Recombination-mediated remodelling of host–pathogen interactions during Staphylococcus aureus niche adaptation

Large-scale recombination events have led to the emergence of epidemic clones of several major bacterial pathogens. However, the functional impact of the recombination on clonal success is not understood. Here, we identified a novel widespread hybrid clone (ST71) of livestock-associated Staphylococcus aureus that evolved from an ancestor belonging to the major bovine lineage CC97, through multiple large-scale recombination events with other S. aureus lineages occupying the same ruminant niche. The recombination events, affecting a 329 kb region of the chromosome spanning the origin of replication, resulted in allele replacement and loss or gain of an array of genes influencing host–pathogen interactions. Of note, molecular functional analyses revealed that the ST71 hybrid clone has acquired multiple novel pathogenic traits associated with acquired and innate immune evasion and bovine extracellular matrix adherence. These findings provide a paradigm for the impact of large-scale recombination events on the rapid evolution of bacterial pathogens within defined ecological niches.


Introduction
Recently, whole-genome sequencing studies have revealed large-scale (macro-) recombination events linked to the emergence of epidemic clones of an array of major bacterial pathogens (Brochet et al., 2008;Chen et al., 2014;de Been et al., 2013;Mostowy et al., 2014;Robinson & Enright, 2004). For example, the epidemic clone defined by sequence type ST258 of Klebsiella pneumoniae responsible for the recent global spread of carbapenem-resistant K. pneumoniae is a hybrid clone resulting from a recombination event of a 1.1 Mbp chromosomal fragment from an ST442-like strain with 4.2 Mbp derived from an ST11-like strain . In addition, large recombination events involving acquisition by conjugation of chromosomal segments up to 300 kb regions have driven the emergence of major human clinical clones of Streptococcus agalactiae (Brochet et al., 2008). Furthermore, large recombination events have been linked to capsule serotype switching and antibiotic resistance among Streptococcus pneumoniae strains in response to vaccination, and Enterococcus faecium clones undergoing hospital adaptation (de Been et al., 2013;Mostowy et al., 2014). Although Staphylococcus aureus is a highly clonal organism, a recent species-wide comparative genomic study identified that chromosomal regions flanking mobile genetic elements and a *750 kb region spanning the origin of replication (oriC) had an elevated recombination rate (Everitt et al., 2014). This is consistent with the previous discovery that the pandemic human hospital-associated meticillin-resistant S. aureus CC239 clone evolved from an import of a 635 kb region from a CC30 donor strain into a CC8 genetic background in a region spanning the origin (Robinson & Enright, 2004;Smyth et al., 2010). Similar, large-scale homologous recombination events of *250 kb in size have resulted in hybrid genomes for human S. aureus clones ST34 and ST42 (Robinson & Enright, 2004). Although the emergence and expansion of hybrid clones has been linked to the recombinant events that define them, our understanding of the influence of such events on the ecological success of bacterial pathogens is not well understood. Here, we report the identification of a widespread livestock-associated hybrid clone of S. aureus which emerged from the major bovine clone of S. aureus CC97 via multiple, large recombination events with other S. aureus strains occupying the ruminant niche. Furthermore, we investigate the consequences of the recombination events on host-pathogen interactions, identifying multiple gains of function that contribute to bovine immune evasion and enhanced host-pathogen interactions. Our data highlight the potential impact of large-scale recombination events on the emergence of new bacterial clones within defined environmental niches.

Methods
Bacterial culture and genomic DNA isolation. All bacterial strains employed are listed in Table S1 (available in the online Supplementary Material). The ST71 S. aureus strains sequenced in the current study were isolated from cases of bovine mastitis in Ireland in 1993 (RF103) and the UK in 2003 (C01122). Strain Phillips and the Cna (collagen adhesin)-deficient mutant were constructed and described previously (Patti et al., 1994). For phylogenetic analysis, the ST71 S. aureus strains were added to a previously defined dataset of CC97 S. aureus (Spoor et al., 2013). For genomic DNA isolation, strains were grown for 16 h on tryptic soy agar (TSA) at 37 uC or in tryptic soy broth (TSB) with overnight shaking at 200 r.p.m. at 37 uC, and DNA isolated as described previously (Spoor et al., 2013). For the in vitro assays, strains were cultured in brain heart infusion (BHI) overnight at 37 uC, shaking at 200 r.p.m.
