The fc fragment of IgMs binds C1q to activate the first step of the classical complement pathway, while inhibiting complement‐dependent cytotoxicity

Abstract

Soluble IgMs, among the most potent activators of the classical pathway, are key mediators of complement‐dependent cytotoxicity, which render them as promising drug candidates for the development of alternative drugs in treating autoimmune or inflammatory diseases. In the present study, we investigated the biochemical and in vitro functional properties of recombinant fragments from IgMs corresponding to the fragment crystallizable region (Fc)‐core in their pentameric or hexameric forms. Biophysical experiments confirmed the crucial role of the IgM Joining (J) chain in favoring homogeneous pentamers, whereas its absence led to a heterogeneous population with a mixture of oligomeric forms. By combining size‐exclusion chromatography with mass photometry, isolation of enriched samples with IgM hexamers or IgM pentamers without the J chain was possible. Biolayer interferometry demonstrated that both IgM‐Fc forms bind C1q, and an ELISA showed that they induce the in vitro C4b deposition when in solid phase. Additionally, our data confirmed the higher efficacy of IgM hexamers compared to pentamers in activating the first component of the classical pathway. Finally, hemolytic assays demonstrate the ability of IgM‐Fc constructs to inhibit Ig‐induced complement‐dependent cytotoxicity, which is likely made possible by the absence of fragment antigen binding region. These findings support a possible mechanism of C1 sequestration in plasma by IgM cores and consumption of the initial complement component C4. Our data thus provide important information for the development of IgM‐based anti‐inflammatory molecules that target specifically complement activation.

Multimeric IgM‐fragment crystallizable region (Fc) fragments retain the ability to bind C1q and initiate the classical complement pathway, leading to C4 activation and deposition in vitro. However, the Fc cores can also inhibit complement‐dependent cytotoxicity by competing with surface‐bound antibodies for C1q engagement. This dual activity reveals a previously unrecognized regulatory role of IgM‐Fc in modulating complement activation in solution, and suggests potential applications as decoy molecules in immune modulation.

Abbreviations

ADCC

antibody‐dependent cellular cytotoxicity

ADCP

antibody‐dependent cellular phagocytosis

AIM

apoptosis inhibitor of macrophage

BLI

biolayer interferometry

CDC

complement‐dependent cytotoxicity

CP

classical pathway

ELISA

enzyme‐linked immunosorbent assay

Fab

fragment antigen binding region

Fc

fragment crystallizable

H

heavy chain

IC

immune complexes

Ig(s)

Immunoglobulins

IgM(s)

Type‐M Immunoglobulins

J

joining chain

L

light chain

MP

mass photometry

NHS

normal human serum

PBS

phosphate‐buffered saline

(SDS)‐PAGE

(Sodium dodecyl‐sulfate) polyacrylamide gel electrophoresis

SEC

size exclusion chromatography

SEC‐MALLS

size exclusion chromatography coupled to multi‐angle laser light scattering

SPR

surface plasmon resonance

sRBC

sheep red blood cell

sv‐AUC

sedimentation velocity‐analytical ultracentrifugation

TEM

transmission electron microscopy

TH₅₀

50% total hemolysis

Tp

tail piece

UV‐vis

ultraviolet‐visible spectroscopy

Introduction

Soluble IgMs stand out among Ig classes with their unique structure and their highest level of oligomeric states. They are primarily secreted by naive B cells, but can also be produced, in some cases, after B cell maturation, to play significant roles in immune responses to infections such as antibody‐dependent cellular cytotoxicity (ADCC), antibody‐dependent cellular phagocytosis (ADCP) and complement‐dependent cytotoxicity (CDC). Like other immunoglobulin classes, they are composed of a heavy chain (H) and a light chain (L) and are submitted to the well‐known VDJ gene rearrangements and the somatic hyper‐mutation processes during maturation, which are crucial for antigen specificity and assembly of the variable domains composing their antigen‐recognition fragment antigen binding (Fab) domains. IgMs structurally differ from the other antibodies in the number of H chain constant domains (comprising four domains, Cμ1 to Cμ4) and in their overall assemblies. They are found in sera with a high degree of oligomerization, forming either pentamers ((H₂L₂)₅J), which include the additional Joining (J) polypeptide or hexamers ((H₂L₂)₆), devoid of the J chain [1, 2, 3, 4, 5].

Those unique assemblies are made possible by the formation of the so‐called fragment crystallizable (Fc) region or Fc‐core, comprising the Cμ2, Cμ3 and Cμ4 domains, and the C‐terminal tailpiece, connected through an extensive disulfide‐bond network. The Fc‐core is essential for IgMs to achieve their ADCC, ADCP or CDC effector functions by binding to cellular or soluble immune receptors. Among them, binding to the polymeric immunoglobulin receptor (pIgR) and its secretory component mediates IgM transport across epithelial layer towards mucosal lumen, aiding in immune tolerance towards commensal bacteria [6]. The IgM‐type immune complexes (ICs) also regulate the immune system though interactions with two Ig receptors, Fcα/μR and FcμR. Fcα/μR is involved in phagocytosis of ICs by B cells to facilitate antigen processing and presentation to helper T cells in the induction of immune response against T‐dependent antigens [7] or to negatively regulate immune responses against T‐independent antigens [8]. IgMs can also be a carrier of the apoptosis inhibitor of macrophage (AIM/CD5L) to regulate thymocytes apoptosis [9]. The AIM‐IgM complexes affect the interaction with Fcα/μR expressed on follicular dendritic cells, slowing ICs internalization to facilitate antigen presentation to B cells and stimulate antibody responses [10]. While Fcα/μR can bind both IgA and IgM subtypes, FcμR (also named FCMR, Toso/Fas apoptotic inhibitory molecule 3 or FAIM3) is only expressed by lymphocytes, specific for IgM, and presumably involved in immune cellular activation and control of autoantibody production [11]. Finally, CDC is triggered by IgMs when bound to surface‐exposed antigens and recognized by the first component of the classical pathway (CP) of the complement system, the C1 complex. They are known to be the most potent activators of the amplifying proteolytic cascade that leads to the production of the main C3 and C5 complement convertases, which regulate the immune system stimulation, and lead to the formation of the membrane attack complex and the elimination of the pathogens or the infected cells [12, 13].

Regarding antigen recognition functions, the IgMs high oligomerization gives them a higher multivalency than IgGs: 10 to 12 Fabs are found in IgMs vs. 2 in IgGs, leading to a greater avidity of IgMs to neutralize their specific targets. Most of all, pentameric and hexameric soluble IgMs are known to be more potent activators of the CDC and the classical complement pathway than hexameric structures of bivalent IgGs [12, 13].

