We previously reported that a portion of the fnbA gene encoding the fibrinogen and elastin-binding A domain (Figure 1) varied substantially in strain P1 compared to the archetypal fnbA gene of strain 8325-4 15, resulting in FnBPA A domains which were 73.5% identical in amino acid sequence. Amino acid substitutions were non-randomly distributed resulting in sub-domain N1 being the most conserved (91.4%), while sub-domains N2 and N3 were more divergent (78.3% and 59.2% identities respectively).

FnBPA A domain sequences from published S. aureus genome sequences (COL, USA300, MRSA252, MSSA476, MW2, N315 and Mu50) were analysed to determine if any further variation within the A domain was apparent. Amino acid sequences were aligned in pairwise combinations and the identities calculated. Strains COL, USA300 and 8325-4 contained highly similar FnBPA A domains (one amino acid substitution in COL). This FnBPA A domain is classified as isotype I. These three strains belong to MLST clonal complex 8 (CC8) 19, indicating a close phylogenetic relationship. Strain MRSA252 (ST-36), which belongs to the CC30/39 clonal complex 20, encoded an FnBPA A domain that was divergent from isotype I (75.3% identity) and is here called isotype II. Two CC-1 strains (MSSA476, MW2) contained fnbA genes specifying identical A domains, but shared only 76.7% and 80.5% residue identity with isotypes I and II respectively, and are here called isotype III. CC-5 strains (N315, Mu50) specified identical A domains that were very similar (94.5% amino acid identity) to the isotype III A domains of ST-1 strains. Variation between the A domains of ST-1 and ST-5 strains was confined to the N1 sub-domains, while the N23 sub-domains were 100% identical. The classification of FnBPA A domains used in this study (for reasons discussed below) is based on the minimal ligand-binding N23 sub-domains, and therefore the ST-1 and ST-5 FnBPA A domains are classed as isotype III.

The FnBPA A domain from S. aureus strain P1 (ST-973), which we analyzed previously 15, was divergent from isotypes I, II and III and is called isotype IV. The divergent N23 sub-domains of isotypes I – IV share between 66.6 and 75.5% identity in any pairwise alignment (Table 3). The N1 sub-domains are more conserved, with average identities of 90.5% between different isotypes. The C-terminal fibronectin-binding motifs 2 – 11 of FnBPA isotypes I – III were highly conserved in the genome sequences indicating that FnBPA A domains alone, in particular the N23 sub-domains, are subject to variation.

A collection of S. aureus strains isolated from carriers and invasive disease patients has previously been analyzed by MLST 18. Eleven strains representing major clonal complexes (CC15, CC22, CC25, CC45), minor groups (CC12, CC59, CC121), and singletons (ST-55, ST-101) were analyzed. Gene fragments encoding the entire A domain plus some flanking sequence were amplified from genomic DNA by PCR with flanking primers (Figure 1 and Table 2), cloned and sequenced. The deduced A domain amino acid sequences indicated that 10 out of 11 strains encoded N23 sub-domains that are very similar to one of isotypes I – IV described above. Strains from ST-15 and ST-55 encoded isotype I (N23 > 95% identity to 8325-4). Strains from ST-30, ST-59, and ST-101 specfied isotype II (N23 > 97% identity to MRSA252). ST-22, ST-25, and ST-45 strains specified isotype III (N23 > 99% identity to MSSA476). ST-121 and ST-123 strains specified isotype IV (N23 > 96% identity to P1) (Table 1). A novel FnBPA N23 sub-domain sequence was found in S. aureus strain 3110 (ST-12) which was between 67.0% and 73.9% identical to A domain isoypes I – IV (Table 3) and is here called isotype V.