Genome sequencing, mapping assembly and de novo genome assembly. Paired-end Illumina sequencing was carried out on a Genome Analyzer IIx, following standard Illumina protocols. Read quality was assessed and genome assembly and variant calling were conducted as described previously (Spoor et al., 2013). Reads were aligned against the reference genome MW2 (GenBank accession number NC_003923), a triple-locus variant of ST97. Core genome was defined as sites shared by all strains included in mapping analysis. Effect of variants were determined using SnpEff 3.0 (Cingolani et al., 2012). Pairwise analysis of SNP density was calculated in 10 000 bp windows across the length of the genome. Prior to de novo assembly, lowfrequency erroneous reads were removed using Quake 0.3.4 (Kelley et al., 2010) and de novo assemblies generated using Velvet 1.2.07 (Zerbino & Birney, 2008) using the VelvetOptimizer.pl script implemented in VelvetOptimizer 2.1.7 (http://bioinformatics.net.au/software.velvetoptimiser. shtml).

Molecular typing and evolutionary analysis of CDSs.
MLST typing was conducted using in silico using SRRS2 (Inouye et al., 2012) and by interrogating genome sequences for the alleles using BLAST (Altschul et al., 1990), and querying the sequences against the MLST database (http://saureus.mlst.net/). For comparative analysis of the evolutionary origin of the ST71 genome, three CDSs situated within and three CDSs situated out-with the SNP-dense region were arbitrarily selected for phylogenetic analysis from selected strains listed in Table S1. Nucleotide sequences for each CDS were extracted from S. aureus genome sequences using BLASTN (Altschul et al., 1990),

Impact Statement
Large-scale recombination events involving the transfer of large fragments of DNA between strains of bacteria have led to the emergence of new epidemic hybrid clones. In some cases, such events have been linked to capsule serotype switching, but the broader functional impact and role in epidemic success is not well understood. Here, we identified a novel hybrid clone of livestock-associated Staphylococcus aureus that has evolved by replacement of a large region of the genome via multiple genetic imports from other S. aureus strains in the same niche. Functional analysis reveals that the recombination has led to the gain of several pathogenic traits associated with immune evasion and host-pathogen interactions. The study demonstrates for the first time how large-scale recombination events can have a profound impact on the phenotype and fitness of bacteria within defined environmental niches. The findings have broad implications for the role of such events in the emergence and success of other hybrid clones of pathogenic bacteria. aligned using the CLUSTAL W method (Larkin et al., 2007) and neighbour-joining trees constructed using MEGA 4.0 with 1000 bootstrap replicates (Tamura et al., 2007).
Core genome phylogenetic analysis. The ST71 consensus sequences were combined with the CC97 consensus sequences determined from a previous study and the core genome redefined (Spoor et al., 2013). Maximumlikelihood phylogenetic trees for the CC97 core genome with and without the predicted recombinant region were reconstructed using RAxML-7.2.6 (Stamatakis, 2006), implementing a GTR model with gamma correction for rate heterogeneity and 1000 bootstrap replicates.
Recombination detection. For input into recombination detection software, an alignment of the study strains along with selected published ruminant-associated S. aureus genomes was created using the progressiveMauve algorithm implemented in Mauve 2.3.1 using default settings (Darling et al., 2010). Locally collinear blocks of at least 1000 bp in length were extracted from the XMFA file using the stripSubsetLCB script distributed with Mauve (at http:// darlinglab.org/mauve/snapshots/2015/2015-01-09/linux-x64/), concatenated and converted to a FASTA alignment file format. Recombination detection was performed on this FASTA file using BratNextGen (http://www.helsinki.fi/bsg/ software/BRAT-NextGen/) (Marttinen et al., 2012), setting the hyper-parameter a to 1, with a cut-off value of 0.1 within the proportion of shared ancestry tree, conducting 40 iterations within the detecting recombination algorithm and 100 permutation runs for estimating significance. In addition, the RDP2 suite of recombination software programs was employed as described (Martin et al., 2005).
