Molecular Immunology, Vol. 27, No. Printed in Great Britain.
I I, pp. 1155-I
161, 1990
0161-5890/90 $3.00 + 0.00 Pergamon Press plc
THE MECHANISM OF ACTIVATION OF THE ALTERNATIVE PATHWAY OF COMPLEMENT BY CELL-BOUND C4b* TIMOTHYC. FARRIES,~~K. LEEKNUTZENSTEURR~and JOHNP. ATKINSON~~ tThe Howard Hughes Medical Institute Laboratories and Department of Medicine, Division of Rheumatology, Washington University School of Medicine, St Louis, MO 63110, U.S.A. (First received 20 October 1989; accepted in revised form 30 April 1990)
Abstract-Investigations into the mechanism of alternative pathway-dependent lysis of C4b-coated cells are reported. Test cells (EAClq4b) were formed by reaction of sheep erythrocytes with antibody, Cl and C4. In (X-deficient serum, more C3b was deposited onto EAClqC4b than onto control cells (EAClq). The possibility that the C4bBb enzyme could form was considered, but no C3 convertase activity was generated when magnesium, properdin and factors B and D were added to EAClqC4b. Binding studies employing radiolabeled components provided evidence that C4b bound the C3 convertase, C3bBbP, through a weak interaction with C3b. These data implied C3 conversion would be localized to the cell surface, thereby amplifying C3b deposition, This could be demonstrated in vitro. C3b, properdin, factor B and factor D were all required and the amplified C3b deposition was not due to deposition onto C4b itself. In serum, CS convertase activity would be consequently expressed and cell lysis would result. This could be the mechanism by which the sera of CZ-deficient patients mediate lysis of antibody coated sheep erythrocytes.
INTRODUCTION Sera from human patients with no detectable C2 activity have unexpectedly significant complementmediated hemolytic activity towards antibody-coated sheep erythrocytes (Steuer et al., 1989). This effect has been shown to be dependent on antibody, on the early components of classical pathway activation, Cl and C4, and on the alternative pathway components B, D and P. It was further demonstrated that the phenomenon did not involve residual C2. The initial stages of the reaction are presumably the same as those of the classical pathway; antibody clusters on the cell activate Cl which in turn cleaves fluid-phase C4. Antibody is required in high concn (compared to classical pathway activation), suggesting that a higher level of C4 cleavage is required. Some of the product, C4b, deposits on the cell surface. If C2 was present, it would also be cleaved to form a bimolecular C3-cleaving enzyme, C4b2a. It is now speculated that C4b can also interact with the alternative pathway to
produce C3 activation without the involvement of C2. This possibility is further suggested by the work of Matsushita and Okada (1986) who reported that C4b-coated erythrocytes are lysed by human serum via the alternative pathway. Cl was not required for this latter reaction. We have therefore sought to further define the mechanism. Strong structural homologies exist between C2a and Bb, between C3b and C4b, and between the C3 convertases, C4b2a and C3bBb. It has therefore been suggested (Matsushita and Okada, 1986) that a hybrid convertase, C4bBb, might form and cleave C3. Another explanation would be that C4b is a favored “acceptor” site for nascent C3b, leading to generation of C3bBb on C4b-coated surfaces. Formation of covalent C4b-C3b dimers has been observed (Takata et al., 1987). However, until now, no detailed experimental analysis of the mechanism has been performed. Our data support a new hypothesis; that C4b binds the C3 convertase non-covalently, and that C3b deposition is thereby localized to the cell surface.
*This work was supported
in part by The American Heart Association. IPresent address: MRC Collaborative Center, London, England. @Author to whom correspondence should be addressed: John P. Atkinson, Division of Rheumatology, 4566 Scott, Box 8045, St Louis, MO 63110, U.S.A. Abbreviations: E, sheep erythrocytes; EA, E coated with anti-E antibody; iC3, thioester-broken form of C3; P, properdin; B, factor B; D, factor D; CVF, cobra venom factor; C5D, CS-depleted human serum; NGPS, normal guinea pig serum; THC or CH,, , total hemolytic complement titer.
MATERIALSAND METHODS Buffers
and reagents
The reagents used were PBS, 8.7 mM K,HPO,, 1.7 mM NaH2P04, 150 mM NaCl, pH 7.4. PBS-dextrose is PBS diluted 1:2 in 5% (w/v) D-glucose. VBS2+ contains 3.1 mM diethyl barbituric acid, 0.9 mM Na barbitone, 145 mM NaCl, 0.83 mM MgCl,, 0.25 mM CaCl,, pH 7.2. GVB2+ is VBS*+ containing 0.1% (w/v) gelatin. 1155
1156
TIMOTHY C. FARRIES et al.
Human complement components properdin (“native” form), C3b, iC3 and factor B, D and H were purified as described elsewhere (Farries et al., 1987; Harrison and Lachmann, 1986). Purified C3 and C4 were the gift of R. P. Levine (St Louis, MO). C4 and CVF, provided by T. Seya (Osaka, Japan), were also used. Sheep erythrocytes (E), rabbit antibodies to E (“hemolysin”), and guinea-pig Cl were purchased from Diamedix (Miami, FL). Antibodies against C3, C4, C5 and alpha 2-macroglobulin were goat IgG preparations purchased from Atlantic Antibodies, Scarborough, MA. Radiolabeling of purified proteins with “‘1 was performed using the Iodogen (Pierce Chemical, Rockford, IL) procedure of Fraker and Speck (Fraker and Speck, 1978). The specific activities were typically 0.1-0.5 pCi/pg. CS-depleted human serum (C5D) was prepared by immunoadsorption with polyclonal goat anti-human C5 IgG. The IgG was coupled to CNBr-activated Sepharose (Sigma Chemical Co., St Louis, MO). After two cycles of absorption at 0°C the depleted serum did not lyse EA in VBS*+.
were preadsorbed with 2.5 x lo9 E/ml, and microfuged (lO,OOOg, 5 min), immediately before use to remove any partially denatured molecules that bind non-specifically. Fixation of radiolabeled C3 was measured similarly, with the addition of a wash with PBS before sedimenting the cells through sucrose.
EAC lq4b cells
C3b deposition via the alternative pathway
EA cells were made by treating washed sheep erythrocytes (2.5 x 109/ml) with a l/l00 dilution of rabbit anti-E antibody (“hemolysin”) for 15 min at 37°C. After washing these EA with GVB*+, they were treated at 37°C for 20min with 10 CHS, units/ml guinea-pig Cl. They were then quickly washed twice more and resuspended in GVB*+ with 70 pg/ml purified human C4. Deposition was allowed to proceed for 30 min at 37°C before washing with PBS containing 10mM EDTA to remove Clr, Cls and unbound C4. Use of ‘2SI-labeled C4 established that approximately 2500 molecules of C4b were fixed per cell. EAClq were prepared as control cells exactly as EAClq4b but with the omission of C4. To determine hemolysis, cells were diluted to 1 ml in PBS/10 mM EDTA at 4°C pelleted by centrifugation and the absorption of the supernatant measured at 415 nm. One hundred % hemolysis was measured by lysing the cells with 0.1% (v/v) nonidet P40.
