Protein PAB, an albumin-binding bacterial surface protein promoting growth and virulence.

The anaerobic bacterium Peptostreptococcus magnus is a human commensal and pathogen. Previous work has shown that strains of P. magnus isolated from patients with gynecological disease (vaginosis) frequently express an immunoglobulin (Ig) light chain-binding protein called protein L. Here we report that strains isolated from localized suppurative infections bind human serum albumin (HSA), whereas commensal isolates bind neither Ig nor HSA. The HSA-binding protein PAB was extracted from the bacterial surface or isolated from the culture supernatant of the P. magnus strain ALB8. Protein PAB was shown to have two homologous HSA-binding domains, GA and uGA. GA is absent in the sequence of a related protein from another P. magnus strain and shows a high degree of homology to the HSA-binding domains of streptococcal protein G. Therefore GA is believed to have recently been shuffled as a module from genes of other bacterial species into the protein PAB gene. This GA module was shown to exhibit a much higher affinity for HSA than uGA and was also found to be present in all of the isolates tested from localized suppurative infections, indicating a role in virulence. Moreover, when peptostreptococci or streptococci expressing the GA module were grown in the presence of HSA, the growth rate was substantially increased. Thus, the HSA binding activity of the GA module adds selective advantages to the bacteria, which increases their virulence in the case of P. magnus strains.

The anaerobic bacterium Peptostreptococcus magnus is a human commensal and pathogen. Previous work has shown that strains of P. magnus isolated from patients with gynecological disease (vaginosis) frequently express an immunoglobulin (Ig) light chain-binding protein called protein L. Here we report that strains isolated from localized suppurative infections bind human serum albumin (HSA), whereas commensal isolates bind neither Ig nor HSA. The HSA-binding protein PAB was extracted from the bacterial surface or isolated from the culture supernatant of the P. magnus strain ALB8. Protein PAB was shown to have two homologous HSA-binding domains, GA and uGA. GA is absent in the sequence of a related protein from another P. magnus strain and shows a high degree of homology to the HSA-binding domains of streptococcal protein G. Therefore GA is believed to have recently been shuffled as a module from genes of other bacterial species into the protein PAB gene. This GA module was shown to exhibit a much higher affinity for HSA than uGA and was also found to be present in all of the isolates tested from localized suppurative infections, indicating a role in virulence. Moreover, when peptostreptococci or streptococci expressing the GA module were grown in the presence of HSA, the growth rate was substantially increased. Thus, the HSA binding activity of the GA module adds selective advantages to the bacteria, which increases their virulence in the case of P. magnus strains.
As a rule, anaerobic infections are caused by bacteria that are part of the indigenous flora of mucosal surfaces and the skin. Peptostreptococcus magnus is such a commensal, and it belongs to the major group of anaerobic bacterial species causing clinically significant infections (1). Still little is known about the virulence factors of this bacterium. Increased oxygen tolerance, as seen among clinical P. magnus isolates, could contribute to the pathogenic potential (2). Other possible virulence factors, apart from the herein discussed bacterial surface proteins, include encapsulation (3) and collagenase production (4).
Numerous Gram-positive bacterial species and human pathogens express structurally related surface proteins that interact mainly with soluble host proteins (5). Protein A of Staphylococcus aureus and protein G of human group C and G streptococci both interact with the Fc region of IgG (6 -8). Protein G also has affinity for human serum albumin (HSA) 1 (9), as do members of the M protein family expressed by Streptococcus pyogenes (10 -12). Albumin binding has also been described for protein PAB (peptostreptococcal albumin binding) of the anaerobic commensal and pathogen P. magnus (13). Protein L is another surface protein of certain strains of P. magnus that binds to immunoglobulin light chains (14). Protein L has been shown to be a virulence determinant in bacterial vaginosis (15), perhaps due to its histamine-releasing activity (16). M proteins contribute to the virulence of S. pyogenes by their antiphagocytic property (17), and also IgGFc-binding proteins of these bacteria have been reported to be virulence factors (18). Experiments with deletion mutants have likewise shown that protein A of S. aureus plays a role in virulence, possibly by inhibiting opsonophagocytosis (19).
