Streptococcal protein H forms soluble complement-activating complexes with IgG, but inhibits complement activation by IgG-coated targets.

Protein H, a surface protein of Streptococcus pyogenes interacting with the constant Fc region of IgG, is known to be released from the streptococcal surface by a cysteine proteinase produced by the bacteria. Poststreptococcal glomerulonephritis and rheumatic fever are conditions in which immune complexes and autoimmune mechanisms have been suggested to play pathogenetic roles. The present study demonstrates that addition of protein H to human serum produces complement activation with dose-dependent cleavage of C3. The activation was IgG-dependent and the result of complexes formed between IgG and protein H. These complexes were size heterogeneous with molecular masses of 400 kDa to 1.4 MDa. Using complement-depleted serum reconstituted with complement proteins, the activation by protein H was found to be dependent of the classical, but independent of the alternative pathway of complement. In contrast to results of experiments based on soluble protein H.IgG complexes, complement activation was inhibited by protein H when IgG was immobilized on a surface. The interaction between C1q and immunoglobulins represents the first step in the activation of the classical pathway, and protein H efficiently inhibited the binding of C1q to IgG immobilized on polyacrylamide beads. Protein H reduced C3 deposition on the IgG-coated beads and inhibited immune hemolysis of IgG-sensitized erythrocytes. Finally, significantly less C3 was deposited on the surface of protein H-expressing wild-type streptococci than on the surface of isogenic mutant bacteria devoid of protein H. The results demonstrate that protein H.IgG complexes released from the streptococcal surface can produce complement breakdown at the sites of infection, whereas complement activation on bacterial surfaces is inhibited. This should have important implications for host-parasite relationships. In addition, soluble protein H.IgG complexes might contribute to immunological complications of streptococcal infections.

Protein H, a surface protein of Streptococcus pyogenes interacting with the constant Fc region of IgG, is known to be released from the streptococcal surface by a cysteine proteinase produced by the bacteria. Poststreptococcal glomerulonephritis and rheumatic fever are conditions in which immune complexes and autoimmune mechanisms have been suggested to play pathogenetic roles. The present study demonstrates that addition of protein H to human serum produces complement activation with dose-dependent cleavage of C3. The activation was IgG-dependent and the result of complexes formed between IgG and protein H. These complexes were size heterogeneous with molecular masses of 400 kDa to 1.4 MDa. Using complement-depleted serum reconstituted with complement proteins, the activation by protein H was found to be dependent of the classical, but independent of the alternative pathway of complement. In contrast to results of experiments based on soluble protein H⅐IgG complexes, complement activation was inhibited by protein H when IgG was immobilized on a surface. The interaction between C1q and immunoglobulins represents the first step in the activation of the classical pathway, and protein H efficiently inhibited the binding of C1q to IgG immobilized on polyacrylamide beads. Protein H reduced C3 deposition on the IgG-coated beads and inhibited immune hemolysis of IgG-sensitized erythrocytes. Finally, significantly less C3 was deposited on the surface of protein H-expressing wild-type streptococci than on the surface of isogenic mutant bacteria devoid of protein H. The results demonstrate that protein H⅐IgG complexes released from the streptococcal surface can produce complement breakdown at the sites of infection, whereas complement activation on bacterial surfaces is inhibited. This should have important implications for host-parasite relationships. In addition, soluble protein H⅐IgG complexes might contribute to immunological complications of streptococcal infections.
Streptococcus pyogenes is an important human pathogen (for a review, see Ref. 1) causing suppurative infections like pharyngitis, tonsillitis, impetigo, and erysipelas. S. pyogenes is also responsible for a hyperacute and serious toxic shock-like syndrome that is sometimes associated with fasciitis and myositis. Rheumatic fever and glomerulonephritis are delayed sequelae of acute S. pyogenes infections, and several observations suggest that immunological mechanisms contribute to these conditions. Antibodies against S. pyogenes surface components have been reported to cross-react with heart sarcolemma (2) and cardiac myosin (3). Patients with acute post-streptococcal glomerulonephritis (APSGN) 1 show circulating immunoglobulin (Ig) complexes (4), complement activation (4), and deposition of complement proteins in the glomeruli (5).