To determine the distribution of FnBPA A domain isotypes I – V in S. aureus strains of different MLST genotypes we used DNA hybridization with isotype-specific probes homologous to a portion of the highly divergent N3 sub-domain. DNA encoding the entire A domain was amplified from strains to be typed (Table 2 and Figure 1). PCR products were spotted onto membranes and hybridized with the DIG-labelled type-specific probes. An example of a hybridization experiment is shown in Figure 2. The probes were shown to be type-specific as each probe only hybridized to the appropriate control fnbA fragment (top row of blots). This allowed for rapid typing of multiple strains without the requirement for sequencing. In the example shown, fnbA gene fragments of any given strain only hybridized with one probe (Figure 2). The hybridization data for A domain typing of strains is summarized in Table 1.

fnbA DNA from S. aureus strains 19 (ST-10) and 3153 (ST-207) did not hybridize with any probe (Figure 2), indicating that they express a novel FnBPA isotype. The fnbA gene fragments from these two strains were cloned and sequenced, and the deduced A domain protein sequences were compared to the sequences of A domains types I – V. The A domain of S. aureus strain 19 (ST-10) was novel and is called isotype VI (N23, 65.9 – 69.1% identical to isotypes I – V; Table 3). It was very similar (97.8% identity) to the FnBPA A domain of bovine strain RF122 (ST-151), the genome sequence of which is available. The A domain of strain 3153 (ST-207) was also novel, and is called isotype VII (N23, 66.6 – 68.8% identical to types I – VI (Table 3)). Isotype VII of strain 3153 is similar (N23, 94.4% identity) to the A domain present in an ST-80 strain 21.

The N1 sub-domain of FnBPA is highly conserved (92.4% amino acid identity between isotypes), while sub-domains N2 and N3 are more divergent (77.5% and 66.2% average amino acid identity, respectively). Neighbour joining (NJ) trees based on DNA sequences of the fnbA allelic variants demonstrate N2 and N3 sequences appear to have co-evolved. For example, strains of ST-36, ST-30, ST-59 and ST-101 have fnbA genes encoding isotype II A domains that are greater than 97% identical. In the fnbA N2 NJ tree (Figure 3b) and the fnbA N3 NJ tree (Figure 3c) sequences from these strains clustered together. Similarly, strains carrying isotype I (ST-8, ST-250, ST-15, ST-55), isotype III (ST-1, ST-5, ST-22, ST-25, ST-45), isotype type IV (ST-121, ST-123, ST-973), or isotype VI (ST-10, ST-151) clustered similarly according to N2 and N3 fnbA sequence (Figures 3b and 3c). Therefore N2 and N3 sub-domains are always associated in a single combination, which is likely to conserve ligand binding functions. For this reason, we have defined isotypes based on the highly variable N23 sub-domain amino acid sequences.

The N1 sub-domain sequences appear to have evolved in distinct manner from the N23 sub-domain sequences (Figure 3). For example, isotype III (N23) A domains of MSSA476 (ST-1) and N315 (ST-5) are 100% identical, while their N1 sub-domains have only 86.2% identity. The DNA sequences encoding N1 sub-domains are generally highly conserved, reflected by the low number of substitutions per site (Figure 3a), and their phylogeny does not appear to match that of the N2 or N3 sub-domain encoding sequences (Figure 3), indicating a different pattern of evolution not related to ligand binding activity. It should be noted that in strains of the same ST (e.g. ST-1 strains MSSA476 and MW2), or strains within the same clonal complex (e.g ST-250 (COL) and ST-8 (8325-4) from CC8) the entire A domains (N123) are very highly conserved.

Figure 4 shows a NJ phylogenetic tree based upon the concatenated sequences of the 7 alleles used for MLST analysis. As MLST reflects the evolution of the stable core genome 22, this tree depicts the phylogenetic relatedness of the S. aureus strains studied here (Table 1). It is separated into two major groupings which reflect previous phylogenetic analysis using MLST alleles plus additional sequences from unlinked loci 23. Some fnbA allelic variants were not restricted to specific clones or lineages, but were widely distributed throughout the population. For example, several strains from CC30/39 (ST-30, ST-36, ST-37, ST-39, ST-2) contained FnBPA isotype II (Figure 4), as did distantly related strains from Group 2 (ST-9, ST-20, ST-101). Also isotype III was found in Group 2 strains of ST-1, ST-5 and ST-25 and in Group 1 strains of ST-22 and ST-45 (Figure 4). Conversely, closely related types such as ST-80 and ST-12 in Group 2 have fnbA genes encoding isotypes VII and V respectively. The fnbA allele of the ST-80 strain H7639 was determined from a recent study of polymorphic genes in S. aureus 21. These data suggest that fnbA genes have occasionally been horizontally transmitted between strains of different lineages.

