Molecular Dissection of Interactions between Components of the Alternative Pathway of Complement and Decay Accelerating Factor (CD55)*

The complement regulatory protein decay accelerating factor (DAF; CD55), inhibits the alternative complement pathway by accelerating decay of the convertase enzymes formed by C3b and factor B. We show, using surface plasmon resonance, that in the absence of Mg 2 (cid:1) , DAF binds C3b, factor B, and the Bb subunit with low affinity ( K D , 14 (cid:2) 0.1, 44 (cid:2) 10, and 20 (cid:2) 7 (cid:3) M , respec-tively). In the presence of Mg 2 (cid:1) , DAF bound Bb or the von Willebrand factor type A subunit of Bb with higher affinities ( K D , 1.3 (cid:2) 0.5 and 2.2 (cid:2) 0.1 (cid:3) M , respectively). Interaction with the proenzyme C3bB was investigated by flowing factor B across a C3b-coated surface in the absence of factor D. The dissociation rate was depend-ent on the time of incubation, suggesting that a time-de-pendent conformational transition stabilized the C3b-factor B interaction. Activation by factor D (forming C3bBb) increased the complex half-life;

organisms or infected cells. The alternative pathway (AP) provides an immediate, antibody-independent means to activate the C cascade, whereas the classic pathway relies in most cases on antibody to initiate activation of the first component, C1. A third pathway, the lectin pathway, is initiated by binding of mannan-binding lectin to sugar residues on bacterial cell walls. Other than the initiating factor (mannan-binding lectin and associated proteases), this pathway is identical to the classic pathway. The AP provides an immediate line of defense to potential infection, it "ticks over" continuously in plasma through low level activation of the central component, C3, probably via hydrolysis of the internal thioester bond (1). Tight regulation of this tick-over ensures that C is not activated to excess in plasma but provides enough active C3b to bind foreign surfaces and amplify the cascade swiftly when appropriate (2). The AP also amplifies the classic pathway by formation of further active convertases on C3b deposited during early amplification steps by the classic pathway convertase. The initiating stimulus of the AP is nucleophilic attack on the internal thioester bond in nascent C3b by an amine or hydroxy group present on a foreign or "activating" surface (3). This results in covalent binding of C3b and forms the nidus for C amplification. The C3 convertase is formed by binding of factor B (fB) to C3b in an interaction that requires a Mg 2ϩ ion. The fB changes conformation such that the C3bB complex is capable of activating the zymogen factor D (fD), a serine protease present in plasma at about 2 g/ml (4). FD cleaves fB into Ba (amino-terminal fragment) and Bb, the latter comprising an amino-terminal von Willebrand factor type A domain (vWFA) and a carboxyl-terminal serine protease (SP) domain. After release of Ba, Bb binds with higher affinity to C3b. It is likely that conformational changes in the vWFA domain transmit an allosteric signal to the SP domain, resulting in its activation and conferring it with the ability to cleave multiple molecules of C3 to nascent C3b, thereby amplifying the cascade (5,6).
The background tick-over of the AP and the potential for rapid amplification of all pathways on self cells means that strict control in both biological fluids and on the membranes is essential. Numerous C regulatory proteins (CReg) have evolved to meet the need to protect "self " from the potentially destructive effects of C (7). These proteins police the body, collaborating to control all pathways. The different CReg belong to various gene families, those encoded in the Regulators of Complement Activation (RCA) gene cluster on chromosome 1 comprise factor H, C4b binding protein, decay accelerating factor (DAF; CD55), membrane cofactor protein (CD46), and complement receptor 1 (CD35) (8). All contain a structural module termed the short consensus repeat (SCR), a domain comprising ϳ60 amino acids that confers C3b/C4b binding affinity to the proteins and contains the regulatory function of the molecule (9). The amino-terminal fragment of fB, Ba, also comprises three SCRs. The number of domains in CReg varies from four (in membrane cofactor protein and DAF) to 37 (complement receptor 1, "B" isoform). DAF is a glycosylphosphatidylinositol-anchored CReg that acts to accelerate decay of the naturally labile enzymes of the AP and CP, C3bBb and C4b2a (10,11). Other than the SCRs, it comprises an elongated, heavily glycosylated "stalk" proximal to the membrane that projects the active site the correct distance from the membrane, allowing it to function efficiently (12,13).
