Bacterial strains, growth conditions

Staphylococcus aureus Newbould 305 (hereinafter referred to as N305) [15] and RF122 [9] were isolated from cases of bovine mastitis. These strains are well characterized and can reproducibly induce severe mastitis (RF122; [25]; JR. Fitzgerald, University of Edinburgh, personal communication) or mild mastitis (N305) under experimental conditions [9],[15]. N305 was kindly provided by Dr F. Gilbert (INRA Tours) and received as a lyophilized vial provided by D.S. Postle (College of Veterinary Medicine, Cornell University, Ithaca, New York) in 1975 and prepared from the original Newbould 305 isolate supplied by F.H.S. Newbould. One can therefore consider it is close to the original isolate (F. Gilbert and P. Rainard, INRA Tours, personal communication). Subcultures prior to invasion assays were performed overnight as follows. S. aureus strains were grown in brain heart infusion medium (BHI; Becton Dickinson, Le Pont de Claix, France) at 37 °C under agitation (180 rpm). The cultures were washed once with phosphate-buffered saline (PBS) and suspended at different concentrations in Dulbecco’s modified Eagle’s medium (DMEM; pH 7.4; D. Dutscher, Brumath, France). Bacterial concentrations in subcultures were estimated by spectrophotometric measurements at 600 nm (OD₆₀₀). They were further confirmed using a micromethod, as previously described [26]. The resulting S. aureus populations (in CFU/mL) were determined on mannitol salt agar (MSA; D. Dutscher, Brumath, France) after 24 h of incubation at 37 °C.

Genome comparison of RF122 and N305

The N305 strain was fully sequenced using the Illumina technique. The whole genome sequencing and assembly strategies are described in Bouchard et al. [14]. The N305 genome sequence was analysed using SurfG + to predict protein locations (potentially surface exposed (PSE), secreted, membrane and cytoplasmic proteins) [27], and PIPS, a software suite designed for the prediction of pathogenicity islands which, in an integrative manner, utilizes multiple features (such as atypical G + C content, codon usage deviation, virulence factors, hypothetical proteins, transposases, flanking tRNA and the absence of this structure in non-pathogenic organisms) to detect pathogenicity islands [28]. For the PIPS analysis, the genome sequences of the strains Staphylococcus xylosus C2a (kindly provided by R. Talon and S. Leroy, INRA Clermont-Ferrand) and Staphylococcus carnosus TM300 [29] were used as references. The assignment of protein function to the coding sequences (CDSs) of the two genomes (N305 and RF122) was performed manually using the results from BLASTP and the COG (Clusters of Orthologous Groups) [30].

These genome sequences are available at DDBJ/EMBL/GenBank under the accession numbers AKYW00000000 (N305) and NC_007622 (RF122).

Phenotype characterization

Dimensional gel electrophoresis for the total proteome and secretome of RF122 and N305

For 2D-PAGE, protein samples were prepared as described previously, with some minor changes [3]. Briefly, S. aureus strains were pre-cultured in RPMI 1640 supplemented with 2.2′-dipyridyl (200 μM) to chelate extracellular iron and consequently reduce intracellular iron stocks. The cultures were then diluted 1:1000 in fresh RPMI 1640 with deferoxamine (152 μM). S. aureus strains were grown without agitation at 37 °C under microaerophilic conditions, in 500 mL flasks for the supernatant fraction or in 50 mL tubes for the total fraction and shaving experiments. These conditions were previously shown to best mimic growth in vivo during mastitis [32]. The cultures were centrifuged at 7000 g for 10 min and the supernatants were filtered through a 0.22 μm filter unit. Total cell lysate proteins were obtained from bacterial pellets, as previously described [32].The supernatant proteins were precipitated with 10% TCA at 4 °C overnight. The samples were centrifuged at 9000 g for 1 h at 4 °C. Protein pellets were washed three times with 96% ethanol and then dried. Proteins were solubilized in urea 8 M. The protein concentration was determined using a Bradford test (Sigma). A constant amount of proteins was used for each acrylamide gel so as to enable determination of a relative abundance of a given protein with respect to the total of secreted proteins. Three biological replicates were used for each analysis.

