Adaptation of Staphylococcus aureus to ruminant and equine hosts involves SaPI-carried variants of von Willebrand factor-binding protein


  • David Viana,

    1. Centro de Investigación y Tecnología Animal, Instituto Valenciano de Investigaciones Agrarias (CITA-IVIA), Apdo. 187, 12.400 Segorbe, Castellón, Spain.
    2. Cardenal Herrera-CEU University, 46113 Moncada, Valencia, Spain.
    Search for more papers by this author
    • David Viana, José Blanco and María Ángeles Tormo-Más contributed equally to this work.

  • José Blanco,

    1. Centro de Investigación y Tecnología Animal, Instituto Valenciano de Investigaciones Agrarias (CITA-IVIA), Apdo. 187, 12.400 Segorbe, Castellón, Spain.
    Search for more papers by this author
    • David Viana, José Blanco and María Ángeles Tormo-Más contributed equally to this work.

  • María Ángeles Tormo-Más,

    1. Centro de Investigación y Tecnología Animal, Instituto Valenciano de Investigaciones Agrarias (CITA-IVIA), Apdo. 187, 12.400 Segorbe, Castellón, Spain.
    Search for more papers by this author
    • David Viana, José Blanco and María Ángeles Tormo-Más contributed equally to this work.

  • Laura Selva,

    1. Centro de Investigación y Tecnología Animal, Instituto Valenciano de Investigaciones Agrarias (CITA-IVIA), Apdo. 187, 12.400 Segorbe, Castellón, Spain.
    Search for more papers by this author
  • Caitriona M. Guinane,

    1. The Roslin Institute and Centre for Infectious Diseases, Royal (Dick) School of Veterinary Studies, University of Edinburgh, Edinburgh EH16 4SB, UK.
    Search for more papers by this author
  • Rafael Baselga,

    1. Exopol, Pol. Rio Gallego D/8, 50840 San Mateo, Zaragoza, Spain.
    Search for more papers by this author
  • Juan M. Corpa,

    1. Cardenal Herrera-CEU University, 46113 Moncada, Valencia, Spain.
    Search for more papers by this author
  • Íñigo Lasa,

    1. Instituto de Agrobiotecnología, CSIC-Universidad Pública de Navarra, 31006 Pamplona, Navarra, Spain.
    Search for more papers by this author
  • Richard P. Novick,

    1. Instituto de Ganadería de Montaña (CSIC-ULE), 24346 Grulleros, León, Spain.
    Search for more papers by this author
  • J. Ross Fitzgerald,

    1. The Roslin Institute and Centre for Infectious Diseases, Royal (Dick) School of Veterinary Studies, University of Edinburgh, Edinburgh EH16 4SB, UK.
    Search for more papers by this author
  • José R. Penadés

    Corresponding author
    1. Centro de Investigación y Tecnología Animal, Instituto Valenciano de Investigaciones Agrarias (CITA-IVIA), Apdo. 187, 12.400 Segorbe, Castellón, Spain.
    2. Cardenal Herrera-CEU University, 46113 Moncada, Valencia, Spain.
    3. Instituto de Ganadería de Montaña (CSIC-ULE), 24346 Grulleros, León, Spain.
    Search for more papers by this author

E-mail; Tel. (+34) 964 71 21 66; Fax (+34) 964 71 02 18.


Staphylococci adapt specifically to various animal hosts by genetically determined mechanisms that are not well understood. One such adaptation involves the ability to coagulate host plasma, by which strains isolated from ruminants or horses can be differentiated from closely related human strains. Here, we report first that this differential coagulation activity is due to animal-specific alleles of the von Willebrand factor-binding protein (vWbp) gene, vwb, and second that these vwb alleles are carried by highly mobile pathogenicity islands, SaPIs. Although all Staphylococcus aureus possess chromosomal vwb as well as coagulase (coa) genes, neither confers species-specific coagulation activity; however, the SaPI-coded vWbps possess a unique N-terminal region specific for the activation of ruminant and equine prothrombin. vWbp-encoding SaPIs are widely distributed among S. aureus strains infecting ruminant or equine hosts, and we have identified and characterized four of these, SaPIbov4, SaPIbov5, SaPIeq1 and SaPIov2, which encode vWbpSbo4, vWbpSbo5, vWbpSeq1 and vWbpSov2 respectively. Moreover, the SaPI-carried vwb genes are regulated differently from the chromosomal vwb genes of the same strains. We suggest that the SaPI-encoded vWbps may represent an important host adaptation mechanism for S. aureus pathogenicity, and therefore that acquisition of vWbp-encoding SaPIs may be determinative for animal specificity.


