MATERIALS AND METHODS

Bacterial Isolates
The study was conducted between September 2007 and September 2009 at the Department of Medical Microbiology of the University Medical Center Utrecht, Utrecht, The Netherlands. Seventy six Staphylococcus aureus isolates from clinical cases of bovine mastitis were included. Thirty isolates were obtained by the Faculty of Veterinary Medicine Utrecht University, Utrecht, The Netherlands, from 26 farms near Utrecht. Isolates from farms sampled twice were taken at least one year apart. Another 46 isolates were collected by the Animal Health Service in Deventer, The Netherlands for the Central Veterinary Institute (CVI), Lelystad and The Netherlands from farms throughout The Netherlands.

Fig. 1:

Virulence factor profiles observed within various typing methods. Place where the isolates were taken. FVM: Veterinary Faculty Utrecht University; CVI: Central Veterinary Institute, Lelystad, The Netherlands. The presence and absence of genes is indicated by red and green spots. Virulence factor profiles were defined based on the presence and absence of the genes, and resulted 5 main profiles, A, B, C, D and E

The sources were individual teat milk samples from dairy cattle from all over the country from clinical or subclinical cases of mastitis. Each isolate from CVI represents one farm, the location of the farms is unknown but they are distributed all over The Netherlands. All isolates were methicillin-susceptible S. aureus (MSSA) and collected between 1988 and 2005.

These isolates were genotyped by Multilocus Sequence Typing (MLST), staphylococcus protein A (spa)-typing, Multiple Locus Variable Number Tandem of Repeat Analysis (MLVA) and Pulsed-Field Gel Electrophoresis (PFGE) (Fig. 1) (Ikawaty et al., 2008). Most of the 76 bovine S. aureus isolates used in this study belonged to 4 dominant MLST types. We observed 3 dominant spa-types and 2 dominant MLVA types. DNA of 76 isolates was digested with SmaI and showed 21 PFGE profiles, 4 dominant profiles, 10 profiles related to the dominant profiles and 7 unique PFGE profiles (Fig. 1). No particular difference was observed between regionally and nationally obtained S. aureus bovine mastitis strains in terms of PFGE profile, MLST and spa-types, except for MT118 that was only present among regionally obtained isolates. Despite the fact that isolates could belong to the same cluster, they were not considered to belong to local outbreaks as at least 90% of the isolates were obtained on different farms whereas the remaining isolates were obtained at least one year apart when sampled on the same farm. MLST generally showed a distinct cluster of bovine S. aureus compared to isolates from the human population, although a little sharing of strains between the bovine and human population was observed. In addition, the ST or Clonal Complexes (CC) of bovine mastitis isolates from The Netherlands differed from those described elsewhere with the exception of 4 isolates (S0337, S0426, S0431 and S0436) that belonged to CC97 which has previously been reported to be obtained from humans.

Genomic DNA Preparation
Preparation of bacterial genomic DNA was performed using the NucleoSpin kit (Macherey-Nagel) following the protocols from the manufacturer with the exception that the bacterial pellet is resuspended in 20 mM Tris/HCl, 2 mM EDTA and 1% Triton X-100 supplemented with 0.2 mg mL^-1 of lysostaphin, achromopeptidase and RNase.

PCR Amplifications
PCRs were performed to detect the genes of 40 potential virulence factors including 19 staphylococcal toxin genes, 12 adhesion genes; 3 genes encoding extra-cellular proteases, 3 immune evasion genes, 2 genes for capsular polysaccharide and SCC_ACME (Table 1). Table 1 lists the oligonucleotide primers, positive and negative control strains used in this study.

Table 1:	Oligonucleotide primers used
1Positive control strains; 2Negative control strains; 3References. UMCU: Primers were designed at Dept. of Medical Microbiology, University Medical Centre Utrecht, Utrecht, The Netherlands

RESULTS AND DISCUSSION

Among the 18 toxin genes tested, hlb, sspA and sspB were detected in all isolates. The genes for enterotoxins A (sea), B (seb), K (sek), O (seo), P (sep), Q (seq), leukocidin encoded by LukS/F, exfoliative toxin A (eta) and B (etb) could not be amplified while the genes encoding the other 8 toxins were present in 1-80% of the isolates. The recently described enterotoxins G and I were the most prevalent (78 and 80% of the isolates, respectively). We observed that 56 of 76 (76%) isolates encoded both SEI and SEG. Forty (52.6%) isolates encoded SEM and SEN, although these two genes were not present simultaneously in 27 (35.5%) isolates. The seh gene was present in only 1 isolate (1%). The SaPIbov pathogenicity island signature combination of sec3, sel and tst was found in 15 (19.7%) isolates. Two (2.6%) isolates encoded sec3 and tst but lacked sel, while 6 (7.9%) other isolates encoded either sec3, sel or tst only. In addition, lukE was detected in 20% of isolates.

