Streptococcal β Protein Has Separate Binding Sites for Human Factor H and IgA-Fc*

The group B streptococcus (GBS) is the most important cause of life-threatening bacterial infections in newborn infants. Protective immunity to GBS infection is elicited by several surface proteins, one of which, the β protein, is known to bind human IgA-Fc. Here, we show that the β protein also binds human factor H (FH), a negative regulator of complement activation. Absorption experiments with whole human plasma demonstrated binding of FH to a GBS strain expressing β protein but not to an isogenic β-negative mutant. This binding was due to a direct interaction between β and FH, as shown by experiments with purified proteins. Inhibition tests and studies with β fragments demonstrated that FH and IgA-Fc bind to separate and nonoverlapping regions in β. Heparin, a known ligand for FH, specifically inhibited the binding between β and FH, suggesting that FH has overlapping binding sites for β and heparin. Bacteria-bound FH retained its complement regulatory activity, implying that β-expressing GBS may use bound FH to evade complement attack. The finding that β protein binds FH adds to a growing list of interactions between human pathogens and complement regulatory proteins, supporting the notion that these interactions are of general importance in bacterial pathogenesis.

The group B streptococcus (GBS) is the most important cause of life-threatening bacterial infections in newborn infants. Protective immunity to GBS infection is elicited by several surface proteins, one of which, the ␤ protein, is known to bind human IgA-Fc. Here, we show that the ␤ protein also binds human factor H (FH), a negative regulator of complement activation. Absorption experiments with whole human plasma demonstrated binding of FH to a GBS strain expressing ␤ protein but not to an isogenic ␤-negative mutant. This binding was due to a direct interaction between ␤ and FH, as shown by experiments with purified proteins. Inhibition tests and studies with ␤ fragments demonstrated that FH and IgA-Fc bind to separate and nonoverlapping regions in ␤. Heparin, a known ligand for FH, specifically inhibited the binding between ␤ and FH, suggesting that FH has overlapping binding sites for ␤ and heparin. Bacteria-bound FH retained its complement regulatory activity, implying that ␤-expressing GBS may use bound FH to evade complement attack. The finding that ␤ protein binds FH adds to a growing list of interactions between human pathogens and complement regulatory proteins, supporting the notion that these interactions are of general importance in bacterial pathogenesis.
Among pathogenic bacteria, the group B streptococcus (GBS) 1 is the most common cause of life-threatening septicemia, pneumonia, and meningitis in the neonatal period. In recent years, GBS has also attracted attention as a significant cause of disease in adults with underlying conditions (1,2). Despite its importance as a human pathogen, relatively little is known about the molecular mechanisms by which this bacterium causes disease and evades attack from the human immune system, and a vaccine against GBS disease is not yet available.
Attempts to develop a GBS vaccine have mainly focused on the polysaccharide capsule, which elicits antibodies that protect against experimental infection (1,2). However, strains of GBS also express surface proteins that elicit protective immunity and these proteins have attracted increasing interest for studies of pathogenic mechanisms and for vaccine development (3)(4)(5)(6)(7)(8)(9). One of these proteins is the ␤ protein (also known as Bac or ␤ C), which is of interest not only because it elicits protective immunity but also because it binds to the Fc part of human IgA (10 -14). The IgA-binding region is situated in the N-terminal part of the ␤ protein (13) and has been localized to a sequence comprising 73 amino acid residues (15). The role of ␤ in infection is not known, but a recent study suggests that the ability of ␤ to bind IgA-Fc may allow it to interfere with IgA effector function (16). Here, we show that ␤ also binds another component of the human immune system, the complement regulator factor H (FH).
Human FH is a 150-kDa single chain plasma glycoprotein belonging to the regulators of complement activation (RCA) family. Like other members of the RCA family, FH is composed of domains designated short consensus repeat (SCR) or complement control protein (CCP) modules, and there are 20 such modules in FH (17)(18)(19)(20). FH plays an important role in the regulation of the alternative pathway of complement activation by downregulating the formation of the alternative pathway C3convertase, resulting in decreased production of C3b, and by acting as a cofactor for factor I (FI) in the degradation of C3b (20).
