Molecular Interactions between MASP-2, C4, and C2 and Their Activation Fragments Leading to Complement Activation via the Lectin Pathway*

Activation of component C3 is central to the pathways of complement and leads directly to neutralization of pathogens and stimulation of adaptive immune responses. The convertases that catalyze this reaction assemble from fragments of complement components via multistep reactions. In the lectin pathway, mannose-binding lectin (MBL) and ficolins bind to pathogens and activate MBL-associated serine protease-2 (MASP-2). MASP-2 cleaves C4 releasing C4a and generating C4b, which attaches covalently to the pathogen surface upon exposure of its reactive thioester. C2 binds to C4b and is also cleaved by MASP-2 to form the C3 convertase (C4b2a). To understand how this complex process is coordinated, we have analyzed the interactions between MASP-2, C4, C2, and their activation fragments and have compared MASP-2-catalyzed cleavage of C4b2 and C2. The data show that C2 binds tightly to C4b but not to C4, implying that C4 and C2 do not circulate as preformed complexes but that C2 is recruited only after prior activation of C4. Following cleavage of C4, C4b still binds to MASP-2 (KD ∼ 0.6 μm) and dissociates relatively slowly (koff ∼ 0.06 s–1) compared with the half-life of the thioester (≤0.7 s) (Sepp, A., Dodds, A. W., Anderson, M. J., Campbell, R. D., Willis, A. C., and Law, S. K. (1993) Protein Sci. 2, 706–716). We propose that the C4b·MASP-2 interaction favors attachment of C4b near to the activating MBL·MASP complex on the bacterial surface so that, following recruitment of C2, the proximity of enzyme and substrate (C4b2) combined with more favorable reaction kinetics drive the formation of the C3 convertase, promoting complement activation.

gen clearance following engagement by the adaptive immune system. In the alternative pathway, spontaneous low level hydrolysis of component C3 effects deposition of protein fragments onto cell surfaces, triggering complement activation on pathogens. Regulatory proteins on host tissues avert activation, preventing self-damage. In the lectin pathway (4,5), mannosebinding lectin (MBL) 3 and serum ficolins bind directly to sugars or N-acetyl groups on pathogenic cells and activate MBL-associated serine proteases (MASPs) to initiate the complement cascade.
Three different MASPs (-1, -2, and -3) bind to MBL and ficolins (6 -8). They are homologues of C1r and C1s of the classical pathway and comprise two CUB domains (domain found in the complement component Clr/Cls, Uegf, and bone morphogenic protein 1) separated by a Ca 2ϩ -binding epidermal growth factor-like domain and followed by two CCP modules and a serine protease domain. MASPs normally circulate as zymogens bound to MBL and ficolins through their CUB and epidermal growth factor-like domains (9). When the lectin components bind to a target cell, MASP-1 and -2 activate through autolysis at a specific site within a linker region at the N-terminal end of the serine protease domain. Only MASP-2 has a clearly defined role in complement activation. It initially cleaves C4 to produce the peptide anaphylatoxin C4a and the C4b fragment, which attaches to the cell surface. C2 then binds to C4b and is also cleaved by MASP-2 to generate C2b and C2a. C2a remains attached to C4b to become the catalytic component of the C3 convertase (C4b2a) (10,11). The roles of MASP-1 and MASP-3 are not known. MASP-1 cleaves C2 but not C4, so it might enhance complement activation triggered by lectin ⅐ MASP-2 complexes, but cannot initiate activation itself (12). MASP-3 does not autoactivate, so is probably activated through the action of an unknown protease (13). Its biological role and physiological substrates remain to be identified.
Recent studies have shown that MASP-2 forms extensive contacts with C4 during complement activation (12,14,15). Indeed, in vitro, zymogen MASP-2 interacts weakly with C4 through accessory-binding sites, even though the catalytic site is disrupted. These accessory sites, which probably include one or both of the CCP modules of the MASP, only become exposed upon activation of the MBL⅐MASP-2 complex, thereby enabling MASP-2 to bind to C4 (12). Additional changes at the catalytic site allow the MASP to cleave its substrates.