The hemolysis described above suggests that C3b is deposited at the cell surface by local activation of the alternative pathway. This was examined using C5depleted human serum (C5D) to prevent complementmediated cell lysis. Initially, trace amounts of [‘251]-labeled C3 were added to C5D, followed by incubation with EAClq4b (as in above experiments). However, no deposition of radiolabel was detected, presumably due to the low specific activity (CSD contains approximately 1 mg/ml unlabeled C3). Therefore a more sensitive hemolytic assay for cell-bound C3b was devised. In this method C3b fixed by C5D is converted by the addition of purified components, to C3bBbP, in order to trigger the terminal complement pathway and thereby elicit hemolysis by serum-
Binding studies
Binding studies were performed by incubating the cells with radiolabeled components in 50 ~1 of PBSdextrose containing 2 mg/ml BSA, for 30 min at 22°C and then pelleting them by centrifugation (lO,OOOg, 30 sets) through 0.2 ml cushions of 20% sucrose in the same buffer. The tubes were frozen, the pellets cut off and both portions counted. After background subtraction, the percentage of bound radiolabel was derived. Binding studies were normally performed as single point determinations, but standard deviations of lo-30% (of bound counts) were obtained when triplicates were used. Aggregated forms of properdin give anomalous results and were removed from the purified proteins before use as previously described (Farries et al., 1987). In addition, purified proteins
RESULTS
Hemolysis
of EAClq4b
via the alternative
pathway
Using our EAClq4b cells we have repeated the experiment of Matsushita and Okada (1986) and obtained the same result. As shown in Fig. 1, these cells were lysed by normal human serum in Mg*+ EGTA buffer, indicating a dependence on the alternative pathway of complement activation. Adding extra antibody (2.5 x original input) to the cells after C4b deposition did not affect lysis. Furthermore, cells pretreated with serum-EDTA, which converts C4b to C4d and C4c, were also not lysed (not shown).
100 90 80
70 2 60 .I 3 50 pD 40 30
1
40
1 20
1
10
1 5
Serum Dilution
Fig. I. EAClq4b cells activate the alternative pathway. EAClq and EAClq4b cells (1.25 x 10s cells/ml) were mixed with serial dilutions of human serum in PBS containing 2 mM MgCl, and 8 mM EGTA. After 90 min incubation at 37°C. cell lysis was determined as described in Materials and Methods. EAClq, n ; EAClq4b, 0.
Alternative pathway activation by cell-bound C4b
1157
Test for EAClq4bBb Formation -I
““-lo1
0
L 20
40
CSD-Serum Dilution
I
Il__ B
Fig. 2. C3b deposition onto EAClq4b by CSD-serum. EACIq and EAClq4b cells (5 x 108/ml) were mixed with serial dilutions of CSD-serum in GVB2+/2 mM MgC1,/8 mM EGTA (50 ~1). After 20 min incubation at 37°C deposited C3b was detected hemolytically as follows. The cells were washed twice and resuspended in 50~1 GVB’+/0.5 mM NiCl, containing 50 fig/ml factor B, 2 pg/ml factor D and 6 pg/ml properdin. These mixtures were incubated for 5 min to form nickel/properdin-stabilized C3 convertases. Then 0.95 ml 1% (v/v) normal guinea pig serum in VBS2+/10 mM EDTA was added and cell lysis determined after a further 15 min at 37°C. EAClq, n ; EAClq4b, 0. (Fig. 2). C3b was deposited onto EAClq4b in a dose-dependent manner and not onto EAClq, as predicted and was inhibited by EDTA. EDTA
Test for C4bBb
“hybrid”
convertase
One hypothetical mechanism for alternative pathway activation by EAClq4b cells is by formation of a “hybrid” C3jC5 convertase, C4bBb (or C4bBbP), on the cell surface. Matsushita and Okada (1986) detected no hemolysis by serum-EDTA after treatment of the EAClq4b cells with factors B and D. This method should detect convertases with activity against both C3 and CS, but not convertases with activity only against C3. C3 cleavage (using purified C3) results in C3b deposition which can be detected as described in Fig. 2. The second cycle of B, D and P addition greatly amplifies the sensitivity. As shown in Fig. 3, there is very little difference between EAClq4b and EAClq cells in this assay. Thus, only background lysis resulting from non-specific associations or incomplete washing occurred. This result argues against the formation of a “hybrid” convertase. Furthermore, the C3 deposition that is observed is not dependent on the first cycle of B, D and P addition, when C4bBb would be formed. This result also implies that there is very little difference (< 25%) in sensitivity to the terminal components between EAClq4b and the control cells. This is insufficient to account for the large difference in hemolysis shown in Fig. 1. Binding studies with EAC 1q4b cells
Additional mechanisms for the interaction of C4b with the alternative pathway were considered. Purified
C
D
Fig. 3. Test for EAClq4bBb formation. EAClq (open bar) and EAClq4b (solid bar) cells (5 x log/ml) were mixed with 50 fig/ml factor B, 2 pg/ml factor D and 6 pg/ml properdin in GVB*+/0.5 mM &I, (50 ~1). After Id min incubation 25 ul of PBS/40 mM EDTA and 25 ,ul C3 (3 . ma/ml in PBS) were added. After 30 min at 37’C. C3b deposition was detected hemolytically as described in Fig. 2. Negative controls involving the omission of one reaction component were included: A, complete reaction sequence; B, no B/D/P in the first incubation; C, no C3; D, no B/D/P added for C3b detection. I,
components were tested for binding to EAClq4b cells under various conditions. Low ionic strength was used to enhance weak affinities. Figure 4 shows that iC3 binds to EAClq4b cells. The binding is inhibited by an anti-C4 antibody and by excess iC3. The Scatchard plot (Fig. 4b) indicates approximately 600 binding sites per cell (compatible with the estimated 2500 molecules of C4b/cell), and a dissociation constant of about IO-’ M under these non-physiological conditions (low ionic strength, 22’C). Two similar but independent experiments both produced consistent [‘251]-iC3 binding to EAClq4b that was 2-3 times that to control cells. iC3 binding could also be demonstrated hemolytically. If the cells were treated with 1.4 mg/ml iC3 (37°C 90min in VBS’+), washed twice and bound iC3 developed hemolytically with B, D and P followed by serum-EDTA (as described in the legend to Fig. 2) 33% lysis of EAClq4b occurred, compared to only 13% of EAClq. It is therefore concluded that iC3 has a weak affinity for C4b. Binding of purified iC3 was clearly detectable under a variety of conditions. However, C3b is probably more significant in uivo. C3b is structurally and functionally homologous to iC3. Therefore it also would be expected to bind to C4b. The results for C3b and properdin binding are shown in Fig. 5. In both cases low, but reproducible, levels of association were detected. (i) C3b binding to EAClq4b was only detectable in the presence of properdin and factor B. Properdin is polyvalent for C3b binding (Farries et al., 1987) and might link C3b molecules into polyvalent (i.e. higher avidity) C4b-binding complexes. The enhancement
TIMOTHY C. FARRSES et al.
1158
(4
0
I 15
I 1 5 1.7 iC3 (pglml)
I 0.56
-
150
450
300
600
Bound (MOlSCulSS/~Sli)
Fig. 4. iC3 binding to EAClq4b cells. Serial dilutions of [1251]-iC3(15, 5, 1.67 and 0.56pg/ml) were incubated with EAClq and EAClq4b cells (109/ml) at 22°C for 30 min, in the presence and absence of 0.6 mg/ml anti-C4 IgG. Binding was then determined as described in Materials and Methods. Figure 4b shows the same data as a Scatchard plot. EAClq, 0; EAClq4b, A; EAClq -tankC4, n; EAClq4b + antX4, *.
