Enterococcus faecium A120 and 17 additional fosB-carrying, vancomycin resistant E. faecium isolates were characterized in our previous paper (Qu et al., 2014). E. avium 19081 is a clinical isolate that was collected during the same time period from fecal specimens carrying fosB and vanA. Using these isolates as donor strains, fosfomycin resistant transconjugants were obtained by filter mating experiments using E. faecium BM4105RF as the recipient strain (Qu et al., 2014).

The MICs of fosfomycin against 19 VRE isolates and their respective E. faecium BM4105RF transconjugants were determined by the agar dilution method. E. faecalis ATCC 29212 was used as the control. The results were interpreted according to the Clinical and Laboratory Standards Institute (CLSI) 2015 guidelines (CLSI, 2015).

Broth mating and filter mating of E. faecium A120 were performed to get the transfer frequency of plasmid pEMA120 using E. faecium BM4105RF as the recipient strain with a donor/recipient ratio of 1:10 as previously described (Zheng et al., 2007).

Single colonies were grown at 37°C with shaking at 200 r.p.m. in 2 ml of brain heart infusion (BHI) broth. Independent overnight cultures were diluted (1:1000) and 200 μl was transferred to a 96-well plate that was incubated at 37°C. The turbidity was measured at OD₆₀₀ using a BioTEK Synergy plate reader (BioTEK, Winooski, VT, United States) every 5 min for 16 h.

A preliminary experiment using isolate A120 was carried out using a fosfomycin concentration gradient from 1/128 the MIC to the MIC. The growth of the 19 transconjugants that were exposed to 1/4 the MIC of fosfomycin was measured and compared with an antibiotic-free control. Each sample was performed in triplicate.

Plasmids were isolated from the 19 transconjugants using a QIAGEN Plasmid Midi Kit (Qiagen, Germany). The manufacturer’s instructions were followed with the inclusion of the lysozyme treatment (10 mg/mL, 60 min at 37°C) prior to the lysis step with buffer P2 [200 mM NaOH, 1% SDS (w/v)].

The restriction enzyme BsaI (New England Biolabs, United States), for which a target site is present within the fosB gene, was used to digest plasmid DNA. The hybridization probe was designed to bind within the latter half of the fosB gene after BsaI digestion (Supplementary Figure S1A) and was synthesized using the primers fosB-latter-F (5′-GTG GTA TAT GGT TAG CTT TGA ACG AAG-3′) and fosB-latter-R (5′-TGA GGT TTA GCC TCT TTA TAA TAA CTC-3′).

Plasmid DNA was first digested by BsaI at 37°C in a water bath overnight. After the digested DNA was agarose gel electrophoresed for 2 h, the DNA fragments were transferred to a positively charged nylon membrane (Millipore, United States), hybridized with a DIG-labeled fosB-latter specific probe, and then was detected with a NBT/BCIP color detection kit (Roche, Germany).

The copy number of fosB was measured via quantitative real-time PCR (qRT-PCR). The chromosomal housekeeping gene purK was chosen as an internal control. The sequences of the fosB and purK primers used are as follows:

Quantitative PCR reactions were carried out in a 10 μl reaction that contained 5 μl SYBR^® Premix Ex Taq^TM PCR kit (Takara Bio, Japan), 2 ng of genomic DNA as template, and 2 pmol of each primer. The Ct values of each sample was measured under appropriate PCR conditions (preheated at 95°C for 5 min; 40 amplification cycles at 95°C for 5 s, 54°C for 30 s and 72°C for 30 s) on an Applied Biosystems ViiA^TM 7 Dx instrument (Applied Biosystems, United States). Data were calculated based on the 2^-ΔΔC_t method.

Single-molecule, real-time sequencing (Pacific Biosciences, United States) was carried out to complete the whole genome sequencing of E. faecium pEMA120 and E. avium pEA19081 plasmids. The plasmid sequences were validated by PCR, annotated using the RAST server (Aziz et al., 2008) and were supplemented by the BLASTP program. Lastly, the circular map of the pEMA120 and pEA19081 plasmids were generated using the CGview server (Grant and Stothard, 2008). A comparison of pEMA120 and three related plasmids was made with Mauve 2.3.1 and modified by Photoshop 7.0.

Inverse PCR was performed using the primers fosBiF (5′-TGTCAGCCCCTAAAATATCTCT-3′, located within the fosB gene) and fosBiR (5′-GTTTCAAATGTACCTAAAGAACT-3′, located within the tnpA gene) (Xu et al., 2013) with genomic DNA as template. The purified PCR products were cloned into the pMD20-T Vector (Takara Bio, Japan) and were sequenced (Biosune, China).