Complement plays an important role in defense against pathogenic microorganisms (6,7). Several functionally interesting interactions have been described for S. pyogenes and components of the complement system. M proteins are antiphagocytic fibrous surface proteins of S. pyogenes (for a review, see Ref. 8). Members of this protein family specifically bind to the complement proteins factor H (9) and C4b-binding protein (C4BP) (10) that regulate complement activation (11)(12)(13). S. pyogenes also expresses a surface-associated peptidase, which degrades C5a (14,15), a chemotactic fragment of C5 (16). Finally, some strains secrete protein SIC, which interacts with terminal complement proteins and inhibits complement-mediated lysis (17).
The starting point for the present investigation was the observation that an extracellular cysteine proteinase of S. pyogenes (SCP) releases a large fragment of protein H from the surface of the bacteria (18). Protein H is a streptococcal surface protein belonging to the M protein family and has high affinity for the constant (Fc) region of IgG (19,20). The structure of protein H is schematically depicted in Fig. 1. We report that complex formation between protein H and IgG in the fluid phase leads to complement activation. By contrast, the interaction of protein H and IgG on surfaces resulted in inhibition of complement function. Possible implications for virulence and for immunological disease mechanisms are discussed. of AP1 lacking surface-bound protein H, the plasmid pBK37 was used. To generate pBK37 an internal PCR fragment, covering about 2 ⁄3 of sph, the gene coding for protein H, was ligated into the XbaI site in pJRS233 (21), a derivative of the temperature-sensitive shuttle vector pG ϩ host4 (22). The PCR fragment was generated using oligonucleotides hybridizing with nucleotides 330 -350 and 1151-1132 (also 1025-1006) in the published sequence of sph (20), using AP1 chromosomal DNA as template. S. pyogenes strain AP1 was transformed with pBK37 by electroporation as described previously (23). To generate insertionally inactivated mutants the procedure of Perez-Casal et al. (21) was followed. Single colonies were established and screened for lack of IgG binding. Southern blot analysis was used to confirm the integration of the plasmid into sph. The sph gene was shown to be interrupted and the adjacent gene on the chromosome, emm1, coding for M1 protein (24), was shown to be intact in the mutant, as analyzed by PCR. This procedure resulted in a mutant, BM27.6, with a truncated sph gene, which lacks the COOH-terminal part necessary for integration into the cell wall (25). The Phadebact typing test (Pharmacia Biotech Inc., Uppsala, Sweden) established that the mutant was a S. pyogenes strain. The occurrence of protein H on the bacterial surface was investigated in binding experiments. Thus, AP1 or BM27.6 bacteria were incubated with 125 I-labeled IgG in 200 l of phosphate-buffered saline (PBS) containing 0.02% NaN 3 and 0.05% Tween 20, for 30 min, and washed in the same buffer. Following centrifugation the radioactivity of the pellets was measured. Dilutions of the bacteria revealed that the binding of IgG to BM27.6 was Ͼ1000 times lower than the binding of IgG to AP1. AP1 bacteria were grown in Todd-Hewitt broth (Difco) supplemented with 0.2% yeast extract, and BM27.6 bacteria in this medium containing 1 g/ml erythromycin (Sigma). Bacteria were heat-killed by incubation for 5 min at 80°C.
Proteins-Recombinant protein H was obtained by expression in Escherichia coli of a PCR-generated DNA fragment corresponding to the mature protein H except for the hydrophobic membrane-spanning region (see Fig. 1). Thus, PCR was performed using oligonucleotides synthesized from nucleotides 451-474 and 1374 -1351 in the published sequence of protein H (20) using AP1 chromosomal DNA as the template. PCR products resulting from amplification with these primers were ligated into the high expression vector pHD389 (26) as described (27). Genomic DNA preparation, ligation, and transformation procedures were as described (28). Plasmids were transformed into E. coli strain LE392 (28). Protein H was purified from cell lysates as described (19). Ig-binding proteins were purified as described previously: M1 protein (24), protein L (29), and protein G (30). Protein A (31) was from Pharmacia. Protein PAB from Peptostreptococcus magnus was purified as described (32). The isolation of C1q, factor D, and properdin was done as described (33)(34)(35). Myeloma IgG was isolated from human sera. Sheep erythrocytes were sensitized (36) with rabbit antiserum (Swedish National Bacteriological Laboratory, Stockholm, Sweden). IgG and IgM fractions of the rabbit antiserum were isolated by gel filtration (Ultrogel AcA22, LKB, Bromma, Sweden). F(abЈ) 2 fragments prepared from anti-C3c antibodies were from Dako (Glostrup, Denmark). Gelatin was from Bio-Rad, and human polyclonal IgG and albumin were from Sigma.