To purify the protein extracts, a 2D Clean Up kit (GE Healthcare, Orsay, France) was used according to the manufacturer’s instructions, and then the samples were treated as described previously [3]. Images of the gels were analysed using SameSpot software (TotalLab Ltd., Proteomics consult, Belgium), as previously described [1],[32],[33]. Spot volumes were determined and normalized with regard to the total volume of the gel. After this normalization, spots with a minimal fold change of 2 and an ANOVA e-value lower than 0.005 were selected for mass spectrometry analysis. The gel pieces were processed exactly as described previously [3],[34]. A shaving technique was used to analyse the PSE proteins. S. aureus cultures were centrifuged at 7000 g for 10 min and the total cell and supernatant fractions were treated separately. The pellets were washed twice with PBS. The bacteria were re-suspended in PBS with sufficient 5 mM DTT to reach 20 units of OD_600nm. Five hundred μL of this suspension were incubated with 20 μg trypsin for 1 h at 37 °C under agitation (180 rpm) and then centrifuged at 10 000 g. The supernatants were filtered through a 0.22 μm filter and incubated overnight with 1 μg trypsin at 37 °C under agitation. Trypsin digestion was stopped by adding 15 μL 5% TFA. The controls did not display any significant loss of viability from the counts of CFUs before and after shaving.

Identification of proteins

Proteins were identified using NanoLC-ESI-MS/MS as described by Le Maréchal et al.[1] with some minor modifications. Peptides were identified using the X! Tandem software. The database used for protein identification was composed of predicted proteins based on the RF122 and N305 genome sequences [9],[14], together with protein data on Staphylococcus aureus (taxon 1280) in the UniProtKB database [35], in order to obtain a statistically significant identification. To achieve valid identification with a high degree of confidence, each protein must have a minimum of two peptides corresponding to a p-value lower than 0.05. An auto-validation of peptides from the X!Tandem search results was performed using X!TandemPipeline [36]. Prediction of the sub-cellular localization of proteins was achieved using PSORTb 3.0.2 software [37]. Validated and identified proteins were then sorted by Clusters of Orthologous Groups (COGs) using EggNOG 3.0 software [38].

Regarding proteins identified by trypsin shaving, only those identified with at least two peptides in one strain and less than two peptides in the other strain were considered as more abundant in the first strain.

Statistical analysis

All experiments were carried out in triplicates (biological repeats). The differences in biofilm formation and cytotoxicity assays were assessed using paired Student’s t tests considering a P value lower than 0.05. In proteome analyses, proteins were considered as over expressed in total proteome or exoproteome when ANOVA e-value given by SameSpot software was lower than 0.005 and the minimal fold change was 2.

Comparison of the gene contents of S. aureus RF122 and Newbould 305

Analysis of the N305 genome using the PIPS pipeline [28] predicted the presence of five putative S. aureus Pathogenicity Islands (SaPIs). SaPI-N305_1 (contig 1) contained 35 predicted CDSs markedly similar to the CDSs of νSaβ already described in RF122 and S. aureus MW2 (Figure 1A). SaPI-N305_2 (contig 1) contained 15 CDSs markedly similar to a SaPIbov found in strain D30 (isolated in the context of human nasal carriage) [39] (Figure 1B). SaPI-N305_3 and SaPI-N305_4 were two adjacent putative SaPIs (contig 2) containing 19 and 28 CDSs, respectively. SaPI-N305_3 shared marked similarity with SaPIbov4 and SaPIbov5, which both carry a bovine variant of von Willebrand factor binding protein (vwb ^Sbo5 ) (Figure 1C). SaPI-N305_4 shared similarity with SaPIn2, which was first described in the human S. aureus strain, N315 [40] (Figure 1D). SaPI-N305_5 (contig 6) contained 35 CDSs, 17 of which were found to be very similar to SaPI2. PIPS analysis also revealed the presence of two putative bacteriophages, on contig 5, containing 34 CDSs and 19 CDSs with homology to Bacteriophage 80alpha and bacteriophage 187, respectively.

Some genes were only found in RF122, encoding for leukocidin M (LukM) and Panton-Valentin leukocidin F (LukF’-PV), toxic shock syndrome toxin 1 (TSST-1), exfoliative toxin B (Etb) and the enterotoxins Cbov/G/I/L/N/O (SECbov, SEG, SEI, SEL, SEN, SEO). All these proteins are secreted toxins.

Particular focus on microbial surface components recognizing adhesive matrix molecule (MSCRAMMs) genes revealed that their number was higher in N305 than in RF122. Although both strains carry the gene coding for fibronectin-binding protein A (fnbA), we found that the fnbA sequence in N305 had two additional fibronectin-binding domains when compared to that of RF122 (Figure 2).