by factor B is consistent with its ability to stabilize properdin binding to C3b (Fairies et al., 1988). The C3b binding to EAClq4b was inhibited by the anti-C4 antibody. Binding of iC3 was also greatly enhanced (lo-20 fold) by properdin (not shown, but see Fig. 6B). As expected, binding of radiolabeled C3b in the presence of B and P could be completely abolished by the presence of 0.4 mg/ml unlabeled C3b. This is consistent with the inhibition of iC3 binding by cold iC3 shown in Fig. 4. iC3 binding was also enhanced by properdin (not shown). Aggregated properdin, which has a higher avidity for C3b than native properdin (Farries et al., 1987) enhanced the binding of C3b to EAClq4b cells in each of three separate experiments. While this amplified effect cannot occur in uiuo, it probably highlights the much weaker natural interaction that is physiologically significant. (ii) With properdin, a weak affinity could be A
detected
but only at five times the ceil concn used for
C3b. The binding to EAClq4b was consistently about twice that to EAClq (and 2-3 times background in four other independent expe~ments). However, as shown in Fig. 5B, anti-C4 IgG was no more inhibitory than a control antibody. Anti-C3 IgG also had no effect (not shown). If properdin was binding to trace amounts of C3b on the cells (no C3b was detectable, e.g. by factor B binding), the affinity would have been greatly increased by the presence of factors B and D, and binding would have been competitively inhibited by fluid-phase C3b (Farries et al., 1988). In fact C3b and factors B and D had little or no effect on properdin binding. Further studies showed that C3b had no effect at any concn between 1 and 140 pg/ml (not shown). An affinity of P for the cells does provide another explanation for its positive effect on C3b-binding. B
06-
0.5 -
3 ;;
0.4 -
0” “s
03-
$
-
P
B
P*B
P*B
+D
P+B +A1
P+B +A2
-
C3b
B
C3b+B C3b*B CJb+B C3b4 +O
+A1
qP.2
Fig. 5. Binding of C3b and properdin to EAClq4b cells. A. C3b. C3b binding was performed using 2.5 x 10s cells/ml in PBS-dextrose containing 2 mg/ml BSA and 2 mM MgCl,, as described in Materials and Methods. The concns of the other components, included as indicated, were: properdin, 4 pgjml; factor B, 250 pg/ml; factor D, 7 pg/ml; antibodies, 60 pg/ml (total IgG). Al, anti-C4 IgG; A2, control antibody. EAClq, q; EAClqrlb, n . B. Properdin. Properdin binding was performed as for C3b, but using I.25 x 10‘)cells/ml. Factor B, 60 pg/ml; C3b, 28 pg/ml; factor D, 7 jug/ml; Al, A2 and symbols as for Fig. SA.
Alternative pathway activation by cell-bound C4b ,
B 0.45 7
P
I C3b
*
P
-
c3
P
C3i
Fig, 5. C3 fixation oato EACtq4b by C3bBbF. A. ~~l-~~nd convertase. EACIq and EACfq4b cells of C3b (7 pg/ml), properdin (7.5 rg/mI), factor B (74 $g/ml) and factor D (3 pg/ml) as indicated in PBS/2 mg/ml BSA/2 mM MgCl, (50 ~1). After 10 min incubation at 22”C, 5Ohi of PBS/2 mg/ml BSA/ZO mM EDTA containing [‘*‘I]-labeled C3 (1 &i) was added. C3b (or iC3) deposition was determined after 30min as described in Materials and Methods. EAClq, a; EACfq4b, 8. B. Comparison of C3b, C3, iC3. EACIq and EAClq4b cells were incubated with or without properdin (as above) and then either radiolabeled C3b, C3 or iC3 (I @Zi each in ~BS~BSA~EDTA) was added. The reaction was completed as above. The iC3 was prepared by multiple @fold) freeze-thaws of the [‘?si]-C3 to ensure identical specific activities. The C3b was iodinated separately. EAClq, a; EAClq4b, W. (I.25 x t@/‘iml) were mixed with combinations
(iii> Even at the higher cell concn, factor 3 binding to EAClq4b was never reproducibly higher than to EACIq. The lack af detectable C4bBb enzymic activity, described above, also suggests that no functional factor B-C4b interaction can occur. These data imply chat the alternative pathway C3 convertase, C3bBbP, may bind to C4b-coated cells through an interaction of C4b with C3b. The functional implications of this hypothesis were therefore examined. Eflect of surface-bound C3 convertase The binding of the C3 convertase complex, CSbBbP, and its precursor, C3bBP, to EAClq4b &Is should localize C3 conversion to the cell surface and hence amplify C3b deposition. This was tested by fh-st treating the ceils with purified C3b, properdin and factors B and D, in the presence of magnesium. Having thus allowed cell-bound C3 convertases to form, EDTA and radiotabeled C3 were added and subsequent deposition of radiolabel determined. The results are shown in Fig. 6A. Approx~mat~iy ten times as much C3 was fixed by EAClq4b as by EAClq. This reaction was reproducible and totally dependent on properdin, but only part was dependent on C%, B and D. This is explained by inevirabfe contamination of the radiolabeled C3 with iC3 that binds to C4b in the presence of properdin (as described above). Thus, about hatf of the C3 fixation actuaffy represents iC3 binding. This was confirmed by comparing active and inactive C3 (iC3) in the same system (Fig. SB). The P-dependent fixation of active C3 was about 50% that of iC3, consistent with this level
of i~3-c~ntamjnation of the C3 p~paration* This reaction was not affected by the presence of Mg2+ or EDTA. The other half of the effect in Fig. 6A is dependent on all the ~mponen~. This impor~nt resutt was highly reproducible with C3 fixation of 4-8 times control values being found in four more independent experiments. One of these is shown in Fig. 7. In this case the EAClq4b were prepared with C4 from an unrelated purification, and the concentrations of components used are siightfy different. The C3 deposition and dependence on P, C3b, 3 and D was very similar to that shown in Fig. 6A. This virtually excludes the possibility that the results are
+B+D
C3b e&D
P +B+D
P*C3b
Fig. 7. C3 fixation onto EAClq4b by CJbBbP. The protocol and reaction conditions are as described in the legend to Fig. 6A with the exceptions that a digerent preparation of C4 was used is the preparation of EAClq4b, and the other components were used at the following concns: factor 8, 12 @g/ml; factor D, 5.5 pgjml; properdin, 5 gggjml; C3b, 12~g/mI; [‘zsftlabeIed C3, OS4pCi@mple. EACIq, a; EACIqilb, m.