To further characterize the previously identified fosB-carrying plasmids, the plasmids from two VRE isolates (E. faecium A120 and E. avium 19081) were completely sequenced. Plasmid pEMA120 is 79,797 bp in size with an average G + C content of 33.8%. It contains 87 open reading frames (ORFs), 47 of which encode proteins with homology to proteins with known functions, which can be classified into several functional modules: antibiotic resistance, mobile elements, plasmid stability, replication recombination and repair, and conjugal transfer (Figure 1).

In the antibiotic resistance region, there are two types of resistance determinants: fosB mediating fosfomycin resistance and vanA-type transposon causing vancomycin resistance. Two copies of fosB gene along with tnpA insert into different position of pEMA120 leading to diverse direct repeats (Figure 2). One fosB gene along with tnpA gene reversely inserts into the vanA-type Tn1546-like transposon between the vanRS and vanHAX intergenic region. Another fosB gene accompanied with tnpA gene is flanked by genes encoding a DNA topoisomerase and a RNA-directed DNA polymerase, which are required for replication, recombination and repair. No same structure was found in existing database, indicating that this is a novel insertion position of the fosB gene.

The surrounding environment of fosB on pEMA120 was compared with fosB in other species (Figure 2) Plasmid-borne fosB in Staphylococcus haemolyticus and Staphylococcus aureus share 99.8 and 99.5% nucleotide identity with fosB in the plasmid of E. faecium, respectively. As to surrounding environment, there is a rep gene downstream fosB on Staphylococcus spp. plasmid instead of a tnpA gene adjacent to fosB on E. faecium plasmids.

The conjugal transfer region of pEMA120 has some genes that are characteristic of a conjugative plasmid. The genes of ORF60 and ORF61, respectively, show similarity with 43 and 58% identity (at the amino acid level) to traB and traA of the conjugative plasmid pMG1 (Tanimoto and Ike, 2008). There is a predicted interrupted gene comprised of ORF29 and ORF31 that encodes a putative ATPase TraE, which belongs to the motor protein family of the type IV secretion system (T4SS) in enterococci (Chandran et al., 2009; Clewell et al., 2014). The traI gene (ORF38) shows 71% identity at the amino acid level with ORF34 of pHTβ encoding the putative DNA relaxase/nickase, which is needed for the initiation of DNA transfer in conjugative plasmids (Francia et al., 2004).

As regards plasmid replicon typing, ORF64 showed 57% identity at the amino acid level with Rep protein of plasmid pHTβ from E. faecium and showed less than 80% identity, at both nucleotide and amino acid level, with any plasmid replicon type according to the classification system of enterococci (Jensen et al., 2010).

To better study the evolution of pEMA120, we compared it to three similar E. faecium plasmids reported previously (Figure 3). Plasmid pMG1 (GenBank accession no. AB183714) is a gentamicin and kanamycin resistance plasmid capable of highly efficient transfer among enterococci and originally isolated from a clinical strain of E. faecium in Japan, 1998 (Ike et al., 1998). Plasmid pHTβ(GenBank accession no. AB206333) is a vancomycin resistance pMG1-like plasmid which was obtained from a hospital in the United States, 2003 (Tomita et al., 2003). Plasmid pZB18 (GenBank accession no. AB611033.1) acquired from Beijing, 2004 (Zheng et al., 2007) is also a vancomycin resistance conjugative E. faecium plasmid but without conserved traA gene or traB specific for pMG1-like plasmid. The distinctive feature of pEMA120 is the structure of the fosB connected by tnpA, which does not exist in other three plasmids. In general, the backbones of the four plasmids have high homology, which is depicted in the figure as the blue shades. In particular, plasmid pZB18 shares 99% identity and 85% query cover with pEMA120. Plasmid pMG1 and pHTβ contain highly efficient conjugative system, especially the similar oriT and transfer-related regions, which have a high degree of homology with the conjugal transfer region of pEMA120.

As shown in Figure 1, the E. avium plasmid pEA19081 was found to share 99.99% nucleotide sequence identity with the E. faecium plasmid pEMA120 after sequence comparison by BLASTN. Thus, the two plasmids are virtually identical, containing the same two copies of fosB as well as a Tn1546 transposon.

As mentioned in above section, a transfer related gene traE which might play an important role in conjugal transfer process, was disrupted by a gene encoding RNA-directed DNA polymerase. To investigate whether this interruption could affect the conjugative transfer, we compared the transfer frequency of plasmid pEMA120 with a disrupted traE and plasmid pZB18 with an intact traE. From Supplementary Table S1, the transfer frequency of pEMA120 was distinctly lower than that of pZB18, using broth mating or filter mating.