Our data show that the ␤ protein has separate binding sites for FH and IgA-Fc and that bacteria-bound FH retains its ability to down-regulate complement activation. These findings focus interest on the role of ␤ in the pathogenesis of GBS infections and suggest that GBS may exploit FH to evade complement attack.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, and Media-The GBS type Ia strain A909 expressing the ␤ protein (10, 21) was obtained from Dr. J. L. Michel (Channing Laboratory, Boston, MA). A series of ␤-expressing GBS isolates used for analysis of FH binding (Fig. 4C) was available in our collections. Plasmid pJRS233 is a shuttle vector in which replication is temperature-sensitive in Streptococcus pyogenes (22) and also in GBS. 2 Plasmid pLZ12Spec is a shuttle vector in which replication is not temperature-sensitive (23). Plasmid pBAC601 is a pUC18 derivative containing an insert that includes the entire bac gene, the structural gene for the ␤ protein, and the chromosomal regions upstream (ϳ1.3 kb) and downstream (ϳ2.9 kb) of bac (13). For construction of a derivative of pLZ12Spec carrying bac, the entire bac gene was amplified from plasmid pBAC601, using synthetic oligonucleotides 5Ј-AAATTTGAAT-TCTGCAGGAAGTTATTATTCCGAATG-3Ј and 5Ј-AAATTTGGATCC-GTATTTTCATTGCCCTCAACATCA-3Ј as primers. The resulting PCR fragment was digested with EcoRI and BamHI, recognition sequences for which had been introduced through the primers. This fragment was ligated into pLZ12Spec, generating plasmid pLZbac. All GBS strains were grown in Todd-Hewitt broth (Oxoid, Basingstoke, Hampshire, UK) at 37°C without shaking. GBS transformed with the pLZ12Spec derivative pLZbac was grown in presence of spectinomycin (70 g/ml).
Purified Proteins and Antisera-The group B streptococcal ␤, Rib, and ␣ proteins were purified from extracts from strains SB35, BM110, and A909, respectively, as previously described (5,24). The N-terminal B6 fragment of the ␤ protein was purified after infection of Escherichia coli with a clone, as described (13). The ␤-derivative ␤⌬XPZ, in which the entire proline-rich XPZ region has been deleted, will be described elsewhere. 3 This ␤-derivative was purified from extracts of whole GBS bacteria expressing the protein (12). Purified human FH, FI, and C3b were purchased from Calbiochem. One preparation of purified human FH was the kind gift of Dr. L. Truedsson (Lund University, Lund, Sweden). Monoclonal human IgA was kindly provided by Dr. J. M. Woof (University of Dundee Medical School, Dundee, UK). Polyclonal human IgA was purchased from Cappel-Organon Teknika (Turnhout, Belgium). Protein G was from Amersham Biosciences. Rabbit antiserum to the ␤ protein was raised as described (5). Antiserum to human FH was purchased from Calbiochem or The Binding Site (Birmingham, UK). Rabbit antiserum to human IgA was from Dakopatts (Copenhagen, Denmark). Rabbit antiserum directed against the GBS type Ia capsular polysaccharide was kindly provided by Dr. D. L. Kasper (Channing Laboratory, Boston, MA).
Construction of a ␤-Negative Mutant of GBS Strain A909 -A mutant of strain A909 was constructed in which bac, the structural gene for ␤, had been replaced with the kanamycin resistance cassette ⍀Km2 (25). The procedure used to construct this mutant employed a derivative of pJRS233, a shuttle vector in which replication is temperature-sensitive in Gram-positive bacteria, allowing efficient selection of recombinants through homologous recombination (22). The pJRS233 derivative, designated pJRS⌬bac, carried the ⍀Km2 kanamycin resistance cassette surrounded by chromosomal regions located upstream and downstream of the bac gene, respectively (see below). Plasmid pJRS⌬bac was transformed into strain A909 as described (26), and a mutant was recovered after homologous recombination between the plasmid and the bacterial chromosome (22). The structure of the mutant was verified by PCR (data not shown). This ␤-negative mutant of strain A909 will be referred to as ⌬bac. Transformation of strain ⌬bac with plasmid pLZbac generated the trans-complemented strain ⌬bac/pLZbac.
The insert in plasmid pJRS⌬bac was constructed as follows. The bac gene in pBAC601 was deleted using two EcoRV sites located 6 bp upstream of the start codon and 30 bp downstream of the stop codon, respectively. This pBAC601 derivative was blunt ligated to SmaIcleaved ⍀Km2 cassette. The resulting plasmid had an insert in which ⍀Km2 was surrounded by the chromosomal regions located upstream and downstream of bac. This insert was isolated after SalI digestion and ligated into SalI-cleaved pJRS233, resulting in plasmid pJRS⌬bac (Fig. 1A).
Purification of a 75-kDa C-terminal Fragment of ␤-A 75-kDa Cterminal fragment of ␤ was prepared by alkaline hydrolysis. A solution (22 ml) of ␤ (35 g/ml in 10 mM NaCl) was mixed with the same volume of 20 mM glycine-NaOH, pH 11.0, and incubated at 60°C for 20 h. After neutralization and concentration by ultrafiltration, the preparation was analyzed by Western blotting. A distinct 75-kDa polypeptide in the hydrolysate lacked reactivity with antiserum to the N-terminal B6 fragment of ␤, suggesting that this 75-kDa polypeptide represented a C-terminal fragment of ␤. The 75-kDa polypeptide was purified by electroelution from gel slices (electroeluter model 422; Bio-Rad) and harvested in elution buffer (25 mM Tris, 192 mM glycine, 0.1% SDS). Amino-terminal sequencing showed that the 75-kDa fragment starts at residue Pro 441 , and the size of the fragment suggests that it corresponds to the entire C-terminal part of the ␤ protein. The 75-kDa fragment contains the XPZ region, as shown by analysis with XPZ-specific antibodies (data not shown).