1160
Tmom
C. FARR~ES et al.
attributable to any trace contaminants in the C4 preparation. The hypothesis that cell-bound C4b activates the alternative pathway by localizing the C3 convertase to the cell surface is therefore supported. Further controls demonstrated that the reaction was blocked by anti-C4 but not by an irrelevant antibody (anti-alpha 2-macroglobulin). Anti-C4 was not inhibitory if added after convertase formation. Therefore the extra C3 deposition is not caused by newly generated C3b binding to C4b. As shown in Fig. 5, the non-covalent binding reaction requires factor B. C3b binding to factor B does not occur in EDTA which is present during radiolabeled C3b fixation. It was still considered possible that the C3 fixation to EAClq4b cells was caused by preferential covalent fixation of nascent C3b to C4b. However, after the reaction, if the cells were lysed with Hz0 and the washed membranes dissolved in 0.1% SDS, only 2% of the radiolabel (C3) could be precipitated with anti-C4 IgG plus Staphylococcus uureus. Furthermore, when a purely fluid-phase convertase, CVFBb, was used, and C3 deposition assayed hemolytically, only 19.3% EAClq4b cells lysed compared to 15.1% of EAClq cells. Therefore fixation of nascent C3b to C4b can only account for a small proportion of the excess deposition onto EAClq4b cells. Instead, the C3bBbP C3 convertase expresses more activity towards the cell surface as a whole, presumably by localization to that surface. DISCUSSION
The results described in this paper have clarified how C4b-coated erythrocytes activate the alternative complement pathway. Firstly, evidence has been obtained against several possible mechanisms. Using a very sensitive assay, no C3 convertase activity was generated by EAClq4b cells in the presence of properdin and factors B and D. Therefore, the hypothetical “hybrid” C3 convertase, C4bBb (or C4bBbP), is not involved. In addition, C4b was found not to be a potent acceptor site for nascent C3b. A combination of the components that form the alternative pathway C3 convertase, C3b, B, D and P, did preferentially deposit C3b onto C4b-bearing cells, but only 2% of deposited C3 was attached to C4b. The alternative explanation is that the convertase, C3bBbP, is localized to the cell surface. Binding studies with radiolabeled C3b supported this hypothesis. A small amount of C3b bound to EAClq4b cells in the presence of B and P. iC3, which is functionally very analogous to C3b (Isenman et al., 1981; Pangburn et al., 1981; Von Zabern et al., 1981), was also shown to bind C4b, both by a sensitive hemolytic assay and by direct association of the radiolabeled protein. This reaction could be detected in the absence of B and P and was specifically inhibited by antibodies to C4. Properdin, a polyvalent C3b-binding protein, greatly enhanced the interaction. Its role might then be to
cross-link several C3b (or iC3) molecules on the surface into polyvalent complexes. Factor B also has a positive effect which can be explained by its stabilization of P-C3b complexes (Farries et al., 1988). Curiously, properdin itself bound slightly but significantly better to EAClq4b cells than to EAClq cells (in the absence of C3b). As C4b is structurally homologous to C3b, a weak P-C4b interaction might occur. However, there was no inhibition by anti-C4 antibodies or, competitively by fluid-phase C3b (fluidphase C4b would presumably be an even weaker competitor than C3b). Perhaps C4b influences the surface in another way, e.g. by affecting the release of Clr,Cls, complexes (Malmehedenyman et al., 1982). Hence P may have a preferential affinity for C4b-coated surfaces without directly binding C4b. Either case provides another explanation for the positive effect of P on C3b binding. There is also considerable circumstantial evidence for C3b-C4b interactions in the published work from other laboratories. Firstly, C5 convertases have recently been shown to be formed by nascent C3b binding covalently to C4b and C3b in C4b2a and C3bBb C3 convertases respectively (Takata et al., 1987; Kinoshita et al., 1988). Preformed C3b, which cannot bind covalently, can also generate C5 convertases (Vogt et al., 1978; Isenman et al., 1980). Therefore, it probably acts by noncovalent binding to C4b or C3b. Secondly, Osterberg et al. (1985) detected dimerization of iC3 and iC4 in the presence of low concentrations of the denaturant lauryl sulfate. As C4 and C3 are highly homologous, iC3-iC4 associations probably occur under the same conditions, reflecting weaker interactions that exist without denaturation. The presence of partially denatured molecules would also explain the much higher apparent affinity of our iC3 preparation for C4b (and C4bP) than that of our C3b. Finally, C3 and C4 are both homologous to the protease inhibitor alpha-2-macroglobulin (SottrupJensen et al., 1985) which exists as tetramers in plasma (formed from two disulfide-linked dimers). The homology extends over nearly all the protein sequence, so a similar, if weaker, binding site could be predicted in C3 and C4. A related protein in plaice plasma, with a similar chain structure to human C3, has been characterized as a non-covalently linked dimer (Starkey, 1983). We hypothesize that EAClq4b cells activate the alternative pathway by binding the C3 convertase C3bBbP, thereby concentrating C3 conversion at their surfaces. Amplified C3 deposition would result. This mechanism would be more effective still if the C3b-bound convertase was protected from inactivation by factor H. Theoretically, there would seem to be little room for factor H to combine with a C3b molecule that is already associated with three other proteins (C4b, Bb and P). However, initial studies have found that EAClq4b/C3bBbP has a similar factor H sensitivity as EC3bBbP, which is known to be unprotected (Fearon and Austen, 1977; Pangburn
Alternative pathway activation by cell-bound C4b
and Muller-Eberhard, fully resolved.
1978). This issue has yet to be
REFERENCES
Farries T. C.. Finch J. T., Lachmann P. J. and Harrison R. A. (1987) Resolution and analysis of “native” and “activated” nronerdin. Biochem. J. 243. 507-517. Farries T. C., Lachmann P. J. and Harrison R. A. (1988) Analysis of the interactions between properdin, the third component of complement (C3), and its physiological activation products. Biochem. J. 252, 47-54. Fearon D. T. and Austen K. F. (1977) Activation of the alternative complement pathway with rabbit erythrocytes by circumvention of the regulatory action of endogenous control proteins. J. exp. Med. 146, 22-33. Fraker P. J. and Speck J. C. (1978) Protein and cell membrane iodinations with a sparingly soluble chloramide, 1,3,4,6-tetrachloro-3a,6a-diphrenylglycoluril. Biochem. biophys. Res. Comm. 80, 8499857.
Harrison R. A. and Lachmann P. J. (1986) Complement technology. In Handbook of Experimental Immunology, 4th edn (Edited by D. M. Weir, L. A. Herzenberg, C. C. Blackwell and L. A. Herzenberg), pp. l-49. Blackwell Scientific Publications, Oxford. Isenman D. E., Kelly D. I. C., Cooper N. R., MullerEberhard H. J. and Pangburn M. K. (1981) Nucleophilic modification of human complement component C3: correlation of confirmational changes with acquisition of C3b-like functional properties. Biochemistry 20, 4458-4467.
Isenman D. E., Podack E. R. and Cooper N. R. (1980) The interaction of C5 with C3b in free solution: a sufficient condition for cleavage by fluid-phase C3/C5 convertase. J. Immun. 124, 326-331.
Kinoshita T., Takata Y., Kozono H., Takeda J., Hong K. and Inoue K. (1988) C5 convertase of the alternative complement pathway: covalent linkage between two C3b molecules within the trimolecular complex enzyme. J. Immun. 141, 389553901. Malmehedenyman I., Stalenheim G. and Sjoquist J. (1982) The effect of C4 on Cl binding and activation. Stand. J. Immun. 15, 587-593.
Matsushita M. and Okada H. (1986) Alternative comple-
1161
ment pathway activation by C4b deposited during classical pathway activation. J. Immun. 136, 2994-2998. Osterberg R., Nilsson U. R. and Eggerstsen G. (1985) Dimerization of human complement proteins C3 and C4 in dilute lauryl sulfate buffer after reaction with methylamine. J. biol. Chem. 260, 12,970-12,973. Pangburn M. K. and Muller-Eberhard H. J. (1978) Complement C3 convertase: cell surface restriction of beta 1H control and generation of restriction on neuraminidase treated cells. &oc. natn. Acad. Sci. U.S.A. 75, 2416-2420. Pannbum M. K.. Schreiber R. D. and Muller-Eberhard H: J. (1981) Formation of the initial C3 convertase of the alternative complement pathway. Acquisition of C3b-like activities by spontaneous hydrolysis of the putative thioester in native C3. J. exp. Med. 154, 856-867. Sottrup-Jensen L., Stepanik T. M., Kristensen T., Lonblad P. B., Jones C. M., Wierzbicki D. M., Magnusson S., Domdey H., Wetsel R. A., Lundwall A., Tack B. F. and Fey G. H. (1985) Common evolutionary origin of alpha2-macroglobulin and complement components C3 and C4. Proc. natn. Acad. Sci. U.S.A. 82, 9-13. Starkey P. M. (1983) The evolution of human alpha-2macroglobulin: further evidence for a common ancestry with complement proteins. Ann. N.Y. Acad. Sci. 421, 112-118.
Steuer K. L. K., Sloan L. B., Oglesby T. J., Farries T. C., Nickells M. W., Densen P., Harley J. B. and Atkinson J. P. (1989) Lysis of sensitized sheep erythrocytes in human sera deficient in the second component of complement. J. Immun. 143, 22562261.
Takata Y., Kinoshita T., Kozono H., Takata J., Takata E., Hong K. and Inoue K. (1987) Covalent association of C3b with C4b within C5 convertase of the classical complement pathway. J. exp. Med. 165, 1494-1507. Vogt W., Schmidt G., Von Buttlar B. and Dieminger K. (1978) A new function of the activated third component of complement: binding to C5, an essential step for C5 activation. Immunology 34, 29-36. Von Zabern I., Nolte R. and Vogt W. (1981) Treatment of human complement components C4 and C3 with amines or chaotropic ions. Evidence of a functional and structural change that provides uncleaved C4 and C3 with properties of their soluble activated forms. C4b and C3b. Stand. J. Immun. 13,413-431.