The C components, from C1 through C9, comprise a fascinating group of proteins whose activities are characterized by differing conformations and presentation of continuously changing faces or binding sites to other C components, receptors, or regulators. C3 and fB are no exception. C3 transformation starts after cleavage and release of C3a, continues after destruction of the internal thioester bond, and moves on by presentation of binding sites for components such as fB, Bb, and eventually C5 (14,15). Along the way, binding sites for a multitude of regulators are formed; their sites may be distinct or overlapping and are often transient as the cascade progresses or C3b becomes inactivated (16). Cleavage and inactivation of C3b by fI results in the production of iC3b, C3c, and C3dg; these inactivation products no longer bind components such as fB, and all differ in their ability to bind various C receptors (17). FB also undergoes a multitude of conformational changes. Within the intact protein, all three domains interact with each other; possibly, Ba holds the vWFA and SP domains in a way that prevents autoactivation (5,6). FB changes conformation after binding to C3b, once more after release of Ba, and presumably again after release to the fluid phase (18,19). The exact mechanisms of AP convertase assembly and regulation are still being determined 50 years after discovery of this pathway (20).
Interaction of DAF with the AP convertase has largely been examined by mutagenesis studies. These have sought to define entire domains in DAF or individual amino acids that are crucial to decay of the convertase and protection of indicator cells (usually erythrocytes) from lysis by C (21)(22)(23). Further information has been gained after resolution of the structure of DAF, both in solution and in crystals (13,23,24). Various sites that extend through SCRs 2, 3, and 4 have been proposed to be important for convertase regulation. It is not yet clear whether DAF undergoes a conformational transition after binding to its ligands; most probably, it acts purely to induce changes in other proteins to which it binds. It is interesting to note that DAF not only binds the convertase enzymes but also a multitude of pathogens, both bacteria and viruses, and also the leukocyte activation antigen CD97 (25). A potential site of interaction for DAF on fB has also been identified by mutagenesis studies and is located distal to the C3b interaction site in vWFA domain (26). Various studies have addressed the binding site on C3b for the regulators of complement activation proteins factor H, membrane cofactor protein, complement receptor 1, and also for fB, but no definitive sites of interaction have been determined and that for DAF has not been examined (27,28). In this study, we examined the interactions between DAF and the AP convertase components by surface plasmon resonance (SPR). We defined the affinity of these associations and examined in real time the assembly and decay of the active AP convertase and its proenzyme and their interaction with DAF. This results of this study provide new insight into these associations and further guide our understanding of the complex mechanisms and protein interactions intrinsic to the C cascade.

EXPERIMENTAL PROCEDURES
Preparation of Complement Components and Regulators-C3, cobra venom factor (CVF), and factor B were prepared by classic column chromatography using established methods (29). Factor D was purchased from Quidel Corporation (San Diego, MA). C3b was prepared from C3 using either of two methods. In the first, CVF was coupled to Sepharose CL-4B using the manufacturer's protocol (Amersham Biosciences). A solid-phase convertase was formed by incubating CVF-Sepharose in normal human serum to form CVFBb. The Sepharose was washed and incubated at 37°C with C3. C3b generated by cleavage was separated from C3a and other minor contaminants by anion exchange on a Source Q column (Amersham Biosciences). In the second method, C3, factor B, and factor D were incubated in complement fixation diluent (Oxoid Ltd., Basingstoke, UK) until total C3 cleavage had occurred. C3b, Bb, and Ba were purified by anion exchange chromatography. The C3b-containing fractions were pooled and concentrated, and monomeric C3b was separated from dimeric C3b by size exclusion on a Superose 6 column (Amersham Biosciences).