The IgM specific properties thus make them attractive candidates for the design of new biotechnological tools and future therapeutics. They have already inspired IgM‐based or Fc‐fusion drug research, although no clinical trials have yet reached completion [3]. Among many, examples are IgM‐like inhalable ACE2 fusion protein [14] and IgG‐IgM Fc chimeras such as the Fc‐μTP‐L309C [15]. Beside biotechnical bottlenecks in recombinant production and the characterization of IgMs or chimeras, the functional uncertainties of newly designed constructs urge for a deeper understanding of their modes of action and potencies.

In the present study, we report the biochemical and functional in vitro characterization of fragments from IgMs corresponding to the rigid Fc‐core in their two forms: pentameric and hexameric. Using data from biophysical methods, and an in‐house ELISA or hemolytic assay for the detection of complement activity, we demonstrated the ability of the IgM cores to bind C1q and to activate the first step of the classical complement pathway when adsorbed to a surface, but with the potentiality to inhibit complement‐induced cell lysis when present in solution.

Results

Protein expression and purification

IgMs have distinct star‐shaped structures with a compact Fc‐core, comprising the heavy‐chain Cμ3 and Cμ4 domains, and the terminal tailpieces. Surrounding this core, Fab‐Cμ2 arms are composed of the heavy‐chain constant domains, Cμ2 and Cμ1, as well as the variable VH domain, paired with the kappa or lambda light‐chain constant and variable domains, and extend outward with flexible conformations (Fig. 1) [9, 16, 17]. We focused on the compact and central assembly of IgMs by restraining the expressed sequence to the last three Cμ2, Cμ3 and Cμ4 constant domains and the terminal piece (Fig. 1). After mutagenesis of a full IgM heavy chain from our previous study [18] to delete the sequences coding for heavy variable and Cμ1 domains, constructs of IgM Fc‐core in expected hexameric oligomeric state (IgM‐Fc) or pentameric state (IgM‐Fc‐J) were obtained recombinantly after expression in eukaryotic cells (HEK293F) with or without the IgM J chain and established IgM purification [19]. Purified samples were analyzed using semi‐native PAGE adapted from Vorauer et al. [20]. At this stage, IgM‐Fc‐J samples migrated as a single homogeneous band with an apparent molecular mass between 800 and 900 kDa, whereas IgM‐Fc samples migrated as a broader band within the same range, suggesting the presence of oligomeric heterogeneities (Fig. 2A).

Fig. 1.

Schematic representation of IgM oligomers. Pentameric (left) and hexameric (right) oligomeric assemblies of type‐M Immunoglobulins are shown. The different domains of the μ chains (Cμ1 to Cμ4), light chains and J chain are depicted in shade of green. Disulfide bonds and glycosylation sites are indicated in black. Circles highlight the domains included in the Fc‐core constructs (IgM‐Fc‐J and IgM‐Fc).

Fig. 2.

Analyses of purified IgM‐Fc and IgM‐Fc‐J by semi‐native PAGE and negative stain TEM. (A) Adapted PAGE analysis of IgM‐Fc and IgM‐Fc‐J constructs. The presented spliced gel is representative of the quality‐control procedure after the final purification strep (n = 1). IgM purified from plasma (Antibodies‐online GmbH, Aachen, Germany) are shown as control along with native markers (Novex NativeMark Unstained Protein Standard; Invitrogen, Walthem, MA, USA). (B, C) Non‐processed images of recombinant IgM‐Fc cores by negative stain transmission electron microscopy. Representative fields of particles (100 nm scale bar) are shown on top of each panel with magnified views of some individual molecules (scale bar = 20 nm) shown on the lower part of each panel: (B) IgM‐Fc‐J and (C) IgM‐Fc.

Furthermore, the structural integrity and oligomeric states of IgM‐Fc samples were also confirmed with negative staining transmission electron microscopy (TEM). As anticipated, both IgM‐Fc cores exhibited the characteristic compact, round shapes typical of IgM cores, with recognizable asymmetrical five‐armed particles for IgM‐Fc‐J (Fig. 2B) and pseudo‐symmetrical six‐armed particles for IgM‐Fc and (Fig. 2C).

Analysis of IgM‐fc core assemblies and their oligomeric distributions by size‐exclusion chromatography coupled to multi‐angle‐laser‐light scattering, analytical ultracentrifugation and mass photometry

The oligomeric distributions of both samples were further characterized using several biophysical methods after SEC purification.

Size‐exclusion chromatography coupled to multi‐angle‐laser‐light scattering (SEC‐MALLS)

As a result of the discrepancy of theoretical molecular masses and the higher apparent masses observed by PAGE, SEC‐MALLS was performed. It revealed a homogeneous and sharp peak with an experimental mass of 454 kDa for IgM‐Fc‐J. This value aligns with the theoretical molecular mass of a fully glycosylated pentameric form (399 kDa from amino‐acid sequence +10 to 15% N‐glycosylation) (Fig. 3A and Table 1). Similarly, the IgM‐Fc was found to have a single experimental mass of 489 kDa matching the theoretical molecular mass of its hexameric form (460 kDa + 10 to 15% N‐glycosylation) (Fig. 3B and Table 1).

Fig. 3.

Analyses of purified IgM‐Fc and IgM‐Fc‐J by SEC‐MALLS, AUC and MP. The top panels show the elution profiles of the purified (A) IgM‐Fc‐J and (B) IgM‐Fc monitored by UV‐vis absorbance at 280 nm (right ordinate axis) and the molecular mass (left ordinate axis) derived from MALLS, refractometry and UV‐vis measurements. The estimated molecular masses of the protein peaks are indicated on the graphs. Data were obtained on a single protein preparation (n = 1) because SEC‐MALLS provides reproducible molecular‐mass determination. The middle panels show the sedimentation distributions of purified (C) IgM‐Fc‐J and (D) IgM‐Fc. Calculated sedimentation coefficients s and s _20,w in brackets are obtained as described in Materials and methods and are indicated on the graphs. Data were obtained on a single protein preparation (n = 1) because sv‐AUC provides reproducible sedimentation coefficients. The bottom panels show the population distributions of purified (E) IgM‐Fc‐J and (F) IgM‐Fc in the SEC peak attributed to the protein fragments. Letters indicate the oligomeric state: H, hexamers; P, pentamers; T, tetramers. The estimated average molecular masses and errors as standard deviation are indicated in the graphs (n = 4). Displayed analyses are representative of replicate experiments.

Table 1.

Summary of theoretical and experimental molecular masses and sedimentation coefficients of IgM‐Fc core constructs. The theorical peptide molecular masses are calculated based on the amino‐acid primary sequences of the subcloned IgM. Experimental molecular masses were determined by SEC‐MALLS, and sedimentation coefficients by AUC and MP. MP measurements were performed in replicates (n = 2–4).