Ebps and efb could be amplified for all isolates, while pls was absent. PCR specific for fnbA and fnbB showed that 73 (96%) isolates possessed fnbA and 33 (43%) isolates had fnbB. Three gene targets involved in biofilm formation, icaB, icaC and icaD, were found in 64 (84%), 63 (83%) and 65 (86%) isolates, respectively. ClfA was detected in 16 (21%) isolates, 33% (n = 16) isolates were positive for sdrE, 49% (n = 37) were cna positive, 18% (n = 14) contained map and most of isolates (99%, n=75) obtained coa.

The three immune evasion genes, sak, scin and chips, could not be amplified in almost all isolates, except for one isolate (S0336). SCC_ACME was not amplified in any of the isolates. Cap8 was detected in 73 isolates, while cap5 was present in 3 isolates.

Based on the presence and absence of virulence factors five main clusters (A, B, C, E and F) were identified among the 76 bovine mastitis S. aureus isolates (Fig. 1). In addition, a unique gene pattern (called D) was observed for 1 isolate (S0366). There is general agreement between the 4 typing methods used (PFGE, MLST, spa-typing and MLVA) and the 6 clusters of isolates, but exceptions were obviously present. Cluster B and C could not be identified based on PFGE profiles, since some of the isolates with an identical PFGE profile were distributed across both virulence profile clusters. These isolates were also considered closely related based on the other typing methods (MLST, spa-typing and MLVA).

Cluster A isolates mostly possessed sdrE, finbB and cna genes which were totally absent in cluster B and C isolates and occasionally present in cluster D and E isolates (Table 2). Identical and closely related isolates identified by MLST, spa-typing and MLVA in cluster B and C had similar virulence gene patterns, except that SaPIbov (tst, sec3 and sel) were predominantly detected in cluster C isolates. Virulence gene patterns detected in cluster E and F isolates mostly lack SaPIbov similar to isolates of cluster A. The gene for enterotoxin sem was absent, while sen, seg and sei were occasionally detected. Almost all isolates from both clusters contained lukE, clfA and map, while other clusters did not. icaB-D and cap8 were detected in all clusters (except cluster E isolates). The only isolate in cluster D showed a unique gene pattern with the presence of seh and a unique combination of other virulence genes.

Thus in general, there is an agreement between the virulence profile and the typing methods and the clusters with exception of clusters B and C. Cluster C contains SaPIbov-positive isolates, whereas this is absent in cluster B isolates, but the overall virulence gene profiles are similar for both clusters.

By analyzing the presence of all virulence factors tested in all isolates fnbA, hlb, ebps, sspA, sspB and coa were always present in Dutch isolates, while cap8, sem, seg, sei, icaB, icaC, icaD and sen were variable present.

The presence of virulence factor encoding genes and variation in their presence in extensively genotyped S. aureus isolates collected from bovine mastitis was investigated. Hlb, ebps, efb, sspA and sspB were found in all isolates tested and coa in all isolates except one, while sea, seb, sek, seo, sep, seq, PV- LukS/F, eta, etb, pls and SCC_ACME were completely absent. Variation in presence was shown for 9 toxin genes: sec3, seg, seh, sei, sel, sem, sen, tst and lukE and 9 adhesion related genes: fnbA, fnbB, icaB, icaC, icaD, clfA, sdrE, cna and map and other genes such as scin, chips, cap8 and cap5. The clustering of tst, sec3 and sel genes is specific for a bovine pathogenicity island known as SaPIbov (Smyth et al., 2005) was present in only part of the isolates. From these data a minimum set of genes which may be required for S. aureus bovine mastitis was distilled which includes hlb, efb, ebps, sspA, sspB and fnbA. However, efb has been reported to be present in only part of the isolates by others (Zecconi et al., 2006). Probably other virulence factors do contribute to the development and severity of disease. Details of the pathogenesis of bovine mastitis may differ for different gene profiles.

