Interaction of Mannose-binding Protein with Associated Serine Proteases

Mannose-binding protein (MBP; mannose-binding lectin) forms part of the innate immune system. By binding directly to carbohydrates on the surfaces of potential microbial pathogens, MBP and MBP-associated serine proteases (MASPs) can replace antibodies and complement components C1q, C1r, and C1s of the classical complement pathway. In order to investigate the mechanisms of MASP activation by MBP, the cDNAs of rat MASP-1 and -2 have been isolated, and portions encompassing the N-terminal CUB and epidermal growth factor-like domains have been expressed and purified. Biophysical characterization of the purified proteins indicates that each truncated MASP is a Ca2+-independent homodimer in solution, in which the interacting modules include the N-terminal two domains. Binding studies reveal that both MASPs associate independently with rat MBP in a Ca2+-dependent manner through interactions involving the N-terminal three domains. The biophysical properties of the truncated MASPs indicate that the interactions with MBP leading to complement activation differ significantly from those between components C1q, C1r, and C1s of the classical pathway. Analysis of MASP binding by rat MBP containing naturally occurring mutations equivalent to those associated with human immunodeficiency indicates that binding to both truncated MASP-1 and MASP-2 proteins is defective in such mutants.

Mannose-binding protein (MBP 1 ; mannose-binding lectin) constitutes the first component of the lectin pathway of complement activation (1,2). This pathway leads to neutralization of pathogenic microorganisms through an antibody-independent mechanism. MBP binds directly to sugars on the surfaces of bacterial, fungal, and parasitic cells and activates comple-ment through two associated serine proteases (MASPs) that are homologues of C1r and C1s of the classical pathway.
MBP in serum preparations copurifies with two distinct MASPs, MASP-1 and MASP-2 (3,4). However, the molecular basis of the interactions leading to MASP activation is poorly understood. The MASPs probably bind to MBP near the N terminus in a collagen-like domain (5). This region of MBP forms a central core that links trimeric subunits to form bouquet-like oligomers ranging from monomers to tetramers of subunits. In human MBP, three separate mutations within the corresponding portion of the collagenous domain are associated with an immunodeficiency disorder characterized by an increased susceptibility to infections (2). When these mutations are recreated in rat MBP, they lead to distinct structural defects that result in a reduced ability to activate the complement cascade (6). The mutation Arg 23 3 Cys causes adventitious disulfide bond formation that reduces the proportion of trimers and tetramers of subunits. The reduced activity of this MBP variant is due to inefficient complement activation by the smaller oligomeric forms. The mutations Gly 25 3 Asp and Gly 28 3 Glu also lead to a small reduction in the proportion of higher order oligomers. However, even the larger oligomers in these mutants are inefficient at fixing complement, probably due to defective MASP binding.
MASP-1 and MASP-2 share a common domain organization comprising two N-terminal CUB domains separated by an EGF-like domain, two complement control protein modules (CCP1 and CCP2), and a C-terminal serine protease domain (Fig. 1). The EGF-like module of each protein has the characteristic sequence motif of a Ca 2ϩ -binding EGF-like domain (7). MASP-1 and MASP-2 are synthesized as zymogens (3,8). Upon activation, each MASP polypeptide is cleaved, generating an active protease that is covalently attached to the N-terminal domains through a disulfide bond. MAP19 is an alternatively spliced form of both human and rat MASP-2 comprising the N-terminal two domains (9 -11). MAP19 copurifies with MBP, suggesting that the N-terminal domains of MASPs can interact with MBP.
In order to study the interactions between MBP and its associated serine proteases, cDNAs encoding rat MASPs have been generated and used to produce the N-terminal CUB and EGF-like domains of MASP-1 and MASP-2. The truncated forms of MASP-1 and MASP-2 bind independently to MBP in a Ca 2ϩ -dependent manner through interactions involving at least two of the three N-terminal modules. Binding of MASPs to variant MBPs associated with immunodeficiency is defective, indicating that aberrant interactions caused by structural defects in the mutant MBPs probably lead directly to inefficient complement fixation.

