ABSTRACT
We identified naturally occurring Staphylococcus aureus mutants of the restriction modification pathway SauI, including bovine lineage ST151. In a model of vancomycin resistance transfer from Enterococcus faecalis, ST151 isolates are 500 times more susceptible than human S. aureus isolates. The eradication of “hyperrecipient” strains may reduce the evolution of vancomycin-resistant S. aureus.
Six cases of fully vancomycin-resistant Staphylococcus aureus (VRSA) in U.S. hospitals have been described since 2002 (1, 16). VRSA strains have acquired vancomycin resistance genes, such as vanA, from vancomycin-resistant enterococci (VRE) (6, 16). Both VRE and methicillin-resistant S. aureus (MRSA) are widespread in hospitals (3, 7), and it is not uncommon for a patient to be colonized or infected with both and treated with vancomycin. In addition, VRE are found in the agricultural setting despite the banning of glycopeptides and S. aureus strains are widespread in animals and are a major cause of dairy cow mastitis. The emergence and spread of VRSA in hospitals is an enormous threat, with resistance to all new antibiotics already reported for S. aureus and no vaccine on the horizon.
The mechanism of the spread of an antibiotic resistance gene from enterococci to S. aureus was first described by Clewell et al. in 1985 (2). Some Enterococcus faecalis strains carry large pheromone-responsive plasmids, which in turn can carry other mobile pieces of DNA, such as transposons encoding resistance genes. These plasmids respond to a lipoprotein signal produced by S. aureus, triggering conjugation. The transposon jumps to the S. aureus chromosome, while the plasmid is unable to replicate in S. aureus and is lost. In this model, the S. aureus recipient strain was 879R4RF, a putative “restriction-deficient” isolate.
A laboratory transfer of vancomycin resistance from enterococci to S. aureus recipient B111 was reported in 1992 (12). While the process was not genetically characterized, it showed that vanA could be transferred to and expressed in S. aureus. When the first naturally occurring VRSA strain was isolated in Michigan (6, 16), the mechanism of transfer appeared to be similar to that described by Clewell et al. (2). The donor plasmid from E. faecalis, pAM830, facilitated the transfer of a resident vanA Tn1546-like element to S. aureus and then was lost. However, pAM830 was not pheromone responsive and was more closely related to the enterococcal broad-host-range plasmid pIP501 (6).
We have recently described the major mechanism that blocks the horizontal transfer of DNA into S. aureus (15). It is the Sau1 (or Sau1I) restriction modification (RM) system. A restriction enzyme composed of subunits encoded by sau1hsdR (restriction) and one of two sau1hsdS (specificity) genes identifies and binds to a specific DNA sequence and digests the DNA. This protects the bacterial cell from deleterious foreign DNA, such as that from a bacteriophage. To protect its own DNA, S. aureus also produces a modification enzyme composed of subunits encoded by one of two sau1hsdM genes and the same sau1hsdS gene. This modification enzyme recognizes the same specific DNA sequence and methylates it, protecting it from restriction. The Sau1 system blocks uptake of DNA from Escherichia coli and reduces uptake from enterococci. In addition, it prevents the transfer of DNA between the dominant lineages of S. aureus (9), which each have unique sau1hsdS gene variants (15). The standard S. aureus laboratory strain that accepts foreign DNA, RN4220, is deficient in sau1hsdR (15).
The aim of this study was to investigate whether all of the strains of S. aureus are capable of accepting resistance genes from enterococci or whether only a select few “restriction-deficient” strains could do this.
S. aureus isolates.
The S. aureus isolates included the standard laboratory strain 8325-4; its sau1hsdR-deficient mutant RN4220 (15); 879R4RF, which has a “restriction-deficient” phenotype (3); and B111, kindly donated by Sue Howell (12).
The human S. aureus isolates represented the major dominant lineages from hospitals and the community, including hospital MRSA. They included 13 epidemic hospital MRSA isolates, representing the lineages CC30, CC22, CC5, CC8, and CC45 (4, 11, 13); 11 hospital methicillin-susceptible S. aureus (MSSA) isolates, representing lineages CC8, CC30, CC45, and CC15 (10); and 15 community-acquired MSSA isolates, representing all 10 dominant human lineages, CC1, CC5, CC8, CC12, CC15, CC22, CC25, CC30, CC45, and CC51 (9).