Sera-Blood was drawn from healthy volunteers. The blood was allowed to clot at room temperature for 60 min. After another 120 min at 4°C tubes were centrifuged and the serum collected and stored in aliquots at Ϫ80°C or used for the preparation of C1q, factor D, and properdin (C1qDP)-depleted serum as described (37). Briefly, serum was dialyzed against 50 mM phosphate buffer, pH 7.3, containing 2 mM EDTA, and NaCl to yield a final conductance of 13 millisiemens/cm and was applied to a Bio-Rex 70 column (Mesh size 200 -400, Bio-Rad). The material not bound to the column was dialyzed against 0.1 M Tris-HCl, pH 7.4, containing 0.4 M NaCl and 2 mM EDTA, and applied to a column of Sepharose-4B coupled with anti-properdin antibodies. The filtrate was dialyzed against veronal-buffered saline (VBS) (5 mM sodium 5,5diethylbarbiturate, pH 7.4, 145 mM NaCl) containing 0.15 mM Ca 2ϩ and 0.05 mM Mg 2ϩ (VBS 2ϩ ) and stored in aliquots at Ϫ80°C. Before dialysis, some of the C1qDP-depleted serum was used for preparation of IgG-depleted C1qDP-depleted serum. This reagent was prepared by absorption of immunoglobulins from the serum on a column of Sepharose-4B coupled with protein LG (38). The material not bound to the column was dialyzed against VBS 2ϩ and stored in aliquots at Ϫ80°C. The IgGC1qDP-depleted serum contained no detectable IgG (Ͻ0.6 g/ ml). Another C1qDP-depleted serum was prepared using buffer with a conductance of 12.5 millisiemens/cm instead of 13 millisiemens/cm during the initial ion exchange chromatography step. The final preparation contained no detectable C4BP or factor H, whereas other complement proteins (C1r, C1s, C2, C4, and C3) were present at concentrations that were similar to those found in the other C1qDP-depleted sera. This reagent, C4BPC1qDP-depleted serum, was used in some C3 deposition experiments. All plasma protein concentrations were determined by electroimmunoassay (37).
Crossed Immunoelectrophoresis, Ouchterlony Immunodiffusion, Gel Filtration, and SDS-PAGE-The capacity of the serum and the serum reagents to support fluid-phase cleavage of C3 during incubation with bacterial proteins was investigated with crossed immunoelectrophoresis (37,39). Normal serum, C1qDP-depleted serum, and IgGC1qDPdepleted serum were used at concentrations of 40%. IgG, C1q, factor D, properdin, and the bacterial proteins were added to yield final reaction mixture volumes of 100 l. The plasma proteins were added at physiological concentrations (IgG, 10 mg/ml; C1q, 70 g/ml; factor D, 1 g/ml; and properdin, 25 g/ml) with regard to undiluted serum. Following incubation for 30 min at 37°C, 5 l 0.2 M EDTA was added and the samples were put on ice, centrifuged, and subjected to 0.6% agarose gel electrophoresis. The gel corresponding to one lane was transferred to the second-dimensional plate. Separated proteins were electrophoresed into a gel containing anti-C3 antiserum. Immunoprecipitates were stained with Coomassie Brilliant Blue.
Double immunodiffusion according to Ouchterlony was performed in 1% agarose (SEA-KEM, FMC, Rockland, ME) in PBS. Diffusion was FIG. 2. Activation of complement factor C3 in serum by Igbinding proteins. Bacterial Ig-binding proteins were added to serum. Following incubation for 30 min at 37°C, the reactions were terminated by cooling and addition of 10 mM EDTA. The samples were analyzed by crossed immunoelectrophoresis utilizing anti-C3 antibodies, and the activation was expressed as the ratio, in percent, between activated C3 and total C3. The proteins analyzed were: proteins A (Ⅺ), G (å), H (E), L (Ç), and M1 (É).