Plasma Absorption Experiments-Plasma proteins binding to whole bacteria were identified in absorption experiments. Bacteria in an overnight culture were washed twice in PBS and suspended to 10 10 cfu/ml. Samples (2 ml) of this suspension were added to 7 ml of fresh human plasma, supplemented with EDTA (9 mM) to avoid complement activation, and the mixture was incubated for 2 h at room temperature with gentle shaking. The bacteria were washed three times in PBS, and bound proteins were eluted with 1 ml of 0.1 M glycine-HCl, pH 2.0, for 15 min at room temperature. After removal of bacteria by centrifugation, the eluates were neutralized with 1 M Tris, pH 8.0, and concentrated ϳ7-fold in a Centricon YM-10 centrifugal filter device (Millipore Corp., Bedford, MA).
Plasma proteins binding to pure ␤ were identified by passing plasma through a column containing immobilized ␤. Pure ␤ (1 mg) was immobilized in a 1-ml HiTrap column (Amersham Biosciences), following the instructions provided by the supplier. Fresh human EDTA-plasma (2 ml) was centrifuged and filtered through a 0.45-m filter to remove particulate material, diluted 4-fold in PBS, and passed through the column. After washing the column with 15 ml of PBS, bound proteins were eluted with 5 ml of 0.1 M glycine-HCl, pH 2.0, and the eluates were immediately neutralized with 1 M Tris, pH 8.0.
Binding Assays with Whole Bacteria-To analyze binding of pure FH to the ␤-expressing strain A909 and its ␤-negative mutant ⌬bac, overnight cultures of bacteria were washed twice in PBSAT (PBS containing 0.02% NaN 3 and 0.05% Tween 20) and suspended to 10 10 cfu/ml. Identical samples (200 l) of bacterial suspension were added to each of a series of tubes, and different amounts of pure FH were added to give the final concentrations indicated. After incubation for 2 h, the bacteria were washed twice with PBSAT and resuspended in 200 l of a 1,000fold dilution of anti-FH in PBSAT. After incubation for 2 h, the bacteria were washed as described above and resuspended with PBSAT (200 l) containing 125 I-labeled protein G (ϳ15,000 cpm). After incubation for 2 h, washing with PBSAT, and centrifugation, the radioactivity associated with each bacterial pellet was determined in a ␥-counter. All incubations were performed at room temperature. Binding of anti-␤ antibodies and 125 I-labeled IgA to whole bacteria was analyzed as described (5,27).
Competitive Inhibition Tests-For competitive binding tests with proteins, the wells of microtiter plates were coated overnight with ␤ protein (50 l; 1 g/ml in PBS). This coating step was performed at 4°C, while all subsequent steps were performed at room temperature. The wells were washed three times with PBSAT and blocked by incubation for 1 h with the same buffer. To analyze the ability of FH or IgA to inhibit the binding of IgA to the immobilized ␤, unlabeled FH or polyclonal IgA was mixed with 125 I-labeled polyclonal IgA in a total volume of 100 l of PBSAT to the final concentrations indicated, and these mixtures were added to the wells coated with ␤. After a 2-h incubation and three washes with PBSAT, bound 125 I-labeled IgA was detected in a ␥-counter. Maximal binding (i.e. binding of 125 I-labeled IgA in the absence of inhibitor) was ϳ7%. Nonspecific binding to wells (Յ0.3%) was determined by analyzing binding of 125 I-labeled IgA to uncoated wells.
To analyze the ability of heparin to inhibit the binding of FH to ␤, unlabeled FH or IgA (negative control) was mixed with different amounts of heparin (final concentrations indicated) in a total volume of 115 l of PBSAT. FH or IgA was used at a final concentration of 4.3 g/ml. After preincubation for 2 h, 50 l of the mixtures were added to wells coated with ␤ protein (see above), and incubation was continued for 2 h. To detect bound FH, the wells were washed three times with PBSAT, and anti-FH antibodies (100 l; diluted 1,000-fold in PBSAT) were added. The wells were incubated for 1 h, and after three washes with PBSAT, 125 I-labeled protein G (ϳ10,000 cpm in 100 l of PBSAT) was added to each well. After incubation for 1 h and three washes with PBSAT, the radioactivity associated with each well was determined in a ␥-counter. Bound IgA was detected with anti-IgA antibodies (diluted 1,000-fold in PBSAT) and 125 I-labeled protein G, as described above. Maximal binding (i.e. binding in the absence of inhibitor) was about 7% for FH and 47% for IgA, respectively. Each of these experiments was performed at least twice, with similar results.