Host protein binding cell wall proteins of Gram-positive bacteria share common primary structure motifs, including (from the distal NH 2 terminus) a signal sequence, a variable NH 2terminal region, a varying number of repeated domains that independently bind different plasma proteins, and a prolinerich region supposedly intercalating the protein in the Grampositive cell wall, followed by a COOH-terminal cell wall sorting signal required to anchor the protein to the cell wall (20). The gene structure of the albumin-binding protein PAB has been shown to contain a centrally located functional domain of 45 amino acid residues responsible for the binding of HSA (Fig.  1). This domain has been subject to module shuffling between bacterial species, and was subsequently named the GA module, protein G-related albumin binding module (13). Such shuffling of modules seems to be a persistent activity among this group of genes, and when a consensus sequence of 15 nucleotides (called recer sequence) flanking the different modules in the P. magnus family of surface proteins was identified, a model for the shuffling of modules was proposed (21). The albumin-binding protein G of group G streptococcal strain G148 carries three GA modules in the NH 2 -terminal part of the protein showing up to 60% identity to the shuffled module of protein PAB, indicating that protein G might be the source of the GA module in protein PAB. The secondary structure and global fold of the GA module in protein PAB have been determined by NMR and were shown to adopt a 3-helix bundle (22), and the third GA module of protein G was recently shown to exhibit the same structure (23). The predecessor of protein PAB has also been identified. This protein, called urPAB, is expressed by a strain that binds less albumin to its surface. Protein urPAB lacks the shuffled GA module (21), but in the NH 2 -terminal region, a domain (uGA) was identified showing 38% identity to the GA module. Likewise, sequence comparison between proteins ur-PAB and PAB revealed an analogous uGA domain in protein PAB, thus indicating the presence of a second HSA-binding region in protein PAB (Fig. 1). HSA binding has been described for a number of additional surface proteins of group C and G streptococci as well as for a P. magnus protein that, in addition to several GA modules, also contains Ig light chain-binding domains closely related to those of protein L (24 -27). Apart from the HSA-binding proteins of S. pyogenes that belong to the M protein family, the other HSA-binding proteins all contain GA-related sequences. In the case of M proteins, the binding of HSA is located to the so-called C repeats (10,11), which show no homology to the GA module, suggesting that these structures have evolved separately.
The aims of the present study were to further characterize the HSA binding properties of protein PAB, to localize its HSA-binding regions, and to compare the affinities of the shuffled GA module with that of the uGA domains, i.e. the NH 2terminal HSA-binding domains of proteins PAB and urPAB. It was believed that this information could help to explain why the GA module has been shuffled into protein PAB. The biological consequences of HSA binding have also been investigated, and the results demonstrate that essential properties such as bacterial growth and virulence are affected by the interaction with albumin.

MATERIALS AND METHODS
Bacteria-P. magnus strains were clinical isolates from the Department of Clinical Microbiology, Lund University Hospital. The peptostreptococci were grown under strict anaerobic conditions at 37°C in Todd-Hewitt broth (Difco).
Proteins, Sera, and Labeling of Proteins-HSA and human IgG were purified from human plasma (28). Other serum albumins and sera were from Sigma. Protein L was prepared from peptostreptococcal growth medium (15), and recombinant protein G was from Escherichia coli lysates (9). Proteins were radiolabeled with 125 I using the Bolton-Hunter reagent (Amersham Corp., Buckinghamshire, Great Britain), chloramine T, or lactoperoxidase.
Binding Assay-Bacteria were suspended, heat killed (80°C, 5 min), and washed in phosphate-buffered saline (PBS) containing 0.02% NaN 3 and 0.5% Tween 20. Bacterial suspensions of different concentrations in a volume of 100 l were mixed with 900 l of Staphylococcus epider-midis strain L603 (10 9 bacteria/ml). 200 l of these mixed bacterial suspensions were then incubated with 10 4 cpm of 125 I-labeled HSA or IgG for 30 min. Cells were spun down, and the radioactivity of the pellet was measured in a ␥ counter and expressed as percentage of added radioactivity.