Complement activation must be tightly regulated to prevent tissue damage. Stable attachment of components to the surface of pathogens is one of the mechanisms employed by the host. MBL⅐MASP complexes bind tightly to pathogens through multivalent interactions between the carbohydrate recognition domains of MBL and arrays of mannose-like sugars on cell surfaces (16). Analogous interactions between fibrinogen-like domains of ficolins and N-acetyl groups on pathogens immobilize ficolin⅐MASP complexes (17). When a lectin⅐MASP complex cleaves C4, the C4b fragment also attaches to the pathogen following exposure of a thioester, which reacts with the cell surface (18,19). For complement activation to proceed, the C4b fragment must be close enough to a lectin⅐MASP complex so that the MASP can cleave C2 once it has bound to C4b. Control of this process is relatively poorly understood. Co-localization is probably achieved, in part, by the high reactivity of the thioester bond of C4b toward hydroxyl, or amino, groups on the pathogen surface. Nevertheless, additional processes probably increase the efficiency of complement activation in vivo.
In this manuscript, we have investigated the steps leading to C3 convertase formation by analyzing the molecular interactions between MASP-2, C4, and C2 and their activation fragments. Based on our findings, we propose a model to explain how interactions between MASP-2 and C4b coordinate the activation process following pathogen recognition.
Amino Acid Sequencing-Amino acid sequencing was carried out using an Applied Biosystems 494A Procise protein sequencer. Samples were run for 10 cycles using standard sequencing cycles.
Production of Proteins and Protein Fragments-Recombinant proteins were used in these studies to ensure that preparations were not contaminated by trace amounts of other complement proteins. Previous work has demonstrated that components are processed correctly during biosynthesis and retain the key properties of native material (12,20).
Recombinant rat complement components C2 and C4 and catalytically active and inactive forms of rat MASP-2, called MASP-2K and MASP-2A, respectively, were produced by expression in Chinese hamster ovary cells and purified as described previously (12,20). MASP-2A comprises full-length MASP-2 in which the active site serine residue at position 613 is changed to an alanine. It was converted from the zymogen to the activated form by incubation with trypsin (0.25% w/w) for 1 h at 37°C in reaction buffer (50 mM Tris-Cl, pH 7.5, containing 150 mM NaCl) containing 1 mM CaCl 2 (Fig. 1). Edman degradation of the C-terminal fragment confirmed that cleavage occurred at the expected site for zymogen activation. Residual trypsin activity was inhibited by the addition of phenylmethylsulphonyl fluoride (0.1 mg/ml final concentration), and protein was dialyzed against reaction buffer to remove excess inhibitor. MASP-2K consists of full-length MASP-2 in which the arginine residue at the cleavage site for zymogen activation (Arg 424 ) is replaced by a lysine residue (12). This change reduces the rate of autocatalysis and thereby prevents activation of the zymogen during biosynthesis, secretion, and purification, allowing preparation of pure zymogen. MASP-2K was activated by incubation at 37°C for 24 h in reaction buffer containing 1 mM CaCl 2 . Complement component C4b was generated from C4 by incubation with activated MASP-2K (0.02% w/w) for 1.5 h at 37°C in reaction buffer (Fig. 1D). The reaction was stopped by the addition of phenylmethylsulphonyl fluoride, as described above. Cleavage of all proteins was monitored by SDS-PAGE, and protein fragments were quantified by scanning gels using a Chemigenius bioimaging system from Syngene. C4(met), in which the thioester has become exposed by reaction with methylamine, was produced by incubation of C4 with methylamine at pH 8 for 1 h at 37°C. Before further analysis, all proteins were dialyzed against reaction buffer. Proteins were separated on a 12% (w/v) gel under reducing conditions and were detected by staining with Coomassie Blue. Edman degradation of the smaller, C-terminal fragment gave the sequence IIGGQPAKPG, confirming that cleavage occurred at the expected site for activation (12). The C-terminal fragment migrates as two bands on the gel, due to differential glycosylation. D, SDSpolyacrylamide gel electrophoresis of C4 and C4b. C4b was generated by digestion of purified C4 with activated MASP-2K. Proteins were separated on a 10% (w/v) gel under reducing conditions and detected by staining with Coomassie Blue.