FIG. 1. Schematic representation of protein H. The protein is expressed at the surface of some strains of S. pyogenes, and the IgGbinding activity is in the NH 2 -terminal A and B domains (57). The COOH-terminal D domain anchors protein H to the bacterial cell wall. Apart from IgG, protein H also shows affinity for albumin (57), factor H (23), fibronectin type III domains (58), and C4BP (50). These proteins have separate binding sites and interact independently with protein H. The E. coli-produced protein H fragment used in this study is indicated as well as the fragment released from the streptococcal cell surface by the SCP. allowed to proceed for 48 h at 4°C. The gels were immersed in PBS for 24 h and in distilled water for 60 min, dried, and stained with Coomassie Brilliant Blue. Gel filtration experiments were performed on an FPLC Superose-6 column (Pharmacia) equilibrated with PBS containing 0.02% NaN 3 . The flow rate was 0.25 ml/min, and 0.5-ml fractions were collected. Before SDS-PAGE (40) fractions were concentrated 10 times by precipitation in 10% trichloroacetic acid (Sigma) and boiled in sample buffer containing 2% SDS and 5% ␤-mercaptoethanol.
Competitive Binding Assays and C1q Binding to Bacteria-Polyclonal human IgG was coupled to polyacrylamide beads (Immunobeads, Bio-Rad) according to the producer's instructions. 125 I-Labeled C1q (33) in 0.1 ml of VBS and 0.1% gelatin, 0.1 ml of Immunobeads coupled with IgG, and 0.2 ml of various Ig-binding proteins in the same buffer were mixed and incubated overnight at 20°C. Two ml of the same buffer containing 0.01 M EDTA was added, beads were spun down, washed, and the radioactivity of the pellets was measured. A concentration of beads resulting in submaximal binding (60%) of 125 I-C1q to IgG was consistently used in these experiments. Data points represented the mean of duplicate determinations in a single experiment. All experiments were performed at least three times. The same procedure was followed in experiments with albumin-coupled polyacrylamide beads. In similar experiments AP1 bacteria (2 ϫ 10 8 cells) were preincubated with an excess of IgG for 30 min at 37°C. The cells were washed, radiolabeled C1q was added, and the radioactivity of the pellets was measured.
C3 Deposition Assay-Polyclonal human IgG coupled to polyacrylamide beads was incubated with bacterial proteins in 0.2 ml of PBS containing 0.25% gelatin, 0.25% Tween, and 0.25% bovine serum albu- min for 20 min, washed twice in VBS 2ϩ containing 0.1% gelatin, and resuspended in 100 l of the same buffer. 100 l of 20% serum was added, and the samples were incubated at 37°C for 20 min. Two ml of cold VBS containing 10 mM EDTA were added. The tubes were centrifuged and the pellets washed. The pellets were resuspended in 200 l of PBS containing 0.25% Tween and 30,000 cpm of F(abЈ) 2 anti-C3c antibodies 125 I-labeled with the chloramine-T method (41). After 3 h of incubation, 2 ml of the same buffer was added, the tubes were centrifuged, and the radioactivity of the pellets was measured. The same procedure was followed in experiments with albumin-coupled polyacrylamide beads. In other experiments the polyacrylamide beads were exchanged for 100 l of AP1 or BM27.6 bacteria (2 ϫ 10 5 cells), omitting the blocking step described for the beads. Data points represent the mean of duplicate determinations of a single experiment. Experiments were performed at least three times.
Hemolytic Assay-Optimally sensitized (IgG or IgM) sheep erythrocytes (2 ϫ 10 9 cells/ml) were incubated for 20 min with the bacterial proteins. The cells were centrifuged at 175 ϫ g for 7 min, washed once in VBS 2ϩ , and resuspended in the same buffer to 2 ϫ 10 9 cells/ml. This suspension (100 l) was incubated with an equal volume of 20% serum for 20 min at 37°C. Three ml of cold VBS containing 10 mM EDTA were added, and the samples were centrifuged at 175 ϫ g for 7 min. The absorbance of the supernatants was measured at 541 nm. Data points represent the mean of duplicate determinations in a single experiment. Experiments were performed at least three times.

