir solution were incubated at 4 °C or 20 °C and imaged under visible and UV light using a RockImager 1000 (Formulatrix, Bedford, MA, USA). Initial hits were scaled up in hanging drops in Linbro plates (Hampton Research, Aliso Viejo, CA, USA) with drops containing 2 μL protein + 2 μL reservoir solution. Crystallization conditions for aghu4D8-C2b were derived from an initial JCSG⁺⁺ screen (Molecular Dimensions, Rotherham, UK). The final composition was 20.00% w/v polyethylene glycol 3350 and 200 mM di-sodium malonate, pH 7.0, incubated at 20 °C. Crystals were briefly soaked in the crystallisation solution supplemented with 15% v/v glycerol before being plunged into liquid nitrogen.

4.13. Data Collection, Model Building, and Refinement

All diffraction data were collected at the Australian Synchrotron on beamlines MX1 & MX2 at a temperature of 100 K using the program Blu-Ice [53]. Diffraction data were integrated using XDS [54], scaled, and merged with AIMLESS [55]. Phase information was generated by the molecular replacement method using PHASER [56]. For the determination of the complex, search models for C2b and Fab were generated using pdb_00002i6q and pdb_00007fab, respectively. The latter model was divided into constant and variable domains to aid molecular replacement. Two copies of the C2b-Fab complex were found in the asymmetric unit with a log-likelihood gain of 5600 and a TFZ score of 400. Model building was carried out using COOT [57], and refinement was carried out using PHENIX.refine [58] and REFMAC5 [59]. Glycosylation sites were automatically built using the Linked Monosaccharide Addition tool within COOT [60].

4.14. C1s Cleavage Assay

Combinations of recombinant huC2 (200 nM), huC1s (2.5 nM), parental hu4D8 (250 nM), and a positive control anti-C2 mAb (250 nM) were mixed at RT for 30 min before being subjected to sodium dodecylsulphate polyacrylamide gel electrophoresis under non-reducing conditions, followed by Coomassie blue staining of the gel. Molecular weight standards (Thermo Fisher Scientific, Waltham, MA, USA) were also loaded to enable accurate determination of proteins and huC2a/C2b fragments on the gel. In a subsequent experiment, 2.5 μg huC2, 4 ng huC1s, and concentrations of hu4D8 to enable 10, 1, 0.1, and 0.01 mAb:C2 fold molar excess were used as above.

4.15. Mouse In Vivo PK/PD Experiments

Mouse studies were performed at the Department of Pharmacology and Toxicology, CSL Behring GmbH, Marburg, Germany. All animals were handled and managed in accordance with animal care protection laws. Ethics approval for the mouse studies was obtained from Regierungspraesidium Giessen, Wetzlar, Germany (approval number A5/2020). Female C57BL/6J wild-type mice and C57BL6 human FcRn homozygous transgenic mice (FcRn Tg; line 32; designated as 32HOM) were obtained from Charles River Laboratories, Erkrath, Germany, and used at an age of 12–20 weeks. WT and CSL305 mAbs were administered to these mice at doses of 10 or 100 mg/kg IV in PBS diluent as required. Different observation periods were selected for each test item/mouse combination: WT mAb in wild-type mice or 32HOM mice—168 h; CSL305 in wild-type mice—48 h; CSL305 in 32HOM mice—168 h. All blood samples were taken retrobulbar at all time points (N = 3 mice per timepoint). The sampling time points selected were as follows: −96 h (for pre-dose endogenous IgG quantification), 5 min, 15 min, 1 h, 6 h, 24 h, 48 h, 72 h, 96 h, 168 h. Blood samples were mixed with 10% EDTA and then processed to EDTA plasma, and stored at −70 °C. Analysis of CSL305 or WT mAbs in samples taken from mice administered with these agents was performed using a human IgG enzyme-linked immunosorbent assay (ELISA). Capture antibody was goat anti-human IgG Fc (Sigma, Germany); blocking solution was BSA-based (Sigma Aldrich, Schnelldorf, Germany); detection antibody was goat anti-human IgG Fc conjugated to horseradish peroxidase (HRP; Sigma Aldrich, Schnelldorf, Germany). Analysis of endogenous IgG from mice was performed using an ELISA-based assay, where the capture antibody was goat anti-mouse IgG (human adsorbed; Southern Biotech, Birmingham, AL, USA); the blocking solution was bovine serum albumin-based (Sigma Aldrich, Schnelldorf, Germany); detection antibody was goat anti-mouse IgG, human adsorbed and conjugated to HRP (Southern Biotech, Birmingham, AL, USA). Exposure data were plotted as a percentage of the drug administered to the dose to better allow for comparison of PK parameters. PK analysis was done by non-compartmental analysis using Phoenix WinNonLin 8.3.4.295, applying the sparse function and linear trapezoidal linear interpolation with uniform weighting. The pre-dose normalized endogenous IgG level was used to compare the impact of CSL305 and WT mAbs on this parameter.

4.16. Cynomolgus Monkey In Vivo PK/PD Experiments

This study was conducted at Charles River Laboratories; UK ethics approval: Home Office Project License number PP7137579, Toxicology of Pharmaceuticals, Protocol number 11. Monkeys received a single dose of 10 mg/kg CSL305 or WT mAbs as an IV bolus injection (N = 3 male animals per group) with a sampling and observation period of 30 days. Blood samples were taken pre-dose, and at the following time points after dosing: 10 min, 1 h, 4 h, 12 h, 24 h, 72 h, then days 6, 8, 10, 12, 14, 16, 22, 30 for CSL305, and days 5, 8, 15, 22, 29 for WT, and processed to serum. An ELISA assay to measure total CSL305 concentrations utilized an internal anti-C2 antibody as a capture reagent, and a commercial antibody (mouse Anti-Human IgG4 Fc-HRP, Southern Biotech, Birmingham, AL, USA) for detection. Acidification was used to allow analysis of total drug levels. Endogenous IgG levels were measured using a commercial ELISA kit (Cell Sciences, Newburyport, MA, USA) according to the manufacturer’s recommendations. CH50 assays were performed as previously described [50]. Individual animal PK analysis was performed by non-compartmental analysis using Phoenix 64 version 8.3.5, applying the Plasma (200–202) model, linear up log down function, and with uniform weighting. For CSL305, related to quick elimination and high LLOQ of assay, the extrapolated AUC exceeded 20% and R²-adjusted <0.85, so clearance and terminal half-life were not reported. Individual pre-dose normalized endogenous IgG levels were used to compare the impact of CSL305 and WT mAbs on endogenous IgG levels.

5. Conclusions

The in vitro and in vivo data presented in this study confirm CSL305 as a dual functional complement inhibitor and FcRn antagonist mAb. It also suggests that the short half-life of CSL305 is due to binding to and antagonism of FcRn in a cell-associated context, given its reduced plasma exposure compared to the WT mAb. However, CSL305 was still able to inhibit ex vivo complement activity after mAb levels fell below the limit of quantitation, likely due to its binding to FcRn in a cell-associated context. Additional studies will be needed to better understand the dose-response PK/PD profile of CSL305 and its ability to ameliorate both complement and FcRn-mediated pathologies in autoimmune diseases and other indications that involve both mechanisms.