Proteolysis of the C4b2 Complex-The catalytic activity of MASP-2K toward C4b2 was measured by incubating activated MASP-2K (1-2 nM) with increasing concentrations of C2 (0.1-2 M) in the presence of excess C4b (2.5 M) in reaction buffer containing 1 mM CaCl 2 , 1 mM MgCl 2 , and ovalbumin (40 g/ml) at 37°C. C2 and C4b were preincubated for 30 min before the addition of enzyme to allow complex formation. At various times, aliquots were removed, and the reaction was stopped by boiling in gel-loading buffer. The extent of C2 cleavage was determined by scanning SDS-polyacrylamide gels. To correct for minor differences in the amounts of sample loaded onto the gels, values were normalized using the amount of ovalbumin in each aliquot as a reference. A similar assay was used to measure cleavage of C2, except that the concentration of activated MASP-2 was 1 nM and the concentration of C2 was between 1 and 10 M. Initial rates were calculated from the first 10% of cleavage or less. Data were fitted to the Michaelis-Menten equation by non-linear regression using the software Origin from Microcal.
Analytical Ultracentrifugation-Equilibrium experiments were carried out as described previously in a Beckman XLA-70 centrifuge using epon charcoal-filled six-hole centerpieces (21,22). Before analysis, all proteins were dialyzed extensively against reaction buffer containing 1 mM CaCl 2 and 1 mM MgCl 2 . To measure the interaction between activated MASP-2A and C4b, proteins were mixed at different molar ratios to give initial absorbances of 0.1-0.6 at 280 nm. Data were analyzed using the program 2C1SFIT, which implements conservation of mass constraints to model the equilibrium distributions globally (23). The apparent equilibrium association constant and the fraction of competent components were allowed to vary as global fitting parameters. Apparent equilibrium association constants were converted to molar dissociation constants as described previously (24). Values are mean Ϯ S.E. from three independent experiments. It was assumed that no changes in the partial specific volumes occur on complex formation and that the molecular mass of the complex is the sum of that of the components. Molecular masses of MASP-2 and C4 were determined from experiments run in parallel (see Table 1) by fitting equilibrium data at three different loading concentrations to a model assuming a single solution species, using software supplied with the centrifuge.
Sedimentation velocity experiments were carried out at 40,000 revolutions/min and at 20°C using aluminum centerpieces. Prior to setting up the experiments, proteins were dialyzed against reaction buffer containing 1 mM CaCl 2 and 1 mM MgCl 2 . Scans were collected at 2-4-min intervals at 230 nm. Data were analyzed using the program DCDT (25). Values are displayed as s 20,w (Table  1) by correcting for the effects of buffers (26).
Surface Plasmon Resonance-Measurements were performed using a BIAcore 2000 instrument (BIAcore). Protein ligands were diluted to 25 g/ml in 10 mM sodium acetate at pH 4.5 (for C4, C4b and C4(met)) or pH 3.5 or 4.0 (for activated MASP-2A) and immobilized onto the carboxymethylated dextran surface of a CM5 sensor chip (BIAcore) using amine-coupling chemistry (BIAcore amine-coupling kit). Binding was measured in reaction buffer at a flow rate of 5 l/min and at 25°C. Regeneration of the protein surfaces between analyses was achieved by injection of 10 l of 5 mM sodium acetate, pH 3.5, containing 1 M NaCl. Data were analyzed by fitting to a 1:1 Langmuir binding model for several protein concentrations simultaneously, using BIAevaluation 3.1 software (BIAcore). The apparent equilibrium dissociation constants (K D ) were calculated from the ratio of the dissociation and association rate constants (k off /k on ).
Kinetics of C4 Binding to Substrate-Binding of C4b to a substrate S can be described by the following reaction scheme: SCHEME 1 Measurements of C4 activation for a variety of species have been made using small substrates such as glycine and glycerol (27

RESULTS
Complement activation normally occurs on the surfaces of pathogens. Nevertheless, solution studies have provided key insights into the activation process. In this manuscript, we have adopted both solution and solid-phase approaches to study the molecular interactions leading to C3 convertase formation in the lectin pathway. Similar processes on the surfaces of pathogens probably help to regulate complement activation in vivo.

TABLE 1 Biophysical properties of complement components
Unmodified molecular masses and partial specific volumes of proteins were calculated from their amino acid compositions (26). Weight-averaged molecular masses were determined by equilibrium ultracentrifugation.

