A Corresponding Tyrosine Residue in the C2/Factor B Type A Domain Is a Hot Spot in the Decay Acceleration of the Complement C3 Convertases*

The cleavage of C3 by the C3 convertases (C3bBb and C4b2a) determines whether complement activation proceeds. Dissociation (decay acceleration) of these central enzymes by the regulators decay-accelerating factor (DAF), complement receptor 1 (CR1), factor H, and C4-binding protein (C4BP) controls their function. In a previous investigation, we obtained evidence implicating the α4/5 region of the type A domain of Bb (especially Tyr338) in decay acceleration of C3bBb and proposed this site as a potential interaction point with DAF and long homologous repeat A of CR1. Because portions of only two DAF complement control protein domains (CCPs), CCP2 and CCP3, are necessary to mediate its decay of the CP C3 convertase (as opposed to portions of at least three CCPs in all other cases, e.g. CCPs 1–3 of CR1), DAF/C4b2a provides the simplest structural model for this reaction. Therefore, we examined the importance of the C2 α4/5 site on decay acceleration of C4b2a. Functional C4b2a complexes made with the C2 Y327A mutant, the C2 homolog to factor B Y338A, were highly resistant to DAF, C4BP, and long homologous repeat A of CR1, whereas C2 substitutions in two nearby residues (N324A and L328A) resulted in partial resistance. Our new findings indicate that the α4/5 region of C2a is critical to decay acceleration mediated by DAF, C4BP, and CR1 and suggest that decay acceleration of C4b2a and C3bBb requires interaction of the convertase α4/5 region with a CCP2/CCP3 site of DAF or structurally homologous sites of CR1 and C4BP.

The complement system consists of about 30 proteins that play critical roles in both innate and adaptive immunity. Three different activation pathways initiate the complement response (1): the classical pathway (CP), 1 the alternative pathway (AP), and the lectin pathway. Although each responds to different activators, all three pathways converge at the assembly of the C3 convertases. These two-component serine proteases (C4b2a and C3bBb) cleave the serum protein C3 at a single site, forming C3b and C3a activation fragments. Nearly all of the biological consequences associated with complement depend on this enzymatic cleavage. C3b binds covalently to and opsonizes activating targets, which marks them for lysis and/or immune clearance (1) and primes them for the production of the high affinity antibody (2)(3)(4). C3a is an anaphylactic agent that focuses inflammatory reactions around foreign substances by inducing local vasodilation, the influx of leukocytes, the upregulation of surface receptors, and the release of inflammatory mediators (5).
There are two structurally different C3 convertases (1): 1) The classical and lectin pathway convertase, C4b2a, traditionally termed the CP convertase and 2) the AP C3 convertase, C3bBb. Each is formed first through the association of C4b or C3b with a zymogen (C2 or factor B) in the presence of Mg 2ϩ . C2 and factor B are homologous proteins composed of three amino-terminal globular domains (complement control protein domains (CCPs)) followed by a type A domain and a serine protease domain. In the classical pathway, C4bC2 is cleaved by C1s at a single site in C2 between the CCPs and the type A domain. Then the amino-terminal portion of C2 (C2b) is released, leaving the active C3 convertase, composed of a complex of C4b, C2a, and a single Mg 2ϩ ion. The AP convertase is generated in a similar assembly process involving the binding of factor B to C3b followed by the cleavage of C3bB to C3bBb by factor D. Each C3 convertase thereby consists of a noncatalytic subunit (C4b or C3b) and a catalytic (enzymatic) subunit (C2a or Bb).
Convertase activity is controlled by the regulators of complement activation (RCA) proteins (6, 7) that irreversibly dissociate (decay accelerate) the convertase subunits or alternatively serve as cofactors for the proteolytic cleavage of C3b and C4b. Among the RCA proteins that mediate convertase dissociation, decay-accelerating factor (DAF), complement receptor 1 (CR1), and C4-binding protein (C4BP) dissociate the CP convertase, whereas DAF, CR1, and factor H dissociate the AP convertase. The purpose of this study was to provide information relevant to the molecular mechanisms that underlie decay accelerating activity.