Solubilization of Peptostreptococcal Proteins-Peptostreptococci were grown as described until the stationary phase was reached and then harvested by centrifugation. Subsequently, 10% solutions of the bacteria were boiled in HCl, pH 2.0, or NaOH, pH 11.0, vortexed in PBS, and treated with mutanolysin or trypsin to extract surface proteins. Acid and alkali extractions were performed by boiling the bacteria for 3 min in 0.1 M HCl or NaOH, whereafter the sample was neutralized by the addition of 0.1 volume of 0.1 M Tris-HCl, pH 8.0. Extraction was done by vortexing the bacterial suspension in PBS. Mutanolysin digestions were performed in 0.01 M phosphate buffer, pH 6.8, and blocked with NaHCO 3 to pH 7.5 and by cooling on ice. Trypsin digestions were done at pH 6.1 in 0.05 M phosphate buffer containing 5 mM EDTA. Reactions were inhibited by adding benzamidine to a final concentration of 5 mM. Enzyme incubations were at 37°C for 1-2 h.
Affinity Chromatography of Peptostreptococcal Protein PAB and an E. coli-expressed Protein PAB Fragment-Culture medium from peptostreptococci (strain ALB8) expressing protein PAB or periplasmic E. coli lysates of clones expressing the 5Ј-end (nucleotides 1-857) of the protein PAB gene (pab) were used as starting materials and subjected to affinity chromatography on HSA-Sepharose CL-4B (Pharmacia Biotech Inc., Uppsala, Sweden). Columns of Sepharose, coupled with 3-5 mg of HSA/ml of packed gel, were equilibrated in PBS. The sample, in 0.01 M phosphate buffer, pH 7.5, was applied, and the column was rinsed with PBS and then eluted with 0.1 M glycine buffer, pH 2.0. Eluates were immediately neutralized by the addition of 0.1 volume of unbuffered 1 M Tris.
Electrophoresis and Blotting of Proteins-Proteins were separated by SDS-polyacrylamide gel electrophoresis (PAGE) (29) and transferred to PVDF membranes (30). Agarose gel electrophoresis was performed as described (31). Transfer of proteins from agarose gels was done by applying PVDF membranes on top of the gels under pressure (1-2 kg).
Absorption Experiments with Peptostreptococci and Human Plasma-P. magnus bacteria of strain 312 (10 9 cells) were incubated in 1 ml of human plasma for 2 h at 37°C. The cells were spun down and washed five times in PBS. The bacterial pellet was then resuspended in 50 l of PBS and boiled in an equal volume of SDS sample buffer. Cell debris was spun down, and the supernatant was subjected to SDS-PAGE and Western blot analysis (9).
Equilibrium Constants-Equilibrium constants were determined by incubating constant amounts of Immunobeads (Bio-Rad), coupled with HSA, together with 125 I-labeled protein PAB or GA and varying amounts of non-labeled proteins for 16 h at room temperature. After washing, the amount of bound radioactivity was measured. Using the formula of Scatchard, the calculations were done as previously reported (32).
Mutanolysin Extraction of an Albumin Binding Surface Component and Slot Binding Experiments-10% (v/v) solutions of strain ALB8 peptostreptococci were incubated for 2 h at 37°C with mutanolysin in 0.01 M phosphate buffer, pH 6.8. The reaction was blocked with NaHCO 3 to pH 7.5 followed by cooling on ice. Cells were spun down, and material from the supernatant was applied to PVDF membranes using a dot blot apparatus from Schleicher & Schuell (Dassel, Germany). Incubation with radiolabeled HSA was followed by autoradiography to visualize binding. These oligonucleotides were used as primers with genomic DNA from P. magnus strains ALB8 and ALB1B as templates to generate inserts for cloning. PCR products were purified by chloroform/phenol extraction and ethanol precipitation and were subjected to restriction enzyme cleavage before a second purification and ligation to the vector. Subsequently, the ligation mixes were transformed into competent E. coli strain JM109 cells. Clones were selectively grown on ampicillin-containing plates and were screened by PCR and for expression of HSA binding peptides. HSA binding peptides were then purified by affinity chromatography on HSA-Sepharose and on a gel chromatography col-  (39,40). Numbers refer to amino acid positions. Accession numbers of the sequences in the GenBank TM /EBI and PIR data banks are: protein PAB, GB X77864 and PIR A53586; protein urPAB, GB Z48975; and protein G, GB X04015. umn, Superose 12 for fast protein liquid chromatography (Pharmacia). The purification of the GA module has also been described elsewhere in more detail (22).