EXPERIMENTAL PROCEDURES
Materials-Restriction enzymes, peptide:N-glycosidase F, and endoglycosidase H were purchased from New England Biolabs. The rat liver cDNA library and cDNA Marathon kit were from CLONTECH. All tissue culture medium was from Life Technologies. Tosylphenylalanyl chloromethyl ketone-treated trypsin was from Worthington Biochemical Corporation. Promix cell-labeling mix (ϳ70% [ 35 S]methionine and 30% [ 35 S]cysteine) was from Amersham Pharmacia Biotech. Nitrilotriacetic acid-agarose, Sepharose 6B, and protein molecular weight markers were from Sigma. Affi-Gel 10 matrix was purchased from Bio-Rad. Immulon 4 microtiter wells were from Dynatech. Polyvinylidene difluoride membranes were from Applied Biosystems.
Analytical Methods-SDS-polyacrylamide gel electrophoresis was performed by the method of Laemmli (12) (Fig. 2). Amino acid sequencing was carried out on a Beckman LF3000 protein sequencer. Polypeptide samples for sequence analysis were separated by SDS-polyacrylamide gel electrophoresis and blotted onto polyvinylidene difluoride membranes (13). Equilibrium analytical ultracentrifugation was carried out as described previously in a Beckman XLA-70 centrifuge (14). Values reported are mean Ϯ S.E. for at least two separate experiments. MALDI-MS was carried out on a FinniganMAT Lasermat mass spectrometer using sinapinic acid as the sample matrix (14). Bovine serum albumin ([M ϩ H] ϩ ϭ 66431 Da) was used as the calibration standard. Protein samples from the column binding assays were concentrated by precipitation with 10% trichloroacetic acid prior to SDS-polyacrylamide gel electrophoresis (14).
Generation of MASP-1 and MASP-2 cDNAs-The cDNAs encoding mature rat MASP-1 and MASP-2 proteins were generated by amplifying overlapping fragments from a rat liver cDNA library using the polymerase chain reaction. Oligonucleotides were derived from the sequence of the corresponding mouse cDNAs (15,16). The 5Ј oligonucleotides, ggagctcacagagcaggaaaatgaggt and caccatgaggctactcatcttcctg, contain the start codons (underlined) and encode the first 3 and 7 residues of the signal sequences of mouse MASP-1 and MASP-2. The 3Ј-end of each cDNA was cloned by 3Ј-rapid amplification of cDNA ends using a rat liver cDNA Marathon kit. Full-length cDNAs were assembled using convenient restriction sites. Standard molecular biological techniques were performed as described (17).
Overproduction and Purification of Truncated MASPs-Truncated cDNAs were generated by introducing stop codons preceded by six histidine codons into the cDNAs by substitution of double-stranded oligonucleotides for restriction fragments at the 3Ј-ends. The position of the C-terminal ends of the encoded proteins are shown in Fig. 1. Restriction fragments encompassing the truncated cDNAs were ligated into the polylinker region of plasmid pED (18). The resulting plasmids were used to transfect adherent Chinese hamster ovary cell line DXB11, and production of the truncated MASPs was amplified by passaging cells into increasing concentrations of methotrexate (19). For large scale protein preparations, cells were grown to confluence in 225-cm 2 tissue culture flasks in 50 ml of minimal essential medium ␣ lacking nucleosides and supplemented with 10% dialyzed fetal calf serum and 0.5 M methotrexate. Once confluence was reached, the medium was replaced with Chinese hamster ovary serum-free II medium supplemented with 50 mM HEPES buffer, pH 7.55, 20 M CaCl 2 , and 0.5 M methotrexate. Medium was harvested and replaced every second day for 8 days. Cell debris was removed prior to purification by centrifugation (3000 ϫ g for 4 min), and medium was stored at Ϫ80°C. Truncated MASPs were purified by loading 200 ml of medium diluted with an equal volume of loading buffer (25 mM Tris, pH 7.5, containing 500 mM NaCl) onto a 1-ml column of nitrilotriacetic acid-agarose that had been preloaded with 5 ml of 50 mM NiSO 4 and equilibrated with 10 ml of loading buffer. Following washing with 10 ml of loading buffer and 10 ml of buffer containing 20 mM imidazole, protein was eluted in 0.5-ml fractions in buffer containing 200 mM imidazole. MASP-containing fractions were identified by SDS-polyacrylamide gel electrophoresis.