Construct	Oligomers	Theoretical (kDa)	SEC‐MALLS (kDa)	AUC (S)	Mass photometry (kDa)
IgM‐Fc‐J	Pentamers	399	454	8.1	471 ± 10
IgM‐Fc	Hexamers	460	489	8.6	546 ± 9
	Pentamers	383			452 ± 11
	Tetramers	307			360 ± 27

Sedimentation velocity‐AUC (sv‐AUC)

Because differential oligomeric distributions were previously observed among our full IgM constructs, which have been recombinantly produced using the same procedure [18], and because PAGE heterogeneities were suspected (Fig. 2), further characterization of IgM‐Fc and IgM‐Fc‐J samples was conducted using sv‐AUC. Migration in the velocity field of IgM‐Fc‐J samples revealed a sharp peak at a sedimentation coefficient of approximately 8.1 S, indicating a highly homogeneous pentamer population (Fig. 3C and Table 1). Similarly, IgM‐Fc samples exhibited a single peak at approximately 8.6 S, albeit broader, indicating a less homogeneous hexamer population (Fig. 3D and Table 1).

Mass photometry (MP)

MP was further used to more precisely assess both IgM‐Fc‐J and IgM‐Fc oligomeric distributions. A single population with an average mass of 471 ± 10 kDa (mean ±SD over replicates) was observed for IgM‐Fc‐J (Fig. 3E and Table 1), aligning with the mass of fully glycosylated pentamers. By contrast, purified IgM‐Fc samples displayed a heterogenous population with observed masses of 546 ± 9 kDa, 452 ± 11 kDa and 360 ± 27 kDa, likely corresponding to hexamers, pentamers without the J‐chain and tetramers, respectively (Fig. 3F and Table 1).

Size‐exclusion chromatography (SEC) coupled to Mass photometry (MP)

As a result of the limited resolution in SEC separation, a more detailed exploration of the oligomeric distributions of IgM‐Fc fragments was conducted by coupling SEC with MP. The approach consisted of combining finely fractionating the size‐exclusion peaks with analyzing each fraction using MP. This strategy was applied to IgM constructs lacking the J chain and for which heterogenous populations were observed: IgM‐Fc from this study and a full recombinant IgM from our previous study, IgM617‐HL [18]. As anticipated, variations in both oligomeric distributions were observed throughout SEC elution, with a predominant population of hexamers detected in the earliest fractions. This hexameric population decreased along the elution volume in favor of pentamers lacking the J chain, tetramers and lower oligomer states (Fig. 4A). Interestingly, oligomeric distributions of both IgM‐Fc core and full IgM617‐HL resemble each other overall, although differences in oligomeric population ratios could be observed along the elution peaks (Fig. 4B).

Fig. 4.

Analysis of purified IgM‐Fc (A) and IgM617‐HL (B) by SEC coupled to MP. The upper part of both panels shows the SEC elution profile monitored by UV‐vis absorbance at 280 nm of purified IgM‐Fc or IgM‐Fc‐J. The lower part of panels shows population distributions monitored by MP of each individual fraction collected during SEC elutions. Histograms are colored according to the corresponding elution volumes. Letters indicate the oligomeric state: H, hexamers; P, pentamers; T, tetramers. Data were obtained on a single protein preparation (n = 1) because MP has shown highly reproducibility in mass distribution in Fig. 3.

In vitro functional analysis of IgM‐fc core assemblies and their oligomeric distributions with ELISA, biolayer interferometry and hemolytic assays

Classical pathway activation by IgM‐fc cores monitored by C4‐deposition ELISA

Using our in‐house ELISA based on the detection of C4b fragment deposition after cleavage of C4 by the C1 complex bound to coated IgM molecules [21], we analyzed the ability of both IgM‐Fc and IgM‐Fc‐J constructs to activate the CP. Notably, both samples triggered C4b deposition in our assays similarly to plasma‐derived full IgMs, with a dependency on C1q (Fig. 5A). Furthermore, ELISA using SEC‐fractionated IgM‐Fc samples, which exhibit different hexamers/pentamers ratios (see above), revealed that the fractions enriched with Fc‐core hexamers showed up to five times more C4b deposition compared to fractions with the Fc‐core pentamers. A strong correlation with the hexamer and pentamer ratio contained in the samples was also observable with a decrease of C4b deposition along with the decrease of hexamer population in the samples (Fig. 5B). Interestingly, a similar trend was observed with the SEC‐fractionated full IgM617‐HL (see above), although the effect of the hexamer/pentamer ratio on C4b deposition was less pronounced (Fig. 5C). All together, these results confirm the influence of the polymer distributions of the IgM oligomers on activating the C1 complex and initiating the first step of the CP proteolytic cascade.

Fig. 5.

In vitro analysis of C1 activation by purified IgM‐Fc and IgM‐Fc‐J. Recombinant IgM‐Fc fragments and IgMs purified from plasma were coated on microplate wells and incubated with either NHS, C1q‐depleted NHS (NHSΔ) or reconstituted NHS (NHSΔ + C1q). C1 activity was monitored via C4b deposition. Reported values are average of replicated normalized experiments and errors are reported as standard deviation between independent replicates (n = 4). (A) C4b deposition comparison between IgMs purified from plasma IgM (pIgM), IgM‐Fc and IgM‐Fc‐J samples obtained in protein SEC peaks (Fig. 3). (B) C4b deposition comparison between the IgM‐Fc of SEC peak (Fig. 1) and the individual SEC fractions (Fig. 4A). (C) C4b deposition comparison between the recombinant full IgM617‐HL of SEC peak and the individual SEC fractions (Fig. 4B). Histograms are colored according to the corresponding elution fractions in Fig. 4.

Binding of C1q to IgM‐fc cores using biolayer interferometry (BLI)

To confirm the dependency of the CP activation on the binding of Fc‐core to C1q, BLI was used to determine the binding kinetics of plasma‐derived C1q to immobilized IgM‐Fc and IgM‐Fc‐J constructs. Because both recombinant Fc‐core forms lack an IgM light chain, the BLI strategy using protein L as a capture molecule could not be employed [18]. Therefore, the streptavidin/biotin system was used and optimized to capture biotinylated IgM‐Fc and IgM‐Fc‐J samples (see Materials and methods). C1q binding to the IgM cores exhibited a concentration‐dependent kinetics with a multiphasic profile, which was fitted with a 2 : 1‐heterogenous model to account for the interaction complexity. Under these conditions, the affinities of C1q for both Fc‐core forms could be determined in the ten‐nanomolar range (Fig. 6 and Table 2) [18].