Comparative genomic analysis of ST71 and ST97 isolates. The population-based de novo assembly software Cortex 1.0.5.20 was used to identify variation in gene content among ST71 and ST97 strains (Iqbal et al., 2012). Cortex utilizes coloured de Bruijn graphs to detect variant sequence among bacterial population datasets. The ST71 strains listed in Table S1 were defined as group 1 and the bovine ST97 strains were defined as group 2. Variant sequences that were present in at least one strain of each group, but none of the strains in the comparison groups were identified, then filtered as representative of genotype only if that variant sequence was present in all strains within a group. Identified variant CDSs were annotated using Prokka 1.5.2 (Seemann, 2014) and BLASTX against the nr GenBank database (Altschul et al., 1990). For each strain, de novo assemblies were aligned against S. aureus Newbould 305 contig 002 (GenBank accession number AKYW01000002) using the Mauve Contig Mover tool implemented in Mauve 2.3.1 (Darling et al., 2010).
Biofilm assay. Biofilms were grown in hydrophilic (Nunclon) 96-well plates, as described previously (Waters et al., 2014). Briefly, overnight cultures were diluted 1 : 200 in either BHI broth, or BHI supplemented with 4 % NaCl or 1 % glucose and grown for 24 h at 37 uC. The plates were then washed three times in distilled water and biofilms stained with 0.5 % crystal violet. Biofilm density was measured at A 490 .
Bacterial adherence to immobilized collagen. Bacterial adherence assays were conducted as described previously (Bannoehr et al., 2011). Microtitre plates were coated with doubling dilutions of commercially available bovine collagen type I (Life Technologies) in PBS at 4 uC for 15 h. After blocking in 4 % milk, exponential phase S. aureus cultures were standardized to OD 600 1.0 in PBS and applied to the plate. After crystal violet staining for 3 min and 5 % acetic acid treatment the plates were analysed at 590 nm (Synergy HT; BioTek). Each assay was performed in triplicate and each experiment was repeated at least twice independently. Inhibition assays were performed by incubating the standardized S. aureus strains in PBS with increasing concentrations of anti-Cna mouse mAb for 1 h at 37 uC before being applied to microtitre plates coated with 1 mg bovine collagen type I ml 21 .
Bovine mammary epithelial cell invasion assays. Bovine mammary epithelial cell line MAC-T was grown in media containing Dulbecco's modified Eagle's medium (DMEM) with 10 % (v/v) FBS (Gibco), 1 % (w/v) penicillin/streptomycin solution (Invitrogen) and 5 mg bovine insulin ml 21 (Sigma), and incubated at 37 uC in a humidified incubator with 5 % CO 2 . Cells were split when confluent with Tryple Express (Invitrogen). For the invasion assays, MAC-T cells were seeded into 24-well plates at 1|10 5 cells per well using the growth media and kept at 37 uC with 5 % CO 2 until confluent. On the day of the experiment, the wells were washed with warm PBS three times, and media containing only DMEM was then added to the wells and kept at 37 uC with 5 % CO 2 . Bacteria were grown to OD 600 0.6 in DMEM and co-cultured with the MAC-T cells for 2 h at m.o.i. 25. The cells were washed with PBS three times and DMEM medium with 150 mg gentamicin ml 21 was added to each well followed by incubation at 37 uC with 5 % CO2 for 30 min to kill any extracellular bacteria. The cells were washed again with PBS and lysed with 0.1 % Triton-X in PBS. The cell lysates were serially diluted, plated on TSA plates and incubated overnight at 37 uC to estimate the number of intracellular bacteria. The percentage invasion was calculated as a ratio of intracellular bacteria compared with the inoculum.