Antibodies against MASP-1 and MASP-2-Rabbit antibodies were raised against synthetic peptides corresponding to the N-terminal 15 amino acid residues of MASP-1 and MASP-2 conjugated to keyhole limpet hemocyanin. Four injections were given at 2-week intervals, and serum was collected 2 weeks after the last injection. The MASP-1 peptide was synthesized by MWG-BIOTECH. Antibodies against MASP-2 were supplied by Eurogentec.
Purification of MBPs and [ 35 S]MBP-Recombinant wild-type and mutant MBPs were purified by affinity chromatography on mannose-Sepharose columns (5,14). Radiolabeled MBP was produced by incubating confluent cells in methionine-free minimal essential medium ␣, supplemented with 10% dialyzed fetal calf serum, 0.5 M methotrexate, and 0.1 mCi/ml [ 35 S]methionine for 16 h. MBP was purified as for unlabeled protein.
MBP/MASP Binding on Mannose-Sepharose and Immobilized MBP Columns-MASP (5 g) was preincubated with MBP (400 g) in 200 l of high salt buffer (50 mM Tris, pH 7.5, containing 1.25 M NaCl, 25 mM CaCl 2 , and 0.01% gelatin) for 16 h at 4°C, diluted to 3 ml in ice-cold high salt buffer, and loaded on to a 1-ml mannose-Sepharose column equilibrated in high salt buffer. Following washes with 3 ml of high and low salt buffers (150 mM NaCl), bound protein was eluted in 1-ml fractions of elution buffer (50 mM Tris, pH 7.5, containing 125 mM NaCl, 2.5 mM EDTA, and 0.01% gelatin).
MBP was covalently attached to an agarose gel by incubating purified MBP (0.5 mg/ml) with 1 ⁄3 volume of Affi-Gel 10 matrix in 100 mM MOPS, pH 7.2, containing 25 mM CaCl 2 for 4 h at 4°C. MASP binding assays were carried out as described above except that protein was loaded on to the column in 2 ml of low salt buffer, washed with 2 ml of high and low salt buffers, and eluted in 1-ml fractions of 50 mM sodium Proteolysis of MASP Truncations-Truncated MASP-1 and MASP-2 were incubated with 2.5 and 5% (w/w) trypsin in 50 mM Tris, pH 7.5, containing 150 mM NaCl, and increasing concentrations of CaCl 2 or EDTA (1 mM) for 1 h at 37°C. Each reaction was stopped by boiling in loading buffer before separation on SDS-polyacrylamide gels. The amount of each MASP remaining after proteolysis was determined by scanning desitometry. Data averaged from three separate experiments were fitted to multiple ligand-binding curves by nonlinear regression using MicroCal Origin (20).
