Carbohydrate recognition and complement activation by rat ficolin-B

Expression and characterization of rat ficolin-B

To produce pure ficolin-B, free of contamination by ficolin-A or other serum lectins, recombinant protein was produced using a mammalian expression system. Initially, it was not clear whether ficolin-B would be secreted from producing cells or retained internally, so studies were carried out to follow the kinetics of biosynthesis, using ficolin-A for comparison. In these experiments, transfected CHO cells were incubated for a brief pulse of ³⁵S-labelled methionine, followed by extended incubation in unlabelled media. By comparing the amounts of labelled protein in the intracellular and extracellular fractions over time, the kinetics of biosynthesis and secretion could be followed. As shown in Fig. 1, labelled ficolin-B appeared in the culture medium of producing cells soon after the initial pulse and was secreted with a half-time of ∼110 minutes. There was no evidence of retention of ficolin-B within cells, nor of degradation. The secretion rate is typical of many secreted proteins, including rat MBL-A and MBL-C (Heise et al. 2000) and was only slightly slower than ficolin-A (half-time ∼90 minutes). Thus, any intracellular retention signals present in the ficolin-B polypeptide are not recognized by CHO cells.

Ficolin-B was harvested from the culture medium of producing cells and purified by affinity chromatography on GlcNAc-Sepharose columns. Yields from several different cell-lines were relatively low, with only 50–100 μg of purified protein per litre of culture medium, compared to 4–5 mg/L of ficolin-A 21. Purified ficolin-B migrated as two bands on SDS-polyacrylamide gels under reducing conditions, with apparent molecular masses of 39 and 40 kDa (Figs. 1 and 2), both of which are greater than the molecular mass of 34 kDa calculated from the amino acid sequence. There is one potential N-linked glycosylation site in the fibrinogen domain and five potential O-linked sites in the collagen-like domain, so the two species observed on gels probably reflect different glycoforms. N-terminal sequencing of each band yielded the sequence: ADTCPEVKVL, indicating that the signal peptide is cleaved between Ala22 and Ala23 of the synthesized polypeptide during biosynthesis.

Under non-reducing conditions, ficolin-B migrated as mixtures of covalently linked polypeptides on SDS-polyacrylamide gels ranging from one to more than 12 chains, so most polypeptides must be linked by interchain disulfide bonds (Fig. 2). It eluted as several overlapping peaks from a gel filtration column, signifying the presence of different oligomeric forms. Oligomers were identified by comparison with the elution positions of the different forms of ficolin-A, which have comparable sizes and have been characterized previously 21. In this way, two major oligomers were identified consisting of six and nine polypeptides (subsequently called dimers and trimers of subunits), together with smaller amounts of monomers and tetramers of subunits. Trace amounts of even larger oligomers were observed, probably comprising pentamers and/or hexamers of subunits. Analysis of fractions collected across the elution profile confirmed that the disulphide-bonding pattern in oligomers is heterogeneous (Fig. 2C). For example, fraction 3 comprises mainly trimers of subunits, but a ladder of covalently linked polypeptides (from 1 to 9) is observed on SDS gels under non-reducing conditions, indicating that the different forms must be assembled from combinations of covalently linked polypeptides, associated through non-covalent interactions. Overall, the oligomeric composition is broadly similar to rat ficolin-A, which also comprises mixtures of disulphide-linked oligomers 21.

Carbohydrate specificity of rat ficolins

Because little detailed information is known about the binding specificities of rat ficolins, we examined their lectin properties towards a broad screen of endogenous and exogenous carbohydrates using the Core H Glycan array technology facility at The Consortium for Functional Glycomics, Emory University (GA, USA). Ficolins were labelled with a fluorescent dye and used to probe a library of 320 ligands (Supporting Information). Ficolin-A is known to bind to N-acetylated ligands such as GlcNAc 21, but specificity towards natural ligands has not been examined. Surprisingly, it bound appreciably to only one of the ligands tested: a trisaccharide containing a terminal α1-6 linked GlcNAc residue (ligand 157) (Fig. 3). No binding was observed towards ligands lacking the terminal GlcNAc (ligand 153) or when the GlcNAc was linked via α1-3 (ligand 156), β1-2 (ligand 158), β1-3 (ligand 165) or β1-6 (ligand 176) glycosidic linkages (Table 1). Thus, the type and mode of attachment of the terminal sugar residue is critical. Although the binding signal from the next best ligand was>10-fold lower, certain sialyated sugars and a disialic acid also bound above background level, neither of which have been previously identified as ligands for a ficolin-A orthologue. Notably, all ligands contained multiple N-acetyl groups, suggesting that internal acetylated residues may be recognized by ficolin-A, in addition to the terminal moiety.