Staphylococcus aureus can infect a wide variety of animals and can be transmitted among diverse species. Nevertheless, most isolates from one species differ at the genome level from most isolates from a different species, suggesting the existence of specific host genotypes (Herron-Olson et al., 2007; Lowder et al., 2009). It has been suggested, moreover, that these genotypes may be determined by only a handful of genes or gene combinations (Sung et al., 2008). Since S. aureus possesses a highly conserved core genome, host-specific adaptations are likely to be carried by mobile genetic elements (MGEs). This implies that the organism may switch host specificities by the acquisition (or loss) of one or more such elements. Although the role of MGEs in overall bacterial biology is well known – e.g. catabolism, resistance, virulence, toxinogenesis, etc. – there is little information on their role in host adaptation, including information on whether such elements contribute to the ability of the organism to colonize a particular animal host or to its ability to cause disease, or both. One example of MGE-mediated host adaptation is provided by prophages that insert into the staphylococcal-β haemolysin gene (hlb). In addition to inactivating hlb, these prophages variably carry genes for one or more putative virulence functions, including staphylokinase (SAK), chemotaxis inhibitory protein (CHIPS), staphylococcal complement inhibitor (SCIN), and either enterotoxin A (SEA) or enterotoxin P (SEP) (Rooijakkers et al., 2006). Such prophages are very common in human isolates and very rare in animal isolates; if β-haemolysin and these phage-carried genes do, in fact, have a role in animal and human virulence, respectively, lysogenization with such a prophage could convert a strain from animal to human specificity, and elimination of the prophage could have the opposite effect.

An additional example of a staphylococcal function that could be involved in host adaptation is the ability to coagulate fibrinogen. In experimental infections, immediately following inoculation, S. aureus rapidly activate fibrinogen clotting, forming a tight fibrin clump (Rothfork et al., 2003). This probably enables bacteria to resist phagocytosis, facilitating the development of a local population sufficient to induce virulence factor production through agr activation by the classical quorum-sensing mechanism (Novick and Geisinger, 2008), although direct proof for this has not been reported. Significantly, most human-adapted strains clot ruminant plasma poorly, if at all, whereas the reverse is true for ruminant isolates and this difference has been used to classify S. aureus strains with respect to animal adaptation (Devriese, 1984). Importantly, it implies that at least one staphylococcal clotting factor has host specificity. In addition to coagulase (Kaida et al., 1987), the vast majority of S. aureus strains secrete a von Willebrand factor-binding protein (vWbp) that interacts with the mature form of von Willebrand factor (vWf), a serum glycoprotein that is involved in the clotting cascade (Bjerketorp et al., 2002; Bjerketorp et al., 2004). Either coagulase or vWbp, or both could be involved in animal specificity. Since all S. aureus strains produce coagulase, which does not effectively induce clotting of ruminant plasma (Friedrich et al., 2006), we have addressed the possibility that vWbp may be the putative clotting factor involved in animal adaptation. We report here that S. aureus strains isolated from ruminants and horses, but not from most other animals, encode vWbp paralogues, carried by mobile pathogenicity islands (SaPIs), which are specific for ruminant or equine plasma. These SaPI-carried paralogues differ from the chromosomally encoded vWbps, which are carried by virtually all S. aureus strains and are not host specific. Since the above-mentioned model relates to infection, this particular host adaptation is more likely to involve internal infection than external colonization, and vwb is one of the few bacterial genes yet described that exploits a known pathobiological mechanism to facilitate host-specific adaptation.