Competition Assay-Polystyrene plates were coated with 100 l of the 3-domain truncated MASP-1 or MASP-2 (0.1 mg/ml) and incubated at 4°C for 16 h. Wells were blocked with 5% (w/v) bovine serum albumin in binding buffer (50 mM Tris, pH 7.5, containing 150 mM NaCl, and 25 mM CaCl 2 ). Following incubation at 4°C for 4 h, wells were emptied and washed three times with binding buffer. 35 S-Labeled MBP (ϳ50 ng at a specific activity of ϳ1 ϫ 10 5 cpm/g) and 2-fold serial dilutions of either wild-type or mutant MBP were mixed in binding buffer containing 1 mg/ml of bovine serum albumin. Aliquots (100 l) were added to the wells and incubated for 2 h at 4°C. Wells were emptied and washed three times in ice-cold binding buffer. Bound protein was removed from the wells in two rinses (75 l each) of 0.1 M NaOH. Samples were neutralized with 50 l of 0.5 M acetic acid containing 6 mM 2-mercaptoethanol and blotted onto a nitrocellulose filter. Levels of radioactivity were determined using a PhosphorImager SI from Molecular Dynamics following exposure for at least 16 h. Data were fitted to multiple ligand-binding curves by nonlinear regression using MicroCal Origin (20). Preliminary studies indicated that the concentration of 35 S-labeled MBP used in the binding assays is below the concentration required for half-maximal binding.

Production and Characterization of Truncated MASPs
Generation of Rat MASP cDNAs and Expression and Purification of Truncated MASPs-The cDNAs encoding rat MASP-1 and MASP-2 were amplified using the polymerase chain reaction. The encoded sequences of the mature polypeptides reveal that rat MASP-1 shares 95 and 86% sequence identity with mouse and human MASP-1, whereas rat MASP-2 shares 90 and 80% identity with its mouse and human counterparts (3,15,16,21). The primary structure of the encoded rat MASP-2 protein corresponds to the recently described sequence with the exception of five residues: Pro 14 (encoded by cct), Gly 15 (ggc), Pro 487 (cct), His 488 (cat), and Ala 603 (gcc) that were reported as Leu, Ala, Leu, Ile, and Arg (10). In each case, the sequences of at least two independent PCR products confirm these differences. Further, residues Pro 14 , Gly 15 , Pro 487 , and His 488 are conserved in both human and mouse MASP-2, suggesting that the discrepancies reflect errors in the original sequence.
In order to produce large amounts of MASP-1 and MASP-2 for detailed biophysical analysis, cDNAs were introduced into an expression vector containing the dihydrofolate reductase gene as a selectable marker, and the resulting plasmids were used to transfect Chinese hamster ovary cells. Preliminary expression studies using the full-size MASP-1 cDNA revealed that the protein is produced only in trace amounts from the expression system. Compared with the much higher yields of other serum proteins produced using this expression system (5,14), the amounts of MASP-1 isolated suggest that expression of the MASP-1 gene may be cytotoxic, probably due to activation of the protease during biosynthesis.
In order to produce sufficient material to study the interactions between the MASPs and MBP, truncated MASPs were created. Because the N-terminal domains of the MASPs are likely to be involved in interactions with MBP, as evidenced by the properties of MAP19, proteins comprising the N-terminal two and three domains of each MASP were produced. The truncated proteins were synthesized with C-terminal histidine tags enabling purification in a single step using affinity chromatography on an immobilized nickel column. Typical yields range from 0.2 to 1 mg from 200 ml of medium, sufficient to enable a detailed study of the structural organization and interactions with MBP. Since the proteins are produced in serum-free medium, they are free from contamination by other components of the complex and by other serum proteins.
Each truncated MASP migrates as broad or multiple bands on reducing SDS-polyacrylamide gels with apparent molecular masses of ϳ52 and 33 kDa for the truncated MASP-1 proteins and 43 and 24 kDa for the truncated MASP-2 proteins. Nterminal sequencing revealed a single sequence in each case ( Fig. 1), confirming that the signal sequences are processed during biosynthesis.
The mobility on SDS-polyacrylamide gels as well as analysis by MALDI-MS indicates that the polypeptides are glycosylated (Table I). There are two potential N-glycosylation sites within the first three domains of each MASP (Fig. 1). Digestion of the three-domain truncated proteins with peptide:N-glycosidase F confirms that they are glycosylated. Following digestion, they migrate with reduced apparent molecular masses of 43 and 35 kDa, closer to those expected for the polypeptides alone. Both truncated proteins are resistant to endoglycosidase H digestion under native and denaturing conditions, indicating that the N-linked oligosaccharides consist of complex structures rather than high mannose or hybrid oligosaccharides.