I I, pp. 1155-I
161, 1990
0161-5890/90 $3.00 + 0.00 Pergamon Press plc
THE MECHANISM OF ACTIVATION OF THE ALTERNATIVE PATHWAY OF COMPLEMENT BY CELL-BOUND C4b* TIMOTHYC. FARRIES,~~K. LEEKNUTZENSTEURR~and JOHNP. ATKINSON~~ tThe Howard Hughes Medical Institute Laboratories and Department of Medicine, Division of Rheumatology, Washington University School of Medicine, St Louis, MO 63110, U.S.A. (First received 20 October 1989; accepted in revised form 30 April 1990)
Abstract-Investigations into the mechanism of alternative pathway-dependent lysis of C4b-coated cells are reported. Test cells (EAClq4b) were formed by reaction of sheep erythrocytes with antibody, Cl and C4. In (X-deficient serum, more C3b was deposited onto EAClqC4b than onto control cells (EAClq). The possibility that the C4bBb enzyme could form was considered, but no C3 convertase activity was generated when magnesium, properdin and factors B and D were added to EAClqC4b. Binding studies employing radiolabeled components provided evidence that C4b bound the C3 convertase, C3bBbP, through a weak interaction with C3b. These data implied C3 conversion would be localized to the cell surface, thereby amplifying C3b deposition, This could be demonstrated in vitro. C3b, properdin, factor B and factor D were all required and the amplified C3b deposition was not due to deposition onto C4b itself. In serum, CS convertase activity would be consequently expressed and cell lysis would result. This could be the mechanism by which the sera of CZ-deficient patients mediate lysis of antibody coated sheep erythrocytes.
INTRODUCTION Sera from human patients with no detectable C2 activity have unexpectedly significant complementmediated hemolytic activity towards antibody-coated sheep erythrocytes (Steuer et al., 1989). This effect has been shown to be dependent on antibody, on the early components of classical pathway activation, Cl and C4, and on the alternative pathway components B, D and P. It was further demonstrated that the phenomenon did not involve residual C2. The initial stages of the reaction are presumably the same as those of the classical pathway; antibody clusters on the cell activate Cl which in turn cleaves fluid-phase C4. Antibody is required in high concn (compared to classical pathway activation), suggesting that a higher level of C4 cleavage is required. Some of the product, C4b, deposits on the cell surface. If C2 was present, it would also be cleaved to form a bimolecular C3-cleaving enzyme, C4b2a. It is now speculated that C4b can also interact with the alternative pathway to
produce C3 activation without the involvement of C2. This possibility is further suggested by the work of Matsushita and Okada (1986) who reported that C4b-coated erythrocytes are lysed by human serum via the alternative pathway. Cl was not required for this latter reaction. We have therefore sought to further define the mechanism. Strong structural homologies exist between C2a and Bb, between C3b and C4b, and between the C3 convertases, C4b2a and C3bBb. It has therefore been suggested (Matsushita and Okada, 1986) that a hybrid convertase, C4bBb, might form and cleave C3. Another explanation would be that C4b is a favored “acceptor” site for nascent C3b, leading to generation of C3bBb on C4b-coated surfaces. Formation of covalent C4b-C3b dimers has been observed (Takata et al., 1987). However, until now, no detailed experimental analysis of the mechanism has been performed. Our data support a new hypothesis; that C4b binds the C3 convertase non-covalently, and that C3b deposition is thereby localized to the cell surface.
*This work was supported
in part by The American Heart Association. IPresent address: MRC Collaborative Center, London, England. @Author to whom correspondence should be addressed: John P. Atkinson, Division of Rheumatology, 4566 Scott, Box 8045, St Louis, MO 63110, U.S.A. Abbreviations: E, sheep erythrocytes; EA, E coated with anti-E antibody; iC3, thioester-broken form of C3; P, properdin; B, factor B; D, factor D; CVF, cobra venom factor; C5D, CS-depleted human serum; NGPS, normal guinea pig serum; THC or CH,, , total hemolytic complement titer.
MATERIALSAND METHODS Buffers
and reagents
The reagents used were PBS, 8.7 mM K,HPO,, 1.7 mM NaH2P04, 150 mM NaCl, pH 7.4. PBS-dextrose is PBS diluted 1:2 in 5% (w/v) D-glucose. VBS2+ contains 3.1 mM diethyl barbituric acid, 0.9 mM Na barbitone, 145 mM NaCl, 0.83 mM MgCl,, 0.25 mM CaCl,, pH 7.2. GVB2+ is VBS*+ containing 0.1% (w/v) gelatin. 1155
1156
TIMOTHY C. FARRIES et al.
Human complement components properdin (“native” form), C3b, iC3 and factor B, D and H were purified as described elsewhere (Farries et al., 1987; Harrison and Lachmann, 1986). Purified C3 and C4 were the gift of R. P. Levine (St Louis, MO). C4 and CVF, provided by T. Seya (Osaka, Japan), were also used. Sheep erythrocytes (E), rabbit antibodies to E (“hemolysin”), and guinea-pig Cl were purchased from Diamedix (Miami, FL). Antibodies against C3, C4, C5 and alpha 2-macroglobulin were goat IgG preparations purchased from Atlantic Antibodies, Scarborough, MA. Radiolabeling of purified proteins with “‘1 was performed using the Iodogen (Pierce Chemical, Rockford, IL) procedure of Fraker and Speck (Fraker and Speck, 1978). The specific activities were typically 0.1-0.5 pCi/pg. CS-depleted human serum (C5D) was prepared by immunoadsorption with polyclonal goat anti-human C5 IgG. The IgG was coupled to CNBr-activated Sepharose (Sigma Chemical Co., St Louis, MO). After two cycles of absorption at 0°C the depleted serum did not lyse EA in VBS*+.
were preadsorbed with 2.5 x lo9 E/ml, and microfuged (lO,OOOg, 5 min), immediately before use to remove any partially denatured molecules that bind non-specifically. Fixation of radiolabeled C3 was measured similarly, with the addition of a wash with PBS before sedimenting the cells through sucrose.
EAC lq4b cells
C3b deposition via the alternative pathway
EA cells were made by treating washed sheep erythrocytes (2.5 x 109/ml) with a l/l00 dilution of rabbit anti-E antibody (“hemolysin”) for 15 min at 37°C. After washing these EA with GVB*+, they were treated at 37°C for 20min with 10 CHS, units/ml guinea-pig Cl. They were then quickly washed twice more and resuspended in GVB*+ with 70 pg/ml purified human C4. Deposition was allowed to proceed for 30 min at 37°C before washing with PBS containing 10mM EDTA to remove Clr, Cls and unbound C4. Use of ‘2SI-labeled C4 established that approximately 2500 molecules of C4b were fixed per cell. EAClq were prepared as control cells exactly as EAClq4b but with the omission of C4. To determine hemolysis, cells were diluted to 1 ml in PBS/10 mM EDTA at 4°C pelleted by centrifugation and the absorption of the supernatant measured at 415 nm. One hundred % hemolysis was measured by lysing the cells with 0.1% (v/v) nonidet P40.