Identification of a ruminant-specific clotting factor

In the hope of identifying bovine strains that were closely related to human strains but differed in their ability to clot bovine versus human plasma, we analysed a series of bovine mastitis isolates by multi-locus sequence typing [MLST (Enright and Spratt, 1999)]. One such strain, BA4 (MLST 97), coagulated ruminant plasma whereas none of five human MLST 97 strains tested did so. The very close relationship between these human and bovine strains suggested that the factor responsible for the difference in coagulation properties might be carried by an MGE, perhaps a prophage or a SaPI. In previous studies, we had observed that many bovine S. aureus strains carry a SaPI integrated at a particular site, the att/intII site (Subedi et al., 2007), and therefore encode a common integrase (Ubeda et al., 2003). On the chance that this was true for BA4, we performed a PCR with primers specific for this integrase and obtained an amplimer whose sequence was closely related to that of the SaPIbov1/2 enzyme, suggesting that BA4 carried a SaPI similar to SaPIbov1 and 2. This was confirmed by sequencing, and the island was accordingly designated SaPIbov4 (Fig. 1 and Table S1). We then introduced into SaPIbov4 a selective marker (tetM), by allelic recombination and prepared a phage lysate from this strain by mitomycin C induction. We were able to transfer the island to RN4220 using this lysate, indicating that BA4 contains at least one prophage capable of generalized transduction. As expected, the transductants contained a new ∼15 kb segment at the att/intII site. Although we were able to transfer SaPIbov4 by generalized transduction, the frequency of this was far below that of typical SaPIs (Lindsay et al., 1998; Ubeda et al., 2005), indicating that SaPIbov4 is not induced by any of the resident prophages present in the original BA4 strain (Tormo-Más et al., 2010). The derivative strain, RN4220 (SaPIbov4), coagulated both cuniculine and ruminant plasmas, whereas RN4220, which we have found to have an inactivating mutation in its chromosomal vwb gene (unpublished data), coagulated only cuniculine plasma (Fig. 2A; Table 1). To eliminate the possibility that the native coa gene contributed to this result, we also transferred SaPIbov4 to DU298, a derivative of RN4220 in which the coagulase gene (coa) had been inactivated (coa::erm). This strain also clotted both ruminant and cuniculine plasmas, indicating that SaPIbov4 carried the bovine-specific coagulating activity associated with strain BA4 (Fig. 2A and Table 1). This was additionally confirmed by deleting SaPIbov4 from BA4, and finding that the resulting strain did not coagulate ruminant plasma (Fig. 2A and Table 1).

Figure 1.

Comparison of SaPI genomes. Genomes are aligned according to the prophage convention with the integrase gene at the left end. Gene colour code: int and xis, yellow; transcription regulators, blue; replication genes, purple; replication origin, red; encapsidation genes, green, with a lighter shade for terS; pif (phage interference) light aqua; superantigen and other accessory genes, pink; vwb genes, orange. Genes encoding hypothetical proteins, brown.

Figure 2.

SaPIbov4-encoded vWbp confers the ability to coagulate ruminant plasma. S. aureus strains BA4, BA4 ΔSaPIbov4, RN4220, RN4220 SaPIbov4, DU298, DU298 SaPIbov4 and DU298 SaPIbov4 Δvwb (A) or DU298 complemented with different vwb and coa genes (B) were incubated in presence of caprine or cuniculine (as a control) plasma. Samples were incubated for 4 h and the clots were stained with safranin. One representative experiment of three is shown.

Table 1.  Role of SaPIbov4 in coagulation of ruminant plasmas.
StrainCuniculine plasmaCaprine plasmaOvine plasmaBovine plasma
  1. Coagulating activity observed after 1 (+++) or 2 h (++). No coagulating activity detected after 5 h (−).

BA4 ΔSaPIbov4+++
RN4220 SaPIbov4++++++++++++
DU298 SaPIbov4++++++++++++
DU298 SaPIbov4 Δvwb
DU516 pCN51-vwbSbo4++++++++++++
DU516 pCN51-vwbCBA4++
DU516 pCN51-vwbCNm+++
DU516 pCN51-coaBA4+++
DU516 pCN51-coaNm+++

To identify the SaPI-carried coagulation gene, we analysed the SaPIbov4 sequence, and found that, in addition to a set of typical SaPI genes, listed in Table S1, it carried an allelic variant of vWbp (encoded by vwb) as well as an allelic variant of scn, encoding the staphylococcal complement-inhibiting protein, SCIN (van Wamel et al., 2006). A map showing the ORFs of SaPIbov4 larger than 60 codons is presented in Fig. 1. SaPIbov4, in contrast to the two other SaPIs found at this site (SaPIbov1 and SaPIbov2), lacks the typical SaPI packaging module (Ubeda et al., 2007; Novick et al., 2010). Consistent with its low transfer frequency, SaPIbov4 also lacks an identifiable homologue of the phage terminase small subunit, which we have previously shown to be essential for SaPI-specific packaging and high-frequency transfer (Ubeda et al., 2007).