To better understand decay accelerating activity, it is necessary to define specific regulator-convertase interactions. In a previous investigation we mapped sites in the type A domain of Bb that are involved in decay acceleration of the AP C3 convertase (C3bBb) (8). Like Bb, C2a is composed of an aminoterminal von Willibrand factor type A domain followed by a carboxyl-terminal serine protease domain (1). Several mutations in Bb centered at the adjacent ␣ helices 4 and 5, especially substitutions of Tyr 338 , caused C3bBb to exhibit marked resist-ance to DAF and CR1, and others at the ␣1 helix and its adjacent loops, especially D254G in the C3b-binding region of Bb, caused substantial resistance to DAF, CR1, and factor H. We hypothesized that the ␣4/5 region of Bb interacts directly with DAF and CR1, whereas the ␣1 helix and its adjacent loops affect the stability and dynamics of the Bb/C3b interface.
The RCA proteins are composed of tandem globular domains termed complement control protein repeats or short consensus repeats (6,7). Only a portion of each protein is necessary for decay acceleration. Of the four CCPs of DAF, CCPs 2-4 are required for the decay acceleration of C3bBb, whereas only CCPs 2 and 3 are necessary to mediate decay of C4b2a (9,10). In both cases about 10 amino acids, located in DAF CCPs 2 and 3 and the inter-CCP segment, appear critical (11). For CR1, CCPs 1-3 (of a total of 30 CCPs) are required for decay acceleration of both C3bBb and C4b2a (12), with a critical site in CCPs 1 and 2 structurally similar to that in DAF CCP2 and 3. For C4BP, CCPs 1-3 (of 8 CCPs) are required for decay acceleration of C4b2a (13).
Based on our previous mutagenesis work with DAF (8, 11), we hypothesized that the ␣4/5 region of Bb interacts with amino acids of DAF in CCP2 and CCP3. However, the ability of DAF to accelerate the decay of C3bBb also involves certain residues in its CCP4, so other models were feasible. Therefore, in the present study we focused on the decay acceleration of C4b2a, where only CCP2 and CCP3 are involved (9,10). We examined the type A domain of C2a, identified C2a residues homologous to those Bb residues critical for decay acceleration of C3bBb, replaced the corresponding C2 amino acids by site-directed mutagenesis, and determined the effect of these mutations on the decay acceleration of C4b2a by DAF, CR1, and C4BP.
We found that the ␣4/5 region of C2a is critical for decay acceleration mediated by DAF, C4BP, and CR1, findings that strongly support the possibility that the ␣4/5 region of both C2 and factor B interact with the same elements of DAF CCP2 and CCP3. These new observations and our previous studies support the hypothesis that in decay acceleration of both C4b2a and C3bBb, the convertase ␣4/5 region interacts with a CCP2/ CCP3 site of DAF or the structurally homologous sites of CR1 or C4BP. The results also indicate that C2 Tyr 327 and homologous factor B Tyr 338 are functional "hot spots" in the convertase/decay acceleration reaction.

EXPERIMENTAL PROCEDURES
Production of Mutant C2 Proteins-The plasmid containing wild type human C2 cDNA, pcDNA3C2SacR (14) originally from Dr. John Volanakis (University of Alabama at Birmingham) was obtained from Suzanne Bohlson (University of California at Irvine) who removed the C2 coding sequence by BamH1/XbaI digestion, cloned it into the pfast-bac1 vector, and introduced a C241A mutation by PCR to avoid in vitro cross-linking. Previous analyses of the C241A protein deriving from this cDNA showed that it is functionally equivalent to wild type C2 (15). We excised the C2 sequence from pfastbac1C2C241A with EcoRI and cloned it into pSG5 (Stratagene, La Jolla, CA). Mutations were introduced into the C2 clone using the QuikChange site-directed mutagenesis method (Stratagene). Consequently, wild type controls and all mutants carried the C241A substitution.