The hemolysis described above suggests that C3b is deposited at the cell surface by local activation of the alternative pathway. This was examined using C5depleted human serum (C5D) to prevent complementmediated cell lysis. Initially, trace amounts of [‘251]-labeled C3 were added to C5D, followed by incubation with EAClq4b (as in above experiments). However, no deposition of radiolabel was detected, presumably due to the low specific activity (CSD contains approximately 1 mg/ml unlabeled C3). Therefore a more sensitive hemolytic assay for cell-bound C3b was devised. In this method C3b fixed by C5D is converted by the addition of purified components, to C3bBbP, in order to trigger the terminal complement pathway and thereby elicit hemolysis by serum-
Binding studies
Binding studies were performed by incubating the cells with radiolabeled components in 50 ~1 of PBSdextrose containing 2 mg/ml BSA, for 30 min at 22°C and then pelleting them by centrifugation (lO,OOOg, 30 sets) through 0.2 ml cushions of 20% sucrose in the same buffer. The tubes were frozen, the pellets cut off and both portions counted. After background subtraction, the percentage of bound radiolabel was derived. Binding studies were normally performed as single point determinations, but standard deviations of lo-30% (of bound counts) were obtained when triplicates were used. Aggregated forms of properdin give anomalous results and were removed from the purified proteins before use as previously described (Farries et al., 1987). In addition, purified proteins
RESULTS
Hemolysis
of EAClq4b
via the alternative
pathway
Using our EAClq4b cells we have repeated the experiment of Matsushita and Okada (1986) and obtained the same result. As shown in Fig. 1, these cells were lysed by normal human serum in Mg*+ EGTA buffer, indicating a dependence on the alternative pathway of complement activation. Adding extra antibody (2.5 x original input) to the cells after C4b deposition did not affect lysis. Furthermore, cells pretreated with serum-EDTA, which converts C4b to C4d and C4c, were also not lysed (not shown).
100 90 80
70 2 60 .I 3 50 pD 40 30
1
40
1 20
1
10
1 5
Serum Dilution
Fig. I. EAClq4b cells activate the alternative pathway. EAClq and EAClq4b cells (1.25 x 10s cells/ml) were mixed with serial dilutions of human serum in PBS containing 2 mM MgCl, and 8 mM EGTA. After 90 min incubation at 37°C. cell lysis was determined as described in Materials and Methods. EAClq, n ; EAClq4b, 0.
Alternative pathway activation by cell-bound C4b
1157
Test for EAClq4bBb Formation -I
““-lo1
0
L 20
40
CSD-Serum Dilution
I
Il__ B
Fig. 2. C3b deposition onto EAClq4b by CSD-serum. EACIq and EAClq4b cells (5 x 108/ml) were mixed with serial dilutions of CSD-serum in GVB2+/2 mM MgC1,/8 mM EGTA (50 ~1). After 20 min incubation at 37°C deposited C3b was detected hemolytically as follows. The cells were washed twice and resuspended in 50~1 GVB’+/0.5 mM NiCl, containing 50 fig/ml factor B, 2 pg/ml factor D and 6 pg/ml properdin. These mixtures were incubated for 5 min to form nickel/properdin-stabilized C3 convertases. Then 0.95 ml 1% (v/v) normal guinea pig serum in VBS2+/10 mM EDTA was added and cell lysis determined after a further 15 min at 37°C. EAClq, n ; EAClq4b, 0. (Fig. 2). C3b was deposited onto EAClq4b in a dose-dependent manner and not onto EAClq, as predicted and was inhibited by EDTA. EDTA
Test for C4bBb
“hybrid”
convertase
One hypothetical mechanism for alternative pathway activation by EAClq4b cells is by formation of a “hybrid” C3jC5 convertase, C4bBb (or C4bBbP), on the cell surface. Matsushita and Okada (1986) detected no hemolysis by serum-EDTA after treatment of the EAClq4b cells with factors B and D. This method should detect convertases with activity against both C3 and CS, but not convertases with activity only against C3. C3 cleavage (using purified C3) results in C3b deposition which can be detected as described in Fig. 2. The second cycle of B, D and P addition greatly amplifies the sensitivity. As shown in Fig. 3, there is very little difference between EAClq4b and EAClq cells in this assay. Thus, only background lysis resulting from non-specific associations or incomplete washing occurred. This result argues against the formation of a “hybrid” convertase. Furthermore, the C3 deposition that is observed is not dependent on the first cycle of B, D and P addition, when C4bBb would be formed. This result also implies that there is very little difference (< 25%) in sensitivity to the terminal components between EAClq4b and the control cells. This is insufficient to account for the large difference in hemolysis shown in Fig. 1. Binding studies with EAC 1q4b cells
Additional mechanisms for the interaction of C4b with the alternative pathway were considered. Purified
C
D
Fig. 3. Test for EAClq4bBb formation. EAClq (open bar) and EAClq4b (solid bar) cells (5 x log/ml) were mixed with 50 fig/ml factor B, 2 pg/ml factor D and 6 pg/ml properdin in GVB*+/0.5 mM &I, (50 ~1). After Id min incubation 25 ul of PBS/40 mM EDTA and 25 ,ul C3 (3 . ma/ml in PBS) were added. After 30 min at 37’C. C3b deposition was detected hemolytically as described in Fig. 2. Negative controls involving the omission of one reaction component were included: A, complete reaction sequence; B, no B/D/P in the first incubation; C, no C3; D, no B/D/P added for C3b detection. I,
components were tested for binding to EAClq4b cells under various conditions. Low ionic strength was used to enhance weak affinities. Figure 4 shows that iC3 binds to EAClq4b cells. The binding is inhibited by an anti-C4 antibody and by excess iC3. The Scatchard plot (Fig. 4b) indicates approximately 600 binding sites per cell (compatible with the estimated 2500 molecules of C4b/cell), and a dissociation constant of about IO-’ M under these non-physiological conditions (low ionic strength, 22’C). Two similar but independent experiments both produced consistent [‘251]-iC3 binding to EAClq4b that was 2-3 times that to control cells. iC3 binding could also be demonstrated hemolytically. If the cells were treated with 1.4 mg/ml iC3 (37°C 90min in VBS’+), washed twice and bound iC3 developed hemolytically with B, D and P followed by serum-EDTA (as described in the legend to Fig. 2) 33% lysis of EAClq4b occurred, compared to only 13% of EAClq. It is therefore concluded that iC3 has a weak affinity for C4b. Binding of purified iC3 was clearly detectable under a variety of conditions. However, C3b is probably more significant in uivo. C3b is structurally and functionally homologous to iC3. Therefore it also would be expected to bind to C4b. The results for C3b and properdin binding are shown in Fig. 5. In both cases low, but reproducible, levels of association were detected. (i) C3b binding to EAClq4b was only detectable in the presence of properdin and factor B. Properdin is polyvalent for C3b binding (Farries et al., 1987) and might link C3b molecules into polyvalent (i.e. higher avidity) C4b-binding complexes. The enhancement
TIMOTHY C. FARRSES et al.
1158
(4
0
I 15
I 1 5 1.7 iC3 (pglml)
I 0.56
-
150
450
300
600
Bound (MOlSCulSS/~Sli)
Fig. 4. iC3 binding to EAClq4b cells. Serial dilutions of [1251]-iC3(15, 5, 1.67 and 0.56pg/ml) were incubated with EAClq and EAClq4b cells (109/ml) at 22°C for 30 min, in the presence and absence of 0.6 mg/ml anti-C4 IgG. Binding was then determined as described in Materials and Methods. Figure 4b shows the same data as a Scatchard plot. EAClq, 0; EAClq4b, A; EAClq -tankC4, n; EAClq4b + antX4, *.
