L- and H-ficolins are serum oligomeric defense proteins consisting of a collagen-like region and a fibrinogen-like recognition domain that bind to pathogen- and apoptotic cell-associated molecular patterns. They share with mannan-binding lectin (MBL) the ability to associate with MBL-associated serine proteases (MASP)-1, -2, -3, and protein MAp19 and to trigger the lectin complement pathway through MASP-2 activation. Recent studies have revealed the essential role of Lys55 in the collagenous region of MBL in the interaction with the MASPs and calreticulin (CRT). To test the possible involvement of the homologous residues Lys57 of L-ficolin and Lys47 of H-ficolin, point mutants of both proteins were produced in which these residues were mutated to Ala, Glu, or Arg. The resulting mutants exhibited oligomerization patterns and ligand binding properties similar to those of their wild-type counterparts. In contrast, all three mutations strongly inhibited the interaction of L- and H-ficolins with MAp19 and MASP-2 and impaired the ability of each ficolin to trigger the lectin pathway. In the case of MASP-1 and MASP-3, replacement of the target Lys residues by Ala or Glu abolished interaction, whereas the Lys to Arg mutations had only slight inhibitory effects. Likewise, binding of each ficolin to CRT was inhibited by mutation of Lys to Ala or Glu, but not to Arg. In conclusion, residues Lys57 of L-ficolin and Lys47 of H-ficolin are key components of the interaction with the MASPs and CRT, providing strong indication that MBL and the ficolins share homologous binding sites for both types of proteins.
Ficolins are members of the defense collagen family that comprises oligomeric proteins with globular recognition domains able to sense danger signals, such as pathogen- or apoptotic cell-associated molecular patterns, and collagen-like stalks providing the link with immune effectors (1, 2). Ficolins are assembled from homotrimeric subunits comprising a collagen-like triple helix and a lectin-like domain composed of three fibrinogen-like (FBG)3 domains. Two cysteines at the N-terminal end of the polypeptide chains form interchain disulfide bonds that mediate assembly into higher oligomerization structures (3, 4). In humans, L- and H-ficolins have been characterized in serum whereas M-ficolin is mainly expressed by the monocytic cell lineage (5, 6, 7). In addition to humans, ficolins have been identified in different mammalian species including rodents and pigs (8, 9), which have two related ficolin genes called A and B and α and β, respectively, orthologous to the human L- and M-ficolin genes, respectively (10). To date, H-ficolin has only been identified in humans and it has been reported recently that the mouse and rat homologues of the H-ficolin gene are pseudogenes, which accounts for the absence of the corresponding protein in rodents (10).
Like mannan-binding lectin (MBL), ficolins are able to activate the lectin complement pathway in response to recognition of neutral carbohydrates and N-acetyl groups on pathogens and damaged cells. This results from the ability of MBL and ficolins to associate with and trigger activation of MBL-associated serine protease (MASP)-2. Activated MASP-2 cleaves the complement proteins C2 and C4, thereby triggering the complement cascade (11, 12, 13). Three other MBL/ficolins-associated proteins have been described, the MASP-1 and MASP-3 proteases (14, 15) and a truncated form of MASP-2 called MAp19 (19-kDa MBL-associated protein) or sMAp (16, 17). MASP-3 has no known physiological substrates whereas MASP-1 cleaves with a low efficiency a few protein substrates, among which are fibrinogen and coagulation factor XIII (18). It has been proposed recently that MASP-1 might contribute to the activation of the lectin pathway, but this issue is still controversial (19, 20). Complement activation results in opsonization of microbes and apoptotic cells with C3-derived fragments, thus promoting their clearance through interaction with C3 receptors on phagocytes (21, 22). In addition, ficolins may themselves function as opsonins, as suggested by the ability of L-ficolin to enhance phagocytosis of Salmonella typhimurium by neutrophils (5) and the ability of L- and H-ficolins to increase adhesion/uptake of late apoptotic cells by macrophages (23, 24). These opsonizing effects of ficolins are exerted through receptors present on phagocytic cells that are likely common to other defense collagens such as C1q and the collectins. A likely candidate is the receptor for the collagenous domain of C1q (cC1qR) or calreticulin (CRT) (25), which is thought to function in complex with the endocytic receptor CD91 (26, 27). This hypothesis is supported by studies showing that L- and H-ficolins bind to cC1qR/CRT (22) and that H-ficolin binding to CRT can be inhibited by MBL, suggesting that both proteins share a common binding site on CRT (24).