Protein
Unmodified monomer molecular mass

Interactions between MASP-2, C2, and C4
Recruitment of C2 by C4b-C2 binds to C4b with high affinity (28). However, binding between C2 and C4 has not been analyzed. Even a weak interaction between these components would probably facilitate C3 convertase formation, because both substrates would be present at the same time for their stepwise activation by a lectin⅐MASP-2 complex. We therefore examined whether C4 binds to C2. Freshly prepared C4 was used in these experiments to minimize hydrolysis of the thioester, which occurs spontaneously in solution but at a low rate. This precaution was necessary, because thioester hydrolyzed, but peptide bond-intact C4 binds to C2, and the resulting complexes would have similar properties to C4b2 complexes. Samples at concentrations comparable with those that occur in vivo were analyzed by velocity analytical ultracentrifugation using the time derivative method g(s*) to display the distribution of apparent sedimentation coefficients (s*). Using this approach, complexes can be distinguished from free components by differences in the rates at which they sediment. In a mixture of C2 and C4, the s* distribution corresponded closely to the sum of the distributions of the components analyzed separately ( Fig.  2A), indicating that C2 and C4 do not interact with each other over the range of concentrations examined. By contrast, when C4b was mixed with C2, a new peak was observed corresponding to a complex, which sedimented faster than either of the free components, confirming that C4b binds to C2 (Fig. 2B). Only small amounts of unassociated species were detected in the mixture, indicating that the K D of the C4b2 complex is considerably lower than the loading concentration of the components (0.2 M), consistent with previous studies in which the K D for the interaction between human C4b and C2 was determined as ϳ1 ϫ 10 Ϫ8 M (28). Thus, we conclude that C2 and C4 do not interact with each other before activation of C4. However, following cleavage of C4, C2 binds tightly to C4b to form the C4b2 complex, which is subsequently cleaved again to form the C3 convertase.
Interactions between Activated MASP-2 and C4b-Binding between MASP-2 and C4b, following activation of C4, would help to ensure that C4b attaches to the pathogen surface close to the lectin⅐MASP-2 complex. Co-localization of these components, in turn, would increase the probability of cleavage of the C4b2 complex by the same MASP and would thus help to coordinate C3 convertase formation. Following cleavage of C4, the C4b fragment undergoes a conformational change that exposes the thioester. Nevertheless, at least some of the MASPbinding sites might still remain after the conformational change. To examine this possibility, we tested whether MASP-2 binds to C4b. In a mixture of activated MASP-2A and C4b, the amount of unassociated MASP-2 decreased, and the average s* increased relative to the individual components measured sep-  arately, indicating that MASP-2A does indeed bind to C4b (Fig.  3). Some of the MASP was still unassociated in the mixture, so the K D is probably greater than the loading concentration of the components (0.4 M).
To investigate binding between MASP-2A and C4b further, samples were analyzed by equilibrium ultracentrifugation. In a mixture of both components, more protein was distributed toward the bottom of the cell compared with the sum of the equilibrium distributions of the components, confirming the presence of MASP-2A⅐C4b complexes (Fig. 4A). The strength of the interaction was quantified by comparing mixtures of MASP-2A and C4b at different molar ratios. Data from three separate experiments, each at two different loading concentrations, fit well to models assuming formation of 1:1 complexes in which the K D was 2.7 Ϯ 1.3 M (Fig. 4B).
Interestingly, the affinity of the interaction between activated MASP-2A and C4b is comparable with the affinity between zymogen MASP-2 and intact C4, characterized previously under comparable conditions (6.8 Ϯ 2.0 M) (12,14). The latter interaction is mediated through accessory-binding sites on the MASP, which become exposed following activation of the MBL⅐MASP complex, thereby lowering the K m for catalysis of C4. Given their comparable affinities, it is likely that MASP-2 binds to C4 and C4b through the same binding sites, implying that the binding surface is maintained upon cleavage of C4, despite the subsequent conformational change.
Although the affinity of the interaction between activated MASP-2 and C4b is relatively low compared with many protein-protein interactions (e.g. binding between C4b and C2), it is unusually high for an enzyme⅐product complex, where affinities are often in the mM range. Binding between enzyme and product might help to localize C4b near the lectin⅐MASP-2 complex on the pathogen by trapping the C4b molecule until it has attached covalently to the cell surface. The probability of such a process is dependent on the relative magnitudes of the rate of dissociation of C4b from MASP-2 (k off ) following cleavage of C4 and the rates at which the thioester is exposed and subsequently reacts with its substrate (see "Experimental Procedures"). When dissociation is of a similar magnitude or slower than the reactivity of the thioester, a significant proportion of C4b molecules are likely to bind covalently to hydroxyl or amino groups on the cell surface before their release by the enzyme.