Mutant and wild type C2 proteins were prepared by transient expression of pSG5 carrying the respective C2 nucleotide sequences in human 293T kidney cells using serum-free medium (16). C2 protein in cell supernatants was identified by Western blot as follows. Supernatants, concentrated 10:1 with Millipore 500-l concentrators, were incubated for 30 min at 37°C in an equal volume of low salt buffer (11 mM Na 2 H 2 PO 4 , 1.8 mM NaH 2 PO 4 , pH 7.4, 25 mM NaCl) containing 10 mM MgCl 2 . 25-l aliquots of the concentrate mixed with NU-PAGE loading buffer were incubated at 70°C for 10 min and loaded on NU-PAGE polyacrylamide gel (Novex), and the separated proteins were transferred to a nitrocellulose membrane (0.45 m; Bio-Rad) that was blocked overnight with 5% Carnation dry milk in TBST (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, supplemented with 0.1% Tween 20). The blot was developed by sequential treatment with a 1:10,000 dilution of goat anti-human C2 monoclonal antibody (Advanced Research Technologies, San Diego, CA) and a 1:30,000 dilution of rabbit anti-goat IgG polyclonal antibody (Jackson ImmunoResearch Laboratories, West Grove, PA) in TBST, and Super Signal Chemiluminescence reagent (Pierce). Native C2 was used as a standard. C2 proteins were quantitated by enzyme-linked immunosorbent assay as follows. Microtiter wells were coated overnight at 4°C with 1:2000 dilution of goat anti-human C2 polyclonal antibody (Advanced Research Technologies, San Diego, CA) in PBS, blocked 1 h at 37°C with 1% bovine serum albumin, 0.1% Tween 20 in PBS, and washed with PBS containing 0.05% Tween 20. Native C2 standards, recombinant proteins and negative controls diluted in PBS containing 4% bovine serum albumin and 0.1% Tween 20 were added to the wells, and the plates were incubated for 2 h at 37°C. After washing, the wells were treated for 1 h at 37°C with 100 l of 200 ng/ml anti-human C2b monoclonal antibody 3A3.3 (17), a gift of Dr. Yuanyuan Xu (University of Alabama at Birmingham). The wells were washed and secondarily incubated for 1 h at 37°C with 1:3000 dilution of peroxidase-conjugated donkey anti-mouse IgG polyclonal antibody (Jackson ImmunoResearch Laboratories, West Grove, PA; catalog number 715-035-150). Color was developed using a 0.2% orthophenylenediamine and H 2 O 2 solution in citrate/phosphate buffer at pH 6.5, and absorbance at 414 nm was read on a microplate reader. Concentrations of recombinant C2 samples were calculated by comparison with standard curves of native C2.
A fluid phase assay was used to monitor the cleavage of recombinant and native C2 by C1s. Twenty-five l of concentrated cell supernatant was incubated for 30 min at 37°C in a total volume of 100 l of low salt buffer (11 mM Na 2 H 2 PO 4 , 1.8 mM NaH 2 PO 4 , pH 7.4, 25 mM NaCl) containing 10 mM MgCl 2 and 250 ng/ml C1s, and cleavage was visualized by Western blotting as above.
Regulators-DAF consisted of a soluble derivative containing its four CCP active regions with a single amino acid substitution (N61Q) that removed the N-linked oligosaccharide situated between CCPs 1 and 2 (11,18). CR1-A, kindly supplied by Richard Hauhart and Malgorzata Krych-Goldberg (Washington University, St. Louis, MO), was a soluble CR1 derivative, consisting of CCP 1-7 as previously described (19). Purified human C4BP was purchased from Advanced Research Technologies.