IgG-dependent Complement Activation in Serum by Protein
H-At the streptococcal surface a large IgGFc-binding fragment of protein H is released by SCP (18). To investigate if soluble protein H is capable of activating complement, purified protein H was added to serum. C3 breakdown was then studied by crossed immunoelectrophoresis. M1 protein is expressed at the bacterial surface together with protein H. M1 protein also FIG. 6. Molecular interactions between IgG, protein H, and C1q analyzed by gel filtration. A, protein H (top), human polyclonal IgG (middle), and a mixture of protein H and IgG (bottom) were run separately on a Superose-6 FPLC column. B, fractions indicated by arrows were concentrated ten times by trichloroacetic acid precipitation and analyzed by SDS-PAGE (10% gel run under reducing conditions). Bands corresponding to monomeric protein H and IgG heavy and light chains are indicated. Their apparent molecular masses are 42, 50, and 25 kDa, respectively. C, gel filtration experiments as in A were performed with the addition of C1q 125 I-labeled with lactoperoxidase (33). 125 I-C1q was also run separately on the column. Fractions of 0.5 ml from the four runs were collected and the radioactivity of each fraction was measured. C1q shows an unspecific binding to the matrix of the column why a substantial fraction of the radioactivity is retained and subsequently eluted at a volume exceeding 18 ml. binds to IgG, but compared with protein H, the affinity is much lower (24). Studies of M1 protein, and of strongly Ig-binding proteins (staphylococcal protein A, streptococcal protein G, and peptostreptococcal protein L) were included in the experiments. Fig. 2 shows that proteins A, G, H, and L all activate complement in a dose-dependent manner, when added to human serum. M1 protein, however, had no effect also at the highest concentration tested (35 M). C3 conversion could result either from activation of the classical or the alternative pathway or from a combination of both. To discriminate between these alternatives a C1qDP-depleted serum was utilized. Reconstitution of the serum with C1q (Fig. 3E) restored complement activation by protein H to the level obtained with intact normal serum (Fig. 3A). Reconstitution with factor D and properdin had no effect (Fig. 3D). The results demonstrated that fluidphase activation of C3 by protein H is mediated through the classical pathway.
The critical importance of IgG for protein H-induced complement activation in serum was shown with an IgGC1qDP-depleted serum (Fig. 4). No C3 cleavage was seen in this serum when protein H and complement proteins were added. The capacity to support protein H-induced complement activation was re-established with polyclonal or monoclonal IgG.
Protein H Forms Soluble C1q-binding Complexes with IgG-In contrast to proteins A, G, and L, protein H did not form distinct precipitation arcs with human IgG, when tested by double diffusion in agarose (Fig. 5). Diffuse staining was observed between the IgG-and protein H-containing wells, indicating the formation of heterogeneous complexes. Gel filtration experiments were performed to demonstrate and analyze soluble protein H⅐IgG complexes. Protein H was incubated with IgG at the equimolar concentrations shown to result in complement activation (Fig. 2). Proteins A, G, and L formed insol-uble precipitates when incubated with IgG in solution; no visible precipitates were formed between protein H and IgG. Protein H, IgG, and a mixture of the two proteins were each subjected to gel filtration on a Superose-6 column (Fig. 6A). Protein H tends to form multimers (42), which explains the broad peak. The protein H⅐IgG mixture gave rise to soluble complexes of predicted molecular masses ranging from 400 kDa to 1.4 MDa. Analysis by SDS-PAGE (Fig. 6B) showed that the complexes contained both protein H and IgG. Finally, when radiolabeled C1q was added to the mixture of protein H and IgG, the label appeared together with protein H⅐IgG complexes (Fig. 6C), reflecting C1q interaction with the complexes.
Inhibition of C1q Binding to Immobilized IgG by Soluble Protein H-Soluble protein H and other Ig-binding bacterial surface proteins were tested for their capacity to interfere with binding of radiolabeled C1q by IgG-coated polyacrylamide beads. Proteins A, G, and H all bind to the C␥2-C␥3 interface region of IgG (43)(44)(45)(46), whereas the binding site for C1q is in the C␥2 domain (47). On a molar basis proteins A and H inhibited the binding of radiolabeled C1q with the same efficiency as unlabeled C1q. Protein G unexpectedly increased the uptake of radiolabeled C1q by solid-phase IgG (Fig. 7). The enhanced binding in the presence of protein G was not due to an interaction between protein G and C1q. Thus, C1q did not bind to protein G immobilized on Sepharose (not shown). M1 protein and the Ig light chain-binding protein L had no effect on C1q binding to IgG (Fig. 7). Control experiments with albumincoated polyacrylamide beads showed that background binding of radiolabeled C1q was low (Ͻ5%) in the assay system.