The MASPs are modular proteases homologous to the C1r and C1s proteases of the classical complement pathway, with an N-terminal CUB (C1r/C1s, Uegf, and bone morphogenetic protein-1) module, a Ca2+-binding epidermal growth factor (EGF) module, a second CUB module, two complement control protein modules, and a C-terminal serine protease domain. MASP-1 and MASP-3 are generated through alternative splicing of the MASP1/3 gene and only differ by their serine protease domains (15). MAp19 is an alternative splicing product of the MASP-2 gene comprising the N-terminal CUB1 and EGF modules of MASP-2 followed by four specific C-terminal residues (16, 17). Using recombinant rat and human proteins, it has been shown that the three MASPs, their N-terminal CUB1-EGF-CUB2 and CUB1-EGF segments, and MAp19 each form homodimers that in turn interact individually with MBL and L-ficolin in the presence of Ca2+ ions (4, 28, 29, 30, 31). Interaction involves a major contribution of the CUB1-EGF module pair of each protease but is reinforced by the following CUB2 module. MBL and ficolins have been shown to compete with each other for binding to the MASPs (30). Resolution of the crystal structure of the CUB1-EGF-CUB2 moiety of MASP-1/3 and site-directed mutagenesis studies have recently allowed us to define a common MBL- and ficolin-binding site in the MASPs, located at the distal end of each CUB module and stabilized by a Ca2+ ion (32). Based on these findings, a model has been proposed in which the MASP dimers interact with homologous sites on four collagen triple helices of tetrameric MBL. Besides, in keeping with previous studies on rat MBL (33), site-directed mutagenesis of human MBL has led to the identification of its MASP-binding site, providing evidence for the essential role of residue Lys55 in the collagenous region (34). In addition, it has been shown recently that mutations of the Lys55 of human MBL not only inhibit binding to the MASPs but also impair interaction with CRT, thus strongly supporting the hypothesis that CRT binds MBL through its MASP-binding site (35).
Sequence alignments reveal that Lys55 of human MBL is conserved in most ficolins of known sequence (supplemental Fig. S1),4 and recent studies have shown that the homologous residue of rat ficolin A (Lys56) is essential for MASP-2 binding (36). To test whether the homologous residues Lys57 and Lys47of human L- and H-ficolins are involved in the interaction with the MASPs and/or CRT, we have produced point mutants of both proteins in which these residues were mutated to Ala, Glu, and Arg. Functional analysis of the resulting ficolin variants reveals that residues Lys57 of L-ficolin and Lys47 of H-ficolin are key components of the interaction with both the MASPs and CRT, thus providing a strong indication that human MBL and ficolins share homologous binding sites for their partners.
Materials and Methods
Human MASP-1, the MASP-1/3 CUB1-EGF and CUB1-EGF-CUB2 segments, MASP-2, MAp19, and MASP-3 were expressed using a baculovirus/insect cell system and purified as described previously (29, 30, 31, 34). Human placenta CRT was purified as described by Pagh et al. (35).
The concentrations of purified recombinant proteins were determined using the following absorption coefficients (A1%, 1 cm at 280 nm) and Mr, respectively: MASP-3, 12.9 and 87,600 (31); MAp19, 11.6 and 19,086 (29); MASP-1/3 CUB1-EGF segment, 10.0 and 21,000 (29); MASP-1/3 CUB1-EGF-CUB2 segment, 10.0 and 34,300 (30). Due to the low amount of material recovered, estimation of the concentration of recombinant MASP-1 and MASP-2 was based on Coomassie blue staining after SDS-PAGE analysis using appropriate internal standards and Mr values of 82,000 and 75,100, respectively (29).
Construction of expression plasmids containing wild-type and mutant L- and H-ficolins
DNA segments encoding the signal peptide and sequence of mature human L- and H-ficolins were amplified by PCR using PCRBac and pGEM-T easy plasmids containing the full-length cDNAs as templates, respectively (5
The expression plasmids coding for the L-ficolin Lys57Ala, Lys57Glu, and Lys57Arg mutants and the H-ficolin Lys47Ala, Lys47Glu, and Lys47Arg mutants were generated using the QuikChange XL site-directed mutagenesis kit (Stratagene). The recombinant pcDNA3.1 plasmids encoding wild-type ficolins were used as templates. Mutagenic oligonucleotides were purchased from MWG Biotec. The sequences of all mutants were confirmed by dsDNA sequencing.
Production and purification of recombinant L- and H-ficolin variants
l-ascorbic acid (Sigma-Aldrich) as described by Teillet et al. (33
The cell culture supernatants containing the L-ficolin variants were dialyzed against 145 mM NaCl, 5 mM CaCl2, and 20 mM Tris-HCl (pH 7.4), concentrated 20-fold by ultrafiltration, and applied to a column of N-acetylcysteine-Sepharose (prepared as described by Krarup et al. in Ref. 37) equilibrated in the dialysis buffer. Elution of the bound L-ficolin was performed by applying the same buffer containing 0.3 M N-acetyl-d-glucosamine (GlcNAc). The purified protein was dialyzed against 145 mM NaCl, 2 mM CaCl2, and 50 mM triethanolamine-HCl (pH 7.4) and concentrated by ultrafiltration to 0.1–0.4 mg/ml. The culture supernatants containing the H-ficolin variants were dialyzed against 5 mM CaCl2 and 50 mM triethanolamine-HCl (pH 8.0) and loaded onto a Q-Sepharose Fast Flow column (GE Healthcare) equilibrated in the same buffer. Elution was conducted using a linear NaCl gradient to 250 mM. Fractions containing the recombinant proteins were identified by Western blot analysis, dialyzed against 145 mM NaCl, 5 mM CaCl2, and 20 mM Tris-HCl (pH 7.4) and further purified by gel filtration on a Superose 6 column (GE Healthcare) equilibrated in the same buffer. The H-ficolin variants eluted in the earlier fractions were concentrated by ultrafiltration to 0.5–1 mg/ml. The concentrations of the purified L- and H-ficolin variants were estimated using absorption coefficients A1%, 1 cm at 280 nm of 17.6 and 19.4 and protomer Mr values of 33,860 and 33,000, as determined by mass spectrometry and estimated from SDS-PAGE analysis, respectively.