Assays of DAF-, CR1-A-, and C4BP-mediated Decay Acceleration-Classical pathway C3 convertase decay accelerating activity was determined using a hemolytic C4b2a decay assay as previously described (11). In this assay, hemolysin-sensitized sheep erythrocytes (E sh A; 1 ϫ 10 7 in 100 l of isoionic gelatin veronal buffer with Ca 2ϩ and Mg 2ϩ (DGVB 2ϩ , (11)) were incubated with ϳ20 site-forming units of human C1 (15 min, 30°C). The cells were pelleted, resuspended in 100 l of DGVB 2ϩ , and incubated for 20 min at 30°C with ϳ10 site-forming units of human C4 (Quidel, Mountain View, CA). The cells were again pelleted and resuspended in 100 l of DGVB 2ϩ and then treated for 5 min at 30°C with an equal volume of DGVB 2ϩ containing an amount of recombinant (above) or native human C2 (Advanced Research Technologies) predetermined to yield ϳ1 C4b2a site/cell after subsequent decay for 15 min at 30°C. The resulting cells (E sh AC4b2a) were then washed and incubated for 15 min at 30°C with DAF, CR1-A, or C4BP (Advanced Research Technologies) diluted in DGVB 2ϩ to a total volume of 200 l of DGVB 2ϩ or with 200 l DGVB 2ϩ alone to allow C4b2a decay. Following the decay, 1.3 ml of guinea pig serum (C3-9) in isotonic gelatin veronal buffer containing 10 mM EDTA (GVB-E) (11) was added, and the mixture was incubated for 1 h at 37°C to develop lysis. Unlysed cells were pelleted, the A 412 value of the supernatant was measured, and residual C4b2a sites were calculated.
C4b2a half-life was determined at 30°C by performing the decay step (above) in DGVB 2ϩ buffer alone for various time intervals. The values obtained for the C241A positive control (19.6 Ϯ 2.1 min) were comparable with native C2 protein (Advanced Research Technologies) and corresponded well to previous values obtained for this same mutant (15.7 min, (15)).
Protein Modeling-C2 (20) and factor B (21) amino acid sequences were aligned using the program SIM (Expasy) (22). The C2 wild type and mutant type A domain models were constructed using the spatial coordinates of the iC3b receptor (CR3) type A domain (23), Protein Data Bank entry 1IDO using the automated program MODELLER (24), with the InsightII software platform (Accelrys, San Diego, CA) using the primary sequence alignment shown in Ref. 25. The factor B type A domain model was previously constructed in a similar fashion (8). Structural alignments of type A domain models were performed with the HOMOLOGY module (Accelrys). In the experiments depicted in Fig.  6, 25 models were constructed for the wild type, and each Tyr 327 mutant C2 protein and amino acids within a 2.5-Angstrom radius of position 327 were determined for each model.

Construction of a Panel of C2 Type A Mutants-As indicated
(in the Introduction): the ␣4/5 region, especially the Tyr 338 residue, and ␣1 and its adjoining ␣1/2 loop, especially the Asp 254 residue, were identified as important for decay acceleration of C3bBb. To determine whether corresponding regions are important in the decay acceleration of C4b2a, we aligned the primary sequences of the factor B and C2 type A domain and connecting linker (Fig. 1). We found conservation of the Tyr 338 homolog, i.e. C2 residue Tyr 327 , and although the analog of factor B Asp 254 is glutamine (C2 residue Gln 243 ), we found that its position remains within a conserved group of five putative Mg 2ϩ coordination residues (23).
Based on the above alignment, we constructed a panel of eight C2 type A domain mutants (Table I) corresponding to mutations in factor B previously shown to affect decay acceleration (8). The recombinant proteins were transiently expressed in human 293T kidney cells and were quantitated as described under "Experimental Procedures." Each protein was examined by Western blotting, assayed for its sensitivity to C1s cleavage, and analyzed for its hemolytic capacity. All of the mutants, except as noted below, assembled similarly to wild type C2 and retained hemolytic activity. Their activities require functioning C4b-binding sites in both the C2a and C2b regions, a Mg 2ϩbinding site located in the type A domain of C2a, an intact C1s cleavage site located at the junction between C2b and C2a, and a working C3/C5 substrate recognition site as well as a functional serine protease catalytic site in the serine protease domain of C2a (Fig. 1). The fact that these sites are distributed throughout C2 argues that the mutations made in each case induced only relatively local structural changes.
