Transposon characterization of vancomycin-resistant Enterococcus faecium (VREF) and dissemination of resistance associated with transferable plasmids

Abstract

Objectives

VanA glycopeptide resistance has persisted on broiler farms in the UK despite the absence of the antimicrobial selective pressure, avoparcin. This study aimed to investigate the contribution of horizontal gene transfer of Tn1546 versus clonal spread in the dissemination of the resistance.

Methods and results

One hundred and one vancomycin-resistant Enterococcus faecium isolated from 19 unrelated farms have been investigated. Tn1546 characterization by long PCR and ClaI-digestions of amplicons showed a very low diversity of Tn types (n = 4) in comparison to the high genotypic diversity demonstrated by PFGE (n = 62). Conjugation experiments were carried out to assess the transfer of vancomycin resistance. Co-transfer of vanA together with erm(B) positioned on the same conjugative plasmid containing a replicon similar to pRE25 was demonstrated and also the presence of different plasmid replicons, associated with antimicrobial resistance on several unrelated farms.

Conclusions

Horizontal transfer of vancomycin resistance may play a more important role in the persistence of antimicrobial resistance than clonal spread. The presence of different plasmid replicons, associated with antimicrobial resistance on several unrelated farms, illustrates the ability of these enterococci to acquire and disseminate mobile genetic elements within integrated livestock systems.

Introduction

Enterococci are lactic acid bacteria that are important in environmental, food and clinical microbiology. They are part of the normal intestinal flora in humans and animals, and have been used as indicators of faecal contamination. They are also relevant in the food industry as fermenters of cheese and sausages. Recently, they have been used as probiotics for animals and humans.¹

Generally, enterococci are not regarded as primary pathogens but they are considered opportunistic pathogens and usually cause infections in patients who have severe underlying disease or are immunocompromised.² A contributing factor to the severity of disease can be their resistance to a wide range of antimicrobials, in particular to the glycopeptide vancomycin, which is used as one of the drugs of last resort for the treatment of multiresistant Gram-positive infections, such as those caused by methicillin-resistant Staphylococcus aureus.

Several studies have shown that the emergence of vancomycin-resistant Enterococcus (VRE) in Europe was associated with the widespread use of the growth promoter avoparcin (another glycopeptide) in animal husbandry.^3,4 Molecular typing of enterococci by techniques such as PFGE, has demonstrated a very high clonal diversity of vancomycin-resistant enterococci,^5–8 indicating that horizontal gene transfer may play an important role in the spread of the vancomycin resistance. Furthermore, the vanA gene cluster responsible for high-level resistance to vancomycin/teicoplanin is known to be located on a transposable element, Tn1546⁹ integrated into conjugative plasmids. Some of these plasmids have been shown to carry resistance to several antimicrobials^10–14 facilitating the persistence of resistance traits in the absence of specific environmental selective pressure.

The results from our previous field studies showed a high-level polyclonality of vancomycin-resistant Enterococcus faecium (VREF) populations both between farms and within the same farm in the UK.⁷ Characterization of VanA elements in previous studies has shown variation mostly due to insertion sequences inside Tn1546.^15–18 The transposon diversity present on UK farms nearly a decade after the withdrawal of avoparcin has not been well characterized. Our study aimed to investigate the dissemination of vancomycin resistance by analysing diversity inside Tn1546 from a panel of isolates collected on independent broiler farms across UK and Wales between 2002 and 2003. The study was designed to identify spread of specific Tn types to genetically different populations of E. faecium, from a situation where diversity of Tn types is found in a diverse genetic background. The transfer of vancomycin resistance and dissemination of antimicrobial resistance in enterococci associated with plasmid replicons was also investigated as part of the study.

Materials and methods

Strain selection

A total of 101 VREF previously found to be carrying vanA genes were selected.⁷ Four to six isolates were selected from faecal (n = 87) and environmental (n = 14) samples on 19 unrelated broiler farms. In four cases, farms were sampled on more than one occasion. Isolates collected from each visit were included in the panel. The total panel of isolates represented 62 different PFGE profiles. All selected isolates were tested with microbroth dilution and MICs were those according to CLSI (former NCCLS) for nitrofurantoin, penicillin, tetracycline, erythromycin, ciprofloxacin, gentamicin, streptomycin, quinupristin/dalfopristin, kanamycin, vancomycin, teicoplanin, chloramphenicol, florfenicol, bacitracin, flavomycin and salinomycin. All of the isolates were resistant to four or more different antimicrobials and represented high phenotypic diversity in antimicrobial resistance patterns.