SDS-PAGE analysis of the oligomerization state of the ficolin variants
Ficolin samples were analyzed by SDS-PAGE under nonreducing conditions using 7.5% (w/v) polyacrylamide gels. Proteins were transferred to a nitrocellulose membrane and detected by Western blotting using rabbit polyclonal Abs directed against the recombinant fibrinogen domains of L- and H-ficolins (38) and an HRP-conjugated polyclonal goat anti-rabbit Ab (Sigma-Aldrich) as primary and secondary Abs, respectively.
Chemical characterization of the recombinant proteins
39). Removal of the N-terminally blocked group of H-ficolin was achieved by incubation of the protein for 6 h at 50°C with Pfu pyroglutamate aminopeptidase (Takara) (2 mU/nmol protein) in the presence of 10 mM DTT. Mass spectrometry analyses were performed using the MALDI technique on a Voyager Elite XL instrument (Perseptive Biosystems), under conditions described previously (39).
Samples were applied to the clean side of carbon on mica (carbon/mica interface) and negatively stained with 2% (w/v) sodium silicotungstate (pH 7.4). A grid was placed on top of the carbon film, which was subsequently air dried. Micrographs were taken under low-dose conditions with a JEOL 1200 EX II microscope at 100 kV and a calibrated magnification of ×39,750 (using a tobacco mosaic virus particle). Selected negatives were then digitized on a Zeiss scanner (Photoscan TD) with a pixel size of 14 μm (3.5 Å at the sample scale).
Surface plasmon resonance (SPR) spectroscopy and data evaluation
Analyses were performed using a BIAcore 3000 instrument (GE Healthcare). Acetylated BSA (Sigma-Aldrich) and purified ficolins were diluted to 30 μg/ml in 10 mM sodium formate (pH 3.0) and 10 mM sodium acetate (pH 5.0), respectively, and immobilized on the surface of a CM5 sensor chip (GE Healthcare) using the amine coupling chemistry as described by Teillet et al. (39). Binding of the ficolin variants to immobilized, acetylated BSA (1,000 resonance units) was measured at a flow rate of 20 μl/min in 150 mM NaCl, 1 mM CaCl2, and 20 mM HEPES (pH 7.4) containing 0.005% (v/v) surfactant P20 (GE Healthcare), and regeneration of the surface was achieved by injection of 10 μl of 1 M sodium acetate (pH 7.20). Binding of MASP-2, MAp19, MASP-1, MASP-3, and their CUB1-EGF and CUB1-EGF-CUB2 segments to the immobilized L- and H-ficolin variants (16,000–20,000 resonance units) was measured at the same flow rate in 145 mM NaCl, 1 mM CaCl2, and 50 mM triethanolamine-HCl (pH 7.4) containing 0.005% (v/v) surfactant P20. Regeneration was achieved by injection of 10 μl of 40 mM EDTA and 2 M NaCl. Equivalent volumes of each protein sample were injected over a surface with immobilized BSA to serve as blank sensorgrams for subtraction of the bulk refractive index background.
Data were analyzed by global fitting to a 1:1 Langmuir binding model of both the association and dissociation phases for at least five concentrations simultaneously using the BIAevaluation 3.2 software (GE Healthcare). The apparent equilibrium dissociation constants (KD) were calculated from the ratio of the dissociation and association rate constants (koff/kon). The χ2 value, which is a standard statistical measure of the closeness of the fit (BIAevaluation 3 Software Handbook), was <3 in all cases. Analysis of the binding data revealed no significant mass transfer limitation as assessed by display of the ln(abs(dY/dX)) function, which yielded a straight line with a negative slope throughout the association phases and by the fact that similar rate constant values were obtained over a flow rate range of 10–40 μl/min.
Assay of the ability of the ficolin variants to trigger the lectin complement pathway
Ficolin-deficient serum was obtained by incubating 3 ml of normal human serum from a healthy donor with 1 ml of acetylated BSA-Sepharose (prepared by coupling 7.5 mg of acetylated BSA to 1 ml of cyanogen bromide-activated Sepharose (GE Healthcare) for 3 h at 4°C. The ficolin-deficient serum contained all factors necessary for complement activation and was <50 μg/L ficolin protein.
Microtiter 96-well plates (Maxisorp Nunc) were coated with 50 μg/ml acetylated BSA in 10 mM NaHCO3 (pH 9.6) for 2 h at room temperature. Wells were washed with PBS (136 mM NaCl, 2.7 mM KCl, 1.45 mM KH2PO4, 8 mM Na2HPO4 (pH 7.2)) and incubated for 1 h at room temperature with PBS containing 0.1% (w/v) Tween 20. Recombinant H- and L-ficolin variants were added at different concentrations to the wells and incubated for 1 h on ice with ficolin-deficient serum diluted 1:25 in 5 mM Na veronal, 145 mM NaCl, 5 mM CaCl2, and 1.5 mM MgCl2 (pH 7.5). The wells were washed with 5 mM sodium veronal, 145 mM NaCl, and 5 mM EDTA (pH 7.5) and C4b deposition was measured as described by Dumestre-Perard et al. (40), except that plates were developed in the dark with tetramethylbenzidine (Sigma-Aldrich), the reaction was stopped with 1 N H2SO4, and absorbance was read at 450 nm.