by factor B is consistent with its ability to stabilize properdin binding to C3b (Fairies et al., 1988). The C3b binding to EAClq4b was inhibited by the anti-C4 antibody. Binding of iC3 was also greatly enhanced (lo-20 fold) by properdin (not shown, but see Fig. 6B). As expected, binding of radiolabeled C3b in the presence of B and P could be completely abolished by the presence of 0.4 mg/ml unlabeled C3b. This is consistent with the inhibition of iC3 binding by cold iC3 shown in Fig. 4. iC3 binding was also enhanced by properdin (not shown). Aggregated properdin, which has a higher avidity for C3b than native properdin (Farries et al., 1987) enhanced the binding of C3b to EAClq4b cells in each of three separate experiments. While this amplified effect cannot occur in uiuo, it probably highlights the much weaker natural interaction that is physiologically significant. (ii) With properdin, a weak affinity could be A
detected
but only at five times the ceil concn used for
C3b. The binding to EAClq4b was consistently about twice that to EAClq (and 2-3 times background in four other independent expe~ments). However, as shown in Fig. 5B, anti-C4 IgG was no more inhibitory than a control antibody. Anti-C3 IgG also had no effect (not shown). If properdin was binding to trace amounts of C3b on the cells (no C3b was detectable, e.g. by factor B binding), the affinity would have been greatly increased by the presence of factors B and D, and binding would have been competitively inhibited by fluid-phase C3b (Farries et al., 1988). In fact C3b and factors B and D had little or no effect on properdin binding. Further studies showed that C3b had no effect at any concn between 1 and 140 pg/ml (not shown). An affinity of P for the cells does provide another explanation for its positive effect on C3b-binding. B
06-
0.5 -
3 ;;
0.4 -
0” “s
03-
$
-
P
B
P*B
P*B
+D
P+B +A1
P+B +A2
-
C3b
B
C3b+B C3b*B CJb+B C3b4 +O
+A1
qP.2
Fig. 5. Binding of C3b and properdin to EAClq4b cells. A. C3b. C3b binding was performed using 2.5 x 10s cells/ml in PBS-dextrose containing 2 mg/ml BSA and 2 mM MgCl,, as described in Materials and Methods. The concns of the other components, included as indicated, were: properdin, 4 pgjml; factor B, 250 pg/ml; factor D, 7 pg/ml; antibodies, 60 pg/ml (total IgG). Al, anti-C4 IgG; A2, control antibody. EAClq, q; EAClqrlb, n . B. Properdin. Properdin binding was performed as for C3b, but using I.25 x 10‘)cells/ml. Factor B, 60 pg/ml; C3b, 28 pg/ml; factor D, 7 jug/ml; Al, A2 and symbols as for Fig. SA.
Alternative pathway activation by cell-bound C4b ,
B 0.45 7
P
I C3b
*
P
-
c3
P
C3i
Fig, 5. C3 fixation oato EACtq4b by C3bBbF. A. ~~l-~~nd convertase. EACIq and EACfq4b cells of C3b (7 pg/ml), properdin (7.5 rg/mI), factor B (74 $g/ml) and factor D (3 pg/ml) as indicated in PBS/2 mg/ml BSA/2 mM MgCl, (50 ~1). After 10 min incubation at 22”C, 5Ohi of PBS/2 mg/ml BSA/ZO mM EDTA containing [‘*‘I]-labeled C3 (1 &i) was added. C3b (or iC3) deposition was determined after 30min as described in Materials and Methods. EAClq, a; EACfq4b, 8. B. Comparison of C3b, C3, iC3. EACIq and EAClq4b cells were incubated with or without properdin (as above) and then either radiolabeled C3b, C3 or iC3 (I @Zi each in ~BS~BSA~EDTA) was added. The reaction was completed as above. The iC3 was prepared by multiple @fold) freeze-thaws of the [‘?si]-C3 to ensure identical specific activities. The C3b was iodinated separately. EAClq, a; EAClq4b, W. (I.25 x t@/‘iml) were mixed with combinations
(iii> Even at the higher cell concn, factor 3 binding to EAClq4b was never reproducibly higher than to EACIq. The lack af detectable C4bBb enzymic activity, described above, also suggests that no functional factor B-C4b interaction can occur. These data imply chat the alternative pathway C3 convertase, C3bBbP, may bind to C4b-coated cells through an interaction of C4b with C3b. The functional implications of this hypothesis were therefore examined. Eflect of surface-bound C3 convertase The binding of the C3 convertase complex, CSbBbP, and its precursor, C3bBP, to EAClq4b &Is should localize C3 conversion to the cell surface and hence amplify C3b deposition. This was tested by fh-st treating the ceils with purified C3b, properdin and factors B and D, in the presence of magnesium. Having thus allowed cell-bound C3 convertases to form, EDTA and radiotabeled C3 were added and subsequent deposition of radiolabel determined. The results are shown in Fig. 6A. Approx~mat~iy ten times as much C3 was fixed by EAClq4b as by EAClq. This reaction was reproducible and totally dependent on properdin, but only part was dependent on C%, B and D. This is explained by inevirabfe contamination of the radiolabeled C3 with iC3 that binds to C4b in the presence of properdin (as described above). Thus, about hatf of the C3 fixation actuaffy represents iC3 binding. This was confirmed by comparing active and inactive C3 (iC3) in the same system (Fig. SB). The P-dependent fixation of active C3 was about 50% that of iC3, consistent with this level
of i~3-c~ntamjnation of the C3 p~paration* This reaction was not affected by the presence of Mg2+ or EDTA. The other half of the effect in Fig. 6A is dependent on all the ~mponen~. This impor~nt resutt was highly reproducible with C3 fixation of 4-8 times control values being found in four more independent experiments. One of these is shown in Fig. 7. In this case the EAClq4b were prepared with C4 from an unrelated purification, and the concentrations of components used are siightfy different. The C3 deposition and dependence on P, C3b, 3 and D was very similar to that shown in Fig. 6A. This virtually excludes the possibility that the results are
+B+D
C3b e&D
P +B+D
P*C3b
Fig. 7. C3 fixation onto EAClq4b by CJbBbP. The protocol and reaction conditions are as described in the legend to Fig. 6A with the exceptions that a digerent preparation of C4 was used is the preparation of EAClq4b, and the other components were used at the following concns: factor 8, 12 @g/ml; factor D, 5.5 pgjml; properdin, 5 gggjml; C3b, 12~g/mI; [‘zsftlabeIed C3, OS4pCi@mple. EACIq, a; EACIqilb, m.