Confirmation of SaPIbov4 vWbp function

To confirm the role of the SaPIbov4-carried vwb paralogue in the coagulation of bovine plasma, we constructed a deletion of the SaPIbov4 vwb gene in DU298 (SaPIbov4) and also cloned the gene to plasmid pCN51, generating pCN51-vwbSbo4. The deletion strain did not coagulate caprine plasma (Fig. 2A), whereas the cloned gene restored the bovine plasma coagulation phenotype to DU298 (Fig. 2B). As noted above, S. aureus strains contain a highly conserved chromosomal vwb gene in addition to their coa gene (Bjerketorp et al., 2002); since human isolates do not coagulate ruminant plasma, it is likely that the SaPIbov4 vwb gene differs from the common chromosomal vwb gene in this specificity. We tested this by cloning the chromosomal vwb gene from strain BA4, and from a human isolate, strain Newman, generating plasmids pCN51-vwbCBA4 and pCN51-vwbCNm respectively. We also cloned the coa gene from these strains (pCN51-coaNm and pCN51-coaBA4) and introduced these clones into strain DU516 (RN4220 vwb-coa::tetM). In contrast to the SaPIbov4 vwb gene, neither the chromosomal vwb norcoa genes from either Newman or BA4 enabled the host strain to clot caprine plasma, although all three clotted cuniculine plasma, confirming their functionality. Similar results were obtained using plasma from bovine and ovine hosts, suggesting that the ability to clot ruminant plasma is not species specific. These results are listed in Table 1 and shown in Fig. 2B. Hereafter, the various vwb genes are designated by a superscript referring to their origin. Thus vwbSand vwbC refer to the SaPI and chromosomal vwb genes respectively; specifically, vwbSbo4 and vwbCBA4 refer to the SaPIbov4 and BA4 chromosomal genes, respectively, while vwbCNm refers to the Newman chromosomal vwb gene. Comparison of the vwb genes of the sequenced human and animal S. aureus strains with that of vwbSbo4 revealed that the former can be separated into two groups (Fig. S1) within which the predicted proteins share about 70% sequence identity and that members of either of these groups share about 70% sequence identity with the predicted vWbp encoded by vwbSbo4 (Fig. 3). Generally, the C-termini are highly conserved (data not shown), and the divergence is N-terminal to this conserved region (see Fig. 3).

Figure 3.

Line-up of mature staphylococcal vWbp protein sequences (residues 1–330), coloured according to relative sequence conservation at each position. Adapted from line-up generated by PRALINE. The scoring scheme works from 0 for the least conserved alignment position, up to 10 (*) for the most conserved alignment position.

Distribution of ruminant plasma clotting activity among staphylococcal strains

To evaluate the generality of this host-specific adaptation, we screened a set of 38 S. aureus strains of different animal origins, most of which were of different MLST types, for plasma clotting specificity. Included in this set were infecting isolates associated with bovine (eight strains), human (four), porcine (three), cuniculine (four), avian (four), simian (three), equine (two), ovine (5), caprine (four) and cervine (one) infections. As shown in Table 2, 16 of the 18 ruminant strains plus one of the equine strains showed caprine plasma clotting activity. The two exceptions were bovine strains V239 and RF122, containing SaPIbov2 and SaPIbov1, respectively, neither of which contains a vwb homologue. Although all but one of the strains coagulated cuniculine plasma, limited preliminary testing revealed that among these strains, the porcine isolates had only weak clotting activity with porcine plasma (not shown).

Table 2. S. aureus strains analysed for their capacity to coagulate ruminant plasma.
StrainHostSaPILocationMLST (CC)Cuniculine plasmaCaprine plasma
  1. Coagulating activity (4 h incubation).

NewmanHuman  254+++
ColHuman  250+++
MRSA252Human  36+++
8325Human  8+++
DL594Porcine  1+++
DL596Porcine  433+++
DL597Porcine  9+
190Cuniculine  121+++
Co9Cuniculine  96+++
Co14Cuniculine  1+++
Co3Cuniculine  1+++
DL577Avian  692
DL579Avian  121++
DL580Avian  101+++
DL590Avian  New+++
DL581Simian  55+++
DL587Simian  55+++
DL588Simian  55+++
DL585Equine  1+++

Ruminant- and equine-specific clotting activity is associated with a second vwb gene

We next tested most of the strains listed in Table 2 for their vwb genotypes by Southern hybridization analysis. For this analysis, we used a probe corresponding to bases 946–1459 of the vwbSbo4 gene, which contains the conserved 3′ region and would be expected to hybridize with all known vwb genes. As shown in Fig. 4, two hybridizing bands were revealed in 16 of 18 ruminant strains tested. One of the two equine strains also showed two bands (Fig. 4), precisely equivalent to the ability of these strains to clot caprine plasma (Table 2). None of the 19 strains from other animals showed a second vwb band or clotted caprine plasma.

Figure 4.

Southern blot hybridization analysis of SaPI-encoded vwb genes in S. aureus strains from different origins. DNA from each strain was digested with HindIII, separated on 1% agarose gel, blotted and probed with a vwb-specific probe. Circles contain the SaPI-encoded vwb genes. Yellow circles: SaPIbov4; red circles: SaPIov2; white circles: SaPIbov5; green circle: SaPIeq1.

This result suggests that the two bands correspond to two different vwb genes, one of which was responsible for coagulating ruminant plasma. The presence of two different vwb genes in each of the strains with two hybridizing bands was confirmed by PCR and sequence analysis, using specific primers for each gene (see Experimental procedures).