Assay of the ability of the ficolin variants to associate with CRT
In the case of H-ficolin, microtiter plates were coated with 10 μg/ml purified human CRT or BSA (Sigma-Aldrich) in PBS overnight at room temperature. The plates were washed with PBS containing 0.05% (w/v) Tween 20 (PBS-T) and saturated for 1 h at 37°C with 1% BSA (w/v) in PBS. Plates were washed with PBS-T and the H- ficolin variants were added at 10 μg/ml in PBS-T and incubated for 3 h at 37°C. After four washes, H-ficolin was revealed by incubation for 1 h at 37°C with the monoclonal anti-H-ficolin Ab 4H5 (HyCult Biotechnology, Tebu-Bio) at 5 μg/ml. After four washes with PBS-T, HRP-conjugated anti-mouse polyvalent Ig (G, M, or A) Ab (Sigma-Aldrich) diluted 1/10,000 was added and incubated for 1 h at 37°C. After 4 washes with PBS-T, plates were developed in the dark with tetramethylbenzidine and the reaction was stopped with 1 N H2SO4. Absorbance was read at 450 nm.
Binding of L-ficolin to immobilized CRT was measured as described above with the following modifications. Plates were coated with 10 μg/ml CRT or StartingBlock (Pierce) diluted 1/10 in PBS overnight at room temperature. Saturation was performed using a 1/10 dilution of StartingBlock in PBS for 1 h at room temperature as described by Faro et al. (41). Revelation was performed using a biotinylated rabbit polyclonal Ab against the L-ficolin FBG domain (diluted 1/2,500) followed by HRP-conjugated streptavidin (Amersham Biosciences) diluted 1/700.
Production and biochemical characterization of the recombinant L- and H-ficolins variants
Stable geneticin-resistant CHO-K1 cell lines were generated for expression of the L- and H-ficolin variants as specified under Materials and Methods and the cells were transferred to serum-free medium for production of the recombinant proteins.
The H-ficolin variants were isolated from the culture supernatants by ion-exchange chromatography and further purified by gel filtration. All variants were recovered in the early eluting fractions corresponding to high m.w. species. SDS-PAGE analysis of wild-type H-ficolin under nonreducing conditions yielded a ladder-like pattern with bands of Mr 66,000, 130,000, and 200,000, along with species of higher Mr, as shown by Coomassie blue staining and Western blot analysis (Fig. 1⇓, lanes 1 in A and B), consistent with previous analyses of recombinant H-ficolin preparations (6, 42). Similar electrophoretic patterns were obtained for mutants Lys47Ala, Lys47Glu, and Lys47Arg (Fig. 1⇓B, lanes 2–4), indicating that the mutations had no significant impact on the oligomerization of the protein. Under reducing conditions the wild-type protein (Fig. 1⇓A, lane 2) and the mutants (data not shown) each yielded a band of Mr 33,000, characteristic of the single polypeptide chain of H-ficolin. Edman degradation of each variant yielded no N-terminal sequence, likely due to the presence of a cyclic glutamine residue at the N terminus (6). Pretreatment of the protein with pyroglutamate aminopeptidase (43) released a single N-terminal sequence Glu-His-Pro-Ser-X-Pro-Gly-Pro-, confirming that mature H-ficolin starts at Gln24 of the unprocessed polypeptide chain.
Wild-type L-ficolin was initially produced in a serum-free medium and purified by affinity chromatography on N-acetylcysteine-Sepharose as described under Materials and Methods. SDS-PAGE analysis of the purified protein indicated that it migrated as a single band of Mr 34,000, characteristic of the polypeptide chain of L-ficolin, both under reducing and nonreducing conditions (data not shown). The lack of the high m.w. species usually observed for both serum and recombinant human L-ficolins (4, 5, 37, 41) suggested that this recombinant material was not properly folded and/or oligomerized. L-ficolin was subsequently produced in the presence of 10% FCS and satisfactorily purified using the same single-step affinity chromatography. By fractionating FCS on N-acetylcysteine-Sepharose, we checked that it contained no detectable ficolins in accordance with previous observations by Ohashi and Erickson (3). This second L-ficolin preparation exhibited a typical electrophoretic pattern under nonreducing conditions with two major bands on Coomassie blue-stained gels, one of Mr 34,000 and the other close to the top of the gel (Fig. 2⇓A, lane 1). Western blot analysis revealed bands of Mr 34,000, 68,000 and 205,000, along with a ladder of higher molecular mass species, corresponding to multimers of the 34,000 chain (Fig. 2⇓B, lane 1). These data indicate that production of human L-ficolin in the presence of FCS allows recovery of a recombinant protein with the expected oligomerization pattern. Mutants Lys57Ala, Lys57Glu, and Lys57Arg were subsequently produced and purified under the same conditions and shown by Western blot analysis to yield similar ladder-like patterns (Fig. 2⇓B, lanes 2–4). SDS-PAGE analysis of each L-ficolin variant under reducing conditions showed a single band with an apparent Mr of 34,000 as illustrated in Fig. 2⇓A, lane 2, for the wild-type protein. Edman degradation of all L-ficolin variants yielded a single N-terminal sequence, Leu-Gln-Ala-Ala-Asp-Thr-X-Pro-Glu-Val-, identical with that of serum L-ficolin.
