Detection of a cfr(B) Variant in German Enterococcus faecium Clinical Isolates and the Impact on Linezolid Resistance in Enterococcus spp.

The National Reference Centre for Staphylococci and Enterococci in Germany has received an increasing number of clinical linezolid-resistant E. faecium isolates in recent years. Five isolates harbored a cfr(B) variant gene locus the product of which is capable of conferring linezolid resistance. The cfr(B)-like methyltransferase gene was also detected in Clostridium difficile. Antimicrobial susceptibility was determined for cfr(B)-positive and linezolid-resistant E. faecium isolates and two isogenic C. difficile strains. All strains were subjected to whole genome sequencing and analyzed with respect to mutations in the 23S rDNA, rplC, rplD and rplV genes and integration sites of the cfr(B) variant locus. To evaluate methyltransferase function, the cfr(B) variant of Enterococcus and Clostridium was expressed in both E. coli and Enterococcus spp. Ribosomal target site mutations were detected in E. faecium strains but absent in clostridia. Sequencing revealed 99.9% identity between cfr(B) of Enterococcus and cfr of Clostridium. The methyltransferase gene is encoded by transposon Tn6218 which was present in C. difficile Ox3196, truncated in some E. faecium and absent in C. difficile Ox3206. The latter finding explains the lack of linezolid and chloramphenicol resistance in C. difficile Ox3206 and demonstrates for the first time a direct correlation of elevated linezolid MICs in C. difficile upon cfr acquisition. Tn6218 insertion sites revealed novel target loci for integration, both within the bacterial chromosome and as an integral part of plasmids. Importantly, the very first plasmid-association of a cfr(B) variant was observed. Although we failed to measure cfr(B)-mediated resistance in transformed laboratory strains the occurrence of the multidrug resistance gene cfr on putatively highly mobile and/or extrachromosomal DNA in clinical isolates is worrisome with respect to dissemination of antibiotic resistances.


Introduction
The oxazolidinone linezolid represents an antibiotic of last resort used to treat severe infections that are caused by multidrug-resistant Gram-positive bacteria such as methicillin-resistant Staphylococcus aureus (MRSA) or vancomycin-resistant Enterococcus spp. (VRE). According to a recent Threats Report from the Centre for Disease Control and Prevention, MRSA and VRE are listed under serious threats which require public health monitoring (http://www.cdc.gov/drugresistance/threat-report-2013/). Considered as an important nosocomial pathogen VRE can cause life-threatening diseases including septicemia, endocarditis or urinary tract infections. After the approval of linezolid in 2000, emergence of resistant strains upon linezolid therapy has been reported for both VRE and vancomycin-susceptible enterococci (VSE) [1,2]. Although linezolid-resistant enterococci (LRE) are generally detected at low prevalence in Europe, the National Reference Centre (NRC) for Staphylococci and Enterococci at the Robert Koch Institute recognized an increasing number of LRE in German hospitals [3].
Linezolid targets the ribosomal machinery by inhibiting formation of the translation initiation complex, hence protein synthesis [4]. In addition to mechanisms such as mutations in the 23S rDNA or ribosomal protein genes rplC, rplD and rplV, resistance to linezolid is mediated by acquisition of the cfr RNA methyltransferase [5][6][7][8]. However, cfr not only confers insusceptibility to oxazolidinones, but also to a broad range of antibiotics collectively known as PhLOPS A (Phenicols, Lincosamides, Oxazolidinones, Pleuromutilins and Streptogramin A) [5].
Unlike staphylococci, where a combination of site-specific mutations within the rDNA and/or ribosomal proteins and the expression of cfr are jointly responsible for the development of a resistance phenotype [9,10], reports concerning cfr-mediated resistance in enterococci remain scarce. Up to now, the majority of studies described the accumulation of mutations affecting the ribosomal machinery [2,11,12]. Although the presence and expression of cfr in enterococci has been reported previously, the impact of conferring resistance to enterococci isolates is still under debate [13][14][15][16][17]. It must be noted that two isoforms of the cfr gene exist in enterococci, one of which is highly similar to cfr from staphylococci and the other one being almost identical to a cfr-like gene from Clostridium difficile [14,16,18]. The latter is now termed cfr(B) and located on the composite transposon Tn6218 [16]. Adding to the complexity of linezolid resistance, enterococci from China and Italy were shown to carry optrA, a gene encoding an ABC-transporter capable of mediating oxazolidinone and phenicol insusceptibility [17,19,20].
It has been reported that C. difficile isolates with elevated MICs harbor cfr [21]; however, it has been shown just recently that episomal expression of the very same gene confers the PhLOPS A phenotype in E. coli [22]. Nevertheless, until today it remains to be determined whether cfr as a sole mechanism is able to mediate PhLOPS A phenotypic resistances in wildtype Clostridium.
In the present study, we analyzed the first clinical cfr(B)-positive E. faecium isolates from German hospital patients. The methyltransferase gene was shown to be a close homolog of cfr (B) from American E. faecium isolates and from C. difficile. As part of a full length or truncated version of transposon Tn6218 the cfr(B) variant described herein showed novel sites for integration and, most importantly, plasmid-association. Expression of this cfr(B) variant from enterococci and of C. difficile failed to recover the PhLOPS A phenotype in E. coli and Enterococcus spp. However, analysis of a closely related strain set of C. difficile clinical isolates demonstrates for the first time that the Cfr-like methyltransferase is able to mediate oxazolidinone and phenicol resistance in the important gut commensal and nosocomial pathogen.