The second vwb genes are SaPI-linked

Since the vwb gene in BA4 corresponding to the upper band in the Southern blot was carried by SaPIbov4, we considered the possibility that one of the two bands in each of the other ruminant strains also represented a SaPI-carried vwb homologue. This possibility was readily confirmed for the ovine strain ED133, for which the genome sequence has been determined (Guinane et al., 2010). Examination of this sequence revealed a well-defined SaPI, closely related to SaPIbov4 (Fig. 1 and Table S2), containing a vwb homologue whose predicted product was 96% identical to that of vwbSbo4 (Fig. 3 and Tables S1 and S2) and had the same coagulating activity for various plasmas (data not shown). This element is designated SaPIov2 (it is noted that ED133 contains two integrated SaPIs, one, SaPIov1, closely related to SaPIbov1, at the att/intII site, the other, SaPIov2, at the att/intV site). This prediction was also confirmed for all of the other ruminant strains displaying two vwb bands, by PCR and sequencing, using oligonucleotides specific for each SaPI (Table S6). Among the SaPIs thus discovered were two that closely resembled SaPIbov4, nine that closely resembled SaPIov2, and four highly similar ones that were significantly different from either. These, designated SaPIbov5, contained a vwb homologue whose predicted product was 100% identical to that of BA4 (SaPIbov4).

Because the equine strain DL584 was our only non-ruminant isolate that showed a second vwb band, we considered it important to determine by complete sequencing whether the second vwb gene in this strain also represented a SaPI. This turned out to be the case, and the SaPI thus identified was designated SaPIeq1. Here, the predicted product of the vwb homologue was 70% identical to that of vwbSbo4 (Fig. 3 and Tables S1 and S4), and strain DL584, coagulated equine as well as ruminant plasma (Fig. 5). Annotations of the ORFs of the four newly identified SaPIs are listed in Tables S1–S4, respectively, and maps showing ORFs ≥ 60 codons are presented in Fig. 1. These SaPIs also carry homologues of three other putative virulence genes, including SCIN (all four), which, as noted, is encoded by a prophage in human strains, adenosine deaminase (SaPIeq1 and SaPIbov5), and a putative protease (SaPIov2). These three genes have not previously been associated with SaPIs.

Figure 5.

SaPIeq1-encoded vWbp confers the ability to coagulate equine plasma. Strain DU298 expressing the SaPIeq1-, the SaPIbov4- or the chromosomally encoded vwb gene (from strain DL584) were incubated in presence of caprine, equine or cuniculine (as a control) plasma. Samples were incubated for 4 h. One representative experiment of three is shown.

Interestingly, SaPIbov4, SaPIbov5 and SaPIov2 are carried by strains of diverse MLST types (Table 2), suggesting that SaPI transfer has occurred widely, and could therefore be responsible for adaptation to ruminant hosts of strains coming from other animals. Additionally, strains containing SaPIov2-like, SaPIbov4-like and SaPIbov5-like elements were isolated from different animal species, suggesting that these strains could spread among animals, especially ruminants, and suggesting further that coagulation specificity was general for the several ruminant species we have studied.

SaPI-carried vwb genes are responsible for coagulation of ruminant plasma

Since SaPI-containing ruminant strains contain a second SaPI-linked vwb gene, and since these strains coagulate ruminant plasma, it seemed highly likely that the SaPI-coded vWbp is generally responsible for coagulation of ruminant plasma, as already demonstrated for SaPIbov4. To confirm this, we cloned the SaPI-linked and chromosomal vwb genes of strains DL601, DL602, ED133, RF122, V329, DL624 and DL584 and tested them in strain DU516 for coagulation activity. As shown in Table 3, this prediction was correct for all the vwb genes tested: the SaPI-coded vWbp clotted ruminant plasma whereas the chromosomal homologue did not. A very interesting variation on this theme was equine strain DL584, which is MLST type 133, as is ovine strain ED133; their chromosomal vwb genes encode 100% identical vWbps. Like the other chromosomal vWbps, neither of these clotted ruminant plasma; however, both clotted equine plasma. Moreover, the SaPI-coded vWbp of DL584 coagulated both equine and ruminant plasmas, whereas the SaPI-coded vWbp of ovine strain ED133 coagulated ruminant but not equine plasma. Presumably this or a closely related strain can infect both horses and sheep; perhaps the SaPIeq1 vwb gene evolved from the SaPIov2 vwb gene of strain ED133, following infection of a horse – for example, by recombination with a chromosomal vwb gene in an equine strain that could clot equine plasma – which could account for the ability of the SaPIeq1 gene to clot both ruminant and equine plasmas.