PCR Screening of P. magnus Isolates-Chromosomal DNA was isolated from 15 HSA-binding strains of suppurative infections and from 10 non-binding, non-pathogenic isolates. These chromosomal preparations were used as templates in PCRs using the following oligonucleotides as primers: RX2N, dTTA AAG AAC GCT AAA GAA GAT GCA AT; 769, dGCT ACC AGC TTT TGG TAA.

HSA-binding Proteins Are Preferentially Expressed by Strains of P. magnus Isolated from Patients with Localized
Suppurative Infections-Forty-eight P. magnus strains were isolated from patients with localized suppurative infections (n ϭ 30) or vaginosis (n ϭ 8). Ten commensal isolates were from healthy carriers. The ability of these strains to bind HSA and human IgG was tested (Fig. 2). Among the 30 suppurative infection isolates, 16 (53%) were HSA binding, whereas only one isolate showed significant IgG binding. In the vaginosis group, on the other hand, five of the eight isolates bound IgG and none bound HSA. Finally, neither HSA nor IgG binding activity could be detected among the commensal isolates. Thus, when the group of strains causing suppurative infections was compared with the commensal strains, there was a significant difference in the prevalence of HSA binding as tested with the 2 test (p Ͻ 0.0001), implying a role for protein PAB in virulence. The binding capacity of the HSA-binding strains was shown to be saturable and varied among the strains between 15 and 71% as tested by direct binding of radiolabeled HSA to bacterial cells (Fig. 3). Among the HSA-binding strains, strain ALB8 showing maximal binding (70%) and strain ALB1B showing intermediate binding (27%) were chosen for further isolation and characterization of the HSA-binding proteins PAB and urPAB, respectively.
Isolation of the HSA-binding Protein PAB from P. magnus Strain ALB8 -After anaerobic growth of P. magnus strain ALB8 bacteria for 3 days in Todd-Hewitt broth, cells were collected by centrifugation and washed in PBS. Different procedures to solubilize HSA binding materials were tested. Boiling of the bacteria at low or high pH in HCl and NaOH, respectively, released substantial amounts of fragmented HSAreactive material in the range of 14 -45 kDa (Fig. 4). Similarly, degraded material could be obtained by treatment of intact bacteria with the muranolytic enzyme mutanolysin, whereas trypsin digestion resulted in extensive degradation. A higher proportion of larger fragments could be obtained by simply vortexing the bacteria in PBS. Still, a full-length molecule of 47 kDa was only seen when the HSA-binding protein was isolated from the culture supernatant by affinity chromatography on HSA-Sepharose. The protein yield from the culture supernatant was approximately 1 mg/liter. The NH 2 -terminal sequence of this 47-kDa material has been shown to correspond to that of the NH 2 -terminal sequence of protein PAB as deduced from the gene sequence (13). Mild treatment of the material coming off the HSA-Sepharose column with trypsin resulted in a single HSA binding fragment of 23 kDa.