Strains used in this study
Five E. faecium strains isolated from German hospital patients and exhibiting linezolid resistance (Table 1) were tested positive for cfr by PCR as described previously [3,23]. Two closely related C. difficile strains were isolated in Oxford, United Kingdom in 2012 and are part of a small cluster of geographically and temporally linked cases (Kate Dingle, personal communication); however, Ox3206 lacks the mobile element cfr-Tn6218 which is present in Ox3196 (HG002389) [24].

Antimicrobial susceptibility testing
The antimicrobial susceptibility of Enterococcus spp. isolates was determined by broth microdilution (determination of minimal inhibitory concentrations, MICs) according to EUCAST and applying EUCAST MIC breakpoints for interpretation of the results (www.eucast.org); Etest 1 linezolid (LZD) (bioMérieux, Nürtingen, Germany) was used to confirm results. In addition, transformants of E. coli Top10, E. faecium 64/3 and E. faecalis JH2-2 were analysed for susceptibility to LZD, chloramphenicol (CMP) and florfenicol (FFC). Please note that according to the latest version of CLSI document VET01-A4 no clinical breakpoints are provided for FFC. In case of C. difficile Etest 1 for LZD and CMP were used to determine the MIC under anaerobic conditions. Please note that there are no MIC clinical breakpoints available for C. difficile.

DNA extraction
Enterococcus strains were grown overnight in BHI broth at 37˚C. C. difficile were routinely grown at 37˚C on Brucella agar supplemented with 5% blood, 1 mg/L vitamin K and 5 mg/L hemin. Air-sealed containers were adjusted to anaerobic conditions using AnaeroGen™ sachets (Thermo Fisher Scientific, Germany). DNA was extracted from cell material obtained from agar plates (Clostridium) or from overnight cultures (Enterococcus) using the DNeasy blood and tissue kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions.