Cofactor Activity of FH Bound to Bacteria-Overnight cultures of bacteria were washed twice in TBS-T (50 mM Tris, 100 mM NaCl, 0.05% Tween 20, pH 7.3) and resuspended to 2 ϫ 10 10 cfu/ml. Aliquots (20 l) were centrifuged, and the bacterial pellets (4 ϫ 10 8 cfu) were resuspended in 19 l of TBS-T containing pure FH (3.15 g). After incubation for 2 h at room temperature, unbound FH was removed by three washes with TBS-T (450 l). To determine the cofactor activity of the bacteriabound FH, 125 I-labeled C3b (ϳ320,000 cpm, corresponding to ϳ8 ng) and FI (44 ng) were added in TBS-T (total volume 20 l). After incubation for 2 h at 37°C, the tubes were centrifuged, and the supernatants were collected. To analyze whether FH that had dissociated from bacteria during the experiment was responsible for cofactor activity, the ␤-expressing strain A909 with bound FH was incubated in TBS-T alone for 2 h at 37°C and then centrifuged. The supernatant was then incubated for another 2 h with FI and 125 I-labeled C3b, as described above. As a positive control, pure FH was mixed with FI and 125 I-labeled C3b and incubated as described above. The amount of FH used was the same as that bound to the bacteria. To analyze whether the ␤-expressing strain A909 can degrade C3b in the absence of FH, a sample of this strain (19 l) was incubated with FI and 125 I-labeled C3b for 2 h at 37°C. For analysis of C3b degradation, samples corresponding to 3 T. Areschoug and G. Lindahl, manuscript in preparation.
ϳ30,000 cpm were subjected to SDS-PAGE under reducing conditions. The gels were dried and analyzed by autoradiography.
Other Methods and Reagents-Radiolabeling of proteins with carrierfree 125 I (Amersham Biosciences) by the chloramine-T method, and Western blot analysis was performed as described (28). High molecular mass heparin (average molecular mass 14 kDa) was purchased from Leo Pharma (Malmö, Sweden). Amino-terminal sequencing of proteins was carried out at the Protein Analysis Center of Karolinska Institutet (Stockholm, Sweden).

Bacteria Expressing ␤ Selectively Bind FH and IgA in Whole
Human Plasma-Several studies have shown that ␤ binds to the Fc part of human serum IgA (10 -14). To analyze whether surface-expressed ␤ can bind human plasma proteins other than IgA, we compared a ␤-expressing GBS strain and an isogenic ␤-negative mutant for ability to bind plasma proteins.
The ␤-negative mutant was derived from the ␤-expressing strain A909, a strain of capsular type Ia that expresses both ␤ and the unrelated ␣ protein (10,21,29). The construction of this mutant is described under "Experimental Procedures" and in Fig. 1A. In this mutant, which is designated ⌬bac, the entire structural gene for ␤, the bac gene, is replaced with the kanamycin resistance cassette ⍀Km2. Surface expression of ␤ was completely abolished in ⌬bac, as shown by analysis with rabbit anti-␤ serum (Fig. 1B). Moreover, the wild type strain A909 bound IgA, while ⌬bac failed to bind IgA, confirming that expression of ␤ is abolished in ⌬bac (Fig. 1C). In contrast, surface expression of the type Ia polysaccharide capsule and the ␣ protein were not affected in ⌬bac, and the in vitro growth rate of ⌬bac was not different from that of the parental strain (data not shown). When the ⌬bac mutant was trans-complemented with the bac gene on a plasmid, the resulting strain, designated ⌬bac/pLZbac, expressed ␤ at a level very similar to that of the parental strain (Fig. 1, B and C).
To compare wild type bacteria and the ␤-negative mutant ⌬bac for ability to bind plasma proteins, bacteria were incubated in human EDTA-plasma, and bound proteins were eluted with 0.1 M glycine, pH 2.0 (see "Experimental Procedures"). The eluates were analyzed by SDS-PAGE. The eluate from the wild type strain A909 contained three protein species of ϳ150, ϳ130, and ϳ60 kDa that were not present in the eluate from the isogenic ␤-negative mutant ( Fig. 2A). The absence of the polypeptides in the eluate from the ␤-negative mutant ⌬bac was not due to a polar effect of the ⍀Km2 cassette on a gene located downstream, because the same three plasma proteins were also found in eluates from the trans-complemented and ␤-expressing strain ⌬bac/pLZbac. Thus, absorption of the three polypeptides from plasma was mediated by surface-located ␤ protein. Determination of N-terminal sequences showed that the sequence of the ϳ150-kDa protein was EDXNELPP, which is identical to that of human FH. The identity of this ϳ150-kDa protein was confirmed by Western blot analysis with specific anti-human FH antibodies ( Fig. 2A). For unknown reasons, the eluted FH migrated as a doublet band, with a minor component of lower molecular weight. A similar observation has been made in another bacterial system (30). The N-terminal sequence analysis revealed that the ϳ130-kDa protein was identical to the ␤ protein itself, indicating that this protein is partially released from the bacterial cell wall during the elution step. This observation was not surprising, because previous studies have shown that ␤ is selectively released from whole bacteria when incubated at nonneutral pH (12). Finally, the ϳ60-kDa protein was identified as the heavy chain (␣H) of human IgA (data not shown). Together, these data indicate In ⌬bac, the entire bac gene, the structural gene for the ␤ protein, is replaced with the ⍀Km2 kanamycin resistance cassette. The procedure to construct this mutant employed plasmid pJRS⌬bac, which contains an insert that includes the ⍀Km2 cassette flanked by the regions upstream (5Ј) and downstream (3Ј) of the bac gene, respectively. This plasmid is a derivative of pJRS233 (22), which carries a gene for erythromycin resistance (erm) and in which replication is temperature-sensitive in GBS. Strain A909 was transformed with pJRS⌬bac, and a ␤-negative mutant was recovered after homologous recombination between the plasmid and the bacterial chromosome (shown in the upper part), resulting in replacement of the bac gene with the ⍀Km2 cassette. B, analysis of the wild type strain A909, its ␤-negative mutant ⌬bac and the trans-complemented strain ⌬bac/pLZbac for surface expression of ␤. The bacteria were analyzed for ability to bind rabbit anti-␤ antibodies, using 125 I-labeled protein G to detect bound antibodies. C, binding of monoclonal 125 I-labeled IgA to the three bacterial strains characterized in B. These experiments were performed twice, with similar results. that ␤-expressing GBS not only bind IgA-Fc but also bind human FH.