Protein PAB-expressing Bacteria and Purified Protein PAB Bind HSA with High Specificity-Protein PAB-expressing bacteria were incubated with human plasma. After incubation and washing, the bacteria were boiled in SDS-PAGE sample buffer. Cells were spun down, and the supernatant was run on a gel that was also blotted onto a PVDF membrane and probed with radiolabeled protein PAB. As seen in Fig. 5, only a single band at 66 kDa representing HSA could be adsorbed to the bacterial surface, and this band also reacted with protein PAB in the Western blot experiment. The conclusion is that neither the bacteria nor protein PAB will interact with any other plasma protein than HSA. Subsequent tests of the binding of 125 I-  labeled protein PAB to a number of purified proteins (i.e. IgG, IgA, fibrinogen, and fibronectin) in slot binding experiments were also negative (not shown). In order to further analyze the specificity of the interaction between protein PAB and HSA, human plasma was run on agarose gels. After transfer to PVDF membranes, the plasma proteins were probed with either protein L or protein PAB. As seen in Fig. 6, protein L bound to the cathodal immunoglobulin region of the plasma sample, whereas protein PAB bound to the anodal albumin region. Proteins PAB and L were also applied to the gel and were found to migrate in opposite directions to their respective ligands. These results reflect that protein PAB is a basic protein (net charge of ϩ7, pI of 9.78), and protein L is an acidic protein (net charge of Ϫ54).
Binding of Protein PAB to Mammalian Sera and Purified Albumins-Sera from 10 different mammalian species were subjected to SDS-PAGE. The separated proteins were blotted onto a PVDF membrane and probed with radiolabeled protein PAB (Fig. 7). Binding to the albumin band in serum from man, baboon, rhesus monkey, rat, and cat was seen. Fig. 8 demonstrates that, also among purified albumins from 12 mammalian species, protein PAB has affinity for albumin from man, baboon, rhesus monkey, and rat (purified cat serum albumin was not available).
Characterization of Protein PAB Fragments Identifies Two Distinct HSA-binding Sites-If P. magnus strain ALB8 bacteria are grown in Todd-Hewitt broth until stationary phase, HSA binding activity can be detected not only on the surface of the bacteria but also in the culture supernatant. This material can be adsorbed to HSA-Sepharose and then eluted at pH 2.0. The eluted proteins were separated by SDS-PAGE and blotted onto two PVDF membranes. One was probed with radiolabeled HSA, and the results demonstrated that the majority of the bands still bound HSA (not shown). The other membrane was stained with Coomassie Blue, and as seen in Fig. 9, the HSA binding material coming off of the column is often size heterogeneous. This could be due to partial hydrolysis at low pH or due to proteolytic activity at the bacterial surface or in the culture medium. The albumin binding bands at 47 (I), 24 (II), and 16 kDa (III) were cut out and subjected to NH 2 -terminal sequencing. All three bands had the same amino acid sequence, identical to the absolute NH 2 terminus of protein PAB. This shows that the fragments are derived from protein PAB and that there must be an HSA-binding site in the most NH 2terminal 130 -140 amino acids, giving an approximate molecular mass of 16 kDa. The same material was then subjected to mild trypsin treatment, yielding a single major albumin-bind-ing protein species at 23 kDa (IV). The NH 2 -terminal sequence of this fragment was MTIDQ which corresponds to the five amino acid residues at positions 213-217 of protein PAB. This sequence is found immediately in the NH 2 -terminal direction of the GA module. From this, a second albumin-binding region in protein PAB is implied as fragments III and IV do not overlap. In the region of protein PAB covered by fragment I, at positions 57-101, a region exhibiting 42% identity to the GA module has been found. This GA-related domain was designated uGA since its counterpart was found in the same position in the protein PAB predecessor, protein urPAB ( Fig. 1) (21).
PCR Screening for the GA Module in Pathogenic P. magnus Strains-To show that the recently introduced GA module contributes to the pathogenicity of the HSA-binding strains causing localized suppurative infections, 25 different P. magnus strains were tested for the presence of the GA module. Chromosomal DNA from the different strains was used as templates in polymerase chain reactions. An oligonucleotide (RX2N) representing the 5Ј-end of the recently introduced GA module of protein PAB and a second oligonucleotide spanning the very well conserved cell wall anchor motif (769) were used as primers in these reactions. DNA from the protein PAB strain, which was used as a positive control, yielded a product of approximately 450 base pairs. All 15 HSA-binding strains yielded products of equal or near equal (Ϯ100 base pairs) sizes, whereas none of the 10 non-binding strains yielded any PCR products with these primers.