Molecular characterization of linezolid resistance and whole genome sequencing
The ratio of 23S rDNA wildtype and mutated alleles was determined as published previously [25]. In brief, amplified PCR products were digested with NheI which discriminates between wildtype and mutated alleles (position G2576T) and fragments were separated thereafter on a microfluidic chip using an Agilent Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA). Densitometric analysis allows a qualitative assessment of wildtype and mutated alleles as well as quantification of both allele types per isolate. To analyse possible linezolid resistance-mediating mutations rplC, rplD, rplV were examined from whole genome sequence data. To this end, 1 ng of extracted DNA was subjected to library preparation using the Nextera XT Library Prep Kit according to the manufacturers' instructions (Illumina). Whole genome sequencing (WGS) was performed on a MiSeq benchtop instrument in paired-end with a maximum of 300 bp read length (MiSeq Reagent Kit v3, 600 cycles, Illumina). For enterococci, obtained sequences were compared to genomic data from E. faecium DO/TX16 (NC_017960), whereas C. difficile R20291 (FN545816) served as a reference for strains Ox3196 and Ox3206.   Genomic reads from C. difficile Ox3196 and Ox3206 were obtained by next generation sequencing (see above) and assembled de novo using the open access assembly pipeline a5-miseq [26]. Subsequently, obtained contigs were concatenated and annotated by RAST [27]. Comparison of whole genome content was carried out using the BLAST Ring Image Generator (BRIG) program [28]. Determination of loci which were absent in C. difficile Ox3206 was performed by mapping of reads to the concatenated and annotated sequence from C. difficile Ox3196 using the Geneious software (version 9.1.3).

Phylogenetic analysis of cfr(B)-positive E. faecium
To determine the phylogeny of the 5 cfr(B)-positive E. faecium, Single Nucleotide Polymorphisms (SNPs) were identified by using a pipeline based on BWA v0.7.12-r1039 [BWA-SW; [29]] for mapping to reference strain E. faecium 64/3 (CP012522) and VarScan v2.3 [30] for variant calling. SNPs located within a 300 bp distance, or less, from each other were excluded as they might represent variants introduced by recent recombination events. A maximum likelihood tree was inferred from a total of 874 SNP positions by utilizing the PhyML algorithm of the seaview program and 1000 iterations for branch validation.

Determination of Tn6218-cfr(B) insertion sites in E. faecium
To differentiate between plasmid and chromosomal insertion of Tn6218 in E. faecium isolates, bacterial genomic DNA was treated with S1 nuclease in order to linearize plasmids which were separated thereafter by PFGE prior to Southern hybridization using a digoxigenin-labelled cfrspecific probe (Roche Biochemicals, Mannheim, Germany) [31]. The cfr-specific probe was generated using primers as published previously [23] and E. faecium UW10882 whole genomic DNA as a template. For further analyses, de novo assembly of reads from Illumina WGS was carried out as described above. In order to determine conserved blocks between isolates, obtained cfr(B) variant-containing contigs were aligned thereafter using the Geneious software. For elucidation of putative Tn6218 insertion sites, transposon flanking regions were extracted and BLASTn search was performed. Insertion sites were verified by PCR using oligonucleotides as stated in S1 Table in a PCR reaction with 12.5 μl of 2 x DreamTaq Green PCR Mastermix (Thermo Fisher Scientific, Germany), 0.1 μM of each primer and 0.5 μl of isolated DNA in a total reaction volume of 25 μl. Thermocycler conditions were set to 2 min at 94˚C for initial denaturation followed by 30 cycles of 30 sec at 94˚C, 30 sec at 55˚C (S1 Table) and 1-2 min at 72˚C. A final extension was carried out at 72˚C for 4 min.

Cloning of cfr(B) variants-expressing plasmids
The cfr(B) variant gene from E. faecium UW10882 and C. difficile Ox3196 was amplified by PCR using Phusion polymerase (Thermo Fisher Scientific, Germany), oligonucleotides cfr_us_BamHI_fw and cfr_ds_SalI_rv3 (S1 Table) and a protocol as follows: 98˚C for 30 sec for initial denaturation, 30 cycles of 10 sec at 98˚C, 30 sec at 54˚C-64.5˚C and 60 sec at 72˚C, and a final extension for 5 min at 72˚C. The primers were designed to bind 220 bp upstream of the start of protein translation and terminate 127 bp downstream of the stop codon. PCR products were digested using BamHI and SalI and ligated to appropriately cut vector pEF-25 before transformation of chemically competent E. coli Top10. Positive transformants were selected on LB agar plates containing 50 mg/L spectinomycin (SPC) and inserts were verified by Sanger sequencing. Designated plasmids pWJB001 (pEF-25 + cfr(B) variant Enterococcus ) and pWJB002 (pEF-25 + cfr Clostridium ) were used to electroporate laboratory strains E. faecium 64/3 and E. faecalis JH2-2 and transformants were selected on BHI agar supplemented with 200 mg/L SPC [adapted from [32]].