Electron microscopy of L- and H-ficolins
Representative electron micrographs of negatively stained wild-type H- and L-ficolins are shown in Fig. 3⇓. Two types of pictures were observed corresponding to side views (Fig. 3⇓, a–c) and top views (d and e). The former showed similar parachute-like structures for both ficolins with, in most cases, four heads connected by a rod-like structure to a small globe at the opposite end, comparable to those described previously for mouse plasma ficolin (44), porcine plasma, and recombinant α-ficolins (3, 45). In contrast, the top views shown in d and e of Fig. 3⇓ suggest that both ficolins comprise six heads surrounding a central globe. The difference between the two views might arise from the technique used, because the staining agent outlines only those parts of the objects that are in direct contact with the carbon film (46), which would be the case for all heads in a top view but for only part of them in side views. Nevertheless, these pictures do not show significant differences between recombinant L- and H-ficolins in terms of their overall shapes.
Ligand binding properties of the ficolin variants
The ability of the L- and H-ficolin variants to bind acetylated ligands was investigated by SPR spectroscopy, using acetylated BSA as the immobilized ligand and native BSA as the blank surface for background subtraction. As shown in Fig. 4⇓A, each L-ficolin variant bound strongly to and dissociated slowly from acetylated BSA in the presence of 1 mM CaCl2. Similar results were obtained with wild-type H-ficolin and the three Lys47 mutants, although dissociation appeared slightly faster than in the case of L-ficolin (Fig. 4⇓B). Complete elution of the bound ficolin variants could be achieved in all instances by a pulse injection of 1 M sodium acetate (pH 7.2). Binding of the proteins to acetylated BSA was strongly inhibited (L-ficolin) or abolished (H-ficolin) when EDTA was substituted for Ca2+ in the running buffer (data not shown). The kinetic parameters of the interactions were determined by recording sensorgrams at varying ficolin concentrations and evaluating the data by global fitting as described under Materials and Methods. As listed in Table I⇓, all L-ficolin variants exhibited comparable kon and koff values, yielding resulting apparent equilibrium dissociation constants of the same order (1.6–2.4 nM). Although slightly higher than in the case of L-ficolin, the apparent KD values for binding of wild-type H-ficolin and its three Lys47 mutants were also in the nM range (6.5–9.4 nM). Consistent with our preliminary observations (Fig. 4⇓), this difference mainly arises from a 6- to 7-fold higher koff value for the H-ficolin variants that is partly counterbalanced by slightly higher kon values (Table I⇓).
Interaction of the L- and H-ficolin variants with MASP-2
SPR spectroscopy was next used to analyze the ability of each ficolin variant to associate with MASP-2 and its truncated form MAp19. As illustrated in Fig. 5⇓, recombinant MASP-2 and MAp19 bound to immobilized wild-type L-ficolin (A and C) and H-ficolin (B and D) in the presence of Ca2+ ions. The kinetic parameters for these interactions were determined, yielding the same resulting apparent KD values for binding of both ficolins to MASP-2 (2.3 nM) and slightly higher values for binding to MAp19 (7.2–13.2 nM) (Table II⇓). However, comparison of the rate constants reveals significant differences, with 4.6-fold lower kon and koff values for H-ficolin binding to MASP-2, and, respectively, 2.4- and 4.4-fold higher kon and koff values for H-ficolin binding to MAp19. The values obtained for L-ficolin are of the same order as those measured previously using serum L-ficolin (30, 47).
Replacement of residues Lys57 of L-ficolin and Lys47 of H-ficolin by Arg, Ala, or Glu abolished their interaction with MASP-2 and MAp19 (Fig. 5⇑, A–D). No binding constants could be determined for these mutants, because no significant MASP-2 or MAp19 binding could be detected at concentrations up to 240 nM.
Interaction of the L- and H-ficolin variants with MASP-1 and MASP-3
Interaction of the immobilized L- and H-ficolin variants with MASP-1, MASP-3, and with their common N-terminal CUB1-EGF and CUB1-EGF-CUB2 segments was analyzed in the same way. All four proteins bound to the wild-type ficolins as illustrated in Fig. 6⇓ by the representative binding curves obtained for the interaction of L-ficolin with the MASP-1/3 CUB1-EGF-CUB2 and CUB1-EGF segments (A and C) and for the interaction of H-ficolin with MASP-3 and the MASP-1/3 CUB1-EGF segment (B and D). Kinetic analyses yielded comparable rate and equilibrium dissociation constants for binding of MASP-1, MASP-3, and their CUB1-EGF-CUB2 interaction domain to each wild-type ficolin, with KD values ranging from 5.6 to 10.7 nM for L-ficolin and from 2.2 to 6.9 nM for H-ficolin (Table III⇓). The values obtained for L-ficolin binding were of the same order as those obtained previously using recombinant and serum L-ficolin (4, 30, 31). The KD value for binding of the shorter MASP-1/3 CUB1-EGF segment to L-ficolin was higher (Table III⇓) mainly because of an increased dissociation rate constant, in keeping with previous results on serum L-ficolin (30). This segment dissociated even faster from H-ficolin (Fig. 6⇓D), but the rate constants could not be determined by fitting the binding curves to the Langmuir 1:1 model used for L-ficolin.