The FnBPA A domain of S. aureus 8325-4 has a similar organization to the fibrinogen-binding A domain of S. aureus ClfA, the X-ray crystal structure of which has been solved 16. A 3D model of the isotype I A domain of FnBPA closely resembles the ClfA structure 15. Variant residues between isotypes I and IV were predicted to be mostly exposed on the surface of the protein model 15. A similar analysis was performed here which showed that the variant residues of isotypes II, III, V, VI and VII were predicted to be exposed on the A domain surface (data not shown). Residues that were identified by mutagenesis as important in fibrinogen and elastin binding by the Type I A domain 15 were highly conserved across all isotypes. Structural models of isotypes II – VII were generated using the web based PHYRE program (data not shown) based on the crystal structure of ClfA. Each showed striking similarities to the isotype I model 15 (data not shown).

The minimum ligand binding N23 sub-domains of isotype I and isotype IV were studied previously 15. Here the N23 sub-domains of isotypes II, III, V, VI and VII were also expressed. Each isotype bound to immobilized fibrinogen and elastin in a dose-dependent and saturable manner (Figure 5). The apparent affinity of each isotype for fibrinogen was very similar with a half maximum binding concentration of ca. 10 nM (Figure 5a). The isotype I A domain bound elastin with a half maximum of 45 nM whereas isotypes II – VII had a higher affinity with a half maximum at 15 – 20 nM (Figure 5b). This is consistent with the previous report for isotype I and IV 15. The conservation of binding activity across all isotypes indicates that ligand binding by the FnBPA A domain is biologically important for S. aureus.

The majority of variant residues between different A domain types were predicted to be exposed on the surface. This might create differences in surface-exposed epitopes that affect antigenicity. We previously demonstrated that polyclonal antibodies raised against isotype I reacted less efficiently with isotype IV by Western immunoblotting 15. Here the ability of polyclonal anti-isotype I antibodies to bind different isotypes was measured by ELISA. Each protein coated the microtitre wells with equal efficiency when detected with anti-hexahistidine monoclonal antibody 7E8 (data not shown). Rabbit anti-isotype I antibodies had a 4 – 7 fold lower affinity for isotypes II – VII compared to isotype I (Figure 6a). Thus amino acid variation creates differences in surface-exposed epitopes on the A domain molecule that affect immuno-crossreactivity.

Several mouse monoclonal antibodies (1F9, 1G8, 7C5, 1E6 and 7B7) directed against isotype I were tested by ELISA for binding to each isotype. Antibodies 7C5 and 1E6 bound efficiently to isotype I but showed little binding to isotypes II – VII (Figure 6d and 6e) indicating that the 7C5 and 1E6 epitope is only present on isotype I. Similarly, monoclonal antibody 7B7 reacted strongly with isotype I, but bound poorly to isotypes II – VII (Figure 6f) indicating that only part of the 7B7 epitope is present on A domains II – VII. 1F9 and 1G8 had a high affinty for isotypes I, III, IV, V and VI and a lower affinity for isotype II and VII (Figure 6b and 6c) indicating that they recognise a relatively conserved epitope which is partially absent in isotypes II and VII. It is possible that the IF9 and 1G8 bind the same or overlapping epitopes.

We previously reported that monclonal antibodies 7C5, 1E6, 1F9 and 1G8 failed to bind to isotype IV by Western blotting and that they bound weakly to isotype I 15. 1E6 and 7C5 did not react with the isotype IV protein by ELISA while 1F9 and 1G8 reacted well (Figure 6). This discrepancy may be explained by inefficient refolding after denaturation in Western immunoblotting.

Escherichia coli strains were cultivated on L-agar and L-broth with shaking at 37°C. E. coli strain XL1-Blue (Stratagene) was routinely used for cloning. E. coli Topp 3 (Qiagen) was used for the expression of recombinant FnBPA A domain proteins. Ampicillin (100 μg ml^-1) was incorporated into growth media where appropriate. S. aureus strains (Table 1) were isolated from individuals from Oxfordshire, U.K, and had previously been characterized by MLST 18. S. aureus strains were cultivated on TSA agar and BHI broth.