Oligomeric Structure of Truncated MASPs-Analytical ultracentrifugation indicates that both the two-and three-domain portions of MASP-1 and MASP-2 are stable homodimers ( Table I). The shape-independent, weight-averaged molecular masses were determined by fitting the data to single species models. The truncated MASPs show no further self-association at concentrations in excess of those normally present in serum (Fig. 3). Similar molecular masses are observed in the presence and absence of EDTA, indicating that dimer formation is not dependent on Ca 2ϩ binding (Table I). Thus, both two-and three-domain MASP-1 and MASP-2 proteins are Ca 2ϩ -independent homodimers in which the interacting domains include the N-terminal CUB and EGF-like modules.
Analytical ultracentrifugation was also used to determine whether the truncated MASPs interact with each other or whether they function independently to bind to MBP. An equilibrium experiment was set up in which three-domain MASP-1 and MASP-2 proteins were mixed in equal molar amounts. The equilibrium distribution of the mixture corresponds precisely to the sum of the equilibrium distributions of the two components run in parallel at identical protein concentrations (Fig.  4). Thus, the truncated MASP homodimers behave independently and do not interact with each other to form a complex even at high protein concentrations and in the presence of Ca 2ϩ .

Independent Interactions of Truncated MASP-1 and MASP-2 with MBP
Ca 2ϩ -dependent Interactions of MASPs with MBP-In order to examine the interactions between MBP and MASP-1 and -2, each of the purified truncated MASPs was incubated with an excess of purified recombinant MBP and passed through a mannose-Sepharose column to which MBP binds. Following several wash steps at both high and physiological salt concentrations, MBP and associated MASP were eluted by chelating the Ca 2ϩ required for binding of MBP to the column. In the absence of MBP, each truncated MASP elutes in the unbound wash fractions, indicating that it does not interact directly with the mannose-Sepharose column (Fig. 5). However, in the presence of MBP, the three-domain MASP-1 protein is largely retained on the column even after significant washing, indicating that it does interact with MBP that is bound to the column (Fig.  5). In contrast, no binding to MBP was detected for the twodomain MASP-1 protein, which elutes in the unbound wash fractions. Thus, the N-terminal three domains of MASP-1 are sufficient for stable binding to MBP in the absence of MASP-2, although the first two domains alone do not support tight binding.
Similar binding results were obtained when the truncated MASP-1 proteins were passed through an affinity column in which MBP was covalently linked to the gel matrix (Fig. 5). The three-domain protein is largely retained on the column, while the two-domain protein elutes in the unbound fractions. When the high salt wash step was replaced by washing at physiological salt concentration, some of the two-domain protein was retained on the column, although the interaction was clearly weaker than binding by the three-domain protein under similar conditions (Fig. 5). Thus, MBP binding by MASP-1 occurs through interactions mediated by multiple modules including the N-terminal three domains. When the assay was carried out in the presence of EDTA, no binding was detected, indicating that the interaction between MBP and MASP-1 requires Ca 2ϩ .
Analysis of MASP-2 using similar assays indicates that the truncated proteins bind to MBP independently of MASP-1. The three-and two-domain proteins are retained on the mannose-Sepharose and MBP-agarose columns through interactions with MBP (Fig. 5). No elution in the unbound fractions was observed by either truncated protein under the conditions examined. As with MASP-1, MASP-2 binding to MBP is Ca 2ϩ -dependent, because no binding to the affinity column is detected in the presence of EDTA. Thus, truncated MASP-1 and MASP-2 bind independently to MBP in a Ca 2ϩ -dependent manner. In each case, binding is mediated through interactions including the N-terminal three domains.