Recombinant human DAF comprising the four SCRs (soluble DAF (sDAF)) was isolated and refolded from Escherichia coli as described previously (30). The structure and function of sDAF has been studied previously, and the purified, refolded protein is known to consist of four correctly folded SCR domains and demonstrates the full range of complement-regulating and pathogen-binding activities associated with DAF purified from erythrocytes (13,30,31). Human DAF-Ig fusion protein comprising four SCR domains fused to human IgG1 Fc (DAF-Ig) was prepared as described previously (32). Recombinant vWF-A domain of human factor B was expressed in E. coli as a fusion protein with glutathione S-transferase using published methods (33) with the addition of a final gel filtration on a S75 column (Amersham Biosciences) to yield protein at a purify of Ͼ98%, as assessed by SDS-PAGE electrophoresis (data not shown).
Biosensor Analysis-All analyses were carried out on a Biacore 3000 machine (Biacore International SA, Stevenage, UK), except for that shown in Fig. 4c, which was carried out on a Biacore 2000. Proteins were coupled to the sensor chip surface using an amine coupling chemistry as instructed by the manufacturers (N-hydroxysuccinimide/ 1-ethyl-3(3-dimethylaminopropyl)-carbodiimide hydrochloride coupling kit). For all kinetic analyses, a CM5 chip (carboxymethylated dextran surface) was used, data were collected at 25°C, the flow rate was maintained at 30 l/min, and data from a reference cell were subtracted to control for bulk refractive index changes. The R max was kept low and the flow rate high to eliminate mass transfer. However, this was controlled for where appropriate by varying flow rate and ensuring observed association rates did not vary. Samples were injected using the KINJECT command to ensure accurate association kinetics. Interactions were analyzed in HEPES-buffered saline (HBS; 10 mM HEPES pH 7.4, and 150 mM NaCl), 0.005% surfactant P20, and either 1 mM MgCl 2 or 3 mM EDTA as stated in the text. Data were evaluated using Biaevaluation software (Biacore International). Concentration of analytes was assessed using absorbance at A 280 , molarities were calculated using the after extinction coefficients (molecular masses and coefficients obtained using Protean software, DNAStar): sDAF (1.

RESULTS
Interaction of DAF with C3b-To study the interaction of DAF with the individual AP convertase components, C3 was incubated with factors B and D until formation of C3b, Bb, and Ba was complete. Individual components were separated by anion exchange chromatography (Fig. 1a). This also removed C3i from the C3b preparation. We routinely found that ϳ50% of the C3b formed by AP activation was in a dimeric form. To eliminate avidity effects from SPR analysis, C3b was further separated into monomeric and dimeric forms by size exclusion chromatography (Fig. 1b). Initial SPR studies indicated that the interaction between DAF and C3b was of low affinity (submicromolar) with very fast on and off rates typical of DAFligand interactions (31,34,35). The affinity of the interaction was studied under steady-state conditions by immobilizing ei-ther DAF or C3b to the sensor chip surface and flowing monomeric C3b or DAF, respectively, over the surface. The most accurate determination as judged by smallest S.E. values for K A and R max was obtained by immobilizing C3b and studying the interaction with soluble DAF, because this analyte was available at concentrations of the same order of magnitude as the K D values determined (Ͼ10 M). C3b was coupled to the chip surface either by amine coupling or through the thioester group. The latter method has the advantage of capturing C3b on the surface in a physiologically relevant orientation. A typical analysis is illustrated in Fig. 2; reference cells were used to correct for any bulk shift effects. Mean K D values are given in Table I.