Pure ␤ Protein Binds FH and IgA in Human Plasma-To analyze whether pure ␤ protein has the same properties as ␤ expressed on the bacterial cell surface, human EDTA-plasma was passed through a column containing highly purified ␤. Elution of bound proteins and analysis by SDS-PAGE demonstrated protein species of ϳ150 and ϳ60 kDa, and the ϳ150-kDa protein was identified as FH using specific antibodies (Fig.  2B). The ϳ60-kDa protein was similarly identified as IgA heavy chain (data not shown). The eluates also contained protein species of ϳ75, ϳ55, and ϳ30 kDa. Amino-terminal sequencing identified the ϳ55-kDa protein as a variant of the IgA heavy chain that for unknown reasons migrated faster than the major species. The ϳ30-kDa proteins were identified as Ig light chains (L) (data not shown). The identity of the minor ϳ75-kDa protein species is unknown. Absorption of these plasma proteins was not seen when plasma was passed through a column containing the GBS surface ␣ protein, which is unrelated to ␤ (29). These results show that pure immobilized ␤ specifically binds FH and IgA present in human plasma.
Pure FH Binds to the Surface of ␤-Expressing GBS-The interaction between FH and protein ␤ was further characterized by analyzing whether pure FH binds to whole ␤-expressing streptococci. The ␤-expressing wild type strain A909 was indeed able to bind pure FH (Fig. 3). No binding of FH was observed to the ␤-negative strain ⌬bac, but binding of FH was restored in the trans-complemented strain ⌬bac/pLZbac. Thus, GBS bacteria were able to bind pure FH, and the binding was mediated by ␤.
Interaction between Pure ␤ and Pure FH-To investigate the ability of pure ␤ to bind pure FH, the ␤ protein was subjected to Western blot analysis using unlabeled FH as probe and anti-FH for detection. As shown in Fig. 4A, pure ␤ bound FH in this analysis, while no binding was observed for two control proteins, the GBS surface proteins Rib and ␣, which are unrelated to ␤ (5, 29, 31). Similar results were obtained when the three GBS proteins were analyzed for FH-binding ability after immobilization in the wells of microtiter plates (data not shown).
Binding of pure ␤ to FH could also be demonstrated in a Western blot, in which FH was present on the blotting membrane and unlabeled ␤ was used as the probe (Fig. 4B). In this analysis, the ␤ protein also bound its other known ligand, human serum IgA, while no binding was detected for the control protein bovine serum albumin (Fig. 4B). Similar results were obtained when ␤ in solution was analyzed for its ability to bind FH, IgA, and bovine serum albumin immobilized in the wells of microtiter plates (data not shown). Together, these data show that ␤ binds directly to FH and that no other components in plasma or on the bacterial cell surface are required for the interaction.
To analyze whether ␤ proteins expressed by different GBS strains have similar properties with regard to FH binding, ␤ was extracted from five different GBS strains of serotypes Ia and Ib. These extracts were prepared by incubating whole bacteria at elevated pH, which causes selective release of the ␤ protein in almost pure form (12). Western blot analysis showed binding of FH to all ␤ proteins studied (Fig. 4C). The ␤ protein expressed by these strains showed slight variation in size, due to size variation in a proline-rich region with periodic structure (13,14). 4 FH and IgA Have Separate Binding Sites in the ␤ Protein-To analyze whether FH and IgA bind to different sites in ␤, competitive inhibition analysis was performed with purified proteins. The binding of radiolabeled IgA to ␤ immobilized in microtiter wells was not inhibited by FH, but as expected, the binding was completely inhibited by unlabeled IgA, indicating that FH and IgA have separate binding sites in protein ␤ (Fig.  5A). The reciprocal experiment could not be performed, because radiolabeling eliminated the ability of FH to bind ␤.