Mutation of Lys57 of L-ficolin and Lys47 of H-ficolin to Ala or Glu abolished binding to MASP-1, MASP-3, and their CUB1-EGF-CUB2 and CUB1-EGF segments (Fig. 6⇑, A–D). In contrast, replacement of these residues by arginine did not prevent binding of MASP-1/3 and its CUB1-EGF-CUB2 interaction domain to either ficolin (Fig. 6⇑, A and B). Kinetic analyses of the binding of MASP-1/3 to L-ficolin Lys57Arg revealed a 1.5- to 2.4-fold decrease of the kon values and a similar increase of the koff values, yielding a moderate 2.4- to 6-fold increase of the resulting KD values (Table III⇑). The effect of the Lys47Arg mutation of H-ficolin was more pronounced, with 8- to 17-fold higher dissociation constants arising mainly from a 6.3- to 9.8-fold increase of the koff values (Table III⇑). No constants could be determined for the Lys57Ala and Lys57Glu variants of L-ficolin and their H-ficolin counterparts due to lack of binding. In contrast to full-length MASP-1, MASP-3, and their CUB1-EGF-CUB2 segments, the short CUB1-EGF segment did not interact with L-ficolin Lys57Arg and with H-ficolin Lys47Arg (Fig. 6⇑, C and D).
Ability of the L- and H-ficolin variants to trigger activation of the lectin pathway
Ficolin-depleted human serum was generated by incubation with acetylated BSA coupled to Sepharose beads and tested for its capacity to induce deposition of C4 fragments when added to microplate wells coated with acetylated BSA, as described under Materials and Methods. No significant C4b binding signal was obtained using the depleted serum alone. In contrast, the addition of increasing amounts of wild-type L- or H-ficolin yielded a signal that reached a value similar to that obtained with normal human serum (data not shown) when using concentrations of 2 μg/ml L-ficolin and 5 μg/ml H-ficolin (Fig. 7⇓). When the L-ficolin mutants were tested in the same assay, no C4 deposition was observed for the Lys57Ala and Lys57Glu mutants, and a marked decrease was observed for the Lys57Arg mutant (Fig. 7⇓A). All three Lys47 mutants of H-ficolin failed to form a functional ficolin-MASP-2 complex (Fig. 7⇓B).
Ability of the L- and H-ficolin variants to interact with CRT
Recent data have revealed that human CRT interacts with MBL through the binding site for MBL-associated proteases and that Lys55 of MBL plays a crucial role in the interaction (35). To check the possible involvement of the homologous lysine residues of L- and H-ficolins, the ficolin variants were tested for their CRT binding ability by using the solid-phase binding assay described under Materials and Methods. Saturation was performed using BSA in the case of H-ficolin and the StartingBlock buffer in the case of L-ficolin due to nonspecific binding of the latter to BSA, as previously reported by Faro et al. (41). As shown in Fig. 8⇓, both wild-type ficolins bound to immobilized CRT in keeping with previous studies demonstrating interaction of L- and H-ficolins with bovine and human CRT (22, 24). Replacement of Lys47 of H-ficolin and Lys57 of L-ficolin by alanine or glutamic acid strongly inhibited binding of both ficolins to CRT, whereas mutation of these lysine residues to arginine had no significant effect on the interaction with CRT (Fig. 8⇓).
A stable mammalian cell expression system was used to produce recombinant human L- and H-ficolins as described previously for production of human MBL (34). Whereas multimeric H-ficolin could be successfully produced in serum-free medium, L-ficolin clearly showed an oligomerization defect when expressed in CHO-K1 cells in the absence of FCS or in HEK293-derived 293F cells adapted in a serum-free medium (data not shown). This problem was solved by producing recombinant L-ficolin in the presence of serum, allowing us to produce a protein with an electrophoretic mobility pattern similar to that obtained previously for the protein isolated from human serum (37) or the recombinant His-tagged L-ficolin produced in the presence of FCS (4). Recombinant L-ficolin was fully functional because it exhibited ligand binding and complement-activating properties similar to those of the protein purified from human serum. Why recombinant human L- and H-ficolins differ in their oligomerization capacity in the absence of FCS is unclear, because both proteins are expected to oligomerize through a similar process likely involving two conserved cysteine residues located at the N-terminal end of the polypeptide chains (3). Based on mass estimates by gel filtration and electron microscopy, it is assumed that human L-ficolin and its porcine counterpart α-ficolin are dodecamers assembled from four structural units (4, 44), whereas H-ficolin is an octadecamer (6). The electron microscopy images of the recombinant proteins produced in this study do not provide evidence for major differences between L- and H-ficolins, which both appear as typical parachute-like structures. However the resolution of the negative staining technique does not allow a precise evaluation of the number of subunits per molecule, which varies from 4 to 6 (i.e., dodecamer to octadecamer) when looking at side and top views, respectively.