Tn characterization

A long PCR (L-PCR) protocol, using primers that target the inverted repeats of Tn1546, as described by Palepou et al.¹⁶ was used. ClaI-digested products were compared with the use of BioNumerics software (Applied Maths, Sint-Martens-Latem, Belgium). Isolates that did not produce an amplicon were tested for vanX by PCR.¹⁹ The amplified products were digested with DdeI (Amersham Bioscience) to screen for point mutations, G to T at position 8234 of vanX, type D associated with pig-related strains.²⁰ Isolates that produced two DdeI fragments were tested for variations on the left and the right sides of Tn1546 by PCR. Primer combinations (Table 1) were P1747 to target the left side of the Tn1546 between the inverted repeat and ORF1, and P1743 to target between vanZ and the inverted repeat on the right.

Genetic bases for resistance to MLS_B and quinupristin/dalfopristin

All isolates were examined for the presence of the gene erm(B) responsible for resistance to macrolide-lincosamide-streptogramin B (MLS_B) by PCR as described previously.²¹ All isolates showing resistance to the streptogramin antimicrobial quinupristin/dalfopristin were examined for the presence of vat(D)−vat(E) genes by PCR, as described previously.²²

Transferability of resistance

A total of 45 VREF isolates from 19 different farms were selected as potential donors. Selection was based on phenotypic heterogeneity of antimicrobial resistance. Conjugation of vancomycin to the recipient E. faecium GE-1 (rifampicin and fusidic acid resistant) was performed using filter-mating, as described by Clewell et al.²³

Plasmid characterization

Replicon typing

All transconjugants were analysed by PCR for seven different plasmid replicons typical of enterococci, associated with sex pheromone production and antimicrobial resistance. As representative of the defined plasmid groups pS86, pMMB1, pRUM (erythromycin, streptomycin, chloramphenicol, streptothricin), pRE25 [chloramphenicol, macrolide erm(B), streptothricin, aminoglycosides], pCF10 [tetracycline tet(M)], pAM373 [tetracycline tet(M)], and pEF418 (streptomycin) were chosen (see Table 1).

Plasmid extraction and Southern blot hybridization

Two representatives of each of the three groups originated by replicon typing were selected. Plasmids were isolated from six of the donors and transconjugants using a NucleoBond^® Plasmid Purification Kit (Macherey-Nagel). Digoxigenin-labelled probes for vanA and erm(B) genes and repA (initiating plasmid replication) of pRE25 and pEF418 (Table 1) were prepared and labelled with DIG DNA Labelling Mix (Roche). Hybridization and colorimetric detection were performed as described by the manufacturer (Roche).

Results

Tn characterization

Analysis of the results identified five different Tn types amongst the 101 isolates. All types contained the 2.009 kb fragment that includes the vanHAX genes. As described in Table 2, Tn type T was the most prevalent (n = 32) followed by Tn type U (n = 26) and type A (n = 26). For two of the isolates, no bands were visualized for the small fragments. These isolates were typed according to the larger fragments of the profile adding two more isolates to the type T. In addition, two other isolates showed a profile that could indicate presence of two copies of different transposon types in the same isolate. These types were named as T1 to indicate relation to transposon type T. Type T1 differed from type T in the presence of an extra genetic fragment of 2.9 kb equal to the second band of the wild-type transposon.

Indistinguishable and closely related PFGE types presented the same Tn types except in four cases. On a single occasion, two isolates from the same farm with identical PFGE profiles showed two different Tn types belonging to the T and T1 types. On three other occasions, one of the isolates with the same PFGE type as another that produced a defined Tn type, failed to produce an amplicon by L-PCR, indicating differences between transposons from isolates with indistinguishable PFGE profiles.

During the field studies VREF with related PFGE profiles were isolated on two different broiler farms.⁷ These farms were spatially separated and managed by different farmers. However, the birds on farm A were sourced from a breeding house that was situated on farm B. When analysing the transposon types of these related isolates, on the first visit to farm A, wild-type transposon was the only type recovered in all isolates (n = 5). At the second visit to farm A, all isolates analysed (n = 3) were type T, which coincides with the type that was present in the isolates recovered from farm B (n = 5). The link between these farms was curtailed during the course of the study. Subsequent visits to farm A did not recover further VREF, while 5 months later, VREF type T could still be found on farm B (four out of five isolates).