Expression analyses of the cfr(B) variants in Enterococcus spp.
Two independent transformants of E. faecium 64/3 (pWJB001 or pWJB002) and E. faecalis JH2-2 (pWJB001 or pWJB002) were analyzed with respect to antibiotic insusceptibility. Gene expression of the cfr(B) variant targets was evaluated by total RNA extraction and reverse transcription as described elsewhere [33]. Quantification of cDNA was carried out using the DyNAmo ColorFlash SYBR Green qPCR kit (Thermo Fisher Scientific, Germany) in a CFX96 qPCR instrument (Bio-Rad, Germany). Primers cfr_UW10882_qPCR_fw and cfr_UW10882_qPCR_rv (S1 Table) were used at a final concentration of 0.1 μM in a 20 μl reaction containing 5 μl of 1:10 diluted cDNA for target amplification or 5 μl 1:1000 diluted cDNA for amplifying the 16S reference gene [33]. Thermocycler conditions were set to: 2 min at 95˚C, 45 cycles of 10 sec at 95˚C and 30 sec at 60˚C and a melting curve was determined in the range from 60-98˚C.

Filter-mating experiments
Transfer capabilities of transposon Tn6218 was determined by filter-mating experiments using E. faecium 64/3, E. faecalis OG1RF or E. faecalis JH2-2RF as recipient strains and as published previously [34]. The following isolates were used as donors: E. faecium strains UW10882, UW11590, UW11733, UW11858, UW12727 and C. difficile Ox3196. The bacterial mixture was placed on a nitrocellulose filter on BHI agar plates and in case of Clostridium under anaerobic conditions to allow overnight transfer of cfr between the two species. Putative transconjugants of filter-mating experiments were then selected on BHI agar plates containing i) 10 mg/L FFC, 30 mg/L rifampicin (RIF) and 20 mg/L fusidic acid (FUS); ii) 5 mg/L CMP, 30 mg/L RIF and 20 mg/L FUS or iii) plates containing 30 mg/L RIF and 20 mg/L FUS only. Further, putative transconjugants were screened for acquisition of cfr by PCR.

Nucleotide accession numbers
Raw reads are available from the SRA under accession number SRP078305.

Description of bacterial isolates
Five clinical LRE isolates, collected between 2013 and 2015 from German hospital patients, were tested positive in cfr-specific PCR screening experiments [3]. The strains exhibited a multidrug-resistant phenotype with varying levels of MICs to linezolid ranging from 6 mg/L to 48 mg/L (Table 1). Three isolates (UW10882, UW11590, UW12712) were of sequence type (ST) 117, but originated from two distinct geographical regions (Table 1). Further, 4/5 strains were positive for either vanA or vanB whereas UW10882 did not harbor a glycopeptide resistance gene ( Table 1). Investigation of the cfr locus (outlined below) revealed a highly similar genomic structure to a cfr gene located on transposon Tn6218 from C. difficile. Thus, we further analyzed the cfr-positive isolate C. difficile Ox3196 (ST37) (HG002389) and compared it to C. difficile Ox3206 (ST37) a close but cfr-negative relative of Ox3196. Phenotypic characterization revealed an 8-fold difference in linezolid MIC between the cfr-positive and the cfr-negative C. difficile isolate (Table 1). Clostridia were also analyzed for resistance to chloramphenicol, thereby showing MICs of 24 mg/L for Ox3196 compared to 2 mg/L for Ox3206, respectively.