The FH-binding region in ␤ was further characterized by using fragments corresponding to different parts of ␤ (Fig. 5B). The recombinant B6 fragment corresponds to the N-terminal part of ␤ and includes the IgA-binding region (13,15). The 75-kDa fragment, which corresponds to the C-terminal part of the molecule, was derived from ␤ by alkaline hydrolysis, as described under "Experimental Procedures." These two fragments were subjected to Western blotting and analyzed for FIG. 3. Binding of pure FH to whole bacteria. Different amounts of FH (final concentrations indicated) were incubated with 2 ϫ 10 9 bacteria. Strains used were the ␤-expressing strain A909, its ␤-negative mutant ⌬bac, and the trans-complemented strain ⌬bac/pLZbac. Surface-bound FH was detected with anti-FH serum and 125 I-labeled protein G. Nonspecific binding (Յ7%) has been subtracted. This experiment was performed three times with similar results .   FIG. 4. Interaction between pure ␤ and pure FH. A, Western blot analysis of purified GBS surface proteins (␤, Rib, and ␣) for ability to bind pure FH used as probe. The blotting membrane was incubated with FH (5 g/ml), and bound FH was detected with anti-FH serum followed by 125 I-labeled protein G and autoradiography. In a control blot with anti-FH serum and 125 I-labeled protein G in the absence of FH, no signal was obtained. B, Western blot analysis under nonreducing conditions of purified FH, IgA, and bovine serum albumin (BSA) for ability to interact with pure ␤ used as probe. The blotting membrane was incubated with pure ␤ (5 g/ml), and bound ␤ was detected by incubation with rabbit anti-␤ serum followed by 125 I-labeled protein G and autoradiography. In a control blot where the incubation step with pure ␤ was omitted, no signal was obtained. C, Western blot analysis of the ability of ␤ proteins extracted from different GBS strains to bind FH. The ␤ proteins were selectively released from the bacterial cell wall by incubation of whole bacteria at elevated pH, as previously described (12). The ␤-expressing strains used were SB35, H36B, SB20, 70339, and A909. The ␤-negative A909 mutant ⌬bac was used as a control. The blotting membrane was incubated with FH (5 g/ml), and bound FH was detected as described above. These experiments were performed at least twice with similar results. ability to bind FH. The 75-kDa fragment, but not the B6 fragment, could bind FH (Fig. 5C, right panel). As expected, intact ␤ protein (positive control) bound FH, whereas no signal was obtained for the unrelated Rib protein (negative control).
The different ␤ fragments were also analyzed for ability to bind IgA. Binding was observed for the B6 fragment but not for the 75-kDa fragment or for the control protein Rib (Fig. 5C,  middle panel). Together, these data show that FH binds to the region of ␤ corresponding to the C-terminal 75-kDa fragment and demonstrate that FH and IgA bind to separate and nonoverlapping sites in ␤.
The C-terminal part of ␤ includes a proline-rich region, designated XPZ, which has a unique periodical sequence in which every third amino acid is proline (13,14). To analyze whether the XPZ region is involved in FH binding, we tested binding of FH to a recombinant derivative of ␤ lacking this region. This protein, designated ␤⌬XPZ, bound FH, showing that the XPZ region is not necessary for binding of FH (Fig. 5C, right panel). As expected, the ␤⌬XPZ construct also bound IgA (Fig. 5C,  middle panel).
Heparin Inhibits the Binding of FH to ␤-FH has been shown to interact with the polyanion heparin and heparin-binding do-mains have been reported to be localized in SCRs 7, 13, and 20 (32)(33)(34). To analyze whether ␤ binds to a site in FH that overlaps with one of the heparin binding sites, heparin was used in various concentrations to inhibit binding of FH to immobilized ␤ protein. Interestingly, heparin inhibited binding of FH to ␤ (Fig.  6). This inhibition was not nonspecific, because heparin was not able to block the binding of IgA to ␤. These data suggest that ␤ and heparin may have overlapping binding sites in FH.
FH Bound to the Surface of GBS Retains Its Cofactor Activity-The cofactor activity of FH bound to the surface of ␤-expressing GBS was analyzed in a C3b degradation assay (Fig. 7). The C3b protein is composed of ␣Јand ␤-chains (lane A). The enzymatic activity of FI in the degradation of C3b is dependent on FH, which acts as a cofactor for FI (20). Thus, when C3b is incubated with FI only, no degradation of C3b is observed (lane A). However, the ␣Ј-chain is cleaved when C3b is incubated with pure FH and FI, generating the 43-kDa fragment ␣ 43 (lane B). When FH was replaced with the ␤-expressing strain A909 that had been preincubated with FH, degradation of the ␣Јchain was also observed (lane C). The degradation of the ␣Јchain was not due to protease activity of the bacteria, as shown by incubation of C3b and the ␤-expressing strain A909 in the absence of FH (lane D). When the ␤-negative A909 mutant ⌬bac was preincubated with FH, followed by incubation with C3b and FI, very little degradation of the ␣Ј-chain was seen (lane E). This slight degradation of the ␣Ј-chain may be due to low levels of nonspecific binding of FH to the ⌬bac strain. As expected, degradation of the ␣Ј-chain was observed with the trans-complemented and ␤-expressing strain ⌬bac/pLZbac, which had been preincubated with FH (lane F). To analyze whether the observed cofactor activity of bacteria-bound FH was due to FH that had dissociated from the bacteria during the incubation, bacteria with bound FH were incubated as in the other assays, and after centrifugation, the supernatant was used as a source of cofactor activity. Only weak degradation of C3b was seen in this case (lane G). Together, these results show that most of the observed cofactor activity was due to FH bound to the ␤ protein on the surface of GBS.

