A Conserved Element in the Serine Protease Domain of Complement Factor B*

Factor B and C2 are serine proteases that carry the catalytic sites of the complement C3 and C5 convertases. Their protease domains are activated by conformational changes that occur during convertase assembly and are deactivated upon convertase dissociation. Factor B and C2 share an 8-amino acid conserved sequence near their serine protease termini that is not seen in other serine proteases. To determine its importance, 24 factor B mutants were generated, each with a single amino acid substitution in this region. Whereas most mutants were functionally neutral, all five different substitutions of aspartic acid 715 and one phenylalanine 716 substitution severely reduced hemolytic activity. Several aspartic acid 715 mutants permitted the steps of convertase assembly including C3b-dependent factor D-mediated cleavage and activation of the high affinity C3b-binding site, but the resulting complexes did not cleave C3. Given that factor B and C2 share the same biological substrates and that part of the trypsin-like substrate specificity region is not apparent in either protein, we propose that the conserved region plays a critical role in the conformational regulation of the catalytic site and could offer a highly specific target for the therapeutic inhibition of complement.

Factor B and C2 are serine proteases that carry the catalytic sites of the complement C3 and C5 convertases. Their protease domains are activated by conformational changes that occur during convertase assembly and are deactivated upon convertase dissociation. Factor B and C2 share an 8-amino acid conserved sequence near their serine protease termini that is not seen in other serine proteases. To determine its importance, 24 factor B mutants were generated, each with a single amino acid substitution in this region. Whereas most mutants were functionally neutral, all five different substitutions of aspartic acid 715 and one phenylalanine 716 substitution severely reduced hemolytic activity. Several aspartic acid 715 mutants permitted the steps of convertase assembly including C3b-dependent factor D-mediated cleavage and activation of the high affinity C3b-binding site, but the resulting complexes did not cleave C3. Given that factor B and C2 share the same biological substrates and that part of the trypsin-like substrate specificity region is not apparent in either protein, we propose that the conserved region plays a critical role in the conformational regulation of the catalytic site and could offer a highly specific target for the therapeutic inhibition of complement.
The complement system consists of about 30 proteins that participate in both innate and acquired immunity (1,2). The complement response is initiated through the activation of several distinct pathways (3). Each pathway converges in the assembly of the C3 and C5 convertases, multicomponent serine proteases that mark targets for immune clearance or cell lysis (3) and direct antigen selection by B lymphocytes (4). Factor B is a zymogen that carries the catalytic site of the complement alternative pathway (AP) 1 convertases (5). Assembly of the AP C3 convertase begins with the association of factor B with C3b in the presence of Mg 2ϩ . This association permits factor B to be cleaved at a single site by factor D, which produces the Ba and Bb fragments. Ba dissociates from the complex, whereas Bb and Mg 2ϩ remain bound to C3b. C3bBb, the active AP C3 convertase, is capable of catalyzing C3 cleavage. Association of C3bBb with additional C3b yields the AP C5 convertase, C3bC3bBb, which cleaves C5. Both the C3 and C5 convertases can be further stabilized by association with properdin. Dissociation of Bb from either complex results in irreversible loss in C3b-binding capacity and C3 and C5 protease activity.
Factor B is a 90-kDa single-chain mosaic glycoprotein composed of three different types of protein modules (6). 1) The amino-terminal region, which constitutes most of Ba, features three complement control protein modules, found in tandem arrays in many complement proteins and frequently carrying binding sites for C3b and closely related C4b (7); 2) the middle region is a von Willebrand factor type A module that also occurs in several members of integrin family, often serving as a ligand-binding site and a metal-binding site (8); and 3) the carboxyl terminus is a serine protease (SP) domain similar to that of trypsin (9). C2, the complement classical pathway analog of factor B, is similar in structure to factor B (10).