Effects of C2 Mutations on Decay Acceleration-C4b2a was assembled with wild type or mutant C2, and the resulting convertases were compared for their sensitivity to DAF, CR1-A, and C4BP (Figs. 2 and 3). Consistent with our previous studies on C3bBb (8), the ␣4 mutation Y327A dramatically reduced the sensitivity of C4b2a to DAF-and CR1-A-mediated decay acceleration to less than 1% of that of wild type C4b2a. Additionally as found with C3bBb, alanine substitutions near Tyr 327 (N324A and L328A) also reduced decay acceleration but to a lesser degree. The Q243G substitution, located in the divalent cation-binding cleft (23) corresponding to the putative C4bbinding region (25), likewise markedly reduced C4b2a sensitivity to DAF (2% of wild type) and moderately reduced C4b2a sensitivity to CR1-A (55% of wild type). Decay acceleration of C4b2a mediated by C4BP was affected by the C2 mutations in much the same way as that by DAF and CR1-A (Fig. 3).
To determine whether the Y327A and Q243G mutations affected intrinsic convertase function, we examined convertase assembly and spontaneous decay. As seen in Fig. 4, both mutant proteins were readily cleaved by C1s. Assembly of C4b2a with Y327A occurred at nearly wild type rate, whereas Q243G enhanced the assembly rate (data not shown). When compared with wild type C2 (19.6 Ϯ 2.1 min), the Y327A mutation had little if any effect on C4b2a half-life (14.9 Ϯ 4.9 min). Interestingly, Q243G increased C4b2a half-life (to 51.5 Ϯ 15.5 min), a result similar to that obtained with the factor B analog D254G (26). Thus, in the case of the CP C3 convertase, Y327A inhibited regulator-mediated dissociation but not spontaneous dissociation, whereas Q243G inhibited both regulator-mediated and spontaneous dissociation.
In contrast to findings for factor B, in which the conservative Y338F mutation reduced DAF sensitivity to 18% and CR1-A sensitivity to 20% (8), the homologous C2 conversion, Y327F, had no detectable effects on DAF sensitivity and only a very modest effect on CR1-A sensitivity (67% of wild type). Also contrary to their factor B counterparts (Table I and Discussion), neither the ␣1 mutation (K254A, E255A) nor the ␣7 mutation (D434A) had detectable effects on decay acceleration by any of the regulators Topology of the C2 and Factor B ␣4/5 Regions-To compare the ␣4/5 structures of the two proteins, we performed homology modeling (24) and constructed three-dimensional models of the wild type C2 and factor B type A domains using as a reference the coordinates of the type A domain of CR3 derived by x-ray crystallography (23). As seen in Fig. 5, analysis of these models indicated that C2 Tyr 327 and factor B Tyr 338 are each part of a cluster of polar and charged amino acids that reside on the ␣4 and ␣5 helices. As noted above, mutation of residue Asn 324 , part of the putative C2 cluster, as well as Leu 328 , adjacent to the cluster, reduced sensitivity to decay acceleration (Fig. 3). In our previous studies, we found that mutation of several residues of the putative factor B cluster (Gln 335 , Ser 339 , and Asp 382 ) (8) similarly led to reductions in decay acceleration.
Because the C2 and factor B type A models are approximations based on the crystallographic structure of the homologous type A domain from the ␣ subunit of CR3 (23), we reasoned that examination of a number of type A models of each key mutation might help clarify why Tyr 327 is critical to decay acceleration. We therefore constructed and compared 25 three-dimensional models of the wild type, Y327A, and Y327F C2 type A domains. In each case the van der Waals' contacts of the position 327 residue were determined.