The same biofilm formation but broader coagulase activity spectrum and higher cytotoxic capacities for N305 during in vitro assays

Based on the differences in gene content revealed by sequence analysis, phenotypic differences could be expected in terms of biofilm formation, coagulase activity, cytotoxicity and invasive capacities. These four properties were tested in vitro at the phenotype level (see Materials and methods for details). In line with the absence of differences in terms of the content of genes involved in biofilm formation, the two strains did not display any significant differences in phenotype. Indeed, RF122 and N305 revealed the same ability to form a biofilm on an abiotic surface, as determined by Crystal violet staining on a plastic surface (data not shown). By contrast, with respect to coagulase activity, the presence of vwb ^Sbo5 (Newbould 305_0962) in the N305 genome (Figure 1C), whose product was also found in the N305 surface exposed protein (Additional file 2), was clearly associated with an increased range of coagulase activity. N305 indeed clotted rabbit, goat and bovine plasma, whereas RF122 only clotted rabbit plasma (Figure 3A). Furthermore, N305 displayed greater cytotoxicity on MAC-T cells than RF122 after co-incubation with their respective supernatants. The viability of cells incubated with the N305 supernatant was 89% lower than under control conditions, whereas the viability of cells incubated with the RF122 supernatant was 36% lower (Figure 3B). N305 supernatant also presented a higher ability to hydrolyse caseins as shown by the translucent halo around the well of N305 deposit (Additional file 3).

Comprehensive analysis of the proteome of S. aureus Newbould 305

A comprehensive analysis of the N305 proteome (total cell, surface and secreted proteins) was carried out on cultures grown under conditions mimicking the context of mastitis, as described previously [32]. The N305 proteome was compared with that of RF122. A total of 215 proteins were categorized as being differentially produced in N305 or RF122, and were further identified. Sixty-eight proteins (31.6%) were identified as being over-expressed by RF122, and 147 (68.4%) by N305. In more detail, a total of 33 proteins (15.4% of total proteins) were identified in the culture supernatant (Additional file 4), 11 proteins being more abundant in RF122 and 22 in N305. Supernatant gels (Figure 4A and B) displayed the most flagrant proteomic differences between N305 and RF122. In the total cell lysate, 36 proteins (16.7% of total proteins) were identified (Additional file 5), ten proteins being more abundant in RF122 and 26 in N305. Trypsin shaving enabled the identification of up to 409 proteins because of the sensitivity of this method, and included 146 proteins that were relatively more abundant in one strain (67.9% of total proteins) (Additional file 2), with 47 proteins more abundant in RF122 and 99 in N305.

The proteins were classified into COGs as a function of their annotation. Most of the secreted and PSE proteins identified here belonged to cellular processes and signalling categories or were poorly-characterized proteins. Among these, a number of exoproteases were found to be common to both strains (Additional files 4 and 2), while some exoproteases such as Spl proteases (SaPI-N305_1), were also specifically found in N305. Virulence factors (classified in the category of poorly-characterized proteins) were also found to be specific to N305. These included haemolysins and leukotoxins (alpha- and gamma-haemolysins, LukS/F) and proteins involved in adhesion to host tissues (e.g. Newbould305_1324, encoding an extracellular matrix and plasma binding protein) and in evasion of the host immune response (e.g. Spa, vWbp^Sbo5). By contrast, RF122 was found to produce leukotoxins (e.g. LukM/F’ and LukE) and enterotoxins (e.g. bovine variant of the staphylococcal enterotoxin C sec-bov) as well as iron metabolism proteins (e.g. IsdB and D). As expected, total proteins (as well as the surface exposed proteins identified by trypsin shaving) comprised many more proteins that those grouped in the Metabolism categories.

An inventory of potential mastitis-associated functions in N305 reflects host adaptation and propensity to chronicity

Interestingly, SaPI-N305_3, one of the SaPIs predicted in silico by PIPS analysis, was absent from the RF122 genome and shared homology with the previously described SaPIbov4 and SaPIbov5 [56]. This putative SaPI carries the Newbould 305_0962 gene whose product was found to be surface-exposed by trypsin shaving in N305. This gene shares homology with SaPIbov4_ORF15, a gene encoding a bovine and SaPI-associated variant of von Willebrand factor binding protein which has been described in the Staphylococcus aureus strain BA4 pathogenicity island SaPIbov4 [57]. These bovine variants of vWbp (identified in SaPIbov4 and SaPIbov5) have been shown to provide carrier S. aureus strains with an ability to specifically coagulate ruminant plasma, whereas strains that do not carry these SaPIs (such as RF122) coagulate rabbit but not ruminant plasma [56]. We were able to show that N305 did indeed clot ruminant (bovine and caprine) plasma. This capacity to coagulate plasma is correlated with an ability to induce abscesses during infections [58]. The formation of abscesses is a mechanism of bacterial resistance against host immune cells. N305 is therefore endowed with ruminant-specific coagulation capacities which reflect its adaptation to the bovine host and correlate with its propensity to induce chronic mastitis.