Interaction of DAF with fB and its Cleavage Products-Mutational analysis of fB has revealed two sites in the vWFA domain that are involved with decay; whether these sites interact directly with DAF or alter the stability of the convertase is not yet clear (26). It is likely, however, that there is a direct interaction of DAF with the Bb portion of the convertase. We immobilized DAF on the chip surface and flowed intact fB, Bb, or Ba over the surface at high concentrations. A direct interaction was seen with intact fB and the Bb subunit but not with Ba (Fig. 3a). The affinity of the interaction between DAF and intact fB was determined by immobilizing either fB or DAF to the surface via amine groups. As with the DAF/C3b interaction, the affinity was low and was analyzed at steady state. The binding profile and affinity data obtained were similar irrespective of the protein partner immobilized (mean K D , 44 Ϯ 10 M in EDTA; Fig.  3) and the presence or absence of Mg 2ϩ (data not shown; mean K D , 74 Ϯ 30 M in 1 mM Mg 2ϩ ). The interaction of DAF with intact fB was of lower affinity than that with C3b (Table I).
Although the interaction of fB was relatively unaffected by the presence of Mg 2ϩ , we noticed that this cation had a marked effect on the interaction of DAF with the Bb subunit (Fig. 4a). In the absence of Mg 2ϩ , the affinity of the Bb subunit for DAF and the binding profile was similar to that of fB (data not shown; average K D , 20 Ϯ 7 M). However, in the presence of Mg 2ϩ , the affinity of the interaction between Bb and DAF was higher. The kinetic profile was different to that of fB, on and off rates were slower, and the affinity was calculated as 1.3 M (using a 1:1 Langmuir interaction model; Fig. 4b, Table I). Further analysis using the isolated vWFA domain demonstrated a similar binding profile (Fig. 4c) and affinity of 2.2 Ϯ 0.1 M for this subunit (Table I).
Interaction of fB with C3b-We sought to assemble the convertase on the sensor chip surface itself, the aim being to assess the affinity of DAF for the intact enzyme in addition to individual components. Low levels of C3b were immobilized to the chip surface and fB was flowed across in the presence of Mg 2ϩ (1 mM). It was immediately apparent that the interaction did not fit a simple 1:1 association model (Fig. 5a). Furthermore, interaction of identical concentrations of fB with C3b for different lengths of time (1, 3, or 20 min) demonstrated that the dissociation became slower the longer the components interacted (Fig. 5b). This can be indicative of a conformational change that results in a tighter binding between components. Indeed, fitting the data from Fig. 5a to a conformational change model (BIAcore software) gave a good fit, although some heterogeneity was evident at high concentrations of fB (Fig. 5c). Kinetic analysis after a 4-min association phase gave the following information: k a1 , 1.36 ϫ 10 4 M Ϫ1 s Ϫ1 ; k d1 , 0.115 s Ϫ1 ; k a2 , 5.53 ϫ 10 Ϫ3 M Ϫ1 s Ϫ1 ; k d2 , 1.78 ϫ 10 Ϫ3 s Ϫ1 ( 2 , 4.2). In this model, the off rate after the conformational change (k d2 ) is much lower.
Formation of the AP Convertase and Interaction with DAF-Either the active or inactive convertase was assembled on the chip surface by flowing fB over C3b in the presence or absence of fD. Either fB (46 g/ml) or a mixture of fB (46 g/ml) and fD (8 g/ml) was flowed across the surface as indicated in Fig. 6a. Several differences in the two complexes were noted. First, cleavage of fB by fD altered the rate of decay of the active enzyme, prolonging its half-life. Second, the Mg 2ϩ ion in the active enzyme was "locked" in place, flowing 10 mM EDTA over C3bB(Mg 2ϩ ) dissociated fB from C3b, whereas it had no effect on decay of C3bBb(Mg 2ϩ ).