To determine whether fosB gene-containing circular intermediates are present in isolate A120, an inverse PCR was performed using previously reported primers (Xu et al., 2013). As expected, a ∼1 kb band was present after agarose gel electrophoresis, indicating the presence of circular intermediates. However, the sequencing electropherogram had imperfect signals, suggesting that this amplified 0.92-kb PCR product was probably a combination of two or more sequences (Supplementary Figure S2). Thus, TA-based cloning and sequencing was employed to obtain unique sequences. All sequences of PCR products that were ligated into pMD20-T vectors aligned similarly except for two regions. The first region was located 26–31 bp upstream of the fosB gene, showing two variants (ATTTG or TAACAT) (Supplementary Figure S2A), matching the corresponding positions of each fosB-carrying ISL3-like transposon within pEMA120. Another polymorphism region, 5-bp in length, was situated between the two IRs of ISL3-like transposon, i.e., the circle-junction region. Three kinds of sequences were observed, ATTTT, TATTT, or TTATC, as shown in Figure S2B. The complete fosB-transposon circular intermediate is 2425 or 2426 bp in length, consisting of the ISL3-like transposon and a 5-bp intervening DNA sequence (Figure 4). In addition, this fosB-transposon circular intermediate could also be acquired by inverse PCR in the other 18 VRE isolates. However, the circular intermediates could not be detected by southern hybridization (using the fosB-latter specific probe) after gel electrophoresis using the uncut or S1 nuclease digested pEMA120 plasmid (data not shown).

Our previous study suggested each of 19 VRE isolates contained one fosB-carrying plasmid by performing S1 nuclease-pulsed-field gel electrophoresis (PFGE) (Qu et al., 2014). The third-generation sequence analysis showed that both pEMA120 and pEA19081 carry two copies of the fosB gene. To determine the fosB copy number in these plasmids from transconjugants of 19 VRE isolates, a special hybridization protocol after a restriction enzyme digestion was used. The principle is depicted in Supplementary Figure S1 using pEMA120 as an example on basis of its clear plasmid sequence.

As Figure 5A shown, the DNA band pattern generated by BsaI digestion showed diverse restriction patterns in the plasmids from the transconjugants of 19 VRE isolates. The hybridization bands of isolate A120 in Figure 5B were well in accordance with the simulating graph of pEMA120 in Supplementary Figure S1, which proved the reliability of BsaI-digestion method to explore the fosB copy number in a plasmid. The isolates IB3, A165, IA110, 19081, A166, 1001, A96, A155, A158, and IA28, which were analyzed but were not shown here, generated very similar two-band signals (16 and 9.5 kb) with isolate A120. This finding suggests that these isolates bear two fosB copies per plasmid. These results were verified by PCR amplification using primers (“between vanSH” and “fosB-2th”) designed to amplify an approximately 3-kb fragment surrounding each fosB (Supplementary Table S2). It was found that all the above isolates contained the two DNA fragments with fosB.

The nylon membranes of isolates A67, A13, A9, B42, IA14, A157 and A92 showed a single band, suggesting that there was one fosB copy per plasmid.

With regards to isolate A3, four bands were observed on the hybridization membrane. BsaI-digestion electrophoresis and hybridization blotting of A3 showed the possibility that more than one fosB-carrying plasmid of similar size existed in at least some isolates. The distribution of the four copies of the fosB gene in isolate A3 warrants further investigation using other technologies, such as third-generation sequencing.

To investigate whether fosB copy number has any relevance to the fosfomycin MICs and bacterial growth rate at sub-MIC levels, the fosB copy number in every transconjugant isolate was determined by qRT-PCR. Table 1 shows the total fosB copy number in every isolate, using the housekeeping gene purK as an internal control.

The total fosB copy numbers [2^{-Δct(fosB-purK)}] reported in Table 1 range from 17.17 to 55.68 in the isolates with two-copies of fosB per plasmid (transconjugants A165, IA110, 19081, A166, 1001, A96, A155, A158, A120, and IA28). In the A3, A13, A9, A92, B42, IA14, A67, and A157 transconjugants, their total fosB copy number was less than 10, except isolate A9 (188.45).

The detailed fosfomycin MICs (without an upper limitation) of 19 VRE isolates and their transconjugants is shown in Table 1. From the susceptibility tests, the MICs of fosfomycin for these isolates and their transconjugants were extremely high, ranging from 16384 to 131,072 μg/ml (Table 1). The MICs of the transconjugants was discussed due to their purer genetic background. As shown in Supplementary Figure S3A, the MIC of fosfomycin did not correlate with the copy number of fosB (Pearson’s R², 0.1550). For example, although the fosB copy number of isolate A67 and B42 were low, their MICs of fosfomycin were higher than most of the other isolates.

From a preliminary analysis of isolate A120, fosfomycin concentrations of 1/4 or 1/2 the MIC notably affected the bacterial growth, including a prolonging of the lag phase and a decrease in growth rate and max biomass (data not shown). The growth of the 19 transconjugants following exposure to 1/4 MICs of fosfomycin was measured (Supplementary Table S3), but no correlation was observed between the copy number of fosB and growth rate, maximum optical density (OD₆₀₀) or lag time (Supplementary Figure S3).