In contrast to ficolin-A, rat ficolin-B bound to a number of ligands in the glycan screen (Table 2), with a marked preference for sialyated structures with terminal N-acetylneuraminic but not N-glycolyneuraminic acid (a sialic acid found in many mammals but not humans; ligands 257–264, Supporting Information). Moreover mono- and biantennary sialated sugars with α2-3 linkages but not α2-6 linkages were preferentially recognized (for example, compare ligands 143, 199 and 54 (Table 2)). Ligands could be divided into four main groups: di- and trisialic acids; mono- or biantennary structures containing the trisaccharide sialyl-N-acetyllactosamine (Sialyl-Gal-GlcNAc), 9-acetylated sialated ligands and more weakly to certain fucosylated structures. It is notable that such sequences are commonly found on endogenous structures including cell surfaces and tumour cells, as well as on exogenous ligands including pathogenic microbes and viruses 22, 23.

Complement activation

Rat ficolin-A binds MASP and can activate complement 21, but the properties of ficolin-B are not known. To test binding to MASP-1 and MASP-2, ficolin-B was immobilized on a sensor chip and interactions with MASP were monitored using surface plasmon resonance. Both MASP associated with ficolin-B in the presence of Ca²⁺ ions to form relatively stable complexes. As with ficolin-A, the binding kinetics were complex. The data fitted best to a two-reaction parallel binding model for MASP-2A (Fig. 4), with apparent equilibrium dissociation constants K_D1 and K_D2 of 752 and 23 nM. Association rate constants (k_on) were 1×10⁴ and 2.6×10⁵ M⁻¹s⁻¹ and dissociation rate constants (k_off) were 5.8×10⁻³ and 6×10⁻³ s⁻¹, respectively. The kinetics of MASP-1 binding was similar to MASP-2 (data not shown). K_D1 and K_D2 were 392 and 8.4 nM, with k_on values of 2.4×10⁴ and 5.7×10⁵ M⁻¹s⁻¹ and k_off values of 9.1×10⁻³ and 4.3×10⁻³ s⁻¹, respectively. The biphasic kinetics are typical of rat MBL-MASP and ficolin-MASP interactions and probably reflect binding of a second MASP to ficolin oligomers at high MASP concentrations, although 1:1 complexes predominate under physiological concentrations 24. No binding of MASP-1 or MASP-2 was detected in the presence of EDTA, confirming that the interactions are Ca²⁺ dependant (data not shown). Overall, the binding data were broadly comparable to the interactions of ficolin-A with MASP-2 21, but markedly different from mouse ficolin-B, which does not bind MASPs despite a high degree of sequence identity 16.

MASP-2 in complex with MBL or ficolin is necessary and sufficient to activate the lectin pathway of complement activation, so we next examined if rat ficolin-B could activate complement. Ficolin-B triggered MASP-2K activation, using GlcNAc-Sepharose as a target, with a half-time of ∼110 min (Fig. 5; compared to 56 min by ficolin-A under comparable conditions (data not shown)). Very little activation was observed in the absence of GlcNAc-Sepharose (∼10% over 16 h), and this was no faster than auto-activation of zymogen MASP-2K alone. Thus, MASP-2 activation is both target and ficolin dependent (Fig. 5B). To confirm that MASP activation by ficolin-B leads to complement activation, we measured deposition of the membrane attack complex, on acetylated BSA as an activating target. Addition of recombinant ficolin-B to human serum depleted of endogenous MBL and ficolins activated complement in a concentration-dependent manner (Fig. 5C) and with similar activity to ficolin-A. Thus, rat ficolin-B, unlike its mouse counterpart, activates the lectin pathway of complement via a comparable mechanism to ficolin-A (and MBL) in which the ficolin increases the rate of MASP-2 autolysis, but only when complexes bind to a target surface.