The Affinity for HSA Differs among GA Modules-To find an explanation to why a second GA module has been introduced into protein PAB, the affinity of this GA module was determined and compared with the relative affinities of the uGA regions of proteins PAB and urPAB. Recombinant fragments of proteins PAB and urPAB covering the three resident GA modules were expressed in E. coli and purified by affinity chromatography on HSA-Sepharose, followed by gel filtration. Fragment AC corresponds to amino acid residues 27-195 of protein PAB, containing the uGA and C domains. A second fragment (GA) covers residues 213-265 of protein PAB containing the GA module, whereas uGA-K is a fragment corresponding to the uGA region (positions 1-102) of protein urPAB (Fig. 1). These fragments together with proteins urPAB and G were run on a Tricine-SDS-PAGE gel to show sizes and purity (Fig. 10A).
They were also radiolabeled and allowed to bind to HSA in slot binding experiments (Fig. 10B). In these experiments, the AC fragment bound weakly, the GA module bound almost as strong as intact protein PAB (compare with Fig. 8), and the uGA-K bound to an intermediate degree.
A competitive binding assay was utilized to determine the equilibrium constant of the interactions between HSA and fragment GA of protein PAB, as well as the relative affinities of the three peptostreptococcal GA modules. The affinity of frag- After 2 h at room temperature, the beads were washed and centrifuged, and the radioactivity of the pellets was determined. Bound and free protein concentrations were calculated and plotted. The equilibrium constant for the interaction between fragment GA of protein PAB and HSA was 6.7 ϫ 10 9 M Ϫ1 . D, competitive binding experiments. The binding of radiolabeled fragment GA of protein PAB to HSA, coupled to polyacrylamide beads, was inhibited with different amounts of unlabeled protein urPAB (E) or fragments GA (Ⅺ) and AC (ࡗ) of protein PAB. Incubations were as described for the Scatchard plots (C). ment GA (the shuffled module) for HSA was determined to be 6.7 ϫ 10 9 M Ϫ1 . Furthermore, HSA was coupled to polyacrylamide beads and incubated with the radiolabeled GA fragment in the presence of different amounts of non-labeled competing proteins (Fig. 10D). The AC fragment could only interfere with the binding of radiolabeled GA when added at a molar excess of Ͼ1000:1. Protein urPAB was a much more efficient inhibitor but still not as efficient as GA itself. Based on the amount of inhibitor needed to elicit the same inhibitory effect on the binding of 125 I-labeled GA to HSA, the relative affinities were 1, Ͻ0.05, and Ͻ0.001 for GA, uGA of protein urPAB, and uGA of protein PAB, respectively.
Growth of HSA-binding Bacteria Is Stimulated by HSA-The growth rates of albumin-binding strains of P. magnus and group G streptococci were markedly increased upon the addition of HSA to the growth medium (Fig. 11). In these experiments, 100 l of overnight cultures were transferred to 12-ml tubes containing plain Todd-Hewitt broth or broth supplemented with 0.5 mg/ml HSA. The growing bacteria were sampled at certain time intervals by plating out aliquots of the culture on blood agar plates for subsequent colony counting or by measuring the optical density at 620 nm. These two methods correlated well with each other, but since group G streptococci grow much faster than P. magnus bacteria in vitro, it was more practical to follow the growth of group G streptococci by measuring the optical density. Under the experimental conditions used, P. magnus reached stationary phase after approximately 30 h as compared with 5 h for the group G streptococci. As seen in Fig. 11, P. magnus strain ALB8 grows faster at the beginning of the exponential phase in the presence of HSA. The culture also reaches a higher cell density at the end of the exponential phase. The group G streptococci also exhibit a faster growth rate when HSA is present, and the doubling time decreases from 30 to 20 min. When non-HSA-binding P. magnus strain (505) was tested, HSA had no effect on the growth rate. Moreover, the growth rates of the Ig-binding P. magnus strain 312 or group G streptococcal strain G148 were not influenced by the addition of 0.5 mg/ml human polyclonal IgG (data not shown).