Cloning, recombinant protein expression, mitogenicity and Vb specificity assays. For gene cloning, forward primer 59-TAGCCTCGAGAGACACAAAATGAT CCAAA-39 was designed to amplify within the coding sequence of the selz gene from strain RF122 (SAB0026), after the signal peptide predicted by the Signal P 3.0 Server (http://www.cbs.dtu.dk/services/SignalP/). The gene in RF122 has an identical derived amino acid sequence in the mature protein (there is a single synonymous mutation) compared with the RF103 gene. Reverse primer 59-CG CCTCGAGCTACTTTTTAGTTAAGT-39 was designed to overlap the stop codon of the gene. Xho I sites were incorporated to facilitate cloning into the pET15b plasmid (Novagen). Cloning and recombinant protein purification were carried out as described previously (Wilson et al., 2011). Peripheral blood mononuclear cells were isolated from the blood of Holstein-Friesian cattle aged 18-36 months via jugular vein puncture by density-gradient centrifugation using Ficoll Plaque Plus (GE Healthcare) as described previously (Wilson et al., 2011). rSElZ protein was incubated with 1|10 6 cells ml 21 for 72 h at 37 uC, 5 % CO 2 in complete cell culture medium (RPMI 1640; Gibco) supplemented with 10 % heat-inactivated FCS, 100 U penicillin ml 21 , 100 mg streptomycin ml 21 , 292 mg Lglutamine ml 21 (PSG) and 50 mM 2-mercaptoethanol (Sigma-Aldrich). Cells were cultured for a further 18 h after the addition of 1 mCi [ 3 H]thymidine. Cellular DNA was harvested onto glass fibre filters and [ 3 H]thymidine incorporation was measured by liquid scintillation counting as described previously (Wilson et al., 2011). Total RNA was extracted before and after stimulation with rSElZ (1 mg ml 21 ) for 96 h. Expansion of cells expressing different bovine Vb gene subfamilies was determined using quantitative realtime PCR as described previously (Wilson et al., 2011).

Results
Identification of a highly divergent subtype of the major livestock-associated S. aureus clone CC97 The S. aureus CC97 lineage is a leading cause of bovine mastitis on a global scale Smyth et al., 2009;Spoor et al., 2013). We previously identified a single-locus variant of ST97 (ST71) among dairy herds in Ireland that had a capsule genotype (capsule type CP8) distinct from other CC97 strains (capsule type CP5) (Guinane et al., 2008). Subsequently, ST71 has been isolated from bovine mastitis in countries in Europe, North America and Asia, indicating successful inter-continental dissemination (Guinane et al., 2008;Ikawaty et al., 2009;Li et al., 2015;Smith et al., 2005Smith et al., , 2009. Here, in order to examine the evolutionary history of the ST71 clone and to examine the molecular basis for the switch in capsule serotype, we carried out comparative whole-genome sequence analysis of ST71 with other CC97 isolates and reconstructed the CC97 phylogeny using a maximum-likelihood approach. Strikingly, the ST71 isolates were highly divergent and clustered together at the tips of a long branch relative to all other CC97 isolates, implying either recombination or an elevated mutation rate for ST71 (Fig. 1a). Network analysis of the extracted SNP alignment using the program SplitsTree (Huson & Bryant, 2006) revealed evidence of a reticulate structure at the junction of the bovine ST97 and ST71 strains (Fig. S1), with the w test providing strong statistical support for recombination (P51.154|10 -7 ). Comparison of whole-genome sequences of isolates RF103 and RF116 representing ST71 and ST97, respectively, by mapping to a reference genome (MW2) revealed a core genome alignment of 2 472 592 nt with 4905 SNPs in total (Table 1). Of note, 4463 (91 %) of the SNPs were located within an *330 kb (SNP-dense) region spanning the origin of replication (oriC) starting within a gene encoding fructose 1,6-bisphosphonate aldolase and ending within a gene encoding a hypothetical protein, representing *15 % of the chromosome, with only 443 SNPs (9 % of the total) identified out-with this region (Figs. 1c and S2, Table 1). It was noteworthy that the SNP-dense region contained a higher proportion of synonymous to non-synonymous mutation (62 % : 21 %) compared with the rest of the genome (25 % : 52 %), consistent with a distinct evolutionary history (Table 1) (Castillo-Ramírez et al., 2011). Strikingly, masking of the SNP-dense region in the CC97 alignment resulted in a phylogeny with a markedly reduced ST71 branch length, consistent with the recent shared ancestry of ST71 and CC97 (Figs. 1b and S1).
In order to examine the evolutionary history of the SNPdense region within the S. aureus ST71 genome, we reconstructed the phylogeny of selected CDSs located within (n53) or out-with (n53) the SNP-dense region (Fig. S3). For each gene tree, the CDSs located within the SNPdense region of the ST71 S. aureus strains was more closely related to non-ST97 than to ST97 sequences, while the CDSs out-with the SNP-dense region co-segregated with ST97 alleles (Fig. S3). Taken together, the phylogenetic and sequence analysis data implied that ST71 has a hybrid genome that contains an *330 kb region with an evolutionary history distinct from that of the ancestral ST97.