MASPs bind to a characteristic motif on the collagenous domain of MBL and ficolins comprising the sequence Hyp-Gly-Lys-Xaa-Gly-Pro, in which Xaa typically represents a hydrophobic aliphatic residue or a methionine, and the lysine residue is essential for binding 21, 24, 25. To confirm that MASP bind to the equivalent site in rat ficolin-B, Lys54 and Ala55 within the sequence Hyp-Gly-Lys-Ala-Gly-Pro were changed to proline and hydroxyproline residues, respectively (K54P, A55O). The resulting mutant (Ficolin-B PO) was expressed and purified from CHO cells and resembled WT protein upon gel filtration chromatography and SDS-PAGE (data not shown), confirming that it assembles correctly during biosynthesis. However, it failed to bind MASP-1 or MASP-2 (Fig. 4) and did not activate MASP-2 (Fig. 5), demonstrating that Lys54 and/or Ala55 form part of the MASP-binding site. Although mouse ficolin-B possesses a similar binding motif, it is unusual compared to other ficolins because it contains an acidic residue adjacent to the key lysine: Hyp-Gly-Lys-Glu-Gly-Pro. In previous work, introduction of a glutamate into rat ficolin-A reduced the affinity of MASP-binding and MASP-2 activation by ∼7-fold, so it is somewhat disruptive 21. To examine the effect of such a mutation within the ficolin-B framework, we made a comparable change in the rat protein (ficolin-B E). The resulting mutant failed to bind to MASP-1 or MASP-2 at the highest concentrations tested (0.5 μM; Fig. 4), despite having equivalent structural and lectin properties to WT ficolin-B. It also failed to activate MASP-2, even after prolonged incubation (up to 16 h), indicating that the glutamate residue disrupts complement activation by ficolin-B by preventing association with MASP-2 (Fig. 5). Thus, mouse ficolin-B differs from its human and rat counterparts probably as a result of a single amino acid substitution in the collagenous domain.

Cloning, expression and purification of rat ficolin-B

The cDNA of rat ficolin-B was amplified from a rat cDNA library kindly provided by Prof. Wilhelm Schwaeble (University of Leicester, UK) using PCR, and was cloned into pGEM-T cloning vector (Promega). The forward and reverse oligonucleotides were GAACTCGAGGCCACCATGGTCCTGGGATCTGCT, where the ribosome-binding site is underlined and the start codon is in bold, and AGAGAATTCCTAGATGAGGCGCACCTTCAT, where the stop codon is in bold. The cDNA was cloned into a mammalian expression vector, pED4 32, the sequence confirmed and the resulting construct was introduced into the CHO cell line DXB11. Production was achieved following amplification using the dihydrofolate reductase inhibitor, methotrexate 21. Culture media were harvested and protein purified by affinity chromatography on GlcNAc-Sepharose columns (1-mL column for 250 mL of media) using the strategy described previously for ficolin-A 21. We also attempted to express mouse ficolin-B using the same strategy. However, only trace amounts of protein was produced from transfected cells, so further analysis was not possible.

Other protein components

Recombinant rat ficolin-A and modified forms of rat MASP-2 were produced in a CHO cell-expression system and purified as described previously by affinity chromatography on GlcNAc-Sepharose and nickel-Sepharose columns, respectively 21, 33. WT rat MASP-1 and MASP-2 are toxic to producing CHO cells because they autoactivate during biosynthesis. We therefore used two modified forms of MASP-2, both of which are secreted as zymogens and have been characterized with regard to their structures, activation and catalytic properties 33, 34. MASP-2A is a catalytically inactive form in which the active site serine at position 613 has been replaced by an alanine. MASP-2K is a catalytically active form, in which the arginine residue at the cleavage site for zymogen activation (Arg⁴²⁴) has been changed to a lysine residue to slow down the rate of spontaneous autoactivation during biosynthesis. A modified form of MASP-1, called MASP-1ent, was used for binding analysis 34. This is a catalytically active form in which residues Lys⁴²⁵–Arg⁴²⁹, immediately preceding the cleavage site for zymogen activation, were replaced with the recognition sequence of the serine protease enterokinase: Asp-Asp-Asp-Asp-Lys. Binding and activation assays were carried out using MASP-2A or MASP-1ent and MASP-2K, respectively. Mutant forms of ficolins were created by PCR and expressed in the same way as the WT protein. All mutations were verified by sequencing of the entire cDNA, within the expression vector, prior to transfection of the CHO cells.

N-terminal sequencing

Samples were separated on 10% Novex BIS-TRIS NuPAGE precast gels using 2-(N-morpholino)ethanesulfonic acid buffer in a Novex×Cell II Mini-Cell gel apparatus. The protein bands were electroblotted onto a Novex 0.2 μm polyvinylidene difluoride membrane (Invitrogen) using the Novex Blot module. The membrane was stained with Coomassie Brilliant Blue and target bands were excised, washed with 10% methanol and sequenced on a 494A Procise protein sequencer (Applied Biosystems, Warrington, UK) for ten cycles using the standard sequencing protocol.