C3b was also bound to the surface using the AP, which resulted in coupling of nascent C3b via its thioester to hydroxy groups present on the dextran-coated surface. We used a modification of a described previously technique in which repetitive cycles of fB/fD with C3 resulted in AP amplification on the chip surface (36). To increase efficiency of activation we deposited the first nidus of C3b on the chip surface using a fluid phase convertase, CVFBb (Fig. 6b). Each subsequent "cycle" illustrated in Fig. 6b comprised incubation with fB and fD to form active convertase followed by incubation with native C3. After C3b deposition, buffer was flowed across the surface for 16 h to allow dissociation of noncovalently bound C3b and of any active convertase. To study the interaction of DAF with the convertase, factors B and D were flowed over the surface. The amount of active/inactive convertase on the chip surface was varied by titrating the amount of fD in the incubation (Fig. 6c). In the presence of fD, the surface reached an equilibrium (Fig. 6c, E), unlike the proenzyme, which displayed complex kinetics during formation (Figs. 5  and 6c, A). When soluble DAF was flowed across the active convertase, decay of the Bb subunit was virtually instantaneous (Fig. 6c, B-E). Decay of the enzyme was so efficient that it was impossible to measure binding and an affinity of DAF for the intact convertase. This contrasted with flow of DAF across the inactive enzyme in which rapid and total decay was not apparent (Fig. 6b, A). Uncleaved fB was dissociated from the surface at the end of the incubation using EDTA as shown in Fig. 6. Alignment of individual sens- orgrams at either the fB/fD injection or the DAF injection demonstrates the variation in decay and identical binding of DAF to C3b on the chip surface (Fig. 6d).
Formation of CVFBb and Interaction with DAF-We have demonstrated that DAF interacts with both subunits of the active enzyme and that decay is rapid and efficient. The convertase enzymes formed from CVF (C3b-like molecule found in cobra venom) and fB are very stable and are assumed to be resistant to decay by human regulators. To analyze the interaction with DAF, CVF was immobilized on the sensor chip surface and DAF was flowed across; no interaction between CVF and DAF was evident even at high concentrations (47 M) (Fig. 7a). To assess whether DAF could decay the CVF-containing convertase, fB and fD were flowed sequentially across the CVF surface and a C3b-coated surface (Fig. 7b). In the absence of fD, the CVFB complex dissociated rapidly and the surface could be regenerated with EDTA. Inclusion of fD resulted in an EDTA-stable surface that efficiently cleaved C3 to C3b (data not shown). Formation of the active enzyme was less efficient than with human C3b, but enzyme that did form was very stable (Fig. 7b). When DAF was flowed across the surface at concentrations (15 M) much higher than those bringing about immediate decay of C3bBb, it was apparent that it could indeed accelerate decay of CVFBb. However, this was far less efficient than with the native, human enzyme. The high concentration of DAF used in this experiment is apparent in Fig. 7b, where binding to the C3b-coated surface can be seen during the DAF inject (indicated in the figure). DISCUSSION We have used SPR technology to analyze the interaction of DAF with individual components of the AP convertase and also with the intact (pro)enzyme and to visualize assembly and decay of the convertase in real time. We demonstrate that DAF interacts with the individual components C3b and fB with low affinity. The equilibrium dissociation constant (K D ) for the interaction of DAF with fB is 44 Ϯ 10 M, demonstrating that this is a significantly weaker interaction than that of DAF with

Interactions in the Alternative Pathway of Complement
C3b where the interaction with amine-coupled C3b is characterized by a K D of ϳ14 Ϯ 1 M and with thioester-coupled C3b with a K D of ϳ7 Ϯ 1 M (Table I and Figs. 2 and 3). The small differences between the K D values for DAF interactions with differently coupled C3b probably reflect the more optimal presentation of binding sites in the natively coupled C3b compared with the amine coupled protein, in which steric hindrance is more likely to occur. DAF did not interact with the Ba subunit of fB but did interact with the Bb subunit. The affinities of DAF for fB and Bb in EDTA were comparable (44 Ϯ 10 and 20 Ϯ 7 M, respectively). However, the affinity of DAF for the Bb subunit was increased Ͼ10-fold to 1.3 M (K D ) in the presence of Mg 2ϩ ; a similar affinity was obtained for the isolated vWFA domain (2.2 M), suggesting that this domain mediates the contact with DAF (Fig. 4). Indeed, a previous report defined two surface