Recombination events have led to the mosaic genome of ST71 S. aureus
To investigate the molecular events which led to the hybrid genome of ST71, we constructed a core genome alignment of the CC97 isolates (including ST97, ST71 and other single-locus variants), and selected non-CC97 ruminant S. aureus strains, and employed the recombination detection software BratNextGen in order to identify predicted recombination events, breakpoints and potential donor sequences (Fig. 1d) (Marttinen et al., 2012). A region of *330 kb in the ST71 chromosome that correlates with the SNP-dense region was predicted to be the result of multiple recombination events (Fig. 1d). In total, 10 predicted recombinant segments were identified in both ST71 strains RF103 and CO1122 (Tables 2 and S2). Similar recombinant regions were identified using the RDP2 suite of recombination detection programs (Martin et al., 2005). There was no evidence of recombination events among the non-CC97 ruminant-associated clones included in the analysis (Fig. 1d). Alignment of each recombinant sequence with genome sequences from a database of 178 S. aureus genome sequences selected to broadly represent the diversity across the species revealed that the majority of ST71 recombinant sequences were most closely related to sequences from a small number of ruminant S. aureus clonal lineages (Table 2, Fig. 2). However, the largest estimated recombination fragment of 224 kb was not closely related to any clonal complex, implying it originated in an unsampled or extinct strain (Fig. 2). Most of the predicted recombinant regions were contiguous, but three intervening regions (IR1 to IR3) were identified which comprised of 29, 22 and 9046 nt, respectively (Tables 2  and S2). It is feasible that IR3 also represents an additional recombinant fragment as it contained sequence not closely related to ST97, but it was not detected as such by the recombination detection programs employed. Taken together, the recombination detection analysis suggests multiple distinct genetic imports into an ST97 progenitor. However, the collinearity of the predicted recombination region and the lack of any vestigial ST97-like sequences identified between predicted recombinant fragments means that a single large import from a currently unsampled or extinct S. aureus genotype cannot be ruled out. Overall, these data indicate that the ST71 clone has emerged through recombination with ruminant-adapted S. aureus clones sharing the same environmental niche.

Recombination-associated diversification of the gene complement mediating host -pathogen interactions
In order to investigate the potential impact of the recombination events on ST71 emergence and niche adaptation, we examined the presence, absence and allelic variation of genes located in the recombinant region of ST71 in Mbp 1000000 1500000 2500000 2000000 Fig. 1. ST71 is a divergent subtype of CC97 containing a 323 kb genomic region with a distinct evolutionary origin. Maximum-likelihood tree constructed from core genome alignment with (a) and without (b) the SNP-dense region. The outgroup is strain MW2; ST71 strains are highlighted in red. CC97 strains included are those listed in Table S1 in addition to the ST71 strains. Scale is indicated for each tree in substitutions per site. (c) Pairwise strain comparisons of core genome SNP density over 10 000 bp windows between ST71 strain RF103 and ST97 strain RF116 (blue line). (d) Identification of multiple large-scale recombination events in the ST71 genome. The names of each strain are shown on the right with MLST sequence genotype indicated in parentheses. On the left side of the diagram is the proportion of shared ancestry tree as determined by BratNextGen (Marttinen et al., 2012). The coloured region indicates detected recombination events along the length of the core genome alignment. The same colour at overlapping genomic locations in different strains indicates those segments are of the same origin. Faint grey lines indicate alignment gaps.
comparison with the ancestral ST97 genetic background. Our analysis revealed that the horizontal import of large genome fragments from other ruminant-associated lineages resulted in a diversification of gene content marked by loss of at least 44 genes and gain of nine genes compared with the ancestral ST97 (Table 3). Of note, many of the variable genes were predicted to influence host-pathogen interactions. For example, the 7 kb operon required for histidine biosynthesis was absent in ST71. The histidine biosynthesis operon was strongly upregulated during human nasal colonization, but its absence indicated its dispensability for ST71 survival in the bovine niche (Krismer et al., 2014). Of note, previous studies of Lactococcus lactis and Streptococcus thermophilus isolates from dairy sources reported a common phenotype of histidine auxotrophy which evolved through frequent independent loss-of-function mutations in the histidine operon (Delorme et al., 1993;Hols et al., 2005). These data suggest that loss of capacity for histidine biosynthesis may be a specific bacterial adaptation to dairy cows or their environment.