Analyses of linezolid resistance-associated DNA mutations and of the cfr (B) variant methyltransferase gene
Genotypic characteristics likely associated with linezolid resistance were determined for each of the five Enterococcus and the two Clostridium isolates. First, mutations in the 23S rDNA were analyzed by NheI restriction. Different ratios of wildtype to mutated alleles were found for E. faecium UW10882 and UW12712 (Table 1). No respective mutation of the 23S rDNA alleles was detected in C. difficile or the remaining E. faecium. Second, WGS was carried out to extract genomic information regarding mutations in loci encoding for ribosomal target proteins. This yielded two mutations within rplD (L4) for E. faecium UW11590 and UW11858 ( Table 1). A triplet nucleotide insertion ( 211 GGT 213 ) and thus addition of amino acid glycine at the respective position was detected in both cases. Interestingly, Enterococcus isolates harboring 23S rDNA mutations showed no additional alterations in genes encoding for the three ribosomal proteins investigated (Table 1). In case of C. difficile no differences were detected between the two isolates. However, synonymous mutations in rplC were found when compared to the reference genome C. difficile R20291 (not shown). Moreover, the entire set of WGS derived reads was mapped to the recently identified oxzolidinone and phenicol resistance gene optrA [19]; however, none of the strains carried optrA (Table 1). In general, no apparent correlation was observed between the occurrence of certain nucleotide mutations and the level of linezolid resistance.
The cfr genes from the investigated E. faecium and C. difficile isolates were extracted from WGS data and the predicted protein sequences were aligned thereafter to previously published Cfr from enterococci and staphylococci [16,35]. Notably, Cfr from E. faecium UW12712 showed 99.7% identity to the recently identified Cfr(B) from E. faecium of human origin [16], hence differing in a single amino acid at position 314 (K314E) only. Cfr of the remaining Enterococcus exhibited 99.4% identity to Cfr(B) due to an additional mutation at position 247 which results in an arginine by lysine substitution (R247K). On the basis of homology, Cfr(B)-like sequences of the five E. faecium isolates were consequently termed Cfr (B) variants.
The Cfr(B) variants were 100% identical between E. faecium UW10882, UW11590 and UW11733. Except for Cfr(B) from E. faecium UW12712, Cfr(B) variants from the clinical E. faecium isolates showed 99.7% identity to Cfr from C. difficile with a single isoleucine to valine (I294V) substitution. Likewise, E. faecium UW12712 displayed this very same mutation, but additionally carried the already mentioned lysine by arginine substitution at position 247 (K247R) compared to Cfr from C. difficile. Similar to Cfr(B), Cfr(B) variants of the here described E. faecium strains were remarkably different to Cfr from staphylococci (74.1% for E. faecium UW12712 and 78% for the other four E. faecium isolates).

Cfr-dependent linezolid resistance in C. difficile
NheI restriction analysis and WGS comparisons of C. difficile Ox3196 (cfr-Tn6218-positive) and C. difficile Ox3206 (cfr-Tn6218-negative) revealed no difference in linezolid-associated genomic mutations (Table 1). It has recently been shown that ectopic expression of cfr from C. difficile in E. coli confers resistance to the recipient bacteria [22]; however, cfr-dependent insusceptibility to linezolid in C. difficile is still under debate [21,22]. In order to verify cfr-mediated linezolid and chloramphenicol resistance in C. difficile (see above) and to exclude the lack of genomic content other than Tn6218 in strain Ox3206 as the source for reduced susceptibility (Table 1) whole genome comparison of both isolates was carried out utilizing the BLAST Ring Image Generator (BRIG) program. As illustrated in Fig 1 only Tn6218 and two Tn916-like elements were absent in C. difficile Ox3206 (Fig 1).
Subsequent RAST annotation and detailed examination of the genomic regions encoded by the two Tn916-like elements produced open reading frames for a Tn916 transcriptional regulator, a putative transposon integrase (only one Tn916-like element), multiple hypothetical protein and two dihydrofolate reductase genes (one Tn916-like element only) (not shown). Thus, it is feasible to attribute elevated linezolid MICs of C. difficile Ox3196 to the presence of the cfrcontaining transposon Tn6218.