Whereas the factor B and C2 SP domains closely resemble trypsin in primary structure, their regulation features an unusual interplay between the SP domain and the adjoining type A domain. Although both factor B and Bb are capable of cleaving small molecule substrate analogs (11), the capacity to cleave C3 is acquired only through C3 convertase assembly (to C3bBb) and is sustained by continuous association with C3b. Moreover, association with additional C3b (C3bC3bBb) is required for C5 convertase activity. This general activation scheme, which also occurs in the case of C2, is in contrast with that utilized by typical serine proteases in which the amino terminus generated by cleavage of the zymogen translocates to the protein interior, rearranging the specificity pocket (12,13). The highly conserved amino-terminal sequence that mediates the typical activation mechanism does not occur in factor B or C2 (see discussion in Ref. 5).
Electron microscopic examination of Bb and C3bBb have indicated that Bb is a dumbbell-shaped structure, with only one lobe binding to C3b (14,15). Mutational studies have strongly implicated the type A domain in binding to C3b (16,17), and genetic or chemical modification of the C2 (18,19) and factor B 2 type A region increases convertase stability. Thus it is likely in C3bBb that the factor B type A domain is in contact with C3b whereas the SP domain is not, and SP activation is regulated by a conformational signal that propagates from the type A ligand-binding site to the SP catalytic site. In this report, a highly conserved sequence in the factor B and C2 SP catalytic region, not observed in other trypsin-like SP, is partially characterized, and its possible role in interdomain signaling is discussed.

EXPERIMENTAL PROCEDURES
Expression of Mutant and Wild Type Recombinant Factor B-As described previously (16), COS-7 cells were transfected with factor B A14 cDNA cloned in expression vector pSG5 (Stratagene). Cells were biosynthetically labeled, and the supernatants were analyzed by immu-* 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 1 The abbreviations used are: AP, alternative pathway; SP, serine protease; PAGE, polyacrylamide gel electrophoresis; ELISA, enzymelinked immunosorbent assay; mAb, monoclonal antibody; PBS, phosphate-buffered saline; CVF, cobra venom factor. noprecipitation followed by SDS-PAGE and autoradiography. Recombinant factor B was then quantitated by ELISA and assayed for hemolytic capacity. Two methods (double-take double-stranded mutagenesis, Stratagene) and transformer site-directed mutagenesis (Ref. 20, CLON-TECH) were used to generate mutant cDNA sequences.
C3b-binding and Cobra Venom Factor-binding Assays-Supernatants derived from transfections in serum-free media were dialyzed in PB (11 mM Na 2 HPO 4 , 1.8 mM NaH 2 PO 4 , pH 7.4) supplemented with 25 mM NaCl. Microtiter wells were C3b-coated as described previously (16). Mixtures of factor B (50 ng/ml), properdin (1 g/ml), and factor D (25 ng/ml) were prepared in PB supplemented with 75 mM NaCl, 10 mM MgCl 2 , 4% bovine serum albumin, and 0.05% Tween 20, incubated in C3b-coated microtiter wells for up to 30 min, washed, and detected by ELISA (16). A similar procedure was followed to assay cobra venom factor (CVF)-binding except that microtiter wells were coated with CVF at 10 g/ml, and properdin was excluded from the reactions (17). Complement proteins were purchased from Advanced Research Technologies, La Jolla, CA.
C3b-dependent Factor B Cleavage-Factor B wild type and mutant forms (500 ng/ml) were incubated with factor D (200 ng/ml), and C3b (2000 ng/ml) in PB, 25 mM NaCl and 10 mM MgCl 2 for 30 min at 37°C. Aliquots of 0.025 ml were mixed with SDS/sucrose nonreducing loading buffer, incubated at 70°C for 10 min, cooled, and loaded on a NU-PAGE polyacrylamide gel (Novex). The gel was run for 55 min and transferred using the recommendations of the manufacturer. The resulting filter was blocked O/N with 5% Carnation dry milk in TBST (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, supplemented with 0.1% Tween 20) and treated successively with a 1:10,000 dilution of mouse anti-human Bb mAb (Quidel) in TBST, and a 1:5,000 dilution of donkey anti-mouse polyclonal antibody (Jackson Immunoresearch Laboratories, West Grove, PA) in TBST. Detection was performed with the Super Signal Chemiluminescence System (Pierce).