Fig. 6.

Kinetics analysis of the interaction between C1q and the purified IgM‐Fc and IgM‐Fc‐J. Fc‐core fragments were captured on Streptavidin biosensors after pooling SEC fractions containing protein samples. The functionalized biosensors were dipped in wells containing plasma‐purified C1q at different concentrations (3.13, 6.25, 12.5, 25, 50, 100 and 200 nm). The binding signals (gray‐scaled sensorgrams) were obtained by subtracting the signals from empty biosensors and from zero‐concentration samples. Fitted curved are depicted as black lines and kinetics values were obtained by global fitting using a 2 : 1 heterogenous ligand model and averaging duplicated independent experiments. Shown kinetics analysis are representative of each binding experiment: (A) IgM‐Fc‐J and (B) IgM‐Fc. Experiments were performed in replicates (n = 2) to assess the reproducibility and experimental variability. Similar profiles were observed and only one representative experiment over replicates is shown.

Table 2.

Kinetics and affinity parameters of C1 binding to IgM‐Fc and IgM‐Fc‐J obtained by BLI. Experiments were performed in replicates (n = 2).

Construct	k a1	k d1	K D1	k a2	k d2	K D2
	104/ms	10−4/s	10−9  m	106/ms	10−2/s	10−9  m
IgM‐Fc	2.78 ± 0.58	3.36 ± 2.15	12.1 ± 0.3	3.33 ± 0.99	6.13 ± 5.41	18.4 ± 1.8
IgM‐Fc‐J	2.06 ± 0.39	3.87 ± 1.62	18.8 ± 0.1	3.24 ± 1.64	4.72 ± 0.66	14.6 ± 0.9

Complement‐induced hemolysis by IgM‐Fc core

Because both previous assays can be considered as antigen/Fab independent methods to evaluate the functionalities of IgM constructs, we also used a hemolytic assay to assess the possibility of IgM‐Fc forms, as well full IgM617‐HLJ, to activate or to inhibit the classical pathway activation in a cellular context [22, 23]. This assay measures the lysis of sheep red blood cells (sRBCs) either sensitized or not with a rabbit anti‐sRBC antibody (hemolysin), followed by incubation with human plasma as a source of all CP components, supplemented with either IgM‐Fc, IgM‐Fc‐J or full IgM617‐HLJ.

As expected, experiments with full IgM617‐HLJ added to plasma showed no effects either on non‐sensitized (Fig. 7A) or sensitized cell lysis (Fig. 7D).

Fig. 7.

Time course of sRBC hemolysis induced by full IgM and IgM‐Fc cores. Direct hemolysis of non‐sensitized cells (A–C) or competitive hemolysis of hemolysin‐sensitized cells (D–F) was followed after mixing either IgM617‐HLJ (A, D), IgM‐Fc (B, E) or IgM‐Fc‐J (C, F) with human plasma and incubation of the total mix sRBCs + IgM + plasma for 25 min. The same control experiments with sensitized sRBCs with hemolysin+plasma, non‐sensitized sRBCs with plasma and non‐sensitized sRBS with either IgM617‐HLJ, IgM‐Fc or IgM‐Fc‐J at the highest concentration are shown in each panel. Experiments were performed in replicates (n = 2) with two different human plasmas. Only one representative experiment over replicates is shown.

When testing the functionality of IgM‐Fc and IgM‐Fc‐J on non‐sensitized sRBCs, a surprising and concentration‐dependent reduction in hemolysis was observed. Compared to non‐sensitized cells incubated with plasma alone, a slight delay of hemolytic activity, as measured by 50% total hemolysis (TH₅₀) between 1 and 5 min, was observed upon addition of either IgM‐Fc (Fig. 7B) or IgM‐Fc‐J (Fig. 7C) at a concentration as low as 0.14 nm. Despite no difference being seen in end‐point lysis (90–99% at 25 min) (Fig. 7B,C), the addition of IgM‐Fc cores led to a diminution of the total complement activity, comparable to the residual activity of plasma against non‐sensitized cells (Fig. 8). Increasing the concentration of IgM‐Fc or IgM‐Fc‐J up to 4.5 nm resulted in a marked decrease of hemolyzed cells amounts (Fig. 7B,C) and a complete loss of total complement activity (Fig. 8).

Fig. 8.

Total complement activities induced by IgM‐Fc and IgM‐Fc‐J. Hemolysis of unsensitized cells (direct assays) or hemolysin‐sensitized cells (competitive assays) was measured after mixing either IgM‐Fc (pink scale histograms) or IgM‐Fc‐J (blue scale histograms) at three different concentrations with human plasma. Controls (gray scale histograms) were performed the same way but omitting one of the components. Total complement activities are calculated as described in the Materials and methods. Experiments were performed in replicates (n = 2) with two different human plasmas.

To investigate the influence of Fc‐cores on IgG‐induced CDC, a second set of experiments was conducted using hemolysin‐sensitized sRBCs and plasma supplemented with varying concentrations of IgM‐Fc or IgM‐Fc‐J. Noticable shifts of TH₅₀ up to 6 min (Fig. 7E,F) and decrease of the residual complement activity (Fig. 8) were measurable when adding either one of both constructs up to 4.5 nm. Although decreases were less pronounced than those observed in direct hemolytic assays, the delayed hemolytic rates (5–9 min) and residual activities (25–38%) tend to be comparable to those measured with non‐sensitized cells (9 min and 11%, respectively).

Taken together, these data suggest that IgM‐Fc cores do not enhance CDC in vitro as antigen‐specific Igs. Interestingly, they might have an inhibitory potential on Igs/complement‐induced hemolysis by reducing the total complement activity.

Discussion

In the pursuit of new biotechnological tools using engineered immunoglobulins, fundamental questions persist, particularly regarding the unique attributes of the IgM class. These questions include the ability to produce the original and innovative molecules in recombinant functional forms, to accurately characterize their molecular properties and sample qualities, and to thoroughly understand their mechanisms of action. Such information is indeed crucial for assessing the potential of macro‐immunoglobulins in the development of future biological tools or therapeutics. In the present study, we focused on expressing and purifying the Fc‐core of IgM at the laboratory scale and on characterizing its quality and functionality using a combination of in vitro biochemical and biophysical methods. We expressed two constructs (IgM‐Fc‐J and IgM‐Fc) of IgM μ chain comprising the last three constant domains (Cμ2, Cμ3 and Cμ4) and the tail piece. By doing so, with or without the J chain in the HEK239F expression system, we expected to obtain either its pentameric or hexameric forms (Fig. 1).