Recent studies have demonstrated the essential role of Lys55 in the collagen-like region of human MBL in the interaction with its partner proteases (34). To test whether the homologous Lys residues at positions 57 of L-ficolin and 47 of H-ficolin are similarly involved in this process, mutated proteins were produced in which these residues were replaced by neutral (Ala), acidic (Glu), or basic (Arg) residues. All purified ficolin mutants exhibited electrophoretic patterns similar to those of their wild-type counterparts, indicative of similar oligomerization states. Likewise, analysis by SPR spectroscopy indicated that all variants bound to immobilized, acetylated BSA in similar ways, confirming that the mutations had no detectable effect on the ligand-binding properties of L- and H-ficolins. Interestingly, this provides evidence that H-ficolin is able to bind certain acetylated derivatives, a property that has been assumed to be restricted to L- and M-ficolins (38, 48). Nevertheless, SPR analysis did not show detectable binding of H-ficolin to immobilized GlcNAc- or N-acetyl-D-galactosamine-BSA (data not shown), in keeping with previous observations that H-ficolin does not bind to GlcNAc-Sepharose (14). In the same way, binding of H-ficolin to polysaccharides from Aerococcus viridans could not be prevented by GlcNAc or N-acetyl-d-galactosamine (49), and binding to this bacterium could not be inhibited by acetylated sugars or noncarbohydrate compounds (37). The fact that H-ficolin binding to acetylated BSA is totally prevented in the presence of EDTA suggests that recognition of acetyl groups involves the external S1 binding site, previously identified by x-ray crystallography, that lies close to the Ca2+ binding site (38). However, kinetic SPR analyses show that the H-ficolin-ligand complex formed is less stable than that achieved by L-ficolin. This is in agreement with the observation that the isolated trimeric FBG domain of H-ficolin does not bind to immobilized acetylated BSA, in contrast to that of L-ficolin (N. Thielens, unpublished results).
We took advantage of the capacity of both L- and H-ficolins to specifically bind to acetylated BSA to set up a functional assay to determine their ability to activate the lectin complement pathway by using acetylated BSA as the activating ligand and ficolin-depleted serum as a source of MASP-2 and C4. L-ficolin was slightly more efficient than H-ficolin, as half-maximal C4 cleavage was obtained at a lower concentration (0.75 vs 2 μg/ml). These results thus disagree with the recent report by Hummelshoj et al. (42) indicating a higher complement-activating potential of H-ficolin. However, it should be emphasized that this latter study was based on a different strategy with direct immobilization of the ficolins on the plate in the absence of ligands, which does not correspond to physiological conditions. Our SPR measurements do not provide evidence for a difference in the affinity of MASP-2 for both ficolins (same KD), but both the association and dissociation rates of the complex were slower for H-ficolin (see Table II⇑).
A major finding of this study is that Lys57 of L-ficolin and Lys47 of H-ficolin are key components of the interaction of both ficolins with their partner proteases. Indeed, mutations of these residues to Ala and Glu abolish interaction of both ficolins with all three MASPs and MAp19. Accordingly, the corresponding mutants of both ficolins lack the ability to trigger complement activation because of their inability to form enzymatically active ficolin/MASP-2 complexes. These results are in agreement with the report of Girija et al. (36) that the homologous Lys56 of rat ficolin-A is essential for MASP-2 binding. Replacement of Lys by Arg abolishes the complement-activating capacity of H-ficolin and strongly inhibits that of L-ficolin. This latter observation does not fully agree with the SPR data, because both the L-ficolin Lys57Arg and the H-ficolin Lys47Arg show no interaction with MASP-2. We have no clear explanation for this discrepancy. In contrast, the Lys to Arg mutations have only a weak inhibitory effect on the interaction properties of both ficolins with MASP-1 and MASP-3 that is slightly more pronounced in the case of H-ficolin due to a decreased stability of the complex. These subtle differences between the two ficolins probably arise from differences in the amino acid sequence surrounding the target Lys residues, considering that additional residues likely contribute to the interaction. In this respect, the sequence alignment in supplemental Fig. S1 shows conservation of hydroxyproline and proline residues at positions −2 and +3 relative to the critical lysine residue and the presence of methionine or an aliphatic residue at position +1. Interestingly, occupancy of the latter position by an acidic residue (Glu) in mouse ficolin-B correlates with the fact that this particular ficolin is not able to bind MASP-2 and activate complement (36). An Arg instead of a Lys residue allows interaction of both ficolins with MASP-1 and MASP-3, but a Lys residue is essential for binding to MASP-2 as previously observed for Lys55 of human MBL (34). This observation also applies to the CUB1-EGF-CUB2 domain of MASP-1/3, providing further evidence that it contains all of the determinants necessary for the interaction. In contrast, the shorter CUB1-EGF fragment does not interact with the L-ficolin Lys57Arg and the H-ficolin Lys47Arg. This is in agreement with the recently reported three-dimensional structure of the MASP-1/3 CUB1-EGF-CUB2 dimer that reveals the presence of MBL/ficolin binding sites in each CUB module, thereby allowing simultaneous interaction of the fragment with four collagen-like helices (32). Although the CUB1 and CUB2 binding sites are homologous, they are not strictly identical, and possibly the Lys to Arg replacement in the collagen-like sequence is better tolerated in the CUB2 binding site. Moreover, the interaction of the longer CUB1-EGF-CUB2 fragment is likely reinforced because of the involvement of its four CUB binding sites, and this could compensate for a weaker affinity of the individual sites.