Determination of the genetic environment of the cfr(B) variant gene locus of clinical E. faecium isolates
It has been reported that cfr from C. difficile and cfr(B) from E. faecium are encoded by transposon Tn6218 and Tn6218-like elements, respectively [16,24]. In order to disclose the genetic environment of the cfr(B) variants from the German clinical E. faecium isolates, next generation sequencing was carried out and cfr(B)-containing contigs were examined after de novo assembly in a way that Tn6218-like structures were extracted and compared to Tn6218 from C. difficile (HG002389) and to cfr(B) and adjacent regions from E. faecium (KR10408) (Fig 2).
E. faecium UW10882, UW11590 and UW11733 harbor a Tn6218-like element highly similar to Tn6218 from C. difficile (HG002389) (Fig 2). Nucleotide alignment revealed an identical and conserved block between the four strains of 6,900 bp and spanning the methyltransferase locus. However, ORFs located upstream of the cfr(B) variant and encoding an integrase, excisionase and putative topoisomerase were less than 97% conserved between Enterococcus and Clostridium (not shown). Interestingly, E. faecium UW11590 and UW11733 displayed an identical stretch of 9,733 bp including the integrase locus that differed when compared to UW10882 (schematic representation in Fig 3). In contrast, examination of regions adjacent to the cfr(B) variant of E. faecium UW11858 and UW12712 indicated that those strains only harbor a truncated version of Tn6218 (Fig 2). The cfr(B) variant locus of E. faecium UW11858 displayed a conserved region of 4,330 bp when compared to Tn6218 from C. difficile and UW12712 exhibited even less identity to the respective mobile element (in total 1,788 bp; Fig  2). It is worth noting that the genetic environment of cfr(B) variants of E. faecium isolates described in this study was strikingly different to the cfr(B) locus of the American E. faecium isolate 448-18961R (KR610408) (Fig 2).

Phylogenetic relationship of German cfr(B)-positive E. faecium isolates
Strain characteristics with respect to sequence type, phenotypic profile, genotype of the vancomycin resistance locus and region of isolation suggests a possible link between isolates UW11590 and UW12712 (Table 1), although the latter only carries a truncated version of the Tn6218-like element (Fig 2). Hence, mapping of sequencing reads to reference E. faecium 64/3 was performed to infer the genetic relationship of the five E. faecium strains. As a result, 874 SNPs served as the basis for Maximum Likelihood analysis by the PhyML algorithm. All the isolates were only distantly related to each other (Fig 3).
Further, a distance matrix extracted from Neighbor-Joining analyses in Geneious demonstrated that E. faecium UW11590 and UW12712 were separated by 56 SNPs, thus excluding an immediate close relatedness. In summary, WGS analyses and epidemiological and genotypic data of the strains as well as their resistance profiles (Table 1)