patches on opposing surfaces of the vWFA domain that may be involved in the decay of C3bBb by DAF; one of these may mediate binding to C3b, whereas the other may be involved in binding DAF (26). It is interesting that DAF has a higher affinity for Mg 2ϩ -bound Bb. It is known that the Mg 2ϩ coordination site of fB is in the active site cleft of the vWFA domain, a site that is involved in the binding to C3b, and that the structure of this domain and of Bb is more stable in the presence of the cation. A conformational dependence of the vWFA domain on the presence of a metal ion has been demonstrated by various spectroscopic techniques (6); in particular, the ␣-helix A7 is conformationally mobile between the metal-free and metal-bound forms of the vWFA domain. Our data show that DAF may bind distinct conformational states of this molecule with different affinities. Although we have demonstrated a direct interaction of DAF with the isolated Bb fragment, the physiologically relevant complex is Bb in association with C3b. Bb has little or no affinity for C3b after dissociation, indicating that fluid phase Bb differs in conformation from C3b-bound Bb (18). 2 It is known that the vWFA domain in Bb, in common with similar domains in other molecules such as CR3 or von 2 C. Harris, unpublished data. Willebrand factor, exists in high and low affinity conformations (37). After cleavage of fB by fD, Bb binds to C3b with higher affinity. The vWFA domain probably transmits an allosteric signal resulting in proteolytic activity in the SP domain of Bb (6). We cannot determine whether the affinity of DAF that we demonstrate here is for a low or high affinity conformation of vWFA domain.
The initial binding of fB to surface bound C3b is complex and clearly does not follow a simple 1:1 Langmuir interaction (Fig.  5). The data fit a model in which a conformational change occurs, resulting in a higher affinity interaction with a slower dissociation rate (k d2 ). This model can only be applied when the change is slow and occurs over the period of the association phase; extremely rapid changes are more likely to fit a simple 1:1 interaction. It is known that fB changes conformation upon binding C3b allowing it to induce a proteolytically active conformation of fD (4). However, inclusion of fD in the incubation resulted in a very different binding profile (Fig. 6), the binding rapidly reached a plateau and an equilibrium was obtained. It is interesting that the resulting complex was resistant to EDTA (Fig. 6a); this is in agreement with previous work, which has shown that a transition in Bb and tight binding of the vWFA active site cleft to C3b protects the Mg ion from chelation by EDTA (19,38). The binding profile of fB in the presence of fD was more typical of a simple interaction, implying that if conformational changes have occurred they were rapid and an "end-point" was swiftly reached. Others have previously reported that fB can bind C3b (and CVF) and form an active enzyme in the absence of fD (39 -41). It is possible that the conformational change evident in the absence of fD in Fig. 5 represents a slow transition to an active conformation in which the SP domain of fB can cleave C3b; fD-mediated cleavage of fB to Bb may act to expedite this transition.
We have shown that DAF binds individual components of the AP convertase with low affinity, the crucial question is how does DAF bind the intact convertase? DAF must "recycle" to protect self cells from C attack. It is clear that if DAF bound to C3b with the same affinity as to the convertase, it would rapidly become saturated with C3b at a site of C activation and would not be available for decay of the active convertase. In 1986, Pangburn examined the ability of DAF and other CReg to decay the zymosan-bound convertases in the presence of fluid phase competitors such as C3b, Bb, C3bB, and C3bBb (42). Apparent association constants (appK A ) of DAF with C3b and Bb were 0.045 and 0.067 M Ϫ1 , respectively. The affinity of DAF for C3bB or C3bBb was Ͼ10-fold higher (0.71 or 0.91 M Ϫ1 , respectively) suggesting that after decay of the enzyme, DAF would be "released," enabling it to bind the next convertase with high affinity. Despite the fact that DAF seemed to bind C3bB almost as well as C3bBb in this study, it was not clear whether DAF decayed both the active convertase and the proenzyme. Early studies using DAF incorporated into erythrocyte membrane showed that DAF did not prevent fB binding surface bound C3b or C4b but rapidly decayed the activated fragment Bb (43). It has been demonstrated more recently, using an enzyme-linked immunosorbent assay-based decay assay, that DAF decays C3bBb(Ni 2ϩ ) but not C3bB(Ni 2ϩ ) (44). This selective decay may be related to the