Inhibition of C3 Deposition and Immune Hemolysis by Soluble Protein H-IgG-coated polyacrylamide beads were incubated with human serum and protein H. C3 deposition was measured with radiolabeled anti-C3c F(abЈ) 2 . Protein H completely inhibited C3 deposition (Fig. 8). The same effect was EDTA was added, beads were washed, and 125 I-labeled F(abЈ) 2 fragments of anti-C3c IgG antibodies were added. Beads were washed, and the radioactivity of the pellets measured. A C1q concentration resulting in 65% of maximal binding of radiolabeled anti-C3c F(abЈ) 2 antibody fragment was used. This value corresponds to 100% binding in the figure. Where not indicated the standard deviation was Ͻ4%. seen with protein A, whereas protein G enhanced deposition, and proteins L and M1 had no effect (Fig. 8). Besides their IgGFc-binding activity, proteins A and G have weak affinities for Fab fragments of IgG (48,49). Judging from control experiments performed in the absence of serum this did not influence the results. In other control experiments polyacrylamide beads coated with albumin were incubated with IgGC1qDP-depleted serum reconstituted with C1q. In this case, C3 deposition was at background level, and no effect was seen with any of the Ig-binding proteins (not shown).
Protein H partially inhibited immune hemolysis of IgG-sensitized sheep erythrocytes (Fig. 9). Protein A blocked hemolysis completely, whereas protein G caused a dose-dependent increase of the hemolytic activity (Fig. 9). The effect of protein G is in line with the observation that the protein enhanced binding of C1q to IgG and surface deposition of C3 (see above). None of the proteins affected immune hemolysis of IgM-sensitized sheep erythrocytes (not shown).
C3 Deposition on Intact Streptococci-No binding of radiolabeled C1q was recorded in experiments with protein H-expressing AP1 streptococci preincubated and saturated with human polyclonal IgG, indicating that Fc regions are not exposed in IgG molecules bound to the bacterial surface. Consistent with this finding, incubation of the AP1 streptococci with serum gave little or no C3 deposition (Fig. 10). An isogenic mutant (BM27.6) devoid of protein H was generated by insertional inactivation of the protein H gene in AP1. The C3 deposition on the mutant bacteria was comparatively high (Fig. 10A). Results obtained with C1qDP-or IgGC1qDP-depleted sera reconstituted with C1q or C1q and IgG (Fig. 10B) showed that C3 deposition depended on C1q and IgG, i.e. the classical pathway. C4BP, a down-regulator of the classical pathway, is known to interact with members of the M protein family (10), including protein H (50). For this reason, C3 deposition experiments were repeated with a C1qDP-depleted serum devoid of C4BP. With the conditions used, C4BP did not appear to influence C3 deposition on the AP1 or BM27.6 streptococci (Fig. 10C). In conclusion, the results were fully compatible with findings using immune hemolysis and IgG-coated polyacrylamide beads (Figs. 7-9) and strongly suggested that surface bound protein H has an anti-opsonizing effect due to interference with IgGFc-C1q interactions.

DISCUSSION
The major conclusion of this investigation is that protein H activates complement when the molecule is part of soluble complexes with IgG, but prevents the activation of complement FIG. 9. The effect of Ig-binding proteins on complement-mediated hemolysis. Sheep erythrocytes were optimally sensitized with rabbit IgG and preincubated with proteins A (Ⅺ), G (å), and H (E) for 20 min and washed twice. An IgGC1qDP-depleted serum (10%) reconstituted with C1q to give 60% of total hemolysis of erythrocytes preincubated with buffer was used as the complement source. The sensitized and preincubated cells were incubated with this reagent. Hemolysis was blocked by EDTA, samples were centrifuged, and the absorbance of the supernatant at 541 nm was measured. Standard deviation was Ͻ3% in all experiments.
FIG. 10. C3 deposition on intact bacteria. Bacteria (2 ϫ 10 5 cells) of the AP1 strain expressing protein H, and the AP1 mutant strain BM27.6 devoid of protein H, were incubated for 20 min with serum (2%) in 200 l. EDTA was added, and the bacteria were washed and incubated with radiolabeled anti-C3c F(abЈ) 2 fragments for 180 min at 4°C. Cells were centrifuged, washed, and the radioactivity of the pellets was measured. In experiments with complement depleted sera the background C3 deposition obtained with unreconstituted serum was consistently 25% or less of the C3 deposition obtained with C1q added. Maximum binding (100%) was defined as the binding obtained with Ն5 ϫ 10 6 BM27.6 bacteria in the assay system. The complement source and the proteins used to reconstitute the serum are indicated above each panel.
when associated with the bacterial surface. These effects should both provide selective advantages to S. pyogenes and help to explain how Ig-binding surface proteins contribute to the virulence (51-53) of this important human pathogen.