DISCUSSION
Like other Gram-positive bacteria, GBS is not sensitive to complement-mediated lysis because it lacks an outer membrane and has a thick cell wall. Therefore, elimination of GBS from an infected individual depends on phagocytosis and killing by leukocytes. Several lines of evidence indicate that both specific antibodies and complement have important roles in the opsonization of GBS for phagocytosis (35)(36)(37). A possible way for the bacterium to avoid such opsonization would be to reduce surface complement deposition. In this study, we have demonstrated that the surface protein ␤ of GBS binds human FH and that bacteria-bound FH retains its ability to down-regulate complement activation. Thus, binding of FH to ␤ may lead to FIG. 5. Localization of the FH-binding region in the ␤ protein.
A, competitive inhibition test with FH and IgA. Different amounts of unlabeled FH or IgA was used to inhibit the binding of 125 I-labeled IgA to ␤ protein immobilized in microtiter wells. B, schematic representation of the ␤ protein and of ␤-derivatives used for localization of the FH-binding region. The numbers refer to amino acid residues. S, signal sequence; C, C-terminal membrane anchor. The B6 and 75-kDa fragments include the N-and C-terminal regions of ␤, respectively. The ␤⌬XPZ construct lacks the proline-rich XPZ region. The table to the right summarizes the binding properties of the different constructs. See "Results" for details. C, Western blot analysis. Proteins analyzed were intact ␤, the B6 and 75-kDa fragments, the ␤⌬XPZ construct, and the unrelated Rib protein (control). These proteins were analyzed for ability to bind FH or IgA, as indicated. For FH binding (right panel), the blotting membrane was incubated with unlabeled FH (5 g/ml), probed with anti-FH serum and 125 I-labeled protein G, and analyzed by autoradiography. In a control blot without FH, no signal was obtained. For IgA-binding (middle panel), the blotting membrane was probed with 125 I-labeled monoclonal IgA and analyzed by autoradiography. An equivalent Coomassie-stained SDS-PAGE gel is shown on the left. As previously reported, the recombinant B6 protein, which was purified after expression in E. coli, is size-heterogeneous (13). These experiments were performed three times with similar results.
FIG. 6. The binding between FH and ␤ is inhibited by heparin. Different amounts of heparin (final concentrations indicated) was used to inhibit the binding of unlabeled FH or IgA (4.3 g/ml) to ␤ immobilized in microtiter wells. The binding of FH or IgA to the immobilized ␤ protein was analyzed with anti-FH serum or anti-IgA serum, respectively, and 125 I-labeled protein G. This experiment was performed three times with similar results. destruction of surface-bound C3b, thereby inhibiting phagocytosis. The interaction between FH and ␤ does not require any other components in plasma or on the bacterial surface, because binding could be reproduced with highly purified components.
Absorption experiments with whole bacteria demonstrated that selective binding of FH occurred in whole human plasma, suggesting that binding of FH to the ␤ protein occurs in vivo and is of relevance to bacterial virulence. This conclusion is supported by the ability of bacteria-bound FH to act as a cofactor to FI in the degradation of C3b. In addition to its effect at the bacterial surface, the ␤ protein may also down-regulate complement activation in the bacterial microenvironment, since ␤ is not only present on the bacterial surface but also is released into the surrounding medium, at least during in vitro growth (12,38).
The ability of ␤-expressing GBS to bind FH adds to a growing list of interactions between human pathogens and complement regulators in the RCA family (39). With regard to bacterial virulence, it is of particular interest that several important pathogens, both Gram-positive and Gram-negative, have been demonstrated to bind one of the two major RCA proteins in plasma, FH and C4b-binding protein (C4BP). Binding of FH has been demonstrated for strains of S. pyogenes (group A streptococcus) (40), Neisseria gonorrhoeae (41,42), Streptococcus pneumoniae (43)(44)(45), and Borrelia burgdorferi (46,47), while binding of C4BP has been demonstrated for strains of S. pyogenes (48 -50), Bordetella pertussis (51,52), and N. gonorrhoeae (53,54). In at least two of these cases, there is strong evidence that the bound FH or C4BP contributes to bacterial virulence (41,55). The role of the bacteria-bound RCA protein is most simply explained by its ability to down-regulate complement activation, but it also seems possible that the RCA protein contributes to virulence by mechanisms that are independent of the effect on complement activation (e.g. by promoting binding to human cells) (48,56). Certain tumor cells have also been reported to bind FH or C4BP, suggesting that binding of a plasma RCA protein may contribute to complement resistance in these cells, providing a possible parallel to the situation in microorganisms (57)(58)(59).