Ca 2ϩ -binding Properties of Truncated MASPs-Because each MASP binds to MBP in a Ca 2ϩ -dependent manner, the Ca 2ϩ binding properties of the isolated N-terminal portions of the MASPs were investigated. Preliminary experiments revealed that each three-domain MASP protein has an increased susceptibility to proteolysis by trypsin in the absence of Ca 2ϩ , confirming that Ca 2ϩ binds to and stabilizes each MASP. The Ca 2ϩ -binding properties of the three-domain MASP-1 and MASP-2 proteins were examined using trypsin digestion in the presence of increasing concentrations of Ca 2ϩ (Fig. 6). The apparent dissociation constants for Ca 2ϩ are 0.35 Ϯ 0.09 mM for MASP-1 and 0.19 Ϯ 0.06 mM for MASP-2, indicating that each MASP binds to Ca 2ϩ at physiological serum Ca 2ϩ concentrations (ϳ1 mM). The order of each binding curve is close to 1 (1.3 for MASP-1 and 0.9 for MASP-2), suggesting that each subunit of the homodimer binds one Ca 2ϩ .
Because the EGF-like domains in MASP-1 and MASP-2 contain the consensus sequences for Ca 2ϩ binding sites, it is likely that Ca 2ϩ is bound to these domains. This suggestion is consistent with the digestion pattern of the three-domain MASP-2 protein. On typsin digestion, a relatively stable proteolytic fragment is produced with an apparent molecular mass of 24 kDa under both reducing (Fig. 6) or nonreducing conditions. Following digestion by peptide:N-glycosidase F to remove the N-linked oligosaccharides, the apparent molecular mass is reduced to 20 kDa. Two N-terminal sequences are detected by Edman degradation when the proteolytic fragment is isolated under under nonreducing conditions: a sequence corresponding to the intact N terminus of the mature polypeptide and a sequence corresponding to residues Leu 189 -Arg 197 within the second CUB domain. Thus, this proteolytic fragment probably corresponds to the N-terminal CUB and EGF-like domains of In order to exclude the possibility that truncated MASP-1 and MASP-2 interact when bound to MBP, for example through formation of a MASP-1⅐MASP-2 heterodimer, each truncated MASP-2 protein was preincubated with the two-domain MASP-1 protein and loaded onto the MBP affinity column under conditions in which the truncated MASP-1 is not retained. If the MASP-1 and MASP-2 fragments formed a stable complex on binding to MBP, they would be expected to coelute in the bound fractions. However, each protein elutes independently from the column within the same fractions as when it is analyzed in isolation (Fig. 7). Thus, the truncated MASP-1 and MASP-2 proteins function independently and do not interact with each other even in the presence of MBP.

Inefficient Binding of Truncated MASPs to MBP Mutants Associated with Immunodeficiency
Previous studies using rat MBP have shown that defective complement fixation is likely to be a major factor in contributing to immunodeficiency caused by three naturally occurring variants of human serum MBP (6). In order to determine whether the defects in these mutants are caused by reduced ability to interact with MASP-1 and -2, a competition assay for MBP-MASP binding was developed. 35 S-Labeled MBP was incubated in wells coated with the three-domain truncated MASPs in the presence of increasing concentrations of either wild-type or mutant MBP. As shown in Fig. 8, wild-type MBP binds to truncated MASP-1 and MASP-2 with inhibition constants of 2.6 and 1.8 g/ml.
Because each MASP contains mainly complex N-linked oligosaccharides, it seemed unlikely that binding is a result of interactions between the carbohydrate recognition domains of MBP and N-linked sugars. However, to exclude this possibility, MASP binding by MBP was monitored in the presence of increasing concentrations of mannose or galactose. No specific inhibition of the MBP-MASP interaction was detected at concentrations of mannose up to 400 mM, levels more than 100-fold higher than the dissociation constant (25). Furthermore, similar binding properties by MBP are observed when the truncated MASPs were pretreated with peptide:N-glycosidase F to remove the N-linked oligosaccharides. Thus, MASP binding is not caused by the lectin activity of MBP and appears to reflect the natural interaction with the collagenous domain.