Elucidation of Tn6218-like insertion sites
Tn6218 of C. difficile was reported to be integrated into the bacterial chromosome. Hence, Southern hybridization experiments of whole plasmid content after S1-nuclease treatment and PFGE and by using a cfr-specific probe was performed for the cfr(B) variant-expressing E. faecium isolates. As depicted in Fig 4, plasmid-association is suggested for E. faecium UW10882 and UW12712 with an estimated plasmid size of approximately 200 kb and 300 kb, respectively (Fig 4A and 4B). It is worth mentioning that hybridization experiments of phenol/chloroform extracted whole plasmid content verified results obtained by S1-PFGE (not shown). For detailed disclosure of insertion sites, sequences of regions adjacent to the mobile element were extracted from cfr(B)-carrying contigs and BLASTn search was performed to reveal respective target genes or intergenic regions. Table 2 summarizes putative insertion sites which were verified by PCR as indicated.
However, insertion sites were questionable in some cases, as for instance in silico analysis of up-and downstream sequences of E. faecium UW10882 Tn6218 yielded two different target regions which, according to reference plasmid pDO3 from E. faecium DO/TX16 (CP003586), are located >95 kbp apart from each other. Nevertheless, both loci showed positive PCR results  Fig 4).
Also, we failed to reveal the integration site of the truncated version of Tn6218 (Fig 2) from E. faecium UW11858 (Table 2). Adjacent regions yielded no hits with BLASTn or showed similarity to conjugational genes from C. perfringens pJIR4150 plasmid sequences or, to a lesser extent of identity (78% -83%), to hypothetical proteins of Tn5386 from E. faecium D344R (not shown). This suggests an environment linked to mobile genetic elements of various origins. It remains elusive whether the ancestor of UW11858, eventually having received the cfr(B) variant, already harbored a Tn5386-like element or related MGEs.
Adding to the complexity, target sites for cfr(B)-Tn6218 insertion in E. faecium UW12712 remain speculative as BLASTn of the entire 38,570 bp cfr-contig revealed homologous regions which could either be allocated to transposon and genomic content or plasmid sequences (not shown). For instance, about half of the contig showed 93% identity to parts of a conjugation module and transposase sequences of C. perfringens plasmid pJIR4150; however, reconstruction of a putative novel plasmid was not achieved which is most likely due to the size and highly plastic nature of enterococcal extrachromosomal DNA. Nevertheless, plasmid-association as proposed in silico is in accordance with results obtained by Southern hybridization which demonstrated plasmid integration of the cfr(B) variant locus for UW12712 (Fig 4). Integrity of the contig was verified by PCR bridging the truncated Tn6218-like fragment and flanking regions ( Table 2). In order to exclude that mutations in linezolid resistance-associated genes (see above) are the sole source for linezolid insusceptibility in cfr(B)-positive E. faecium isolates, the cfr(B) variant gene locus of both E. faecium UW10882 and of C. difficile Ox3196 was cloned and expressed in E. faecium 64/3 and E. faecalis JH2-2, respectively. Acquisition of the cfr(B) variant was verified by colony PCR (not shown) and cfr-mediated resistance to various antibiotics was determined by broth microdilution (S2 Table). However, introduction of the cfr(B)variant-containing vectors did not confer resistance to the transformants or elicited elevated MICs to the antibiotics tested (S2 Table). Quantitative PCR was carried out from reverse transcribed RNA to exclude insufficient transcription as causative for the lack of a resistance phenotype. Thereby, cfr(B) variant transcripts were detected at high levels in all strains analyzed (not shown).

Conjugational transfer of E. faecium and C. difficile Tn6218
Whole genome sequencing revealed the presence of putatively intact Tn6218 and Tn6218-like structures in C. difficile Ox3196 and E. faecium UW10882, UW11590 and UW11733, respectively (Fig 2). Also and according to results from Southern hybridization utilizing a cfr-specific probe, E. faecium UW10882 and UW12712 seem to carry the cfr(B) variant on a plasmid (Fig  4). In order to examine transfer capabilities and thus spread of the resistance cassette, filtermating experiments were carried out using E. faecium 64/3 and E. faecalis OG1RF or JH2-2RF as susceptible recipient strains. Conjugation between C. difficile and enterococci was investigated under anaerobic conditions. The antibiotics chloramphenicol or florfenicol were used as selective markers; however, putative transconjugants were also examined from nonselective plates in case transfer and expression of the cfr(B) variant fails to mediate a resistance phenotype as seen in previous experiments (see above). Unfortunately, screening of putative transconjugants by cfr-specific PCR yielded no positive candidates after repeated filter-mating experiments and multiple conditions tested.