DISCUSSION
The Gram-positive obligate anaerobic peptostreptococci are part of the human indigenous flora on the skin, in the oropharynx, and in the gastrointestinal and genitourinary tracts. Among these commensals, P. magnus is the species most often associated with clinically significant infections. It is second in frequency only to Bacteroides fragilis among obligate anaerobic bacteria recovered from clinical isolates, accounting for about 10% of anaerobic human infections. The isolates are often part of mixed infections with other anaerobes and facultative anaerobes such as S. aureus, but P. magnus can also be found in pure isolates, a fact stressing its potential as a pathogen (1). However, like most anaerobes, it is a low grade pathogen that will probably not invade tissues unless preceded by trauma or other more invasive pathogens (34). P. magnus will generally cause localized suppurative infections or vaginosis. Localized suppurative infections are understood as infection entities such as abscesses, soft tissue, and wound infections. In the present study, 16 out of 30 isolates from localized suppurative infections were found to bind significant amounts of HSA, whereas none of the vaginosis or commensal isolates showed affinity for HSA (Fig. 2). Statistical analysis of these figures, using the 2 test, demonstrated a correlation between the albumin binding phenotype and suppurative infections (p Ͻ 0.0001), implying a role for this phenotype in virulence. Previously, P. magnus strains expressing the Ig light chain-binding protein L have similarly been associated with bacterial vaginosis, a condition characterized by bacterial overgrowth in the vagina (15). In the current study, all strains were tested for both HSA and Ig binding, and only one strain expressing both activities could be identified. Such a strain expressing a protein with sequence homologies both to the Ig light chain-binding domains of protein L (35) and the GA module of protein PAB was recently described (25). By PCR, the genes for proteins PAB, urPAB, and L have been amplified with chromosomal DNA from the corresponding strains as templates and by using the same sets of oligonucleotides, representing both intra-and extragenic sequence homologies (13,21). These sequence homologies in the framework regions of the genes suggest a common evolutionary origin. Applying the same PCR reactions to the collection of non-binding strains even at low stringency yielded no specific products. This indicates that there are no silent or down-regulated copies of the genes in non-binding strains. Moreover, the occurrence of these proteins appears to be a constant and stable property as the level of expression has not changed over several years of subculturing in our laboratory.
It was shown by examining fragments of peptostreptococcal protein PAB (Fig. 9) and by studying recombinant fragments of protein PAB produced in E. coli that there are two binding sites for HSA. This fits well with the finding that the PAB gene sequence contains two GA modules (21). When the sequences of the GA modules of proteins PAB, urPAB, and GA3 of protein G are compared, it can be noted that the pairwise alignments yield an identity of 29 -42% of all pairs with one exception; the GA modules (positions 214 -265) of protein PAB and GA3 of protein G are 60% identical at the protein level (Fig. 12). The close relatedness of these two sequences is also functionally FIG. 11. Growth curves of the HSA-binding P. magnus strain ALB8 (left) and group G Streptococcus strain G148 (right). HSA binding bacteria were grown in plain Todd-Hewitt broth (Ⅺ) or Todd-Hewitt broth supplemented with 0.5 mg/ml of HSA (ࡗ). Samples were taken at regular time intervals, and culture density was measured by colony counting (colony forming units) or by measuring optical density at 620 nm.