Another major outcome of this study is that the Lys57 of L-ficolin and the Lys47 of H-ficolin also play a role in the interaction of these proteins with the cC1q receptor CRT. Preliminary SPR experiments did not provide evidence for a CRT/ficolin interaction by using either of these proteins as the immobilized ligand. Nevertheless, it has been suggested previously that the binding of CRT to C1q and MBL may require a conformational change in the recognition proteins and/or CRT (35, 50). Covalent immobilization of one or the other interactant on a BIAcore sensor chip might therefore restrict the conformational changes associated with the binding process. Using a solid-phase binding assay, we were able to demonstrate specific interaction of both ficolins with CRT and significant inhibition after mutation of the Lys57 of L-ficolin and the Lys47 of H-ficolin to Ala or Glu, but not to Arg. These findings thus provide strong indication that both ficolins share homologous binding sites for the MASPS and for CRT, as previously demonstrated in the case of MBL (35). This binding likely involves a major ionic interaction between the conserved Lys residues of the collagen triple helix of MBL and the ficolins and acidic residues contributed either by the MASP CUB modules (32) or by CRT. However, a significant difference between the MASPs and CRT lies in the Ca2+ dependency of their interaction with the ficolins. Indeed, binding of the ficolins to CRT could be observed without the addition of Ca2+ ions (this study and Refs. 22 and 24) as shown previously for MBL (35). In contrast, the binding sites in the MASPs likely involve acidic residues also acting as Ca2+ ligands (32). In addition, CRT is considered to be a monomer under physiological conditions (51), and its interaction with MBL and the ficolins likely involves interaction with a single collagen-like helix. In contrast, interaction of the MAp19 and MASP dimers involves simultaneous binding to two and four MBL or ficolin triple helices, respectively.
Whereas the ficolin binding site of CRT is unknown, the MBL/C1q binding site has been identified in the so-called S-domain encompassing most of the central proline-rich P-domain and part of the N-terminal N-domain (25, 52). Based on sequence alignments, it was also proposed that the S-domain is related to the CUB modules found in the C1r/C1s/MASP family (25). However, resolution of the x-ray structure of the homologous chaperone calnexin (53) did not support this hypothesis because it showed no homology to the various CUB module crystal structures solved during the past years (32, 47, 54, 55). Nevertheless the recombinant S-domain of CRT was shown to inhibit C1q-dependent complement activation, likely through blocking the C1s-C1r-C1r-C1s binding site on C1q (52, 56).
The fact that the MASPs and CRT likely share similar or overlapping binding sites for MBL and the ficolins has important potential physiological implications. Although the recognition proteins circulate in serum in association with the MASPs, it may be hypothesized that, following complement activation, MASP-2 is released in the same way as is the C1s-C1r-C1r-C1s tetramer from the C1 complex. This would make the MBL/ficolin binding site available for interaction with other partners such as CRT. Alternatively, locally synthesized MBL and ficolins could act as opsonins and CRT binding could then allow them to interact with cell receptors on phagocytes. In this respect, it has been shown that L- and H-ficolins bind to apoptotic cells, and it was proposed that their binding to CRT may mediate apoptotic cell clearance through the CRT/CD91 receptor complex on the surface of macrophages (22, 24).
In summary, our results reveal that residues Lys57 of L-ficolin and Lys47 of H-ficolin, which are homologous to Lys55 of human MBL, are key components of the interaction with the MASPs and with the cC1q/collectin receptor CRT, thus strongly supporting the hypothesis that MBL and ficolins share homologous binding sites for both types of proteins. These residues likely participate in a major ionic interaction with Ca2+-binding residues contributed by the MASP CUB modules and hitherto unknown residues of CRT. However, both types of binding sites are not identical as judged from their differential sensitivity to a Lys to Arg mutation and requirement for Ca2+ ions. The interaction between CRT and ficolins is similar to that observed in the case of MBL and C1q, suggesting a common CRT-dependent effector mechanism mediated through the collagen-like regions of these innate immune recognition proteins.
The assistance of Philippe Frachet for the development of stably transfected CHO cell lines is gratefully acknowledged. We thank Jean-Pierre Andrieu for performing N-terminal sequence analyses and Bernard Dublet for mass spectrometry measurements. We thank the CIBB joint electron microscopy platform for help with electron microscopy.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 This work was supported by the Commissariat à l’Energie Atomique, the Centre National de la Recherche Scientifique, the Université Joseph Fourier (Grenoble, France).
↵2 Address correspondence and reprint requests to Dr. Nicole Thielens, Laboratoire d’Enzymologie Moléculaire, Institut de Biologie Structurale Jean-Pierre Ebel, 41 Rue Jules Horowitz, 38027 Grenoble Cedex 1, France. E-mail address:
↵3 Abbreviations used in this paper: FBG, fibrinogen-like; cC1qR, receptor for collagenous domain of C1q; CRT, calreticulin; EGF, epidermal growth factor; GlcNAc, N-acetyl-d-glucosamine; MASP, MBL-associated serine protease; MBL, mannan-binding lectin; SPR, surface plasmon resonance.
↵4 The online version of this article contains supplemental material.
- Received August 29, 2008.
- Accepted November 3, 2008.
- Copyright © 2009 by The American Association of Immunologists, Inc.