To analyze the kinetics of the interaction between C4b and MASP-2, each component was immobilized in turn, and binding by its soluble partner was measured using surface plasmon resonance. Data for MASP-2A binding to immobilized C4b are shown in Fig. 5A, and the kinetic parameters from all experiments are summarized in Table 2. Reassuringly, the association (k on ) and dissociation (k off ) rate constants were comparable irrespective of whether MASP-2A or C4b was attached to the sensor chip. Furthermore, the K D values (at 25°C; calculated from k off /k on ) were broadly comparable with the K D measured by equilibrium ultracentrifugation (at 20°C), demonstrating that immobilization of either component did not impair binding greatly. The k off values were 7.4 Ϯ 4 ϫ 10 Ϫ2 (t1 ⁄ 2 ϭ 11.5 Ϯ 4.8 s) and 4.6 Ϯ 1.4 ϫ 10 Ϫ2 s Ϫ1 (t1 ⁄ 2 ϭ 16.7 Ϯ 5.1 s) using immobilized C4b and MASP-2A, respectively. By comparison, the half-life of the thioester in human C4B is Յ0.7 s (27) (see "Kinetics of C4 Binding to Substrate" above). Therefore, assuming that the activation kinetics of rat C4 are broadly comparable with those of human C4B, then a significant proportion of C4b molecules are likely to bind to substrate before release from the MASP. Thus, the relatively slow dissociation of C4b from MASP-2 probably helps to co-localize these components on an activating surface, thereby enabling subsequent cleavage of C2 by the same MASP.
To further probe the activation mechanism, we also measured the interactions of MASP-2A with C4 and C4(met) (Fig. 5  and Table 2). MASP-2A bound to C4 4 -10-fold more tightly than to C4b, probably because of additional contacts between the enzyme and substrate at or near the active site. Interestingly, very little binding was detected between MASP-2A and C4(met), even though the amount of C4(met) bound to the sensor chip was comparable with the amounts of C4 and C4b (Fig. 5C). Likewise, only minimal binding was detected between soluble C4(met) and immobilized MASP-2A at concentrations of C4(met) up to 0.5 M. We conclude that MASP-2 binds more weakly to C4(met) than to C4b, despite both proteins undergoing similar conformational changes upon cleavage of the thioester. The most likely explanation for this observation is that the C4a fragment, which is still covalently attached in C4(met) but not in C4b, impairs MASP-2 binding.
Catalysis of the C4b2 Complex by MASP-2-During formation of the C3 convertase, C4b, the product of the first MASP-2-catalyzed reaction, becomes part of the substrate of the second reaction (C4b2). Any persisting interactions between the MASP and the C4b component of C4b2 would be likely to affect the reaction kinetics. To investigate the role of C4b in the cleav-age of C4b2, we compared the cleavage of C4b2 with the cleavage of C2 alone. Kinetic parameters are shown in Table 3, and examples of cleavage of C4b2 and C2 by MASP-2K on polyacrylamide gels are shown in Fig. 6. Both reactions followed Michaelis-Menten kinetics over the concentration ranges examined. There is a modest increase in the catalytic efficiency of C4b2 cleavage compared with C2 cleavage, which is characterized by a 7-fold lower K m value and a smaller reduction in k cat (Table 3). Thus, C4b binding modifies the kinetics of C2 cleavage catalyzed by MASP-2. The significance of these findings are discussed below.

DISCUSSION
The data presented here suggest novel mechanisms for the regulation of C3 convertase formation during complement activation by the lectin pathway. By combining these findings with existing knowledge, we revised the current model for the early stages of complement activation (Fig. 7). When MBL⅐MASP-2 or ficolin⅐MASP-2 complexes bind to the surface of a pathogen, conformational changes trigger MASP autoactivation. Activated MASP-2 binds to C4 through accessory-binding sites on the CCP domains as well as at the active site within the protease domain, leading to cleavage of the C4 polypeptide (12). The newly exposed thioester on C4b reacts with hydroxyl or amino groups on the bacterial cell surface before dissociation from the MASP, thus binding proximally to the activating lectin⅐MASP-2 complex. Co-localization of these components facilitates subsequent cleavage of incoming C2 molecules by the same MASP, thereby coordinating activation.
Intriguing questions still remain concerning the order of events during the initial stages of complement activation. For example, it is uncertain whether C4a is released from C4b immediately after cleavage by the lectin⅐MASP complex (as shown in Fig. 7) or at a later stage. Interestingly, during C3 activation, C3a remains associated with C3b and may not be released until the larger C3b fragment is broken down to iC3b (29).