Discussion
Determination of resistance development to antibiotics of last resort, such as resistance to the oxazolidinone linezolid, is of global importance with respect to remaining treatment options for severe infections caused by multidrug-resistant bacteria. As for VRE, linezolid therapy represents a valuable alternative since the general prevalence of LRE has remained constant 1% for the last few years (http://www.ars.rki.de) [36]. We herein describe the detailed investigation of the first cfr(B) variant-containing E. faecium strains isolated from German hospital patients [3]. Cfr and recently cfr(B) have already been shown to be distributed across various bacterial species [16,22,37] and outbreaks involving cfr-encoding Gram-positive bacteria haven been reported [38]. However, the impact of cfr on linezolid insusceptibility in E. faecium is still questionable, as resistance development is based on multiple mechanisms including mutations affecting the ribosomal machinery [8].  (Table 2) [17,19]. As mentioned above, Hansen and Vester confirmed that cfr from C. difficile is able to confer resistance to E. coli transformants [22]; however, they did not pursue to examine cfr-mediated resistance in the parental Clostridium. Also, a previous study could not clearly link the presence of cfr to elevated levels of linezolid MICs in C. difficile [21]. By analyzing a closely related set of C. difficile isolates differing only in the presence or absence of cfr-encoding Tn6218 we demonstrated for the first time that upon acquisition of the transposon C. difficile exhibits elevated MICs to linezolid (Table 1) and chloramphenicol (see above). Importantly, linezolid is not prescribed for the treatment of C. difficile infections. Nevertheless, clostridia are part of the human gut microbiota which represents a reservoir for active and mobile resistance determinants [39]. Moreover, C. difficile are known to harbor various conjugative and non-conjugative transposable elements, some of which encode for chloramphenicol or tetracycline resistance determinants [40]. Hence, antimicrobial selective pressure imposes a risk to clinical settings and raised the question whether linezolid resistance encoded by the putatively mobile genetic element (MGE) Tn6218 is transferable between gut commensal organisms. Successful transfer of MGEs other than Tn6218 has frequently been demonstrated to occur between C. difficile and Enterococcus spp. [41][42][43]. Three of the five E. faecium strains analyzed in this study harbor a Tn6218-like transposon highly similar to the one from C. difficile and thus could potentially disseminate the resistance to receptive bacteria. Also, isolates UW10882 and UW12712 carry the cfr(B) variant on plasmids (Fig 4) which were reported to easily facilitate transfer of linezolid resistance [18,44,45]. Also, plasmid-encoded cfr was shown to transfer frequently between enterococci or staphylococci [15,18,37,45] and in some cases even within bacterial phyla [35]. A series of filter-mating experiments using all five clinical cfr(B)-positive E. faecium isolates and C. difficile Ox3196 as donors were conducted but failed to transfer the resistance to laboratory strains of E. faecium or E. faecalis, respectively. The most obvious explanation is represented by the inability of the cfr(B) variant to confer elevated MICs to florfenicol (see above) which was used as a selective marker in filter-mating experiments. Further, Tn6218 belongs to the family of non-conjugative transposons and thus lacks a functional conjugation machinery. Mobilization of MGEs, though possible [46], would require specific host determinants which might be absent in our donor strains tested In accordance, some studies also report about ineffective transfer from Enterococcus spp. to staphylococci or vice versa [18,47]. Thus, the questions of whether or not the cfr(B) variant could to be disseminated amongst enterococci by conjugation requires further investigation.
In general, cfr is an integral part of multiple MGEs incorporated into plasmids or the chromosomal backbone [37]. The various Tn6218-like elements of the German E. faecium strains revealed target sites for transposon integration which were entirely different to the ones reported for cfr and cfr(B) from enterococci so far [16,37]. Phylogenetic investigations did not indicate clonal relatedness of the five E. faecium isolates and thus suggest an independent acquisition of the Tn6218-like and cfr(B) variant-containing element. In addition, chromosomal rearrangements preceding or following the uptake of the resistance determinant are highly likely for isolate UW10882, as cfr-flanking regions yielded two different insertion sites which, according to reference pDO3, would be located far apart from each other. Last but not least, the two truncated variants of the Tn6218-like element investigated herein document the modular structure of such elements [24] and add to the pool of newly composed cfr-containing MGEs.
In conclusion, we could demonstrate for the first time that cfr is able to mediate linezolid resistance in a clinical C. difficile isolate. Most importantly, this is the first report of cfr(B) in German E. faecium strains and the first mentioning of plasmid-localization of a cfr(B) variant in enterococci of clinical origin. As a consequence, surveillance of cfr-positive strains and detailed investigation of the genetic requirements for resistance acquisition and expression is essential to prevent emergence and spread of untreatable multidrug-resistant pathogens.
Supporting Information S1