Through recombination-mediated replacement, ST71 has lost the gene encoding the cell wall-anchored protein SasD, in addition to the entire ica operon (icaABCD operon and transcriptional repressor icaR), responsible for the biosynthesis of PIA, an essential component of polysaccharide-mediated biofilm formation (O'Gara, 2007). Lack of production of PIA by ST71 strains in comparison with ica-positive ST97 strains was confirmed by Western blot analysis with PIA-specific sera (Fig. S4), indicating that a polysaccharide-dependent biofilm is not required in the niche occupied by ST71 (O'Gara, 2007). We carried out a biofilm assay in the presence of NaCl or glucose which would support the production of a polysaccharide or protein-based biofilm, respectively. Considerable variation in the capacity to produce biofilm was observed among CC97 strains (Fig. S5). Of note, neither ST71 strain produced a biofilm in the presence of NaCl, consistent with lack of a PIA biosynthetic operon. However, one of the ST71 strains produced biofilm in the presence of glucose, demonstrating the capacity for production of a protein-based biofilm (Fig. S5).
Previously, by whole-genome microarray and PCR, we identified a capsule serotype genetic difference between ST71 and ST97 strains (Guinane et al., 2008). Here, we confirmed that ST71 has undergone a recombinationmediated capsule serotype gene switch from CP5 to CP8. Expression of CP5 and CP8 by selected ST71 and ST97 strains was examined by Western blot analysis. Several   (Fig. 2).
strains did not express detectable levels of capsule, consistent with the high prevalence of non-capsular bovine S. aureus evolved through loss-of-function mutations in capsular biosynthesis genes (Cocchiaro et al., 2006). One of the ST97 isolates examined was strongly positive for CP5 in contrast to both ST71 strains, consistent with loss of CP5 capsule serotype genes. However, considerable cross-reactivity was observed for the CP8-serotype antibody among CP8 gene-negative strains, limiting the capacity to draw conclusions regarding the gain of CP8 expression by ST71 (Fig S4). Capsule serotype switching is widely reported for Streptococcus pneumoniae strains and reported to be the result of selection for humoral immune evasion, particularly in the context of vaccine  Polysaccharide deacetylase Newbould305_0480 Metallo-b-lactamase Newbould305_0633 ATPase Newbould305_0558 Cobalt ABC transporter ATP-binding protein Newbould305_0544 Lipase Newbould305_0633 RNA helicase Newbould305_0638 Tandem lipoprotein Newbould305_0562 Lactonase Drp35 Newbould305_0533 Methionine sulfoxide reductase A Newbould305_0555 Polysaccharide deacetylase Newbould305_0632 Membrane spanning protein Newbould305_0563 Rhodanese domain sulfur transferase Newbould305_0534 Acetyltransferase Newbould305_0564 Pyrrolidone carboxylate peptidase Newbould305_0641 Amidohydrolase cysG Newbould305_0486 Precorrin-2 dehydrogenase Newbould305_0793 Hexitol dehydrogenase Newbould305_0736 RND transporter Newbould305_0757 NADH-dependent dehydrogenase escape (Croucher et al., 2011(Croucher et al., , 2013(Croucher et al., , 2015Wyres et al., 2015). Recent studies of Streptococcus pneumoniae and the Klebsiella pneumonia hybrid clone ST258 suggest that macro-recombination events may be driven, at least in part, by immune selection for a switch in capsule type Wyres et al., 2015). To the best of our knowledge, this is the first identification of a recombination-mediated serotype switch for S. aureus.
In addition to gene loss or replacement events, the import of large chromosomal fragments from other lineages into ST71 has resulted in acquisition of genes encoding proteins involved in host-pathogen interactions, including a gene encoding nitric oxide reductase which may enhance survival within macrophages in response to the bactericidal killing activity of nitric monoxide radicals (NO 2 ). Furthermore, ST71 has gained genes for a novel lipoprotein, novel putative superantigen and Cna, a cell wall-anchored protein that mediates binding to collagen and inhibition of the classical activation pathway of the complement system (Kang et al., 2013). Taken together, the large-scale recombination events of ST71 resulted in extensive diversification of the complement of genes which would influence host-pathogen interactions.