Ͼ10-fold higher affinity of DAF for Bb compared with fB in the presence of Mg 2ϩ . Our data support the suggestion that DAF only decays the active enzyme. To monitor the decay of the convertase using SPR, we formed the convertase on the sensor chip surface in the presence or absence of fD (Fig. 6). DAF-mediated decay of the active convertase, C3bBb, was visualized in real time. It was immediately apparent that dissociation of Bb from C3b was virtually instantaneous, and it was not possible to measure an affinity of DAF for the active convertase directly. In support of the ELISA-based assay it was evident that DAF decayed the active enzyme and had little effect on the inactive convertase. By titrating the amount of active convertase on the chip surface it was possible to see this differential decay (Fig. 6c). In the absence of any fD, a small amount of DAF-mediated decay was evident (Fig. 6c, A), we do not know what this represents, but it is possible that a portion of C3bB has undergone a subtle transition such that it is in an "active" conformation, and the affinity for DAF is enhanced as discussed above rather than this being the observation of decay of the inactive convertase. When DAF was flowed across the C3b-coated surface in the absence of any fB or fD, an increase in resonance units was evident (Fig. 6d, alignment at DAF injection). This represented binding of DAF to C3b and decreased again rapidly after the DAF injection was completed; the off rate for this interaction is extremely rapid, as illustrated in Fig. 2. After an identical injection of DAF over the C3bB-or C3bBb-coated surface, the same decrease in resonance units was evident at the end of the inject and dissociation was rapid (Fig. 6d). The dissociation of DAF from the surface was evident above the background decay of fB from C3b, which remained unaltered. These preliminary data imply that in our assay system, the affinity of DAF for C3b was unaffected by the presence or absence of proteolytically inactive fB. We are currently conducting further investigation of the kinetics of this interaction. Attempts to stabilize the C3bBb surface using chemical cross-linking to measure the affinity of DAF for the active enzyme were not successful. Although the release of Bb from the surface was prevented, an interaction with DAF could not be visualized. Either the crosslinking reagent destroyed the interaction sites with the convertase, or DAF did indeed decay the components, but they remained loosely tethered to each other via the cross-linker without any physical protein/protein interaction. There are various reasons why DAF might decay the active but not the inactive convertase. First, the SCR-containing subunit of fB, Ba, might directly compete for the same binding site on C3b. Second, the presence of Ba might stabilize the C3bB complex; a direct interaction of Ba with C3b has been demonstrated (45). 2 Several points of contact will increase the avidity of the interaction. Third, as suggested above, Bb might adopt a conformation in the absence of Ba that binds DAF with a higher affinity. Finally, it is also possible that the Ba domain partially blocks the binding site on Bb for DAF; it has been shown previously that Ba contacts both the vWFA and SP domains in inactive fB (5).
It is interesting that DAF accelerates decay of the CVFBb convertase, albeit with much lower efficiency than decay of C3bBb (Fig. 7b). DAF had no affinity for the isolated C3b-like component of the enzyme, CVF (Fig. 7a). In the study described above, Pangburn (42) demonstrated that the affinity of DAF for CVFBb was identical to that of DAF for Bb. Although we show here that the accelerated decay of CVFBb mediated by DAF was very inefficient, it was the only way we could regenerate the surface of the chip. Neither EDTA nor another powerful CReg, soluble recombinant complement receptor 1, could regenerate the surface back to uncomplexed CVF. The inefficiency of this decay implies that either the conformational change in Bb resulting in its release is slowed by the presence of CVF or that DAF works by also inducing a transition in C3b (not seen with CVF) that accelerates decay of the enzyme. FIG. 5. Interaction between C3b and factor B. a, C3b was immobilized on the chip surface via amine coupling. Interaction with factor B was analyzed at the indicated concentrations in HBS-Mg 2ϩ allowing association for 4 min. b, factor B was flowed over C3b for different lengths of time (1, 3, and 20 min), data were normalized using BIAcore evaluation software and sensorgrams overlaid at the start of the dissociation curve. c, the affinity of the interaction was analyzed using the "conformational change" model; dashed lines are modeled data, 2 ϭ 4.2; gray lines are actual data.