Each of the mutants associated with immunodeficiency has a lower affinity for both truncated MASP-1 and MASP-2, since higher concentrations of mutant protein are required to displace labeled MBP in the competition assays (Fig. 8). Because complement fixation is dependent on activation of MASPs bound to MBP, defective MASP binding probably leads directly to inefficient complement fixation in these mutants.
For each mutant MBP, the competition assays indicate that binding to the N-terminal domains of MASP-1 and to MASP-2 is affected to a similar extent. The broadly similar relative inhibition constants are summarized in Table II. Thus, it seems likely that MASP-1 and MASP-2 both bind to the same region of MBP, within the first part of the collagen-like domain where the mutations cause structural changes.
The effect of each of the mutations on MASP binding by MBP parallels the defects in complement fixation activities. The similar rank order of the changes suggests that defective MASP binding leads directly to aberrant complement fixation (Table II). The Gly 25 3 Asp and Arg 23 3 Cys mutations lead to a similar reduction in complement fixation activities, whereas the Gly 28 3 Glu mutation has a greater effect (6). Because the defect in the Arg 23 3 Cys is due almost entirely to a change in its oligomeric composition, the finding that MASP binding is defective suggests that the smaller oligomers that predominate in this mutant have a lower affinity for the MASPs.
The defect in complement fixation in the Gly 25 3 Asp and Gly 28 3 Glu mutants is also due in part to a reduction in the proportion of the higher oligomers, although this effect alone is not sufficient to account for the activity observed in either mutant (6). Because the oligomeric composition of each mutant is very similar, the observation that binding of the Gly 28 mutant to both truncated MASPs is more severely impaired indicates that this mutation must directly affect the MASP binding sites on MBP. Thus, these data suggest that the structural changes caused by mutations associated with immunodeficiency directly affect the binding sites for MASPs on MBP, which in turn leads to a reduced ability of the mutants to activate the complement cascade.

Interactions between MBP and MASPs-
The results reported here indicate that the isolated N-terminal domains of MASP-1 and MASP-2 are stable homodimers formed through interactions involving the first two modules. The first three domains of each protein are sufficient for independent, stable binding to MBP that occurs in a Ca 2ϩ -dependent manner. The biophysical properties of the N-terminal regions of MASPs suggest that the interactions leading to complement activation are likely to differ significantly from those between C1q, C1r, and C1s of the classical pathway.
In the classical pathway, components C1r and C1s form a C1r 2 C1s 2 tetramer that binds to C1q (26). The core of this complex is formed from the C-terminal complement control protein modules and the serine protease domain of C1r that interact to form homodimers (27). Binding to C1s occurs in a Ca 2ϩ -dependent manner through interactions involving the N-terminal CUB and EGF domains of each protease (28). The C1r 2 C1s 2 complex in turn binds to C1q through interactions involving the N-terminal domains of C1s (29). The finding that the isolated N-terminal domains of both MASP-1 and MASP-2 interact with MBP independently suggests that each MASP binds directly to MBP. Furthermore, because these domains of the MASPs are homodimers and do not interact with each other, they probably do not mediate the formation of MASP-1⅐MASP-2 heterocomplexes in the intact proteins.
The evolutionary relationship between the serine proteases is compatible with the suggestion that MASP-1 and MASP-2 may function differently from C1r and C1s. Analysis of the gene organization of MASPs together with their chromosomal locations suggests that their precursors probably diverged relatively early compared with those of C1r and C1s, before the emergence of primitive vertebrates (30,31). The C1r and C1s genes are more closely related to the MASP-2 gene and are likely to have arisen as a result of a gene duplication event before the emergence of cartilagenous fish.