DdeI digestion

Fifteen isolates did not yield an amplicon for Tn1546 by L-PCR. DdeI-digestion differentiated a group of four isolates that only generated two fragments. Additional analysis by PCR of the flanking regions to the right and to the left of Tn1546, showed amplification on the left from the inverted repeat to ORF1 with a deletion of 50 bp when compared with the prototype. In contrast to the type D already described²⁰ these isolates failed to amplify the right flanking region. This work has revealed a new variation of Tn type, which presented a G to T point mutation at position 8234 of the vanX gene, but showed a deletion between vanZ and the inverted repeat on the right side of Tn1546. We designated this Tn type type Y; a new type not defined previously.

Genetic bases for resistance to MLS_B and quinupristin/dalfopristin

Out of 101 VREF, 88 isolates (87%) were concurrently resistant to the macrolide erythromycin (MIC ≥ 16 mg/L). Of these macrolide-resistant VREF, 87.5% were erm(B)-positive by PCR.

Quinupristin/dalfopristin resistance (MIC ≥ 4 mg/L) was found in 48% of the VREF investigated. Among the quinupristin/dalfopristin-resistant isolates, the vat(D) gene was never detected, while 32 out of 48 isolates (67%) produced the vat(E) amplicon. The genetic background for the rest of the isolates (33%) was not determined. MICs for vat(E)-negative isolates ranged from 4 to 8 mg/L, clustering around the official breakpoint of 4 mg/L with 94% of isolates (15/16) having an MIC of 4 mg/L and 6% of isolates (1/16) having an MIC of 8 mg/L. MICs for the vat(E)-positive isolates ranged from 4 to 16 mg/L with 59% of the isolates (19/32) having an MIC of 4 mg/L, 13% (4/32) having an MIC of 8 mg/L, 25% (8/32) having an MIC of 16 mg/L, and 3% (1/32) having an MIC of 32 mg/L.

Transfer of vancomycin resistance

The results of the filter-mating experiments showed that 8 out of 45 donor strains (18%) were able to transfer vancomycin resistance genes. All of the successful donors presented different PFGE patterns as well as resistance traits and each donor was isolated from a different farm. Six of the donors concurrently transferred resistance to macrolides. The recipients were confirmed to be erm(B)-positive by PCR. Furthermore, co-transfer of streptomycin and bacitracin resistance was also observed for one of the transconjugants (Table 3). Transfer of resistance to quinupristin/dalfopristin was never achieved.

Plasmid characterization

PCR analysis of the plasmid replicons differentiated the eight transconjugants into three groups. The first group (isolates 85, 87 and 147) contained two replicons similar to pEF418 and pRE25. The second group (isolates 39 and 65) contained only the pRE25 replicon and the last group (isolates 97, 134 and 157) did not test positive for any of the seven replicons (Table 3). In addition, this last group included the two erm(B)-negative transconjugants.

Southern hybridization analysis of plasmids extracted from six pairs of donor/transconjugant revealed hybridization to the same high molecular weight plasmid with probes for the pRE25 replicon, vanA and erm(B) in three of the pairs (donors 39, 65 and 85). Donor 85 also co-transferred a plasmid of smaller molecular weight (4.3 kb) that hybridized with the probe for the pEF418 replicon.

Discussion

This study investigated the mechanisms for the dissemination of vanA-mediated resistance in broiler farms. Tn1546 characterization showed very low diversity of VanA elements. However, 11 isolates did not produce an amplicon by L-PCR and were confirmed to have a conserved vanX gene by DdeI-digestion. The remaining four isolates showed deletions of the inverted repeat region on the right side of the transposon, in contrast with the previously described type D, which is associated with vanA enterococci from pig faeces and hospital patients.¹⁹ These four isolates originated from dust in the environment of a single farm. We named this new Tn type as type Y. VREF were never detected from any of the faecal samples collected from this farm. Whether these VREF had persisted in the environment from previous unknown inhabitants or had evolved by process of homologous recombination between transposons of different animal species such as porcine strains and poultry strains cannot be determined.

Overall, four main transposon types were found spreading in isolates from the 19 broiler farms. By comparison to the high polyclonal distribution observed by PFGE (62 types), polymorphisms in internal regions of Tn1546 obtained from this study remained low. This is suggestive of horizontal transfer of the transposons carrying the resistance genes between genetically unrelated E. faecium.^8,24,25 PFGE has become the reference method for epidemiological investigations;^26–29 however, this may not provide effective discrimination for the study of the genetic relationship between epidemiologically related enterococci in terms of resistance to vancomycin.³⁰ The acquisition or the loss of mobile DNA elements may not be reflected by the enzyme target site used for PFGE typing, resulting in the same PFGE profile for isolates with differences inside Tn1546. This may also explain why on three occasions, two different isolates presented indistinguishable PFGE profiles but different Tn types.¹⁵ On the other hand, on the one occasion where two isolates from the same farm showed indistinguishable PFGE patterns but different Tn types, T and T1, this may indicate the presence of a second copy of the transposon.³¹ On the individual farms, closely related PFGE types presented the same Tn types. However, in the case of the two related broiler farms encountered in this study, PFGE was not sufficient to indicate epidemiological relatedness between two isolates, but together with transposon characterization has proved to be a useful combined-tool to study the dissemination of vanA within integrated livestock systems.