FIG. 6. Assembly and decay of the AP C3 convertase in real time. a, C3b was amine-coupled to the chip surface. fB was flowed across in HBS-Mg 2ϩ as indicated; incubation in 10 mM EDTA decayed the complex. In contrast, C3bBb was not decayed by 10 mM EDTA. b, C3b was deposited on to the acceptor surface using a method similar to that already described previously (36). The AP was amplified by repeated cycling (in HBS-1 mM Ni 2ϩ ) at 30°C of C3 followed by fB and fD. Convertases and non-covalently bound C3b was allowed to decay for 16 h before using the surface. c, different amounts of active convertase were assembled on the chip surface by varying the concentration of factor D in the incubation. As the quantity of active convertase increased, the ability of sDAF to decay the components was greater. FB, fD, sDAF, and EDTA were injected at the indicated time-points. Decay mediated by (0.8 M) sDAF is indicated by a double-headed arrow. FB was injected at 450 g/ml and fD at 0, 2.5, 5, 10, and 20 ng/ml (lettered A-E, respectively). d, sensorgrams from c are aligned at the start of the fB/fD injection to allow for comparison of convertase assembly or at the end of the DAF inject to allow comparison of DAF-mediated decay. Current concepts regarding AP convertase assembly and decay are as follows. Ba mediates initial binding of fB to C3b, and the avidity of binding may be enhanced by sites on the Bb subunit which interact with C3b. If sites in Bb interact with C3b, the conformation differs from that of the dissociated subunit, Bb, which has much decreased affinity for C3b. After binding of fB, a conformational change occurs such that the complex activates the zymogen fD, which in turn cleaves fB into Bb and Ba. Ba is released from the complex (although this interaction is reversible). Release of Ba results in a higher affinity complex between C3b and Bb, the Mg 2ϩ ion in the active site cleft of the vWFA domain is locked into the enzyme through association with C3b, the complex becomes more stable with a longer half-life, and the SP domain in Bb is activated such that it can proteolytically cleave C3. However, this complex is subject to decay by DAF. We show that DAF binds individual components of the convertase in the following order of affinity: Bb Ͼ C3b Ͼ fB. We assume that it associates weakly with C3b deposited covalently on a cell surface. However, to selectively bind and decay active convertase, binding to C3bBb must be greater than that to C3b. A simple explanation would be that a second point of contact with the Bb subunit increases the avidity of the interaction. We have noted that apparent affinities of bivalent interactions can be 10-fold that seen with the same monovalent interaction. 3 It is also possible that complexing of C3b and Bb favors a particular conformation in one or both of the components that bind DAF with higher affinity. Increased avidity of DAF binding to the multimolecular complex would in effect stabilize the complex further and prevent decay/dissociation. It is likely, therefore, that DAF promotes a further change in conformation, resulting in dissociation of Bb from C3b, possibly by inducing the low affinity conformation of the vWFA domain. Decay of the complex results in a low (submicromolar) interaction with the individual components, allowing DAF to recycle to other active convertase. It is noteworthy that the AP convertase is stabilized in vivo by binding of properdin to the C3bBb complex (46). We are currently examining the effect of properdin on DAF binding to the AP convertase and to individual components. Although the exact mechanisms of assembly and decay of the AP convertase are still unclear, we have come a long way in deciphering the ways in which C is activated and regulated. Data presented here and similar analyses of other C regulators will provide a more informed picture of their 3 C. Harris, unpublished data. FIG. 7. Interaction of DAF with CVF. CVF was immobilized on the chip surface via amine groups. a, CVF did not bind sDAF although it formed a C3 convertase when factor B and factor D were flowed across the chip surface (b). sDAF was injected as indicated, during which time the rate of decay of the convertase increased. The C3bBb sensorgram is indicated by the dashed line, and the CVFBb sensorgram is indicated by the solid black line. CVF convertase was assembled in HBS-Mg 2ϩ .