FIG. 12. Alignment of four GA modules of proteins PAB, urPAB, and G. GA modules of proteins PAB (ALB8-uGA and ALB8-GA) urPAB (ALB1B-uGA), and G (G148-GA3) are designated by strain numbers. The protein sequence alignment of the different GA modules is shown, with identical residues in shaded boxes. The consensus sequence is shown at the bottom. The secondary structures of ALB8-GA (22) and G148-GA3 (23) are comparable and schematically shown on top. The relative affinities of the peptostreptococcal GA modules derived from Fig. 10D are shown to the far left. Three residues (TSR) in ALB8-uGA and one (P) in ALB1B-uGA are shown in bold because they are suggested to be responsible for the lowered affinities for HSA in these modules as compared with ALB8-GA. The equilibrium constant for the interaction between a protein G fragment containing G148-GA3 and HSA has been determined previously to be 2.6 ϫ 10 9 M Ϫ1 (36). emphasized since these two GA modules show equally strong affinity for HSA, 6.7 ϫ 10 9 and 2.6 ϫ 10 9 M Ϫ1 (36), respectively. Assuming that the different GA modules compete for the same binding sites on HSA, the NH 2 -terminal uGA domains of proteins PAB and urPAB were shown to have much weaker affinities for HSA. This fact has implications for the interpretation of why the GA module was probably shuffled into protein PAB more recently. The uGA domains are located in analogous positions in proteins PAB and urPAB and can be assumed to have evolved divergently in situ from a common ancestor. uGA in protein PAB can therefore be said to be "older" than the more recently acquired GA. A reason for the organism to acquire a new GA module could be that the old one had lost most of its function, hence the difference in affinities.
uGA of protein PAB has low affinity for HSA and differs in four amino acid positions that are all conserved in the other three GA modules with higher affinities (Ala-4, Glu-11, Lys-30, and Thr-31). A fifth conserved position (Val-32) is changed to a proline residue in the uGA domain of protein urPAB. The octapeptide region constituting residues Ile-26 to Glu-33 is the only longer conserved region in 16 GA modules from seven proteins compared. The GA modules of protein PAB (GA) and protein G (GA3) have been investigated by NMR and were found to adopt antiparallel 3-helix bundle structures (22,23). In these structural determinations, it was shown that residues 30 -32 are part of a loop between helix II and III and should therefore be accessible for solvent. Taken together, these findings indicate that the binding site for HSA is to be found in this octapeptide region of the GA module.
Bacteria are well adapted to exploit their nutritional environment, and their growth rates can vary over more than a 10-fold range. The results of Fig. 11 show that HSA-binding strains of P. magnus and human group G streptococci grow with higher growth rates and to higher maximum cell densities when HSA is present in the culture medium. However, addition of IgG to the growth medium of comparable Ig-binding strains (P. magnus strain 312 and group G streptococcal strain G148) had no effect on the growth rates. Moreover, parallel experiments with non-binding strains of P. magnus demonstrated that the mere presence of HSA in the growth medium was not enough to stimulate growth.
HSA is the sole or major transporter of such diverse ligands as long and short fatty acids, tryptophan, thyroxine, calcium ions, etc. HSA-binding bacterial surface proteins could therefore make these ligands available for the bacteria. In this context it is noteworthy that P. magnus grows better in the presence of Tween 80 (37), a detergent composed of 70% fatty acids. The plasma concentration of free fatty acids (FFAs) is approximately 0.5 mM, and 99% of it is tightly bound to the FFA-binding sites of HSA (three major sites per HSA molecule). A rapid turnover, t1 ⁄2 in plasma is 2 min, sees to it that 50% of all lipid transportation in the human body is carried out this way. The average plasma concentration of HSA is 45 g/liter, and the protein is also present extravascularly in most tissues and at mucosal surfaces in inflammatory exudates. Albumin-binding bacteria will therefore have ample access to HSA and its ligands. It is also likely that FFA associated with HSA bound to the bacterial surface can be utilized as nutrient. Thus, in the case of hepatocytes, the binding of HSA to a surface protein is followed by the rapid diffusion of FFA across the plasma membrane (38).
Previous studies have described the dynamic on-going evolution of HSA-and Ig-binding surface proteins of P. magnus (13,21). These studies have also identified interdomain sequences, so-called recer sequences, with intron-like function that have promoted the evolution of this protein family (21). Here we find that the binding of HSA stimulates bacterial growth. This growth stimulation could constitute the selective pressure behind the evolution of protein PAB as the shuffling of an additional GA module into protein PAB has dramatically increased its affinity for HSA. Moreover, the presence of GA in all of the isolates from patients with localized suppurative infections supports the notion that this module adds to the pathogenicity of these strains. Thus, as a consequence of the shuffling of the GA module and the subsequent growth stimulation, protein PAB-expressing strains have become more virulent.