Isolates from animals were kindly collected by David Lloyd and colleagues at the Royal Veterinary College, United Kingdom. These isolates were from cows (n = 19; predominantly mastitis), horses (n = 13), sheep (n = 2), goats (n = 2), and a camel (n = 1). A further 18 United Kingdom bovine S. aureus strains were kindly donated by Chris Teale, Veterinary Laboratories Agency. The lineage of animal strains was determined by microarray (9, 17), and representative isolates of each lineage were confirmed using multilocus sequence typing (5); a more complete description of this population will be published separately. RF122 was kindly donated by Ross Fitzgerald. Three animal isolates were naturally tetracycline resistant and not studied further. Antibiotic resistance was tested on Mueller-Hinton agar with discs according to CLSI (formerly NCCLS) guidelines.
Conjugation assays.
We used the conjugation assay previously described by Clewell (2, 15). One group of isolates accepted enterococcal DNA from JH2-2 at an extremely high transfer frequency (Table 1; Fig. 1). These isolates were all of the same lineage, ST151, and all came from dairy cows in the United Kingdom. The sequencing of the two sau1hsdS genes required for Sau1 activity in five ST151 strains revealed that they each had exactly the same two stop mutations, one in each of the two sau1hsdS gene copies. This genotype is predicted to prevent the restriction (and modification) of foreign DNA and would explain the enhanced ability to accept resistance genes from enterococci. The RF122 sequence shows that it is ST151 and carries two identical mutations. Thus, all strains from this lineage are probably “hyperrecipient.” In our collection, one-third of 39 bovine isolates were from the ST151 lineage. ST151 isolates have been reported in cows in Norway (www.mlst.net), but in a study of U.S. and South American bovine isolates, they were not found (14). There are no reports of ST151 in humans.
Susceptibility of strains to conjugational transfer. Data are presented as a log scale of numbers of transconjugants per 108 donors. Strains are grouped according to source, with laboratory strains in the order of 8325-4, RN4220, 879R4RF, and then B111, followed by bovine, equine, and other animal and human sources as indicated below the figure. The lineage of each isolate is indicated in color (see key), with ST151 isolates in red. All isolates from human lineages are navy blue and include one ST1 and three ST188 from bovine sources and four ST1 and one ST22 from equine sources.
Conjugation frequency
The S. aureus B111 recipient strain used by Noble et al. (12) was hyperrecipient (Table 1). However, after sequencing the five hsd genes in this strain, we had not identified any obvious mutations. B111 belongs to CC1, and another hyperrecipient CC1 isolate was identified from a horse. Other CC1 isolates were not hyperrecipient. The 879R4RF isolate used by Clewell et al. (2) was also found to be hyperrecipient. We have not identified an hsd mutation in this strain. It belongs to CC51, but a second CC51 isolate was not hyperrecipient. Thus, some hyperrecipient strains have developed independently of their lineages. Furthermore, there is likely to be a second pathway in S. aureus that blocks the horizontal transfer of foreign DNA or an unknown but necessary step that is essential for Sau1 activity. Other RM pathways have been described for some isolates of S. aureus, including on mobile genetic elements, and they may be implicated (8, 15).
Conclusions.
The discovery that certain S. aureus lineages and strains have deficiencies in the dominant RM pathway and a hyperrecipient phenotype is key for predicting how VRSA may arise in the future. The high incidence of antibiotic resistance gene transfer into animal strains strongly supports the decision to ban glycopeptide antibiotics, such as avoparcin, for agricultural use (18). The incidence also supports the surveillance of S. aureus populations for hyperrecipient strains, particularly when they are in close contact with VRE, so that high-risk situations can be identified and contained.
Nucleotide sequence accession number.
During the course of this project, S. aureus RF122, an isolate from bovine mastitis in Ireland, was sequenced and deposited in GenBank under accession number AJ938182.
ACKNOWLEDGMENTS
We thank our strain donors and Josh Cockfield and Denise Waldron for comments. We thank The Wellcome Trust-funded Bacterial Microarray Group at St. George's (BμG@S) (Jason Hinds, Kate Gould, Adam Witney, Lucy Brooks, and Philip Butcher) for assistance with microarray studies.
This work was supported by a grant from the Department for Environment, Food and Rural Affairs to J.A.L.
FOOTNOTES
- Received 17 November 2006.
- Returned for modification 26 December 2006.
- Accepted 12 March 2007.
↵▿ Published ahead of print on 19 March 2007.
- American Society for Microbiology