Cultures were grown in 5 ml BHI broth at 37°C with shaking for 24 h. Genomic DNA was prepared using the Edge Biosystems Bacterial Genomic DNA purification kit with the addition of 100 μg ml^-1 lysostaphin at the cell lysis step. DNA pellets were washed with 70% v/v ethanol, air-dried, and resuspended in 100 μl of TE buffer (10 mM Tris-HCl, pH 8.0; 1 mM EDTA).

Primers pfnbA Adom F1 and pfnbA Adom R1, corresponding to DNA encoding the signal sequence and fibronectin binding domain 2 respectively (Table 2 and Figure 1), were designed from conserved sequences in fnbA genes from publicly available S. aureus genomes. DNA encoding the entire FnBPA A domain along with flanking sequences was amplified, cleaved with BamHI and HindIII (Boehringer Mannheim) at restriction sites incorporated into the primers, ligated using T4 DNA ligase (Roche) to pBluescript KS+ DNA cleaved with the same enzymes, and finally transformed into E. coli XL1-Blue cells using standard procedures 34. Transformants harbouring cloned fnbA gene fragments were selected on L-agar with ampicillin and 80 μg ml^-1 of the chromogenic substrate 5-bromo-4-chloro-3-indolyl-β-D-galactoside (Melford Laboratories, Ipswich, U.K.). Plasmid DNA was purified from transformants using the Favorgen Plasmid DNA Extraction Mini Kit (Favorgen Biotech Corp). The cloned fnbA gene fragments were sequenced using T3 and T7 primers by GATC Biotech AG (Germany). The EMBL accession numbers for the sequences reported in this paper are AM749002 – AM749015.

Genomic DNA samples from clinical isolates were used as the template in PCR reactions with primers pfnbA Adom F1 and pfnbA Adom R1 (Table 2 and Figure 1) to amplify the entire coding region of the FnBPA A domain. PCR amplimers were purified using the Wizard SV Gel and PCR Clean-Up System kit (Promega) Five ng of fnbA DNA was spotted onto positively-charged nylon membranes (Roche) and allowed to air-dry. Membranes were incubated for 5 min on blotting paper soaked in denaturation solution (1.5 M NaCl, 0.5 M NaOH), 5 min in neutralization solution (1.5 M NaCl, 1 M Tris-HCl, pH 7.4), and finally for 15 min on blotting paper soaked with 2× SSC solution (300 mM NaCl, 30 mM tri-sodium citrate). DNA was fixed on the membranes by incubation at 120°C for 30 min. Membranes were incubated for 2 h at 68°C in pre-hybridization solution (5× SSC, 0.1% w/v N-lauroylsarcosine, 0.02% w/v SDS and 1× Blocking reagent (Roche)). DIG-labelled probes complementary to DNA encoding isotype I – V N3 sub-domains were synthesised by PCR using type-specific probe primers (Table 2 and Figure 1) and PCR DIG labelling mix (Roche). Probes were denatured by boiling at 95°C for 10 min, diluted in pre-hybridization solution and incubated with nylon membranes for 18 h at 68°C. Following hybridization, the membranes were washed twice with 2× SSC/0.1% w/v SDS at room temperature followed by two washes with 0.5× SSC/0.1% w/v SDS at 68°C for 20 min. Membranes were equilibrated for 30 min in maleic acid buffer (100 mM maleic acid, 150 mM NaCl, pH 7.5), and bound DIG-labelled probes were detected by incubation for 30 min with alkaline phosphatase-conjugated anti-DIG antibody (Roche) diluted 1:10,000 in maleic acid buffer. After washing twice with maleic acid buffer containing 0.3% v/v Tween 20, the chemiluminescence substrate CSPD (Roche) was used to detect bound anti-DIG antibodies and membranes were exposed to X-OMAT UV Plus Film (Kodak).