Biotinylated C3-17 g of freshly dissolved sulfosuccinimidyl-6-(biotinamido)hexanoate (EZ-link sulfo-NHS-LC-Biotin; Pierce) was mixed with 125 g of C3 in 150 l of water and incubated for 5 min at 25°C. The result was treated with 250 l of PBS, and the mixture was spun in a Millipore 500-l concentrator with a Biomax 10K NMWL membrane. After two washes with 500 l of PBS, the biotinylated C3 product was eluted with 100 l of PBS and stored in 5-l aliquots at Ϫ70°C.
C3 Convertase Assay-Factor B wild type and mutant forms (0.5 g/ml) were incubated with factor D (0.2 g/ml), CVF (2 g/ml), and biotinylated C3 (0.043 g/ml) in PB, 25 mM NaCl, and 10 mM MgCl 2 (total volume of 100 l) for 3 h at 37°C. Aliquots of 25 l were treated with 2.5 l of 0.5 M dithiothreitol and heated for 10 min at 70°C. Samples were run alongside standard molecular markers for 75 min on a 10% NU-PAGE gel. Filter transfer and blocking was performed as described above. Blocked filters were incubated in 20 ml of TBS (50 mM Tris-HCl, pH 7.5, 150 mM NaCl) supplemented with 0.1% Tween 20 and 17 l of ExtrAvidin peroxidase conjugate (Sigma) for 1 h at 25°C and developed as above. The position of C3b was determined with biotinylated C3b, a gift of M. Kathryn Liszewski.
Anti-Bb ELISA-A mouse anti-human Bb-specific mAb was used to screen mutations of the Bb region. 500 ng of factor B was cleaved by factor D in the presence of C3b (as above). The products were assessed by Western blot, and the detection of a mouse anti-human Bb-specific mAb antigenic site was performed with the Bb fragment EIA kit (Quidel, San Diego, CA). Recognition values were calculated as the percent of the equivalent wild type recombinant Bb value.
Phage Display Analysis of the Bb-specific mAb Epitope-The Ph.D.-12 M13 peptide library (New England Biolabs), which carries 1.9 ϫ 10 9 independent linear 12-mer clones, was screened using the directions of the manufacturer. Microtiter wells were coated with the mouse anti-human Bb specific mAb (Quidel) that functioned in the anti-Bb EIA kit (see above) (100 mg/ml in PBS (11 mM Na 2 HPO 4 , 1.8 mM NaH 2 PO 4 , pH 7.4, 145 mM NaCl)) overnight at 4°C and washed. Wells were incubated with 0.01 ml (1.4 ϫ 10 13 plaque-forming unit) phage, washed once in 0.1% Tween 20 in TBS, and washed twice in 0.5% Tween 20 in PBS. Phage were eluted by incubation in 1 mg/ml bovine serum albumin, 0.2 M glycine-HCl, pH 2.2, for 15 min. Eluate was removed and neutralized with 15 ml of 1 M Tris, pH 9.0. Phage were amplified, the resulting preparations titered, and 1 ϫ 10 11 plaque forming units were incubated in the microtiter well as above. After the enrichment procedure was performed a total of three times, phage clones were picked, DNA isolated, and the 12-mer insert sequenced. All seven 12-mer regions identified yielded the identical nucleotide sequence, CAT-ATG-CGT-CTT-TCT-CAG-TGG-CCT-CTT-TTG-AAG-C-CT, which was translated into the peptide sequence HMRLSQW-PLLKP. Computer analysis employing the GeneWorks program (Intel-liGenetics, Mountain View, CA) revealed only one homologous site in the human factor B sequence.

RESULTS
In an effort to understand the structural basis for the regulation of the factor B and C2 SP domains, we compared the amino acid sequences of factor B and C2 to those of other trypsin-like serine proteases. In the case of trypsin and chymotrypsin, important determinants of substrate specificity are clustered near the COOH terminus (21,22). In factor B and C2, at least one of the typical trypsin-like specificity determinants is absent (Fig. 1), and in its place a common sequence of eight identical amino acids (RDFHINLF) was observed. That conserved factor B/C2 element was not found in any other serine protease examined (9,21).