Pulse-chase labelling of ficolins

Pulse-chase experiments were carried out essentially as described in 35. Briefly, CHO cells producing rat ficolin-A or -B were grown to confluence in 35×10-mm tissue culture dishes containing 2 mL of minimal essential medium-α, lacking nucleosides and supplemented with 10% dialyzed foetal calf serum and 0.5 μM methotrexate. Culture medium was changed daily for an additional 4 days. Before labelling, all culture media were pre-warmed to 37°C in 5% CO₂. Cells were incubated with 2 mL of methionine-free medium for 5 min to diminish any intracellular methionine. Labelling was initiated with 1 mL of fresh methionine-free medium containing ³⁵S-methionine (56 μCi/mL). After 15 min, the pulse was terminated by removing the ³⁵S-methionine-containing medium and the cells were incubated with medium containing a 20-fold excess of unlabelled methionine. After various time intervals, cells were released from the culture dish by scraping, pelleted by centrifugation at 2000 rpm in a bench-top centrifuge, and disrupted by resuspension in cell lysis buffer (50 mM Tris-HCl, pH 7.4, containing 500 mM NaCl and 1% Triton X-100). GlcNAc-Sepharose (10 μL) was added to each fraction and mixtures were incubated on ice for 10 min. Bound ficolin was pelleted by centrifugation and separated by SDS-PAGE. Ficolins were quantified by scanning autoradiographs using a ChemiGenius (Syngene).

Glycan array analysis

Alexa Fluor^® 488 tetrafluorophenylcarboxylate ester (Invitrogen, Paisley, UK) was dissolved in DMSO to a final concentration of 1 mg/mL, and 20 μL was added to purified ficolin (250 μL of a 400 μg/mL solution) and incubated for 2 h at room temperature. Excess label was removed by dialysis. The labelled ficolin was then screened using the Glycan array technology facility at The Consortium for Functional Glycomics, Emory University. This facility uses printed glycan microarray chips with 320 different carbohydrate targets of natural and synthetic glycans attached to the chip via amino linkers. Binding was detected by measuring the fluorescence associated with each glycan. In each case, screening of the whole chip was performed with a total volume of 70 μL of 200 μg/mL ficolin in 20 mM HEPES, pH 7.4, with 140 mM NaCl, 5 mM CaCl₂, and repeated six times. Error bars represent the average±SE.

Surface plasmon resonance

Measurements were performed using a BIAcore 2000 instrument (GE Healthcare). Protein ligands were diluted into 10 mM sodium acetate, pH 4.0, and immobilized on to the carboxymethylated dextran surface of a CM5 sensor chip (GE Healthcare), using amine coupling chemistry. Binding was measured in 10 mM HEPES (pH 7.4) containing 150 mM NaCl and 5 mM CaCl₂ at a flow rate of 5 μL/min and at 25°C. After injection of ligand, the protein surface was regenerated by injection of 10 μL of 10 mM HEPES buffer (pH 7.4) containing 1 M NaCl and 5 mM EDTA. Data were analyzed by fitting association and dissociation curves to Langmuir binding models for several protein concentrations simultaneously, using BIAevaluation 4.1 software (GE Healthcare). Increasingly complex models were tested until a satisfactory fit to the data was achieved. Apparent equilibrium dissociation constants (K_D) were calculated from the ratio of the dissociation and association rate constants (k_off/k_on). Errors were determined from two separate experiments.

MASP-2 activation assays

Activation was measured by following MASP autolysis as described previously 34. Proteins were separated by SDS-PAGE under reducing conditions and the amount of MASP cleaved was quantified by densitometry. Data are the mean±SE from at least two separate experiments using different protein preparations, unless otherwise stated.

Complement activation

Complement activation was measured using the Weislab Complement system MBL pathway kit (Euro-diagnostic). Assays were conducted as described, with the exception that wells of the microtitre dish were coated with acetylated-BSA (100 μL of 0.1 mg/mL), in 50 mM sodium bicarbonate (pH 9.6), overnight at 4°C, instead of mannan. Whole human serum was depleted of endogenous MBL and ficolins before setting up the assay by passage through two GlcNAc-Sepharose columns (0.5 for 1 mL of serum). Data represent the average±error from duplicate observations.