Table 3.  Strains used to characterize the chromosomal and SaPI-encoded vWbp proteins.
StrainHostMLSTvWbpCaprine plasma
StrainHostMLSTvWbpEquine plasma

Differential regulation of the vwb genes

It is well known that where there are two paralogues of a given gene in a particular organism, these are often regulated differentially and often have different functions (Sanchez-Perez et al., 2008). We considered this possibility for the vwbS and vwbC genes. The chromosomal vwb genes contain a conserved octanucleotide sequence within the promoter region, which is required for their expression (Harraghy et al., 2008). This octanucleotide sequence is absent in the promoter region of each of the vwbS genes, suggesting the possibility of differential regulation. We tested this by analysing the effect of the global regulator saeRS on the expression of vwbS and vwbC, by cloning the promoter regions of the SaPI-carried vwbSeq1 and of the chromosomal vwbC584 to the β-lactamase reporter vector pCN41. We chose to test regulation by saeRS because this system has been reported to be essential for innate immune evasion by S. aureus (Voyich et al., 2009). In addition, the SaPI-encoded vwb genes contain the consensus binding motif for SaeR (data not shown) (Nygaard et al., 2010; Sun et al., 2010). These reporter constructs were transferred to SH1000 and to its derivative ΔsaeRS. Remarkably, as shown in Fig. 6A, the SaPI-carried vwb genes were substantially downregulated in the absence of saeRS whereas the chromosomal genes were unaffected, as previously reported (Rogasch et al., 2006). This result is remarkable because it implies that absence of the above-mentioned octanucleotide is required for saeRS-mediated regulation. Experiments to test this are in progress.

Figure 6.

Effect of SaeRS on transcription of the SaPIeq1-encoded vwb gene.
A. SH1000 and SH1000 Δsae strains containing plasmids pCN41-vwbC584 or pCN41-vwbSeq1 (blaZ transcriptional fusions) were assayed for β-lactamase activity under standard conditions. Samples were normalized for total cell mass.
B. Strains JP6873 and JP6874 (JP6873 Δsae), containing SaPIeq1, were incubated in presence of equine plasma. Samples were incubated for 4 h and the clots were stained with safranin.
One representative experiment of three is shown.

To confirm that the result of this reporter assay was applicable to coagulation activity, we transferred SaPIeq1 to JP6871 (wbp-negative, coa::ermC) and its derivative ΔsaeRS (JP6872) and tested the resulting strains (JP6873 and JP6874) for the ability to coagulate horse plasma. As shown in Fig. 6B, JP6873 coagulated equine plasma whereas JP6874 did not.


Recent studies have revealed the existence of MGE which are specific for strains of S. aureus colonising different host species (Lowder et al., 2009; Guinane et al., 2010). Of note, sequencing of the small ruminant strain ED133 revealed a unique complement of MGE encoding proteins with enhanced activity in ruminant hosts including SaPIov2, characterized in the current study (Guinane et al., 2010). Here, we have observed that ruminant-adapted strains of S. aureus contain two genes encoding variants of the coagulation factor, vWbp, one chromosomal, the other carried by a new family of SaPIs. The SaPI-encoded protein activates clotting of ruminant and equine plasmas, and is responsible for the previously reported differential coagulation specificity of ruminant S. aureus isolates (Devriese, 1984). As shown here, the chromosomally encoded vWbp in ruminant and equine isolates, however, does not show species specificity – indeed, with the exception of one vWbp encoded by an equine strain, does not induce clotting of its host animal's plasma. Further, most animal strains, with the exception of those carrying the above-mentioned SaPIs, do not clot host plasma efficiently (our unpublished results). This raises the question of the role of plasma clotting in the animal pathogenicity of S. aureus in general. Since most ruminant isolates are from mastitis, it may be that plasma clotting is not important in this particular type of infection; milk from mastitic animals is often partially clotted (Milner et al., 1996), probably owing to digestion of casein by bacterial proteases, which may have the same role as plasma clotting in internal infections. Of note, one of the four newly described SaPIs, SaPIov2, contains a putative protease gene.

SaPIbov5 was identified in a ruminant strains belonging to clone ST398. In contrast to other adapted clones, strains belonging to ST398 have been isolated from different animals, including humans, cows, horses and pigs, suggesting that this clone has a broad range of hosts. In a preliminary analysis, we have observed that around 50% of the ST398 isolates from pig (6 out 11) or human (6 out 13) infections carried a SaPIbov5-like element. Whether this is simply a consequence of SaPI mobility or represents any specific animal adaptation remains to be determined.