ST71 has acquired new pathogenic traits associated with bovine immunomodulation and extracellular matrix interactions
In order to investigate the potential impact of recombinationmediated gene acquisition by ST71 on ecological success and host adaptation, we examined the phenotypic consequences of selected imported genes implicated in pathogenesis. First, we investigated the functional effect of the import by recombination of the gene (cna) encoding Cna on the interaction of ST71 with the bovine extracellular matrix and intra-mammary epithelium. The cna gene has been reported to be present at a higher frequency in bovine strains of S. aureus compared with human clinical isolates, but the role of Cna in pathogenesis of the bovine udder has not been investigated (Delgado et al., 2011;van Leeuwen et al., 2005). Expression of Cna by S. aureus ST71 isolates, and lack of expression by ST97 isolates, was demonstrated by Western blot analysis with a Cna-specific mAb (Fig. 3a). Furthermore, adherence assays revealed that ST71 isolates, but not ST97 isolates, had the capacity to adhere to immobilized bovine type I collagen and this binding activity was reduced in a dosedependent manner by pre-incubation with the Cna-specific mAb (Fig. 3b, c). A previous study of bovine Streptococcus uberis reported a role for collagen binding in bovine mammary epithelial cell interactions (Almeida et al., 1999). We measured the capacity for ST71 strains to invade bovine mammary epithelial cells (MAC-T) in vitro and found a strain-dependent invasion capacity that did not correlate with the abundance of Cna in cell wall fractions detected by Western blot analysis (Fig. 3d). Of note, a previous study highlighted the role of capsular polysaccharide in masking Cna expression and binding activity in vitro (Gillaspy et al., 1998), consistent with the capsule-negative, higher collagen-binding phenotype of strain C01122 compared with the capsule-positive, lower collagen-binding activity of RF103 observed in the current study (Figs. 3d and S4). As a control in the invasion assay, we measured the ability of S. aureus strain Phillips and its isogenic mutant deficient in Cna expression to invade bovine mammary epithelial cells (Patti et al., 1994). In contrast to the parental strain Phillips, the cna isogenic mutant demonstrated a significant decrease in invasion, supporting a role for Cna in promotion of bovine epithelial cell invasion (Fig. 3d). In summary, molecular functional analysis has demonstrated that recombination-mediated acquisition of the cna gene has resulted in a gain of function by ST71 strains facilitating adherence to bovine collagen, the major protein component in bovine mammary tissue. Furthermore, the previous report that Cna can bind to the collagenous domain of C1q and interfere with complement activation suggests that acquisition of Cna may also confer an innate immune evasion strategy (Kang et al., 2013). Additional experimental work is required to elucidate the role of Cna in S. aureus-epithelial cell interactions.
Secondly, ST71 has acquired the gene for a putative novel superantigen, that we named staphylococcal enterotoxinlike toxin Z (selz). Superantigens contribute to disease pathogenesis by activating specific subpopulations of T-cells, resulting in T-cell anergy and dis-regulation of the acquired immune response (Spaulding et al., 2013). In particular, bovine isolates typically have multiple genes encoding superantigens that demonstrate host-specific activity, suggesting an important role in host adaptation (Deringer et al., 1997;Fitzgerald et al., 2001). SElZ belongs to the superantigen phylogenetic group and has 65 % amino acid identity with its closest homologue SEG (Fig. 4a). We purified recombinant SElZ and demonstrated that SElZ had dose-dependent mitogenic activity for bovine T-cells (Fig.  4b). Furthermore, we identified that bovine T-cell subpopulations with Vb3 and Vb11 receptors were preferentially activated, indicating that SElZ is a novel bovine superantigen (Fig. 4c). These data indicate that recombination-mediated acquisition of the selz gene by ST71 enhanced the capacity for non-specific stimulation of bovine T-cell subpopulations, suggesting a role in modulation of the acquired bovine T-cell immune response.
Taken together, the genomic and molecular functional analyses indicate that large-scale recombination events that shaped the evolution of ST71 have resulted in loss, acquisition and allele replacement of an array of pathogenic traits enhancing the capacity of S. aureus to survive in the bovine host. Our findings support the idea that large-scale recombination events can have a profound influence on the success of emergent bacterial clones.

Discussion
We propose a model for the emergence of the ST71 clone that involved large-scale recombination events influencing host-pathogen interactions and ecological success (Fig. 5).