pQE30 expressing the minimal ligand-binding FnBPA A domain truncate (N23: residues 194 – 511) from S. aureus 8325-4 and S. aureus P1 (residues 194 – 512) were described previously 15. pQE30 expressing His-tagged N23 A domains of isotypes II, III, V, VI and VII were constructed with pQE30 vector primer pairs (Table 2 and Figure 1) and genomic DNA templates from strains MRSA252 (ST-36), MSSA476 (ST-1), 3110 (ST-12), 19 (ST-10) and 3153 (ST-207), respectively. Each construct was verified by sequencing (GATC Biotech AG, Germany) and transformed into E. coli Topp 3 for protein expression and purification. Proteins were purified by Ni²⁺ chelate chromatography 34. Concentrations were determined using the BCA Protein Assay Kit (Pierce). Proteins were dialysed against PBS for 24 h at 4°C, aliquoted and stored at -70°C.

Human aortic elastin (Elastin Products Company) and human fibrinogen (Calbiochem) were coated onto Nunc 96-well microtitre dishes (1 μg protein well^-1) for 18 h under UV light and at 4°C respectively. Wells were washed three times with phosphate-buffered saline (PBS) containing 0.05% v/v Tween 20 (PBST) and blocked with 2% w/v bovine serum albumin (BSA; Sigma) in PBST buffer (BSA-PBST) for 2 h at 37°C. Following three washes with PBST, various concentrations of purified FnBPA isotype proteins in 2% w/v BSA-PBST were added to the wells and incubated at room temperature for 1 h with shaking. Wells were washed three times with PBST buffer and bound proteins were detected with monoclonal antibody 7E8 (90 ng well^-1) that recognises the N-terminal hexahistidine fusion tag. Following incubation for 1 h with shaking at room temperature, the wells were washed three times with PBST followed by 100 μl per well of goat-anti-mouse antibodies conjugated to horseradish-peroxidase (HRP, Dako; Denmark) diluted 1:2000 in 2% w/v BSA-PBST. After incubation for 1 h at room temperature, wells were washed three times with PBST, and bound HRP-conjugated antibodies were detected with 10 μg per well of 3,3',5,5'-tetramethylbenzidine (TMB; Sigma) in 0.05 M phosphate-citrate buffer containing 0.006% (v/v) hydrogen peroxide. After incubation at room temperature for 5 min the reaction was stopped by adding 50 μl of 2 M H₂SO₄. The absorbance at 450 nm was measured with an ELISA plate reader (Multiskan EX, Labsystems).

Various concentrations of recombinant FnBPA A domain proteins in PBS were coated onto Nunc 96-well microtitre dishes for 18 h at 4°C. Wells were washed and blocked with BSA for 2 h as described above. Following three washes with PBST, 100 μl of anti-FnBPA A domain antibodies diluted in BSA-PBST (1.8 μg polyclonal IgG ml^-1; 2.5 μg monoclonal IgG ml^-1), were added to each well and incubated for 1 h at room temperature with shaking. Polyclonal and monoclonal antibodies 7C5, 1E6, 1F9 and 1G8 raised against the isotype I A domain of FnBPA from S. aureus 8325-4 were described previously 81115. Monoclonal antibody 7B7 was generated by immunizing mice with recombinant Type I FnBPA A domain (N123 sub-domains) as described previously 11. After 1 h incubation the wells were washed three times with PBST. Goat anti-rabbit IgG-HRP conjugated antibodies or goat anti-mouse IgG-HRP conjugated antibodies (Dako, Denmark), each diluted 1:2000 in BSA-PBST, were added to the wells and incubated for 1 h. After washing three times with PBST, bound HRP-conjugated antibodies were detected as described above.

Protein sequences were aligned in pairwise combinations to calculate amino acid identity using the ExPASY SIM alignment tool 35. Alignments of multiple protein sequences to view areas of conservation amongst A domains were performed using T-Coffee 36. Theoretical 3D models of A domain isotpyes I – VII were generated using PHYRE Protein Fold Recognition Server 37. Protein structure files were viewed using PyMOL 38, which allowed visualization of the distribution of variant residues between isotypes. The concatenated MLST allele sequences of S. aureus strains were downloaded from the MLST database 39. Phylogenetic and molecular evolutionary analyses of fnbA gene sequences and concatenated MLST allele sequences were conducted using MEGA version 3.1 40. The overall relatedness of the MLST population of S. aureus was visualized using eBURST 4142.