1160
Tmom
C. FARR~ES et al.
attributable to any trace contaminants in the C4 preparation. The hypothesis that cell-bound C4b activates the alternative pathway by localizing the C3 convertase to the cell surface is therefore supported. Further controls demonstrated that the reaction was blocked by anti-C4 but not by an irrelevant antibody (anti-alpha 2-macroglobulin). Anti-C4 was not inhibitory if added after convertase formation. Therefore the extra C3 deposition is not caused by newly generated C3b binding to C4b. As shown in Fig. 5, the non-covalent binding reaction requires factor B. C3b binding to factor B does not occur in EDTA which is present during radiolabeled C3b fixation. It was still considered possible that the C3 fixation to EAClq4b cells was caused by preferential covalent fixation of nascent C3b to C4b. However, after the reaction, if the cells were lysed with Hz0 and the washed membranes dissolved in 0.1% SDS, only 2% of the radiolabel (C3) could be precipitated with anti-C4 IgG plus Staphylococcus uureus. Furthermore, when a purely fluid-phase convertase, CVFBb, was used, and C3 deposition assayed hemolytically, only 19.3% EAClq4b cells lysed compared to 15.1% of EAClq cells. Therefore fixation of nascent C3b to C4b can only account for a small proportion of the excess deposition onto EAClq4b cells. Instead, the C3bBbP C3 convertase expresses more activity towards the cell surface as a whole, presumably by localization to that surface. DISCUSSION
The results described in this paper have clarified how C4b-coated erythrocytes activate the alternative complement pathway. Firstly, evidence has been obtained against several possible mechanisms. Using a very sensitive assay, no C3 convertase activity was generated by EAClq4b cells in the presence of properdin and factors B and D. Therefore, the hypothetical “hybrid” C3 convertase, C4bBb (or C4bBbP), is not involved. In addition, C4b was found not to be a potent acceptor site for nascent C3b. A combination of the components that form the alternative pathway C3 convertase, C3b, B, D and P, did preferentially deposit C3b onto C4b-bearing cells, but only 2% of deposited C3 was attached to C4b. The alternative explanation is that the convertase, C3bBbP, is localized to the cell surface. Binding studies with radiolabeled C3b supported this hypothesis. A small amount of C3b bound to EAClq4b cells in the presence of B and P. iC3, which is functionally very analogous to C3b (Isenman et al., 1981; Pangburn et al., 1981; Von Zabern et al., 1981), was also shown to bind C4b, both by a sensitive hemolytic assay and by direct association of the radiolabeled protein. This reaction could be detected in the absence of B and P and was specifically inhibited by antibodies to C4. Properdin, a polyvalent C3b-binding protein, greatly enhanced the interaction. Its role might then be to
cross-link several C3b (or iC3) molecules on the surface into polyvalent complexes. Factor B also has a positive effect which can be explained by its stabilization of P-C3b complexes (Farries et al., 1988). Curiously, properdin itself bound slightly but significantly better to EAClq4b cells than to EAClq cells (in the absence of C3b). As C4b is structurally homologous to C3b, a weak P-C4b interaction might occur. However, there was no inhibition by anti-C4 antibodies or, competitively by fluid-phase C3b (fluidphase C4b would presumably be an even weaker competitor than C3b). Perhaps C4b influences the surface in another way, e.g. by affecting the release of Clr,Cls, complexes (Malmehedenyman et al., 1982). Hence P may have a preferential affinity for C4b-coated surfaces without directly binding C4b. Either case provides another explanation for the positive effect of P on C3b binding. There is also considerable circumstantial evidence for C3b-C4b interactions in the published work from other laboratories. Firstly, C5 convertases have recently been shown to be formed by nascent C3b binding covalently to C4b and C3b in C4b2a and C3bBb C3 convertases respectively (Takata et al., 1987; Kinoshita et al., 1988). Preformed C3b, which cannot bind covalently, can also generate C5 convertases (Vogt et al., 1978; Isenman et al., 1980). Therefore, it probably acts by noncovalent binding to C4b or C3b. Secondly, Osterberg et al. (1985) detected dimerization of iC3 and iC4 in the presence of low concentrations of the denaturant lauryl sulfate. As C4 and C3 are highly homologous, iC3-iC4 associations probably occur under the same conditions, reflecting weaker interactions that exist without denaturation. The presence of partially denatured molecules would also explain the much higher apparent affinity of our iC3 preparation for C4b (and C4bP) than that of our C3b. Finally, C3 and C4 are both homologous to the protease inhibitor alpha-2-macroglobulin (SottrupJensen et al., 1985) which exists as tetramers in plasma (formed from two disulfide-linked dimers). The homology extends over nearly all the protein sequence, so a similar, if weaker, binding site could be predicted in C3 and C4. A related protein in plaice plasma, with a similar chain structure to human C3, has been characterized as a non-covalently linked dimer (Starkey, 1983). We hypothesize that EAClq4b cells activate the alternative pathway by binding the C3 convertase C3bBbP, thereby concentrating C3 conversion at their surfaces. Amplified C3 deposition would result. This mechanism would be more effective still if the C3b-bound convertase was protected from inactivation by factor H. Theoretically, there would seem to be little room for factor H to combine with a C3b molecule that is already associated with three other proteins (C4b, Bb and P). However, initial studies have found that EAClq4b/C3bBbP has a similar factor H sensitivity as EC3bBbP, which is known to be unprotected (Fearon and Austen, 1977; Pangburn
Alternative pathway activation by cell-bound C4b
and Muller-Eberhard, fully resolved.
1978). This issue has yet to be
REFERENCES
Farries T. C.. Finch J. T., Lachmann P. J. and Harrison R. A. (1987) Resolution and analysis of “native” and “activated” nronerdin. Biochem. J. 243. 507-517. Farries T. C., Lachmann P. J. and Harrison R. A. (1988) Analysis of the interactions between properdin, the third component of complement (C3), and its physiological activation products. Biochem. J. 252, 47-54. Fearon D. T. and Austen K. F. (1977) Activation of the alternative complement pathway with rabbit erythrocytes by circumvention of the regulatory action of endogenous control proteins. J. exp. Med. 146, 22-33. Fraker P. J. and Speck J. C. (1978) Protein and cell membrane iodinations with a sparingly soluble chloramide, 1,3,4,6-tetrachloro-3a,6a-diphrenylglycoluril. Biochem. biophys. Res. Comm. 80, 8499857.
Harrison R. A. and Lachmann P. J. (1986) Complement technology. In Handbook of Experimental Immunology, 4th edn (Edited by D. M. Weir, L. A. Herzenberg, C. C. Blackwell and L. A. Herzenberg), pp. l-49. Blackwell Scientific Publications, Oxford. Isenman D. E., Kelly D. I. C., Cooper N. R., MullerEberhard H. J. and Pangburn M. K. (1981) Nucleophilic modification of human complement component C3: correlation of confirmational changes with acquisition of C3b-like functional properties. Biochemistry 20, 4458-4467.
Isenman D. E., Podack E. R. and Cooper N. R. (1980) The interaction of C5 with C3b in free solution: a sufficient condition for cleavage by fluid-phase C3/C5 convertase. J. Immun. 124, 326-331.
Kinoshita T., Takata Y., Kozono H., Takeda J., Hong K. and Inoue K. (1988) C5 convertase of the alternative complement pathway: covalent linkage between two C3b molecules within the trimolecular complex enzyme. J. Immun. 141, 389553901. Malmehedenyman I., Stalenheim G. and Sjoquist J. (1982) The effect of C4 on Cl binding and activation. Stand. J. Immun. 15, 587-593.
Matsushita M. and Okada H. (1986) Alternative comple-
1161
ment pathway activation by C4b deposited during classical pathway activation. J. Immun. 136, 2994-2998. Osterberg R., Nilsson U. R. and Eggerstsen G. (1985) Dimerization of human complement proteins C3 and C4 in dilute lauryl sulfate buffer after reaction with methylamine. J. biol. Chem. 260, 12,970-12,973. Pangburn M. K. and Muller-Eberhard H. J. (1978) Complement C3 convertase: cell surface restriction of beta 1H control and generation of restriction on neuraminidase treated cells. &oc. natn. Acad. Sci. U.S.A. 75, 2416-2420. Pannbum M. K.. Schreiber R. D. and Muller-Eberhard H: J. (1981) Formation of the initial C3 convertase of the alternative complement pathway. Acquisition of C3b-like activities by spontaneous hydrolysis of the putative thioester in native C3. J. exp. Med. 154, 856-867. Sottrup-Jensen L., Stepanik T. M., Kristensen T., Lonblad P. B., Jones C. M., Wierzbicki D. M., Magnusson S., Domdey H., Wetsel R. A., Lundwall A., Tack B. F. and Fey G. H. (1985) Common evolutionary origin of alpha2-macroglobulin and complement components C3 and C4. Proc. natn. Acad. Sci. U.S.A. 82, 9-13. Starkey P. M. (1983) The evolution of human alpha-2macroglobulin: further evidence for a common ancestry with complement proteins. Ann. N.Y. Acad. Sci. 421, 112-118.
Steuer K. L. K., Sloan L. B., Oglesby T. J., Farries T. C., Nickells M. W., Densen P., Harley J. B. and Atkinson J. P. (1989) Lysis of sensitized sheep erythrocytes in human sera deficient in the second component of complement. J. Immun. 143, 22562261.
Takata Y., Kinoshita T., Kozono H., Takata J., Takata E., Hong K. and Inoue K. (1987) Covalent association of C3b with C4b within C5 convertase of the classical complement pathway. J. exp. Med. 165, 1494-1507. Vogt W., Schmidt G., Von Buttlar B. and Dieminger K. (1978) A new function of the activated third component of complement: binding to C5, an essential step for C5 activation. Immunology 34, 29-36. Von Zabern I., Nolte R. and Vogt W. (1981) Treatment of human complement components C4 and C3 with amines or chaotropic ions. Evidence of a functional and structural change that provides uncleaved C4 and C3 with properties of their soluble activated forms. C4b and C3b. Stand. J. Immun. 13,413-431.