It is important to acknowledge that complement activation normally occurs on the surfaces of foreign cells in serum, which is a very different environment from those typically used for in vitro studies. In vivo, activation is controlled by a variety of regulators and inhibitors, and many different serum proteins and carbohydrates interact with components to modulate the process. For example, in the classical pathway, C1 inhibitor plays a major role in the activation efficiency of C1s, the classical pathway homologue of MASP-2, by binding and inhibiting C1 and the subsequent activation of C2 and C4 (30). C1 inhibitor also binds and inhibits MASP-2 (11) and so might play a comparable role in the lectin pathway, although the kinetics of    (30). Nevertheless, this rate of turnover is sufficient to trigger activation, which is subsequently amplified further by latter steps in the reaction cascade. Complement is also regulated by polysaccharides, such as heparin and heparan sulfate, which bind to many complement components, including C4, C1 inhibitor, and more weakly to C2 (31). Although these and other protein-protein or protein-carbohydrate interactions might modulate enzyme activities or compete for substrate binding, they are unlikely to change the underlying activation mechanism proposed here, because the key step of C4b deposition onto the activating surface (such as a bacterial cell wall) is faster than its release from MASP-2 and therefore happens while C4b is still attached to the enzyme.
It is relatively unusual for the product of an enzyme-catalyzed reaction to bind to its enzyme with appreciable affinity, because such interactions are likely to reduce the efficiency of catalysis by blocking the catalytic site. During formation of the C3 convertase, however, interactions between the product of the first reaction (C4b) and the enzyme (MASP-2) probably help to coordinate the two-step reaction on the pathogen sur-face by reducing the probability of additional C4 cleavage while increasing the chances of C4b2 cleavage. In addition, the proximity of C4b2 and MASP-2 will increase the effective concentration of substrate (C4b2) greatly, thus driving formation of the convertase.
Interestingly, although the K m value for C2 cleavage decreases by 7-fold when C4b binds to C2, the increase in the catalytic efficiency of the reaction in solution is relatively modest (ϳ2-fold). As discussed above, on the surface of a bacterial cell, cleavage of C4b2 is likely to be driven by its high effective concentration. Nevertheless, it is interesting to consider the kinetics of the reaction in solution and the changes that occur when C2 binds to C4b. Cleavage of C4 and C4b2 by MASP-2 are both characterized by relatively low K m values, which, for C4 at least, is because of a relatively low K D value for the enzyme⅐substrate complex. Generally, a relatively stable enzyme⅐substrate complex is likely to improve substrate specificity but at the expense of catalysis (because of the increase in the resulting energy barrier). However, such properties might be advantageous in the complement cascade, where control of activation is particularly important. The data would be consistent with this model in which a decrease in the K m value for C4b2 FIGURE 6. Cleavage of C4b2 complexes and C2 by MASP-2K analyzed by SDS-PAGE. Ovalbumin (0.6 g in each lane) was included in the reaction mixtures to prevent nonspecific interactions and to correct for minor differences in the amounts of protein loaded onto the gels. Proteins were detected with Coomassie Blue. A, C2 (1.5 M) with an excess of C4b (2.5 M) was incubated with activated MASP-2K (2 nM) at 37°C. Aliquots were removed and proteins were separated on a 12% (w/v) polyacrylamide gel. Gel electrophoresis was carried out under non-reducing conditions, so that the cleavage products (C2a and C2b fragments) could be distinguished from the three C4b polypeptides. B, C2 (10 M) was incubated with activated MASP-2K (1 nM) at 37°C. Proteins were separated on a 15% gel (w/v) under reducing conditions. Only the initial rates were measured. FIGURE 7. Model of C3 convertase formation in the lectin pathway of complement. A, binding to the surface of a pathogen induces autoactivation of MBP⅐MASP-2, exposing accessory C4-binding sites on the MASP and leading to recruitment of C4. B, MASP-2 cleaves C4. Anaphylatoxin C4a is released, whereas interactions between the MASP and the C4b fragment lead to covalent attachment of C4b proximal to the activating lectin⅐MASP-2 complex on the pathogen surface. C, C2 binds to C4b, and the resulting complex is cleaved by the same MASP-2 molecule to form the C3 convertase (C4b2a). D, C2b probably remains attached to the convertase through non-covalent interactions (10). Proteins are represented in schematic form for simplicity.