Previous studies have shown that a binding site for human IgA-Fc is located in the N-terminal part of the ␤ protein (13,15). The data presented here indicate that the binding site for FH is located in the C-terminal half of ␤. Thus, ␤ has separate and nonoverlapping binding sites for two important components in the human humoral immune system. Although the exact function of these ligands is not known, one may speculate that they both contribute to phagocytosis resistance. As mentioned above, FH may exert this function by down-regulating complement deposition on the bacterial surface. With regard to IgA, the ␤ protein might interfere with IgA-mediated phagocytosis, due to its ability to block the binding of IgA-Fc to the human IgA receptor CD89 on phagocytes (16).
Interestingly, the ability to bind both IgA-Fc and a human complement regulator is not only a property of the ␤ protein but is also a common property among M proteins, which are major virulence factors of S. pyogenes (48). Many M proteins have nonoverlapping ligand-binding domains that allow them to simultaneously bind with high specificity to IgA-Fc and the complement regulator C4BP (48,49,60,61). Because M proteins and the ␤ protein are expressed by different bacterial species and are structurally unrelated, the ability to simultaneously bind IgA-Fc and an RCA protein has apparently arisen independently during the evolution of surface proteins of S. pyogenes and GBS. These data suggest that the ability to simultaneously bind the two ligands may confer a selective advantage. In this context, it is of interest to note that the PspC/SpsA-like proteins of S. pneumoniae have been reported to bind both the secretory component of secretory IgA and FH, providing yet another example of a bacterium binding both an RCA protein and a polypeptide found in immunoglobulins (44,45,62). In this case, there is evidence that the bacteria may use free secretory component as a receptor on the surface of epithelial cells (63).
The location of the binding site for ␤ in FH is not known, but the observation that binding of ␤ to FH is inhibited by heparin suggests that ␤ and heparin may have overlapping binding sites in FH. These data focus interest on SCRs 7, 13, and 20, which have been implicated in the binding of heparin to FH (32)(33)(34). Previous studies have shown that some M proteins of S. pyogenes bind to SCR 7 in FH, a binding that is inhibited by heparin (30,64). Because M proteins bind to SCR 7, they not only bind FH but also bind the naturally occurring splice variant FHL-1, a ϳ42-kDa plasma protein that corresponds to the first seven SCRs of FH (65,66). There is even evidence that M proteins selectively bind FHL-1, rather than FH, in human plasma (65). However, we did not detect any binding of FHL-1 to the ␤ protein in plasma absorption experiments, suggesting that ␤ does not bind to SCR 7 but may bind to SCR 13 or 20 in FH. It is noteworthy that the C-terminal part of FH has previously been implicated in the binding of FH to two bacterial surface structures, the sialylated lipo-oligosaccharide of N. gonorrhoeae and the OspE protein of B. burgdorferi, which bind in the regions corresponding to SCRs 16 -20 and 15-20, respectively (41,46).
In summary, we have shown that the ␤ protein of GBS has separate binding sites for human FH and IgA-Fc and that bacteria-bound FH retains its ability to down-regulate complement activation. These data focus interest on FH as a target for pathogenic microorganisms and identify a novel ligand for this human plasma protein. Further studies of this interaction are of interest for analysis of the function of FH and for analysis of pathogenetic mechanisms in infections caused by ␤-expressing GBS. Moreover, studies of the interaction between ␤ and FH are of interest for vaccine development, since the ␤ protein elicits protective immunity and has been used for the prepara- FIG. 7. Cofactor activity of FH bound to protein ␤ on the surface of GBS. Analysis of the ability of bacteria-bound FH to act as a cofactor for FI in the degradation of 125 I-labeled C3b. Samples were subjected to SDS-PAGE, and the gels were dried and analyzed by autoradiography. Lane A, 125 I-labeled C3b incubated with FI. The ␣Ј and ␤ polypeptides of C3b are seen. Lane B, 125 I-labeled C3b incubated with FI and FH. Lane C, 125 I-labeled C3b and FI incubated with the ␤-expressing wild type strain A909 that had been preincubated with FH and carefully washed. Lane D, 125 I-labeled C3b and FI incubated with strain A909. Lane E, 125 I-labeled C3b and FI incubated with the ␤-negative bacterial mutant ⌬bac, which had been preincubated with FH. Lane F, 125 I-labeled C3b and FI incubated with the trans-complemented ␤-expressing strain ⌬bac/pLZbac, which had been preincubated with FH. Lane G, 125 I-labeled C3b and FI incubated with a supernatant from strain A909 preincubated with FH. The supernatant contained any FH that had dissociated from strain A909 during the incubation. This analysis was performed three times with similar results. tion of a protein-polysaccharide conjugate intended for use as a human vaccine (6).