In a previous study, we observed more heterogeneities in the oligomeric states of full IgMs expressed without the J chain compared to those with the J chain [18]. Consequently, we also applied the panel of biophysical analytical methods, including SEC‐MALLS, sv‐AUC and MP (Fig. 3), to determine the oligomeric distribution and assess the homogeneity of the two new IgM‐Fc constructs. Similar to our findings on full IgMs, IgM‐Fc‐J core could be purified as highly homogeneous pentamers, whereas IgM‐Fc still exhibited heterogeneities, notably containing lower assemblies, presumably pentamers lacking the J chains and tetramers in addition to the hexamers. The difference in heterogeneity is most likely a result of the absence of the J chain in the IgM‐Fc and IgM617‐HL constructs. Without the J chain, IgM assembly may probably become less constrained, leading to a mixture of oligomers and not only hexamers. This decreased homogeneity has been consistently observed and our results align with most of the previous attempts to produce recombinant full IgM hexamers using different expression systems [24, 25, 26, 27]. Altogether, these results suggest a crucial role of the J chain in stabilizing the IgM Fc‐core, likely in a more favorable manner than a sixth IgM protomer. The structural details by which the additional protomer fails to totally replace the J chain require further investigation. Nonetheless, our study, along with our previous work on full recombinant IgM models [18, 28], reinforces the importance of extensively investigating the polydispersity of large macromolecular complexes such as IgMs using advanced biophysical techniques. Conventional methods such as SEC and PAGE, although adapted for IgMs, have not proven completely suitable for detecting oligomeric status and heterogeneities in full IgM and IgM‐Fc samples.

In particular, semi‐native PAGE significantly overestimated the apparent molecular masses of both IgM‐Fc and IgM‐Fc‐J, showing migration in the range 800–900 kDa, well above the expected masses based on theoretical calculations and SEC‐MALLS results (Fig. 2A). This discrepancy is likely due to the complex migration behavior of large, glycosylated and non‐globular proteins under semi‐native conditions, which affects both their mobility and interaction with the matrix. In addition, in the case of IgM‐Fc cores, we were also unable to resolve the various oligomers during the purification and quality controls, likely because of the limitations in terms of separation resolution or detection capabilities of those methods.

In our studies, the use of MP has emerged as a valuable and reliable method for quickly characterizing the polydispersity of full IgMs and IgM‐Fc cores with low sample amounts. To address the limitations of SEC and leverage the advantages of MP, we combined both methods with a strategy involving in depth characterization of the SEC protein elution peak. By applying this approach to both full IgM and IgM‐Fc lacking the J chain, we were thus able to isolate different fractions containing different ratios of oligomers, with the first eluted fractions enriched with hexamer populations (Fig. 4). This approach appeared to be faster and easier than previous methods, which have used the combination of sucrose density gradient centrifugation with SDS‐PAGE [25, 29].

For assessing complement activation levels, several techniques and assays are available. Hemolytic assays are traditionally used by measuring lysis of sRBCs opsonized with anti‐sheep erythrocyte antibodies (hemolysin) or not sensitized rabbit blood cells. These assays are increasingly replaced by liposome immunoassays, where sRBCs are substituted with liposomes coated with antigens and containing a reporter molecule [30]. Alternatively, ELISAs using plate coated IgMs, mannose or lipopolysaccharide allow for a more accurate and reproducible quantitation of the different complement components produced after either classical, lectin or alternative pathway activations, respectively [30]. During our study, we evaluated the IgM‐Fc core ability to activate CP using both ELISA and sRBC hemolytic assays. Notably, our data revealed in vitro deposition of C4b in a C1q‐dependent manner as detected by ELISA (Fig. 5A). The activation of CDC through CP is assumed to be triggered by IgMs only when they are bound to their specific antigens at pathogen surfaces. The antigen binding may be required to induce quaternary structural changes within IgM molecules to allow exposure of C1 binding sites, C1 binding and subsequent proteolytic mechanisms [31]. The bias induced by immobilizing/coating IgMs onto in vitro surface is well acknowledged for the quantification of complement component depositions and CP activation [32]. Our ELISA data with IgM‐Fc and IgM‐Fc‐J constructs confirm this bias. Moreover, they suggest the Fab independency of CP activation under such conditions, which might provoke the necessary structural exposure of the C1q binding motif in the Cμ3 domain and hidden by the Cμ2 domains when IgM are unbound [31, 33].

Furthermore, our method for separating enriched fractions of either IgM‐Fc or full IgM hexamers from pentamers and lower oligomeric states allowed us to explore the differences between the forms in C1 activation. In our previous study, no obvious differences could be observed between pentamer and hexamer constructs. This observation was likely a result of the presence of high and variable proportions of pentamers in the purified samples of IgMs lacking the J chain [18]. Recently, disparities of up to two‐fold in in vitro C4b depositions have been observed between supposed hexameric and pentameric IgM preparation [34]. Our ELISA data clearly demonstrated that the enrichment in hexamers in both IgM‐Fc and full IgM‐HL samples is associated with an increase of C4b deposition and thus activation of the first CP step (Fig. 5B,C). Our latest observations are consistent with previous hemolytic assays, which have demonstrated the higher efficacy of hexameric IgMs in inducing CDC compared to pentameric IgMs [25, 35].

The dependency of the activation of CP on direct C1q binding to IgM cores was further confirmed by our new kinetics data obtained via BLI, which showed affinities of C1q for IgM‐Fc constructs in the 20‐nanomolar range (Fig. 6). Although the affinity values for Fc cores fall within a similar range as those measured for full IgMs [18, 21, 36], differences of up to one order of magnitude can be easily observed between them (Table 2). These differences can be explained by the different methods and strategies employed. For example, in the case of IgM‐Fc and IgM‐Fc‐J constructs, we had to adapt the biotinylating protocol to achieve the molecule capture on the streptavidin BLI biosensors, whereas capturing via protein L has been shown to be more efficient for full IgMs [18]. The Fab independency of C1q binding to IgMs and IgGs in optical surface‐based methods such as surface plasmon resonance and BLI remains an open question. Indeed, we or others have measured kinetics and affinities between partial or full immunoglobulins and C1q under diverse conditions and without any pre‐coated specific antigens. The influence of Igs coating in inducing and stabilizing favorable Ig conformation for C1q binding has been previously suggested [18]. Zhou et al. [37] also reported that, for coated IgGs, the binding of a specific antigen can influence the C1q binding kinetics. Although the C1q binding to immobilized trastuzumab followed by Her2 binding or immobilized preformed trastuzumab/Her2 complex could be prevented by the presence of HER2 antigen, the opposite was observed for the adalimumab/TNFα complex. Their observation is strongly related to the different CDC activities of these IgGs [37].