The role of homologous recombination in adaptation of bacteria to different environments is not well understood. Previously, Sheppard et al. (2013) reported introgression in a specific Campylobacter coli clone with sequences from Campylobacter jejuni occupying the same agricultural niche, via genome-wide homologous recombination. The authors highlighted the acquisition of genes involved in the transport and metabolism of fucose, a major component of mucin, and speculated that this may confer a survival advantage in the gut (Sheppard et al., 2013). In  contrast to the introgressed C. coli clone that was the result of genome-wide homologous recombination of short DNA sequences, the hybrid S. aureus clone ST71 was the result of multiple large-scale recombination events affecting a region representing *15 % of the genome. To date, several hybrid clones of S. aureus have been identified, but the mechanism of import of large chromosomal fragments leading to recombination is unclear. S. aureus pandemic clone ST239 was the result of a single genetic import from a ST30 donor lineage into an ST8 background, and hybrid clones ST34 and ST42 also likely evolved from single large-scale recombination events (Robinson & Enright, 2004). It is speculated that these events may have been promoted by conjugative transfer of large chromosomal fragments (Robinson & Enright, 2004). However, some of the smaller predicted recombination events in ST71 could alternatively have arisen by phage-mediated transduction or pathogenicity island-mediated transfer (Chen et al., 2015;Moon et al., 2015). Pan-genus analysis of the Staphylococcus genome highlights the plasticity of the region spanning oriC (oriC environ) in different staphylococcal species, in contrast to the synteny observed across the rest of the genome. The oriC environ appears to be a hotspot for recombination across the Staphylococcus genus, suggesting a role in ecological specialization of different species (Takeuchi et al., 2005).
It is noteworthy that one of the differences in gene content effected by the recombination observed in ST71 is the replacement of genes encoding a type I restriction modification system with genes for a transposase and five hypothetical proteins (Fig. S6). We speculate that loss of a restriction barrier may support the capacity of S. aureus strains to import foreign DNA sequences leading to recombination. However, additional experimental work would be required to test this hypothesis. The apparent enrichment of synonymous SNPs in this region of the genome in ST71 compared with the ST97 background is consistent with the previous observation by Feil and colleagues that non-core and core sequences affected by recombination have lower ratios of non-synonymous to synonymous substitutions reflecting the longer time-frame for purifying selection to purge mildly deleterious mutations (Castillo-Ramírez et al., 2011). Importantly, the CC97 lineage is predicted to have diverged from a human-associated ancestor *1200 years ago, making it considerably younger than several of the predicted donors of the recombinant region in ST71, including CC133 and ST151, that are estimated to have originated in humans *5400 and *3000 years ago, respectively (Weinert et al., 2012). It is reasonable to infer that the import of genes from S. aureus donor clones which have been under ruminant host-adaptive selection for several thousand more years could enhance the fitness of the younger recipient clone in the bovine host. The predicted function of the genes affected and the phenotypic effects observed in the current study, including immunomodulation, enhanced adherence and intracellular invasion, lead us to speculate that the ST71 clone may be evolving towards a less pathogenic association with the bovine host. Of note, we previously identified very low levels of RNAIII expression among ST71 strains in vitro and reduced bacterial burden relative to ST97 after experimental murine infection, consistent with reduced virulence (Guinane et al., 2008). The possibility that some bovine S. aureus strains may be evolving towards a more intracellular lifestyle has been proposed previously (Herron-Olson et al., 2007) and may reflect increased antibiotic selective pressures in the dairy industry driving the bacteria into niches which are less accessible to most antibiotic classes. This may also explain the apparent paucity of resistance determinants among most S. aureus isolates from bovine sources (Spoor et al., 2013). However, the very recent emergence of ST71 strains that are meticillinresistant is a significant cause for concern, and surveillance is required to monitor the dissemination of the ST71 hybrid clone, and to comprehensively assess its threat to veterinary and public health.
Overall, our analysis suggests that the recombination events that shaped the mosaic genome of the ST71 clone have attenuated host-pathogen interactions, enhancing the capacity of S. aureus to evade the host immune response and adhere to host tissue. These findings provide a paradigm for the potential impact of large-scale recombination events on the rapid adaptive evolution of bacterial pathogens within defined ecological niches.  L. E. Spoor and others