Transfer of vancomycin resistance by conjugation^32–35 was demonstrated for eight isolates, illustrating the presence of transferable elements and their ability to spread under the correct conditions.^36,37 Transferability did not appear to be dependent on Tn type, since representatives from all Tn types successfully conjugated. Southern blot hybridization revealed that in three isolates, vanA was transferred on a high molecular weight plasmid as suggested in other studies.^{10,24,25,38,39} Furthermore, hybridization experiments provided evidence of vanA and erm(B) being co-transferred on the same conjugative element, most likely a large plasmid containing a pRE25-like replicon. In isolate 85, transfer of a smaller plasmid with a different replicon (pEF418) was detected. Macrolide resistance encoded by erm(B) has previously been linked to large conjugative plasmid harbouring the vanA gene cluster.^10,40 Persistence of VRE in the pig industry has also been associated with co-selection by the use of tylosin.^41,42 Persistence of vanA-mediated glycopeptide resistance in the conventional broiler industry may in part be a consequence of co-selection by the use of lincomycin–spectinomycin (lincomycin hydrochloride and spectinomycin sulphate combination) as a therapeutic drug. This has become a common practice in the UK for disease control since the withdrawal of antimicrobial growth promoters (AGPs). Unfortunately, information on management practices was not available for all the farms involved in this study. However, three of the farms reported medication of the birds with this compound during the course of this study. In addition, 77 out of 101 isolates were concurrently resistant to vanA and erm(B). Larger epidemiological studies involving farms with different management practices and medication regimens would help clarify whether there is a significant association between the use of this combined compound and the persistence of VREF in farms due to co-selection.

Interestingly, plasmids with replicons pRE25 and pEF418 were found in five isolates originating from five different farms, suggesting that spread of the same plasmids has probably occurred across unrelated farms. These plasmids may have spread through pyramidal integrated livestock systems as in a ‘plasmid epidemic’. It would be of interest to conduct studies of the possible occurrence of VREF within parent flocks and transportation companies to investigate the source of contamination and to provide evidence, which can then be used to reduce the risk of dissemination between farms.

Although co-transfer of quinupristin/dalfopristin and vancomycin resistance has been previously described,¹¹ in this study it could not be demonstrated. However, the number of isolates resistant to this compound remained high (48/101). Resistance to streptogramins may in part be related to the former use of virginiamycin growth promoter. In 32 of the 48 streptogramin-resistant isolates the vat(E) gene was detected. The remaining 16 isolates all had MICs around 4 mg/L; the officially defined breakpoint, indicating lower susceptibility and not presence of an unknown genetic mechanism for streptogramin resistance. Virginiamycin was banned in the EU in 1999 but the prevalence of quinupristin/dalfopristin resistance still remains relatively high in the panel of strains examined. The use of macrolides can also co-select for streptogramin resistance since erm(B) is one of the MLS_B system genes, encoding resistance to these antimicrobials.

The results from this study suggest that vanA-mediated glycopeptide resistance is co-located on high molecular weight transferable plasmids together with erm(B) genes. Since resistance to lincosamide antimicrobials is generally mediated by the erm(B) gene,⁴³ the use of lincomycin on conventional broiler farms may indirectly be an important factor for the persistence of VREF and streptogramin-resistant enterococci. Furthermore, the ability of these organisms to exchange and disseminate genetic information has been demonstrated, and this may help explain the high level of phenotypic resistance found in this study. The relevance of plasmid transfer to the acquisition of new resistance traits and dissemination of vancomycin resistance should also be considered. Recent studies have identified clonality among enterococci isolated from patients,⁴⁴ but this study also indicated that dissemination of vancomycin resistance by horizontal transfer could play an important role. The likelihood of animal strains entering the food chain should be determined to assess the risk that these multiresistant strains may pose to the public health.

Acknowledgements

We also would like to thank Christina Svendsen at the National Food Institute, Danish Technical University (DTU), for her technical assistance. This study was funded by Defra project OD2006. We gratefully acknowledge Dr Rob Davies, Professor Tony Hart and Chris Teale for reviewing this manuscript. L. G. M. received a Med Vet Net grant supporting part of the research of the study and contributing towards her PhD.

Transparency declarations

None to declare.

References