To verify whether our IgM‐Fc constructs can modulate CDC, hemolytic assays were conducted using either non‐sensitized or IgGs‐coated sRBCs. With non‐sensitized cells, both IgM‐Fc and IgM‐Fc‐J prevent complement induced hemolysis of sheep sRBCs with a concentration dependency: the lysis levels and rates drastically decreased as the amount of Fc cores increase (Fig. 7). Interestingly, the total residual complement activity appeared to be abolished by the presence of the IgM‐Fc cores with levels that were even lower than the residual complement activity of plasma on non‐sensitized sRBCs (Fig. 8). The ability of the recombinant IgM‐Fc cores to inhibit Ig‐induced CDC was confirmed with hemolysin‐sensitized cells. Although less marked, decrease of the total complement activity was observed when adding IgM‐Fcs to plasma to reach residual activities comparable to the one on non‐sensitized cells (Fig. 8). Of note, no relevant differences could be reported between inhibitory effects of IgM‐Fc and IgM‐Fc‐J (Fig. 7). Because only IgM‐Fc samples obtained prior fine fractionation and hexamer enrichment could be used for those assays, the mixed content of hexamers and pentamers might mask a putative differential effect between oligomeric states.

This capability to modulate and counterbalance the Ig‐induced CP activation has been previously documented for IgG‐based molecules inspired by the oligomeric structuration of IgMs. For example, Fc‐μTPL309C [15] and GL‐2045 [38] have been reported to have significant impacts on reducing complement activation in solution, whereas surface‐bound Hexa‐Fc [39] retains aptitude to fully activate the complement. Interestingly, some of these Fc constructs designed so far keep the ability to bind C1q, leading to C4b generation, but fail to induce C2 cleavage [15] and/or result in reduced production of C3a and no C5b [15, 38]. Thus, it is not surprising to observe CDC inhibition by our IgM‐Fc cores in solution, whereas they still bind C1q and induce C4b deposition when bound to an in vitro surface. The detailed molecular mechanism by which the Ig‐Fc multimers can block CP remains to be determined. A clue might reside in the absence of the Fab region from all the studied constructs and in our results of assays performed with a non‐relevant recombinant full IgM. Contrasting with the IgM‐Fc cores, no reduction in hemolytic abilities and in total complement activity could be observed on either non‐sensitized or hemolysin‐sensitized‐cells experiments when full IgMs have been added in plasma (Fig. 7). As mentioned before, it is well established that IgM binding to antigen through Fabs is required for activating the complement cascade. One can speculate that the absence of Fab might favor sequestration of the C1 complex in solution by the IgM Fc‐cores or Ig‐based molecules and thus consumption of the initial complement components such as C4, preventing recruitment of C2 and the subsequent C3 and C5 at the target surface.

In the present study, we demonstrate that the IgM‐Fc cores, both in their J chain‐containing and J chain‐lacking forms, can compete with IgG and reduce CDC in hemolytic assays. This inhibitory capacity likely depends on the preserved ability of the IgM constant region to bind C1q, as confirmed by BLI. Interestingly, despite this inhibitory effect in hemolysis, the IgM‐Fc constructs retain the ability to activate the classical pathway and induce C4b deposition in ELISA‐based assays, consistent with findings for other oligomeric Ig‐Fc‐based molecules. These apparently paradoxical observations raise new questions about the molecular regulation of the classical pathway in solution and suggest that the Fc region of IgM alone is sufficient to mediate C1q binding and complement activation, a property prevented by the Fab arms.

Importantly, the finding that IgM‐Fc fragments can both engage C1q and inhibit CDC opens up promising avenues for the development of novel immunomodulatory strategies. The ability of these constructs to modulate complement activity could be harnessed to develop IgM‐Fc‐based therapeutics for diseases involving excessive or dysregulated complement activation, such as systemic lupus erythematosus, myasthenia gravis, ischemia–reperfusion injury or age‐related macular degeneration. Once the underlying mechanisms are fully elucidated, engineered multimeric IgM‐Fc scaffolds could be optimized to act as soluble complement decoys, selectively regulate antibody‐driven complement responses or function as complement‐redirecting biologics with therapeutic precision.

Finally, the constructs and approaches described in the present study provide a versatile platform to dissect the structural determinants of C1q recognition and classical pathway activation by immunoglobulins, contributing valuable tools for both basic and translational immunology.

Materials and methods

DNA construct generation

Mammalian expression vector for a truncated construct comprising the Cμ2, Cμ3 and Cμ4 domains (Fig. 1) and the terminal tailpiece of the IgM μ chain (pcDNA3.1(+)‐Cμ2Cμ3Cμ4Tp) was obtained using site‐directed mutagenesis (QuickChange II XL kit; Agilent, Santa Clara, CA, USA) to delete the sequence of the variable and Cμ1 domains from the pcDNA3.1(+)‐IgM012 H chain vector, a subcloned full length and optimized IgM heavy chain sequence previously published in Chouquet et al. [18].

The vector construct for the mammalian expression of the human J chain (pcDNA3.1(+)‐J‐chain) was also described previously in Chouquet et al. [18].

Protein expression

First, to generate a stable HEK293F cell line expressing IgM‐Fc, FreeStyle 293F cells (Thermo Fisher Scientific, Waltham, MA, USA; RRID: CVCL_D603) were purchased directly from the provider and no additional authentication or mycoplasma testing was carried out in our laboratory. They were transfected with pcDNA3.1(+)‐Cμ2Cμ3Cμ4Tp using 293fectin in accordance with the manufacturer's protocol (Thermo Fisher Scientific) and generation of the stable line achieved with culture in FreeStyle 293 expression medium supplemented with 400 μg·mL⁻¹ G418 (Thermo Fisher Scientific).

These cells were then transfected with pcDNA3.1(+)‐J‐chain in the same way and the stable transfectants producing IgM‐Fc‐J were generated using culture in medium supplemented with an additional 100 μg·mL⁻¹ hygromycin (Sigma‐Aldrich, St Louis, MO, USA). The stable cells expressing each Fc‐core construct were then cultured in Freestyle 293 expression medium under antibiotic pressures and maintained every 3–4 days when cell density approached 3 × 10⁶ cells·mL⁻¹.

Recombinant expressions of full IgM617‐HL and IgM617‐HLJ have been previously described in Chouquet et al. [18].