The binding studies indicate that the isolated N-terminal two domains of MASP-1 interact with MBP relatively weakly compared with the N-terminal three domains. It is unlikely that this fragment is misfolded, because it is a stable homodimer, like its three-domain counterpart. The difference in affinities of the two truncations suggests that the second CUB domain may be important for stable MBP binding by MASP-1. This domain may either form part of the binding site for MBP or may stabilize interactions mediated through the N-terminal CUB and EGF-like domains.
Role of MAP19 -Because MAP19 consists essentially of the N-terminal two domains of MASP-2, it probably binds directly to MBP. It has been reported that MAP19 also binds to MASP-1 (9). The data described here suggest that binding does not occur through interactions involving the N-terminal three domains, because fragments of MASP-1 and MASP-2 comprising these regions do not bind to each other. MAP19 lacks the serine protease domain necessary for activation of downstream components of the complement cascade. However, it may modulate complement fixation, either by competing for MASP binding sites on MBP or by forming part of the MBP⅐MASP complex that triggers activation of complement.
Defective MASP Binding by MBP Containing Mutations Associated with Immunodeficiency-The data indicate that MASP binding is weakened in MBP containing any of the three mutations leading to human immunodeficiency. Thus, it seems likely that aberrant interactions between the variant MBPs and their associated proteases lead directly to inefficient complement fixation.
There are two distinct mechanisms that contribute to the reduced complement fixation activities of mutant MBPs (6). Defective oligomerization leads to an increase in the proportion of monomers and dimers and a corresponding decrease in the amounts of the larger oligomers. The smaller oligomers are less efficient at fixing complement. Structural changes to the Nterminal domains also lead to defective complement activation. These changes occur as a direct consequence of the mutations within the collagenous domain. Both mechanisms probably lead to inefficient complement activation through aberrant interactions with MASPs. Defective complement fixation in the Arg 23 3 Cys mutant is due almost entirely to a decrease in the proportion of the larger oligomers. Because MASP binding is disrupted in this mutant, the smaller MBP oligomers are probably less efficient at binding to the MASPs. The defect in the Gly 25 3 Asp mutant is also in part due to a change in the distribution of oligomers. However, this effect alone is insufficient to account for the loss in activity, suggesting that disrup- tion to the MASP-binding site within the collagenous domain contributes to defective complement activation. The structural defects of the Gly 28 3 Glu mutant are similar to those of the Gly 25 3 Asp mutant. However, because both complement fixation and MASP binding are further disrupted, it is likely that this mutation leads to a more specific change within the MASPbinding site.
The binding assays indicate that each mutation associated with MBP deficiency leads to defective MASP binding by MBP (Table II). However, the decrease in affinity is always smaller than the overall reduction in complement fixing activity, suggesting that the defects are amplified by the proteolytic cascade. One mechanism to explain this amplification would be if two MASP dimers are required to initiate complement fixation. Although the downstream molecular events triggered by the activated MASPs are poorly understood, it is possible that MASP-1 and MASP-2 function cooperatively to fix complement. This effect could, but would not necessarily have to, occur through a direct interaction between MASP-1 and MASP-2.
Binding analysis indicates that MBP forms relatively stable complexes with the MASP fragments. Because the concentration of 35 S-labeled MBP used in the competition assays is well below that required for saturation of binding, the K I values are probably similar to the overall binding constants. At typical serum concentrations of human MBP (ϳ1 g/ml; Ref. 2) and MASP-1 (ϳ5 g/ml; Ref. 22), it is likely that MBP circulates largely as a complex with its associated proteases. Defective MASP binding by the variant MBPs may lead directly to defective complement activation through a decrease in the concentration of circulating MBP⅐MASP complex. However, the relatively large changes to complement fixation activities suggest that activation of MASPs by the mutant MBPs is probably itself inefficient, as a result of the structural perturbations caused by the mutations.