Regarding the molecular basis of species specificity, it has been observed that vWbp, like coagulase, is a prothrombin activator (Bjerketorp et al., 2004). Prothrombin proteins from different hosts are not highly conserved (Fig. S2), which is consistent with the existence of the host specific SaPI-carried paralogues of the vWbp. Prothrombin activation by either protein occurs by the newly described ‘sexual’ activation mechanism where a specific short N-terminal amino acid segment of the activator (‘male’) physically inserts into a cleft in the receptor (prothrombin) molecule (female) (Friedrich et al., 2006; Kroh et al., 2009). As shown in Fig. 3 (multiple sequence line-up of 18 vWbps), all the known chromosomal vWbps have the N-terminal VVS triplet, whereas all of the predicted SaPI proteins have IVT. It is uncertain, however, whether this difference represents the determinant of species specificity. Since the ruminant isolates described in this article each coagulate plasma from at least the three ruminants tested, owing to their SaPI-carried vwb genes, it appears that the species specificity of vWbp seems to cover all ruminants in common. This implies that ruminant isolates would be essentially interchangeable with respect to host-specific adaptation; indeed, strains carrying SaPIbov4, SaPIbov5 or SaPIov2 have each been isolated from different animals (Table 2). It is noted, however, that the species specificity of S. aureus strains, with respect to pathobiological potential, can be inferred only by post hoc analysis, since it is very difficult or impossible to demonstrate species specificity in experimental infections – virulent staphylococci are essentially equally pathogenic for different animals, including humans, even though bacterial pathogens have co-evolved with their host species – for example, canine infections are mostly caused by Staphylococcus pseudintermedius (Bannoehr et al., 2007). However, although species specificity must have a long developmental history, the result of this is more complex than whether a given animal strain can infect a different animal.

From the standpoint of evolution, we have made the striking observation that the equine SaPI-carried vwb paralogue can clot both ruminant and equine plasma, while the ruminant SaPI-carried genes only coagulated ruminant plasma. Since SaPIeq1 was present in a strain belonging to the ruminant clone ST133, and since all vwbs are clearly co-ancestral, it could be argued that the SaPIeq1 vwb gene evolved from the SaPIbov or SaPIov counterpart by acquiring the capacity to coagulate equine plasma – perhaps by recombination – enhancing its adaptation to an equine host.

Although coagulase has not been shown by standard mutational analysis to be involved in staphylococcal pathogenesis, there is the very interesting finding that depletion of host fibrinogen attenuates staphylococcal virulence (Rothfork et al., 2003). This is thought to be because following entry into host tissues, the organism rapidly surrounds itself with a fibrin clump that protects it from phagocytosis and facilitates the attainment of a local population density that is sufficient to activate its virulence genes by the classical agr-dependent quorum-sensing mechanism. It is suggested that in the absence of fibrinogen, the bacteria are more available to opsonization/phagocytosis, do not generate a local ‘quorum’ and are more easily controlled by the host's innate immune system. It therefore seems odd that coagulase-defective mutants are not detectably less virulent than coagulase producers (see, for example, Moreillon et al., 1995). Perhaps the results in this article may help to resolve this apparent paradox. We here, and others previously (Bjerketorp et al., 2002; 2004; Kroh et al., 2009) have observed that either coagulase or vWf (activated by vWbp) can activate fibrin clotting, albeit by different mechanisms, suggesting that there may be functional redundancy between these two proteins.

Overall, we have discovered a family of pathogenicity islands encoding a host-specific functional activity likely to be central to S. aureus host adaptation. Our data are consistent with a central role for MGEs in the adaptation of bacteria to different host species.

Experimental procedures

Bacterial strains and growth conditions

Bacterial strains used in these studies are listed in Table S5. Strains were obtained from farms situated in different regions of Spain, except strain RF122 (from Ireland) and BA4 (from Argentina). All were from infected animals. Bacteria were grown at 37°C overnight on TSA agar medium, supplemented with antibiotics as appropriate. Broth cultures were grown at 37°C in TSB broth with shaking (240 r.p.m.).

DNA methods

General DNA manipulations were performed by standard procedures (Sambrook et al., 1989; Ausubel et al., 1990). Oligonucleotides used in this study are listed in Table S6. Labelling of the probes and DNA hybridization were performed according to the protocol supplied with the PCR-DIG DNA-labelling and chemiluminescent detection kit (Roche).

To produce the vwb mutations we used plasmid pMAD (Arnaud et al., 2004) and oligonucleotides listed in Table S6, as previously described (Ubeda et al., 2005).