Protein purification

All IgM protein constructs were purified from harvested supernatants according to Hennicke et al. [19]. Briefly, POROS CaptureSelect™ IgM Affinity Matrix (Thermo Fisher Scientific) was used for affinity chromatography. Culture supernatants were directly applied to a packed column and the IgM molecules were eluted with 1 m arginine, 2 m MgCl₂, pH 3.5 or pH 4.0. The collected fractions were immediately neutralized with 1 m Tris pH 8.5. The IgM‐containing fractions were then pooled, dialyzed against 0.025 m Tris‐Base, 0.137 m NaCl, 0.003 m KCl, pH 7.4 and concentrated. For a second purification step consisting of SEC, they were applied on a Superose™ 6 increase 10/300 or 16/600 column (Cytiva, Marlborough, MA, USA) equilibrated in the dialysis buffer at a flow rate of 0.5 mL·min⁻¹. Proteins were eluted as a single peak. Identification of purified proteins by SDS‐PAGE was performed as described in Vorauer‐Uhl et al. [20] using NativePAGE 3–12% Bis‐Tris gels followed by Coomassie Blue staining.

C1q was purified from human serum according to a well‐established protocol. Briefly, IgG‐ovalbumin insoluble immune aggregates were prepared as described by Arlaud et al. [40]. After clarification by centrifugation, human serum, obtained from the Etablissement Français du Sang Rhône‐Alpes (agreement number 21‐001 regarding its use in research), was incubated with immune aggregates on ice for 45 min. C1/immune complexes were then collected by centrifugation and extensively washed with 20 mm Tris, 120 mm NaCl and 5 mm CaCl₂ at pH 7.4. Following C1r and C1s release by washing with buffer containing EDTA, C1q was eluted from immune aggregates with 50 mm Tris and 700 mm NaCl at pH 10.0 and recovered by centrifugation. C1q samples were further purified to homogeneity by carboxymethyl‐cellulose chromatography.

TEM

TEM imaging was performed similarly as in Hennicke et al. [28] or Chouquet et al. [18]. Briefly, approximately 4 μL of diluted IgM‐Fc‐J or IgM‐Fc samples (60–80 ng) were applied to a carbon film evaporated onto a mica sheet. The carbon film was then floated off the mica in approximately 100 μL of 2% sodium silicotungstate (SST; Agar Scientific, Sheffield, UK) and transferred onto a 400 mesh Cu TEM grid (Delta Microscopies, Mauressac, France). Images were acquired with a CETA camera on a Tecnai F20 TEM microscope operating at 200 keV.

SEC‐MALLS analyses

SEC combined with online detection by MALLS, refractometry and ultraviolet‐visible spectroscopy (UV‐vis) was used to measure the absolute molecular mass in solution as in Hennicke et al. [28] or Chouquet et al. [18]. The SEC runs were performed using a Superose™ 6 increase 10/300 column (Cytiva) equilibrated in 0.025 m Tris‐Base, 0.137 m NaCl and 0.003 m KCl, pH 7.4. Protein sample (50 μL), concentrated to approximately 1 mg·mL⁻¹, was injected with a constant flow rate of 0.5 mL·min–¹ at room temperature. Online MALLS and differential refractive index detection were performed using a DAWN‐HELEOS II detector (Wyatt Technology Corp., Santa Barbara, CA, USA) with a laser emitting at 690 nm and an Optilab T‐rEX detector (Wyatt Technology Corp.), respectively. Averaged molar masses determination was performed via astra6 (Wyatt Technology Corp.), using the “protein conjugate” module. The following refractive index increments and UV‐vis absorbance values were used: dn/dc protein = 0.185 mL·g⁻¹; dn/dc glycosylation = 0.15 mL·g⁻¹; A ₂₈₀ (0.1%, 1 cm) = 1.38.

Analytical ultracentrifugation

sv‐AUC experiments were conducted as in Hennicke et al. [28] or Chouquet et al. [18], in an XLI analytical ultracentrifuge (Beckman, Palo Alto, CA, USA) using an ANTi‐60 rotor, double channel Ti center pieces (Nanolytics, Germany) of 12‐ or 3‐mm optical path length equipped with sapphire windows, and the reference channel being typically filled with the sample solvent. Acquisitions were conducted overnight at 4 °C and at 32 000  g using absorbance (280 nm) and interference detection. Data processing and analysis were completed using sedfit [41] from P. Schuck (NIH, Bethesda, MD, USA), redate [42] and gussi [43] from C. Brautigam (University of Texas, Dallas, TX, USA), as well as using standard equations and protocols described previously [44, 45, 46].

MP

Mass determinations was performed similarly as in Hennicke et al. [28] or Chouquet et al. [18]. In detail, coverslips (high precision glass coverslips, 24 × 50 mm², No. 1.5H; Marienfeld, Lauda‐Königshofen, Germany) were cleaned by sequential sonication in Milli‐Q H₂O, 50% isopropanol (HPLC grade)/Milli‐Q H2O and Milli‐Q H₂O (5 min each) (MilliporeSigma, Burlingotn, MA, USA), followed by drying with a clean nitrogen stream. To keep the sample droplet in shape, reusable self‐adhesive silicone culture wells (reusable CultureWell gaskets; Grace Bio‐Labs, Bend, OR, USA) were cut into four to 10 segments. To ensure proper adhesion to the coverslips, the gaskets were dried well by using a clean nitrogen stream. To prepare a sample carrier, gaskets were placed in the center of the cleaned coverslip and fixed tightly by applying light pressure with the back of a pipette tip. Protein landing was recorded using a Refeyn One (Refeyn Ltd, Oxford, UK) MP system by forming a droplet of each sample at a final concentration of 8 nm in 0.025 m Tris‐Base, 0.137 m NaCl and 0.003 m KCl, pH 7.4. Movies were acquired for 120 s (12 000 frames) with acquire, version 2.1.1 (Refeyn Ltd) software using large camera acquisition settings. Contrast‐to‐mass calibration was performed using a mix of proteins with molecular mass of 66, 146, 500 and 1046 kDa. Data were analyzed using discover, version 2.1.1 (Refeyn Ltd) and analysis parameters were set to T1 = 1.2 for threshold 1. The values for number of binned frames (nf = 8), threshold 2 (T2 = 0.25), and median filter kernel (= 15) remained constant. The mean peak contrasts were determined in the software using Gaussian fitting. The mean contrast values were then plotted and fitted to a line. The experimental masses were finally obtained by averaging replicates using independent recombinant IgM preparations (n = 4) and errors were the standard deviation.

BLI

BLI experiments were performed on an OctetRED96e (Sartorius, Göttingen, Germany) and were recorded with the manufacturer's software (data acquisition, version 11.1). IgM‐Fc and IgM‐Fc‐J samples were buffer exchanged against either 0.01 m Na₂HPO₄, 0.0018 m KH₂PO₄, 0.137 m NaCl or 0.0027 m KCl at pH 7.4 (phosphate‐buffered saline, PBS) with Zeba Spin Desalting columns (Thermo Fisher Scientific) and were biotinylated using normal human serum (NHS)‐PEG4‐biotin EZ‐link kit (Thermo Fisher Scientific) prior to loading. The recommended manufacturer conditions were adapted to achieve a higher coupling degree with one biotin molecule coupled to each protomer (5–6 biotin per Fc molecule).