S. aureus N305 is suitably armed to invade host cells

A genomic comparison of S. aureus N305 and RF122 also revealed differences in colonization and invasion factors. Most of these factors belong to surface proteins such as Secretable Expanded Repertoire Adhesive Molecules (SERAMs) or MSCRAMMs, which participate in both adhesion and the internalization of S. aureus in host cells [59]. SERAMs or MSCRAMMs interact with host proteins such as fibronectin, collagen and elastin, and trigger invasion. Numerous studies have reported the importance of these factors during an infection. In particular, fibronectin-binding proteins are considered to be the principal staphylococcal proteins involved in S. aureus internalization in host cells [60]. The analysis of the N305 genome using the SurfG + package [27] enabled the identification of PSE proteins. When the genome comparison of RF122 and N305 was focused on these PSE proteins, it revealed that N305 is better equipped than RF122 to achieve host invasion. S. aureus N305 indeed possesses two fibronectin-binding proteins (fnbA and fnbB) rather than one for RF122 (fnbA). Furthermore, a comparison of the predicted FnbA protein sequence revealed that N305 FnbA presented two additional fibronectin-binding domains, and RF122 FnbA only one. This suggests a greater affinity of N305 for fibronectin. It should be noted that fibronectin-binding proteins have been shown to be indispensable to the internalization of S. aureus into host cells [60],[61]. Other surface protein genes were found in N305 while they were missing from the RF122 genome.

Biofilm formation was not significantly different in the two strains, in the conditions used (on abiotic surface). Although bap gene is reportedly associated with chronic mastitis isolates [8], its absence does not hinder N305 to induce chronic mastitis. Moreover, the sasG and pls genes, known to participate in host colonization and modulate pathogen internalization, were also found in N305 only. These observations correlate well with previous studies by our laboratory which compared the adhesion and internalization capacities of both strains, in vitro, on bovine MEC cultures (MAC-T line). The adhesion and internalization rates of N305 were indeed significantly higher than those of RF122 [24]. These capacities seem to favour host invasion by S. aureus N305, and together with the smaller number of toxin genes and host-adaptation features such as production of the bovine variant of vWbp, may account for a milder but chronic phenotype in N305 mastitis.

Some moonlighting proteins were also found in the proteomic profiles. Proteins identified as being relatively more abundant in the N305 secretome formed part of the “Carbohydrate transport and metabolism” COG category and were also involved in adhesion and colonization of host tissue, as has been reported for enolase [62],[63], glyceraldehyde-3-phosphate deshydrogenase (GAPDH) [64] and phosphoglycerate kinase (pgk) [65]. Other cytoplasmic proteins reportedly involved in adhesion, such as glucose-6-phosphate isomerase (pgi) [66], fructose-biphosphate aldolase (fda) [67], superoxide dismutase (sod) [68] and elongation factor Tu (tuf) [69], were also found, but in the N305 supernatant only (Figure 4). The fact that N305 expresses more of these proteins than RF122 may also account for N305′s better capacities to adhere, internalize and cause chronic infections.

Conclusions

N305 harbours fewer toxin genes but more colonization and invasion factors than RF122. In vitro, coagulation, proteolysis and cytotoxicity were significantly higher in N305 compared to RF122 as well as adhesion and internalisation, as previously demonstrated in Bouchard et al. [24]. This correlated well with its gene content and proteomic profile. The range of bovine-adaptive features might account for the propensity of N305 to persistence and mild and chronic mastitis, which is more difficult to detect and cure. By contrast, RF122 appeared to be less well equipped in terms of its genes related to host-adaptation and tissue invasion, but better equipped in toxin genes, and therefore more prone to inducing a severe inflammatory response. Whether these two profiles correspond to the evolutionary strategies of S. aureus bovine strains towards a commensal or strictly pathogenic lifestyle is not yet known, and this point deserves further investigation, including epidemiological studies. In line with this, based on MLST data, Smith et al. suggested that N305 should be considered as a teat skin associated strains [70]. S. aureus N305 nevertheless represents a strain of choice for further study to improve our understanding of the pathogenesis of S. aureus in the context of chronic mastitis. A complete demonstration of the involvement of the determinants identified here would require further experiments such as gene disruption and virulence determination in vivo. The genome sequence and fine characterization of this strain is a first necessary step towards developing strategies to understand, prevent and combat this disease.