Table 2:	Presence and absence of toxin and adhesion gene profiles in different bovine PFGE-types
aBovine staphylococcal pathogenicity island characterized by the presence of sec, sel, and tst. The (+) means sec, sel and tst genes were amplified, while the (-) means none of the genes were detected. Written gene/s depicted their presence. bEnterotoxin gene cluster (egc) consists of 5 SEs, seg, sei, sem, sen, and seo. Since all of our isolates lack of seo, the (+) indicated the presence of seg, sei, sem and sen

The virulence factor gene profiles in general agreed with the clustering obtained from the 4 genotyping methods. The exception of cluster B and C was caused by the presence of SaPIbov in a part of the isolates belonging to PFGE types D, D1, D2 and E. This result indicates that SaPIbov was acquired independently by several isolates belonging to different genetic lineages. SaPIbov was present in 19.7% of the isolates and10.5% of all isolates carried either 1 or 2 genes representative for SaPIbov. The tst gene has also been reported on other pathogenicity islands (Lindsay et al., 1998). Whether the presence of sec3 and sel in two isolates and sec3 or sel in other isolates implicates that SaPIbov is incomplete or that the sec3 and sel can be present on other locations is unknown. Although SaPIbov may contribute to mastitis it not required, because other lineages appear to be equally widespread as cluster C (Fig. 1).

The presence of hlb, coa in nearly all isolates is in agreement with other reports (Dinges et al., 2000; Zecconi et al., 2006). The genes that were not identified among Dutch bovine mastitis isolates were also generally absent from isolates from other countries with the exception of sea, sek, seo and eta and etb (Fournier et al., 2008; Haveri et al., 2007, 2008; Monecke et al., 2007; O’Riordan and Lee, 2004; Vautor et al., 2009; Zecconi et al., 2006). Sea and seo for example have been reported to be present in 77 and 52% of mastitis isolates in Switzerland (Fournier et al., 2008). However, Turkish isolates also lacked sea, but eta, etb, sek and seo were not tested (Karahan et al., 2009). The variability of the presence of other toxin genes reported here is in general agreement with other reports (Fournier et al., 2008; Haveri et al., 2007; Kenny et al., 1993; Monecke et al., 2007; Zecconi et al., 2006). The coexistence of the toxin genes seg and sei has been described before and further explained by the presence of both genes in a tandem orientation and belong to an operon of enterotoxin gene cluster (egc) (Jarraud et al., 2001; Karahan et al., 2009). In addition, we detected sem and sen within seg- and sei- harboring S. aureus isolates without the presence of seo. This finding is in contrast to those studies which demonstrated that sem, sen, seg, sei and seo predominate in S. aureus from animal hosts (Haveri et al., 2007; Smyth et al., 2005). The reason for these discrepancies is not known, although the number of different genetic clusters observed among the Dutch isolates was low. In addition, many of the virulence factors are encoded on mobile genetic elements which can easily be lost or exchanged.

None of our isolates were lukPV-positive in agreement with the study done by Fueyo et al. (2005), although, Zecconi et al. (2006) identified pvl in more than 50% of isolates from bovine mastitis (Zecconi et al., 2006). However, also among human methicillin-susceptible S. aureus only a few percent of the isolates carry the pvl genes.

One of the common features of invasive bacterial pathogens is a capsule. In this study, the cap8 was predominant compared to the cap5. Poutrel et al. (1988) described the difference in distribution of CP5 and CP8 between clinical S. aureus isolates collected from humans and bovines. He observed 51.4% of S. aureus from bovine origin were CP5 positive and 18% were CP8 positive, while the percentages of human S. aureus harboring CP5 and CP8 were 20 and 60%, respectively. Several studies described that CP5 and CP8 were commonly found in the clinical human S. aureus isolates (Sordelli et al., 2000). Since CP 5 and 8 could be recovered from either human or animal derived S. aureus isolates, we conclude that CP5 and 8 are non-host specific.

Only one isolate carried the genes for CHIPS and SCIN, which interfere with activation of the human complement system but not the bovine complement system. Both genes are located on the same bacteriophage that is integrated in the hlb gene (Van Wamel et al., 2006). The hlb product is poorly effective against human erythrocytes, but highly effective against bovine erythrocytes and bacteriophage insertion disrupts the hlb gene. Therefore, we consider this isolate as human derived. ST97 isolates have been obtained from both humans and bovines (Smith et al., 2005). The gene profiles of these isolates resemble those of bovine mastitis isolates and lack the human complement specific interference factors (Table 2 and Fig. 1). Based on these data we believe that ST97 S. aureus isolates are from bovine origin.

CONCLUSION

In summary the virulence factor profiles with the exception of SaPIbov were generally in agreement with the typing results. This suggests independent mobilization of this pathogenicity island. Five major virulence factor profiles were observed. Of the tested virulence genes hlb, ebps, sspA, sspB, fnbA and coa were present in all isolates from bovine mastitis.