Similarly to as described in Chouquet et al. [18], analyses were performed in 0.2 mL per well in black 96‐well plates (96 well PP; Greiner Bio‐One, Kremsmünster, Austria) at 25 °C at 1000 rpm agitation and using SA (Streptavidin) Biosensors (Sartorius). These tips were pre‐wetted in 0.2 mL of PBS, 0.02% Tween‐20 for 10 min, followed by equilibration in pre‐wetting buffer for 120 s. Biotinylated IgM‐Fc and IgM‐Fc‐J samples were applied at concentrations between 50 and 100 μg·mL⁻¹ and loaded for 600 s until reaching a spectrum shift between 3.5 and 4.5 nm, followed by an additional equilibration step of 120 s or more in analysis buffer composed of TBS complemented with 0.002 m CaCl₂ and 0.02% Tween‐20. Kinetics analyses were performed with association phase of plasma C1q monitored for 300 s and with sample diluted in analysis buffer at concentrations between 0 and 200 nm, followed by a dissociation phase in analysis buffer for 600 s. To assess and monitor unspecific binding of analytes, measurements were performed with biosensors treated with the same protocols but replacing ligand solutions with analysis buffer. All measurements were performed in replicates (n = 2) using independent recombinant IgM‐Fc and IgM‐Fc‐J loadings. Kinetics data were processed with the manufacturer's software (data analysis ht, version 11.1; Sartorius). Signals from reference biosensor and zero‐concentration sample were subtracted from the signals obtained for each functionalized biosensor and each analyte concentration. Resulting specific kinetics signals were then fitted using a global fit method and 2 : 1 heterogeneous ligand model. Reported kinetics parameter values were obtained by averaging the values obtained with replicated assays and reporting errors as the standard deviation (n = 2).

Complement activation detection by C4‐deposition ELISA

The activation of the classical complement pathway was monitored by an ELISA based on the detection of C4b deposition according to Bally et al. [21], Hennicke et al. [28] or Chouquet et al. [18]. Briefly, 200 ng of IgM‐Fc or IgM‐Fc‐J diluted in PBS were adsorbed on a MaxiSorp 96‐well plate (Thermo Fisher Scientific) by incubating overnight at 4 °C. After washing, unspecific binding was prevented by saturation with PBS complemented with 2% bovine serum albumin (Sigma‐Aldrich, St Louis, MO, USA) for 1 h at 37 °C. Replicate wells were then incubated with either NHS diluted 25 times in a buffer containing 5 mm barbital, 150 mm NaCl, 5 mm CaCl₂ and 1.5 mm MgCl₂ at pH 7.4, C1q‐depleted serum (NHSΔ; Complement Technology, Tyler, TX, USA) diluted 25 times, or NHSΔ diluted 25 times and reconstituted with purified human C1q (4 μg·mL⁻¹) for 1 h at 37 °C. NHS was obtained from the Etablissement Français du Sang Rhône‐Alpes (agreement number 21‐001 with respect to its use in research). The reaction was stopped by washing with a buffer containing 5 mm barbital, 150 mm NaCl and 5 mM EDTA at pH 7.4. Deposition of cleaved C4 form was detected with a rabbit anti‐human C4 polyclonal antibody (Siemens, Munich, Germany), an anti‐rabbit‐HRP antibody conjugate (Sigma‐Aldrich), addition of TMB (Sigma‐Aldrich) and a Clariostar plate reader (BMG Labtech, Ortenberg, Germany). Polyclonal IgM isolated from human serum (Sigma‐Aldrich) was used as a control. Blank wells were prepared and processed similarly to wells coated with IgMs but incubated with buffer instead of NHS samples. Reported values were obtained by normalizing each data set (polyclonal IgM/NHS or IgM‐Fc or IgM617‐HL defined as 100) after blank subtraction and by averaging data obtained in replicated assays using independent recombinant IgM preparations (n = 4); reported errors were the standard deviation of the replicates.

Hemolytic assays induced by classical complement pathway

Hemolytic assays to evaluate classical pathway activation were performed similarly as previously described in [47]. sRBCs (Atlantis France, Voulmentin, France) were purchased directly from the provider and no additional authentication or mycoplasma testing was carried out in our laboratory. They were washed three times with DGVB++ buffer (glucose 2.5%, barbital 2.5 mm, CaCl₂ 2.1 mm, MgCl₂ 0.5 mm, NaCl 72.5 mm and gelatin 0.05%, pH 7.4), with each wash step followed by cell sedimentation (800  g for 10 min at 4 °C). Cell concentration was adjusted to 0.5 × 10⁸ cells·mL⁻¹. Buffer (for non‐sensitized cells) or hemolysin (anti‐erythrocyte antibodies; Sigma‐Alrich) (for sensitized cells) was then added with a final dilution of 1 : 5000 and the mixture incubated at 37 °C for 15 min. A mixture of 0.9% (v/v) human plasma, non‐sensitized or sensitized sRBCs, and dilutions of IgM‐Fc, IgM‐Fc‐J or IgM617‐HLJ were incubated at 37 °C. Absorbance at 660 nm was recorded during incubations using a spectrophotometer (model 190DES; Safas, Monaco) to measure turbidity as rate of erythrocyte lysis. After blank subtraction, defining time zero as the starting time of hemolysis of sensitized RBCs mixed with human plasma and normalization, percentage of hemolysis values were reported after 25 min of incubation and TH₅₀ as the specific time at 50% of lysis. Total residual complement activities are expressed as the ratio: (TH50 of control)/(TH50 of sample) × 100. Values were averaged over two replicates and two different human plasmas. Errors were reported as standard deviations over these replicates.

Conflict of interest

The authors declare that they have no conflict of interest.

Author contributions

J‐BR, AP and WLL were responsible for conceptualization. NT, J‐BR and RK were responsible for funding acquisition. NT, CG and RK were responsible for team supervision. Investigation and data: IB, AC, AP, VR, RK, NT and J‐BR were responsible for protein expression and purification; AP and J‐BR were responsible for mass photometry, ELISA and BLI measurements; AP and CDP were responsible for hemolytic assays; and AP and WLL were responsible for electron microscopy. AJP, WLL and J‐BR were responsible for writing the article.

Data availability statement

Raw or processed data are available from the corresponding author upon reasonable request. All resources produced for the publication are available under a material transfer agreement from CEA or CNRS.