Based on its level of conservation, we hypothesized that RDFHINLF plays an important role in the function of the factor B/C2 serine protease. To test this hypothesis, each of the eight conserved residues of human factor B was replaced by site-directed mutagenesis, and the resulting mutant proteins were examined (Fig. 2). Whereas substitutions were permitted at most sites, the substitution of Asp-715 severely reduced hemolytic activity in all five cases. Activity was also affected in two of three Phe-716 substitutions: F716Y had normal hemolytic activity levels, F716W was reduced to 23% that of wild type, and hemolytic activity of F716A was nearly abrogated.
Three Asp-715 substitutions were selected for further examination: D715E was chosen because it retained the wild type negative charge, D715N because it retained some of the wild type side-chain structure, and D715A because it removed most of the wild type side-chain. Factor B forms were treated with factor D and C3b in the presence of Mg 2ϩ in a fluid phase reaction, and the results were analyzed by Western blot (Fig.  3A). All of the mutants underwent C3b-dependent cleavage, the first indication that the earliest steps of convertase assembly, the low affinity interaction between factor B and C3b and subsequent recognition and cleavage by factor D, were proceeding normally. Mutant F716A, which lacked hemolytic activity, was also examined by this method, and C3b-dependent cleavage was also observed (Fig. 3B) The cleavage of factor B by factor D results in a higher affinity between ligand and the Bb fragment. This is indicated by a factor D-dependent increase in factor B binding to immobilized ligand (17). Wild type factor B bound C3b in the presence of properdin, which stabilizes C3bBb (3), with binding increasing severalfold when factor D was also supplied (Fig. 4). Asp-715 mutants also exhibited factor D-dependent increases in binding activity (Fig. 4), a strong indication that both factor D cleavage and high affinity bond formation occurred normally. C3b-binding was not detected with F716A (Fig. 4).
Once factor D-mediated cleavage occurs and the high-affinity complex is established, assembly of the C3 convertase is complete, and the serine protease domain is capable of catalyzing the cleavage of C3. To determine whether the Asp-715 mutants are capable of C3 cleavage, a fluid phase assay was employed. CVF, an analog of human C3b, was used as ligand because the resulting C3 convertases are 50-fold more stable than those that incorporate C3b (23), and the 715D mutants undergo factor D-dependent CVF-binding at the same rate as wild type factor B (Fig. 5). Biotinylated C3 was used as substrate, and the C3b product was resolved from C3 by gel electrophoresis. As seen in Fig. 6, C3b was formed with wild type recombinant factor B but not with the Asp-715 factor B mutants.
The activation of the factor B SP domain likely involves conformational changes precipitated by factor D mediated cleavage, Ba release, and/or the transition to high affinity ligand binding. At least one mAb is available that recognizes Bb but not factor B zymogen. In an effort to map the anti-Bb neo-antigenic site, an M13 peptide display library was utilized. The library was suspended in microtiter wells coated with antibody, and wells were washed. Bound phage were eluted and amplified, the selection procedure was repeated two more times, and individual isolates were sequenced. All isolates (seven of seven) yielded the identical sequence, HMRLSQW-PLLKP, which is similar to a single factor B site overlapping the SP conserved region (Fig. 7A).
Several Asp-715 and Phe-716 substitutions were screened for recognition by the Bb-specific mAb. Mutant proteins were cleaved by factor D, Bb generation was assessed by Western blot, and Bb samples were analyzed using a commercially available Bb-specific ELISA that utilizes the anti-human Bb mAb. Of six mutant Bb forms, the five hemolytically defective mutants were recognized at lower efficiency than wild type, and the one hemolytically normal mutant was recognized at the wild type level (Fig. 7B). Most affected was Bb F716A , which was recognized about 25% as effectively as wild type Bb. DISCUSSION Factor B and C2 share an 8-amino acid conserved sequence (RDFHINLF) near the SP carboxyl terminus that is not seen in other trypsin-like serine proteases. A panel of factor B mutants was assembled, each with a single amino acid substitution in this region. Whereas most substitutions were functionally neutral, all five different substitutions of Asp-715 as well as one Phe-716 substitution exhibited severely reduced hemolytic activity.