It has been demonstrated that an IgG-binding fragment of protein H is released from the streptococcal surface by SCP (18). This fragment covers the entire surface-exposed part of the molecule. The fragment used throughout this study has a similar size excluding only the bacterial cell wall-associated COOH-terminal region (Fig. 1). IgG is one of the most abundant soluble extracellular human proteins, and the high affinity between protein H and IgG, 1.6 ϫ 10 9 M Ϫ1 (19), suggests that protein H in vivo is always complexed with IgG. When protein H⅐IgG complexes are released by SCP, the data of this study show that the complexes are soluble and capable of activating the classical complement pathway.
This activation will lead to breakdown of complement in the vicinity of the bacteria, thus preventing assembly and activation of complement at the bacterial surface. On the other hand, complement activation will generate C5a and thereby attract phagocytic cells. From the bacterial point of view, this should be an unwanted effect of the release of protein H⅐IgG complexes by SCP. However, SCP efficiently and simultaneously releases also a large and biologically active fragment of a C5a peptidase (18) associated with the streptococcal surface (14). This enzyme cleaves and destroys C5a as a chemoattractant for polymorphonuclear leukocytes (15).
To initiate complement activation away from the bacterium and to inactivate the C5a that is generated represents sophisticated and rational microbial defense mechanisms. The inhibition of complement activation at the surface of protein Hexpressing bacteria should also be beneficial for the microbe. Despite that large amounts of IgG are bound to the bacterial surface (54) through Fc-protein H interactions, the blocking of C1q binding to IgGFc by protein H was found to be highly effective. As a result, the deposition of opsonic C3 fragments was significantly lower in protein H-expressing S. pyogenes than in the isogenic mutant devoid of protein H. The blocking of the C1q-binding region of IgGFc by surface-associated protein H is in contrast to the soluble IgG-protein H complexes. Here the stoichiometry of complexes apparently results in the exposure of IgGFc regions capable of interacting with C1q and subsequently activation of the classical pathway of complement.
In contrast to protein H and staphylococcal protein A, protein G of group C and G streptococci unexpectedly enhanced the binding of radiolabeled C1q to IgG. The binding site(s) in IgGFc for protein H has not yet been defined by x-ray crystallography or NMR. In the case of proteins A and G, it was demonstrated that apart from a shared binding region, the proteins have unique binding sites in IgGFc. Compared with protein A, the unique protein G-interacting site is located away from the predicted C1q-binding site (43,47,55), which explains why protein G does not inhibit the binding of C1q to IgGFc. Protein G was not found to interact with C1q, and it remains unclear why complexes between protein G and IgG appear to have higher affinity for C1q than IgG itself.
Protein H is expressed by S. pyogenes strains of the M1 serotype. This serotype is associated with severe complications of suppurative S. pyogenes infections, i.e. the toxic shock-like syndrome, rheumatic fever, and APSGN (for references, see Ref. 56). The data of this study raise the possibility that release of complement-activating IgG-protein H complexes from the bacterial surface could be involved in development of these complications. Like human IgG, rabbit IgG has affinity for protein H (20). When two rabbits were given protein H intra-venously (0.5 mg at times 0, 12 h, and 24 h), the animals developed anuria and died within 48 h, 2 demonstrating the toxic property of protein H. It is tempting to speculate that protein H⅐IgG complexes may cause localized inflammation and tissue damage as a result of deposition in for instance the heart and the kidneys. In APSGN, the patients typically develop renal symptoms 1-3 weeks after a throat or skin infection with S. pyogenes. Glomerular deposition of protein H⅐IgG complexes in the course of infection could be an interesting mechanism by which antigen is "planted" in a target organ. In this case, a classical immune response directed against protein H and/or protein H⅐IgG complexes could then initiate local pathophysiological events in the glomeruli, explaining the latency period following the triggering infection. These and other possible consequences of the molecular host-microbe interactions described here are currently under investigation.