A SaPIbov4 derivative with tetM inserted into a non-coding region of the island was constructed by allele replacement with a plasmid constructed by cloning the tetM gene flanked by SaPIbov4-specific sequences into plasmid pRN6680. The SaPIbov4 sequences were amplified using oligonucleotides listed in Table S6. The PCR products were digested with appropriate restriction endonucleases, ligated with tetM and cloned into plasmid pRN6680. The resulting plasmid (pJP729) was digested at native EcoRI and HindIII sites, and ligated into the multiple cloning site of the temperature-sensitive plasmid vector pMAD, generating pJP730. Plasmid pJP730 was introduced by electroporation into S. aureus strain RN4220 before transduction into strain BA4 using phage 80α, followed by allele replacement. The temperature-sensitive phenotype of the plasmids facilitated integration by homologous recombination, and a double-cross-over event was detected by plating on appropriate antibiotics followed by confirmation of a stable mutant by PCR and directed sequencing. Strain BA4 ΔSaPIbov4 was obtained by plating strain BA4 SaPIbov4-tetM on TSA, followed by replica plating onto tetracycline-containing medium to identify strain sensitive to tetracycline, and deletion of SaPIbov4 was confirmed by PCR and Southern blot analysis. Strain RN4220 SaPIbov4 was generated by transduction of SaPIbov4-tetM from strain BA4 SaPIbov4-tetM to RN4220, after SOS induction of resident prophages, as previously described (Ubeda et al., 2005).

Plasmids expressing the different vWbps were constructed by amplifying the specific genes from the appropriate strains with high-fidelity thermophilic DNA polymerase (Dynazyme Ext, Finnzymes) with the primers listed in Table S6. The PCR product was cloned into the BamHI and EcoRI sites of pCN51 (Charpentier et al., 2004) and the resulting plasmids were transformed by electroporation into S. aureus RN4220 derivatives. All clones were sequenced by the Institute Core Sequencing Lab.

Identification and isolation of the vwb-encoding genetic elements

Once identified the int genes carried in the different SaPIs, PCR experiments, using oligonucleotides hybridizing in the flanking regions of the attB site for each SaPI (Table S6), were performed. The different reaction products were cloned and sequenced.

To verify that the deduced sequence of the different SaPIs represented the native gene without additions or deletions during manipulation of the various clones, sequence information was verified by PCR amplification of selected regions, specific for each SaPI (Table S6) from their original S. aureus strains DNA and restriction mapping.

To prove that the novel identified elements present in the different strains corresponded to one of the SaPIs identified here (SaPIov2, SaPIbov4, SaPIbo5 or SaPIeq1), we performed PCR analysis using the primers listed in Table S6. The PCR-amplified fragments were cloned and sequenced. SaPIs were considered identical if they produce the expected PCR products and were of identical sequence.

Coagulase assays

Cuniculine, caprine, ovine, bovine and equine plasmas with EDTA (Rockland Immunochemicals) were used for the coagulation experiments. The tube coagulation assay was performed in glass tubes by mixing 300 µl of plasma with 1 × 108 of PBS-washed S. aureus bacteria from an overnight culture. The tubes were incubated at 37°C, and the level of coagulation was observed by tilting the tubes. The test was regarded as positive if the tube content formed a coherent clot after 4 h of incubation.

Enzyme assays

β-Lactamase assays, using nitrocefin as substrate, were performed as described (Ji et al., 1997). Cells were obtained in exponential phase. β-Lactamase units are defined as (Vmax)/OD650.

Nucleotide sequence accession number

The SaPIbov4, SaPIbov5 and SaPIeq1 sequences shown in Fig. 2A have been assigned GenBank Accession No. HM211303, HM228919 and HM228920 respectively. Chromosomally encoded vWbps sequences for the strains BA4, DL601, DL602, DL624, DL584, V329, DL581, DL190, DL577, DL586, DV482 and DL585 were deposited in the GenBank database under Accession No. HM234505, HM234506, HM234507, HM234508, HM234509, HM234510, HM240414, HM240415, HM240416, HM240417, HM240418, HM240419 respectively.


This work was supported by Grants Consolider-Ingenio CSD2009-00006, BIO2005-08399-C02-02, BIO2008-05284-C02-02 and BIO2008-00642-E/C from the Ministerio de Ciencia e Innovación (MICINN), and grants from the Cardenal Herrera-CEU University (PRCEU-UCH25/08 and Copernicus-Banco Santander programme), from the Conselleria de Agricultura, Pesca i Alimentació (CAPiA) and from the Generalitat Valenciana to J.R.P., by Grant AGL2008-00273/GAN from MICINN to J.M.C., and by Grant BB/D521222/1 from the Biotechnology and Biological Sciences Research Council to J.R.F.