The initial association between factor B and C3b is a low affinity interaction. The C3b-dependent cleavage of factor B by factor D produces a high affinity ligand-binding site in the type A domain. All of the Asp-715 mutations examined permitted C3b-dependent factor D cleavage. In vitro binding studies indicated that the high affinity ligand-binding conformation occurred in at least several of these cases. High affinity binding was dependent on factor D activity, and the formation of high affinity complexes proceeded at rates similar to those of wild type factor B. Nevertheless, whereas C3 convertases appear to assemble, C3 cleavage was not detected with any of the Asp-715 substitutions examined. This shortcoming is likely the primary defect of the Asp-715 mutations. It is not clear from the present study whether the Asp-715 mutations would also interfere with C5 cleavage directly because assembly of the C5 convertase depends on a functioning C3 convertase. The F716A mutant also permitted C3b-dependent cleavage, but high affinity C3b-binding was abolished. Thus, F716A is defective in convertase assembly.
The conserved SP sequence, RDFHINLF, distinguishes the human complement proteases factor B and C2 from other serine proteases. It is encoded by a single exon (24) that encodes the translational stop codon as well as 214-SWG, found in the S1 binding pocket of trypsin and chymotrypsin (Fig. 1). Several non-human factor B and C2 forms were examined for the presence of the conserved sequence. Remarkably, the sequence was found in all reported factor B/C2 forms of mammals, amphibians, and fish, in several cases without substitution (Fig. 8). Preserved since the evolution of bony fish, RDFHINLF is likely involved in a fundamental role that has little tolerance for subtle structural changes, especially in the first three positions. Given that factor B and C2 share the same biological substrates, C3 and C5, and that parts of the trypsin-like S1 pocket and/or its neighboring loop are not apparent in factor B and C2 amino acid sequences, we propose that the conserved region has a critical role in substrate specificity. Residue Asp-715 could be directly involved in substrate recognition.
Residue Phe-716 appears to be of structural significance: Its replacement with alanine inhibits mAb recognition at a proximal site and high-affinity ligand binding at a distal site. Interestingly, substitution of Phe-716 with aromatic amino acids Tyr and Trp permitted at least partial functional activity, which suggests the possibility of its participation in cationinteractions (25).
Although it is unclear why factor B/C2 might have evolved with a different specificity determinant, this novel element may facilitate the unusual conformational regulation of SP function. By this model, RDFHINLF is a key part of a conformationally regulated SP specificity determinant that is controlled by ligand interactions in the type A domain. Its absence in lamprey factor (Fig. 8) suggests its approximate evolutionary origins and indicates that it is likely unique to factor B and C2. In any case, the conserved region could offer a highly specific target for the therapeutic inhibition of complement.
Several SP mutations reduced Bb-specific mAb recognition, with the most extreme effects observed with the F716A mutation. The corresponding neoepitope, partially characterized using phage display methods, overlaps the SP conserved sequence. Two simple interpretations are possible. 1) Expression of the mAb recognition site and the possible activation of the conserved sequence requires conformational changes that occur upon factor B cleavage and the release of the Ba region; and 2) Ba could physically block an already existing mAb recognition site and possibly the site of the conserved sequence. Given the likely close proximity between the mAb neoepitope, the conserved sequence, and the SP active site, either explanation could also account for the impact that factor B cleavage and Ba release has on SP catalytic activity.
Recently, Tuckwell et al. (17) described a factor B type A domain mutant that is unable to make or sustain the high affinity Bb conformation, although capable of ligand-dependent factor D-mediated cleavage. That mutant, Ch1, is a sequence substitution of a putative Mg 2ϩ -binding loop located distal to the type A/SP connection (26,27). Similar functional behavior is reported here in the case of the SP mutant F716A. The cleavage of C3bB by factor D precipitates a conformational change from a low affinity ligand-binding catalytically inactive state to a high affinity ligand-binding catalytically active state. Both the Ch1 and F716A mutations appear to block this transition. We propose that these two physically distant mutants indicate an inter-domain signaling mechanism that alters the structure of the SP specificity determinants in response to the ligand-binding state of the type A domain.