In 74 of the 75 models, the position 327 side chain was in van der Waals' contact with at least one ␣5 side chain, although the precise ␣5 contact depended on the specific position 327 residue (Fig. 6A). Contact between position 327 and ␣5 residue Ile 378 occurred in 92% of the wild type and Y327F models but only 44% of the Y327A models, and contact between position 327 and the ␣5 residue Glu 374 occurred in 40 and 24% of the wild type and Y327F models but only 4% of the Y327A models (Fig.  6, B and C). Differences in position 327 contacts within the ␣4 helix were also observed. Contact between position 327 and ␣4 residue Asn 324 was seen in 48% of the wild type models and 56% of the Y327F models but in none of the Y327A models. In summary, when compared with wild type Tyr 327 by these modeling methods, the Y327A replacement permits fewer side chain contacts both between ␣4 and ␣5 helices and within the ␣4 helix, whereas the conservative Y327F replacement permits an array of side chain contacts similar to the wild type Tyr 327 models. DISCUSSION The C3 convertases, the central amplification enzymes of the complement cascade, initiate several potent inflammatory processes (see the Introduction). Their strict regulation is essential to permit the elimination of foreign agents while at the same time protecting self-tissues from autologous complement-mediated injury. The RCA family of regulators has evolved to serve this purpose. The mechanism by which these proteins control the activity of the convertases is by accelerating their irreversible dissociation (decay acceleration). As indicated (in the In-FIG. 2. DAF-mediated decay acceleration of C4b2a. C4b2a was assembled on the surface of sheep erythrocytes (E sh A) using mutant or wild type C2, and the cells were incubated with control buffer or with DAF at various concentrations for 15 min and then subjected to lysis via the complement terminal pathway (see "Experimental Procedures"). Decay acceleration (Inhibition of hemolysis) was calculated as the percentage of loss in the Z value (the average number of lytic sites per erythrocyte) caused by incubation with DAF. In this experiment, the decay accelerating activity of the wild type recombinant C2 protein was compared with that of the two C2 type A mutants, Y327A and N324A.
FIG . 3. The effects of C2 mutations on decay acceleration of C4b2a. In the case of C4b2a assembled with wild type C2, decay acceleration (inhibition of hemolysis) caused by the activity of regulator, was defined as 100%. The effects of each mutation on sensitivity to decay acceleration were compared with the wild type (wt) for each regulator. Regulator concentrations were: CR1-A, 172.7 ng/ml; C4BP, 700 ng/ml; DAF, 6 ng/ml. Approximate molar equivalents based on molecular masses of 69, 550, and 27.8 kDa, respectively, were 2.50, 1.27, and 0.22. At these regulator concentrations, hemolysis of wild type convertase was inhibited by 40 -50%.  troduction), we previously showed that for factor B, mutations in two sites, the parallel ␣4 and ␣5 helices and the ␣1 helix and its adjacent ␣1/2 loop, are important in DAF-and CR1-mediated decay acceleration of C3bBb (8). In the current study we found that homologous regions of C2 are similarly important in DAF-and CR1-mediated as well as C4BP-mediated decay acceleration of C4b2a. One major finding was that alanine substitution of the C2 type A residue Tyr 327 results in a complex that is highly resistant to the decay acceleration activity of DAF, CR1, and C4BP, and alanine substitution of the neighboring C2 residues Asn 324 and Leu 328 renders the complex partially resistant (Fig. 3). These mutants otherwise functioned normally, as did their factor B homologs (8). The results strongly support the proposition that the ␣4/5 region of the CP C3 convertase interacts with all three regulators. Taken together with our previous finding that the homologous factor B Tyr 338 residue is involved in both DAF and CR1-mediated decay acceleration, the results highlight a central importance for this site in the decay acceleration of both the CP and AP C3 convertases.
In general, protein-protein interfaces have been described as large bodies of relatively weak interactions that are punctuated by a few "hot spots" (27), amino acids whose interactions are responsible for most of the binding energy. Hot spot residues are relatively conserved and often polar (28,29). The residues surrounding hot spots are usually less conserved and have a more subtle influence on binding (27)(28)(29), possibly serving to exclude bulk solvent and retard the attack of water molecules on the interface core (27,28). The ␣4/5 regions of C2 and factor B each define such a cluster of polar/charged surface residues (Fig. 5) that could be part of the structural and functional core of a convertase-regulator interface.
Previously we showed that substitution of Tyr 338 with Phe reduced the sensitivity of C3bBb to both DAF and CR1-A, but substitution of Tyr 338 with Ala resulted in a much greater loss, indicating that both the 4-hydroxyl group and the phenyl group are important (8). In this study, the C2 Y327F substitution had less effect on CR1-A sensitivity of C4b2a and no detectable effect on DAF or C4BP sensitivity, indicating that the role of the phenyl group is more critical than that of the 4-hydroxyl group. Modeling studies provided evidence that the phenyl group (but not the hydroxyl group) plays a direct role in the contacts between the ␣4 and ␣5 helices and within the ␣4 helix (see under "Results" and Fig. 6 (A and B)). Although the Tyr 327 phenyl group could be a contact point for regulator interaction, the modeling studies suggest two additional nonexclusive mechanisms to account for the insensitivity of Y327A to decay acceleration: 1) the Y327A substitution alters the relative orientation of the ␣4 and ␣5 helices, thus changing the relative positions of regulator contact points in ␣4/5 (such as N324A and L328A), and 2) the Y327A substitution interferes with the dynamics between the ␣4 and ␣5 helices, which could be critical to convertase-regulator interaction (via an "induced fit" model) or to the propagation of allosteric changes promoted by the convertase-regulator interaction.
The interaction of C4b2a with DAF may afford the simplest model for decay acceleration because as indicated (see the Introduction) only two DAF CCPs, CCP2 and CCP3 (9, 10), are involved, whereas decay acceleration by all other RCA proteins appears to involve at least three CCPs (12,13,30). Moreover, unlike the other regulators, DAF does not have cofactor activity for the factor I-mediated cleavage of C4b or C3b. Although the affinity of DAF for C4b2a and C3bBb appears very low (indeed it has never been demonstrated directly), indirect evidence supports physical interaction with Bb, C3b, and C4b (31)(32)(33) and by inference, with C2a. Recent three-dimensional structures of DAF CCP2-4 have also suggested that DAF might interact with both convertase subunits (34,35).
Based on our findings in this study, the ␣4/5 region of C2 plays a key role in DAF-mediated decay acceleration. Among 24 residues previously tested within DAF CCP2 and CCP3, 10 residues have been found to be important for that reaction (11). One simple interpretation would be that the ␣4/5 region interacts directly with certain of these implicated DAF CCP2 and CCP3 residues. Alternatively, the ␣4/5 region could play an essential role in the propagation of allosteric changes from the binding site of DAF to the C4b-C2a interface.
We previously showed that substitution D254G in the type A ␣1-2 loop of factor B results in the production of an AP C3 convertase that is relatively resistant to DAF, CR1-A, and factor H (8). In the current study we found that substitution Q243G, the homolog to D254G in C2, likewise increases resistance to DAF and C4BP, and to a lesser extent, CR1-A. These residues (C2 Gln 243 and factor B Asp 254 ) lie close to the centers of the putative Mg 2ϩ -binding (23) and C4b(C3b)-binding (25) sites, and their replacements likely affect decay acceleration by altering the structure of the C4b-C2a (C3b-Bb) interface. Indeed, these substitutions also increase convertase stability in the absence of regulator (this study and Ref. 8). Unlike Y327A and Q243G, two other C2 substitutions, K254A/E255A in ␣1 and D434A in ␣7, did not affect decay acceleration in our experiments. In the case of the homolog D445A in factor B, a small but statistically significant difference in the sensitivity to CR1-A (71% wild type) was found (8). In the case of the factor B homolog K265A/K266A, the differences were greater (about 18% sensitivity to DAF, CR1-A, and factor H) and may be attributable to effects on the C3b-Bb interface (8). Parallel effects were not detected with our K254A/E255A C2 mutants, the simplest interpretation being structural differences between C4b and C3b in their respective convertases.
In summary, our findings indicate that the ␣4/5 region of C2a is critical to decay acceleration mediated by DAF, C4BP, and CR1. Importantly, the results indicate that C2 Tyr 327 and homologous factor B Tyr 338 constitute a functional "hot spot" in the convertase/decay accelerator reaction. The available evidence suggest two simple interpretations: 1) that decay acceleration of C4b2a and C3bBb requires interaction of the convertase ␣4/5 region with the mapped CCP2 and CCP3 sites of DAF, or structurally homologous sites of CR1 and C4BP and 2) that decay acceleration requires the propagation of an allosteric signal from regulator-convertase interface to the C4b2a (C3bBb) interface, which involves participation of the ␣4/5 region. In either model, the region plays an important role in a common mechanism to destabilize C4b2a and C3bBb.