Horizontal transfer of vanA between probiotic Enterococcus faecium and Enterococcus faecalis in fermented soybean meal and in digestive tract of growing pigs

Background The aim of this study was to investigate the intergeneric transfer of vancomycin resistance gene vanA between probiotic enterococci in the fermentation progress of soybean meal and in the digestive tract of growing pigs. One vanA genotype vancomycin resistant E. faecium strain, Efm4, and one chloramphenicol-resistant E. faecalis strain, Efs2, were isolated from twenty-nine probiotic basis feed material / additive samples. For in vitro conjugation, Efm4 and Efs2 were used as starter to ferment soybean meal. For in vivo conjugation, thirty growing pigs were randomly assigned to five groups (n = 6), treated with a basic diet, or supplemented with 10% fermented soybean meal, 1% Efm4, 5% Efs2 or a combination of 1% Efm4 + 5% Efs2 for 7 d, respectively. Fecal samples of pigs in each group were collected daily for the isolation and dynamic analysis of Efm4, Efs2 and transconjugants. The sequence types (STs) of Efm4, Efs2 and transconjugants were analyzed by multilocus sequence typing (MLST). The vanA harboring plasmid in Efm4 and transconjugants was analyzed by S1-pulsed field gel electrophoresis (PFGE) and further verified by multiple alignments. Results The results showed that, in FSBM, transconjugants were detected 1 h after the fermentation, with a conjugation frequency of ~ 10− 3 transconjugants / recipient. Transconjugants proliferated with Efm4 and Efs2 in the first 8 h and maintained steadily for 10 d till the end of the experiment. Additionally, in vivo experiment showed that transcojugants were recovered in one of six pigs in both FSBM and Efm4 + Efs2 groups, with conjugation frequency of ~ 10− 5 and ~ 10− 4, respectively. MLST revealed the ST of Efm4, Efs2 and transconjugants was ST1014, ST69 and ST69, respectively. S1-PFGE confirmed the existence of the vanA-harboring, 142,988-bp plasmid, which was also a multi-drug resistant plasmid containing Tn1546-like transposon. Conclusions The findings revealed the potential safety hazard existing in the commercial probiotic enterococci in China, because the horizontal transfer from farm to fork could potentially pose a safety risk to the public. Electronic supplementary material The online version of this article (10.1186/s40104-019-0341-x) contains supplementary material, which is available to authorized users.


Background
Enterococci species are widely distributed in the environment and enable to colonize in the gastrointestinal tract in humans and most animals [1]. For decades, some enterococci strains like Enterococcus faecium and Enterococcus faecalis are used as probiotics in human and farm animals as a kind of lactic acid bacteria [2]. Previous studies have proven that enterococci can maintain a balance among the intestinal flora [3], promote the absorption of nutrients [4][5][6] and improve the host immunity [7][8][9]. In spite of those probiotic properties, however, enterococci have also been known as one of the multiresistant pathogen, ranking among the leading causes of hospital-acquired infections worldwide [10][11][12].
Vancomycin therapy is one of the most efficient treatments for enterococci infections. While the emergence and rapid spread of vancomycin-resistant enterococci (VRE) represents a particular challenge [13,14], as there are few remaining therapeutic options for infections caused by VRE [15,16]. Glycopeptide resistance genes van is proven to be the major vancomycin resistant mechanism known in the enterococci. Within all the nine van gene types [17], vanA is the most frequently indentified one among enterococci isolates from clinical settings [18,19]. The ability of enterococci to acquire and exchange plasmids and mobile genetic elements that carry antimicrobial resistance gens (like vanA) has contributed to their role as multiresistant pathogens within both human and animals.
The spread of VRE as well as vanA is not restricted only clinically but they appear also in the community and natural environment as well as animal sectors. For instance, the emergence of VRE in food animal in Europe is proven to have a high association with the use of avoparcin, a glycopeptide antibiotic as growth promoting agent in animal feed [20]. Avoparcin was never approved as a feed additive in China, and worldwide reports on VRE in the livestock industry is limited.
Recently, growing attention has been focused on the worrisome situation of VRE in hospitals or community care centers, which might be traced back to the consumed food, such as yogurt and pork, and eventually to the diet that farm animals intake [21]. Therefore, the aim of this study was to determine the VRE strains in commercial available probiotic enterococci in Beijing, China. Moreover, to assess the potential security issue of these bacteria as an accumulator of antibiotic resistance genes (ARGs), horizontal dissemination of vanA between E. faecium and E. faecalis in fermented soybean meal (FSBM) and the digestive tract of growing pigs were also analyzed.

Bacterial isolates
Enterococci based FSBM and silage corn samples, as well as microbial feed additive samples were purchased in swine farms and local husbandry and veterinary stations in Beijing, China, during June to December, 2016. For FSBM and silage corn samples, enterococci enrichment was performed by adding 25 g sample into 225 mL of Buffered Peptone Water (Beijing Land Bridge, China), vortexed, and then incubated at 37°C for 24 h. Then 1 mL of the enrichment was added into 9 mL Bile Esculin Azide (BEA; Beijing Land Bridge, China) broth and then incubated 37°C for 24 h. The changing color of BEA broth from transparent dark brown to opaque black was a sign of the existence of enterococci. A loop of enrichment was streaked on BEA agar plate, incubated 37°C for 24 h. For feed additive samples, 1 g or 1 mL of the sample was dissolved in 9 mL normal saline, then a loop of the solution was streaked on BEA agar plate, incubated 37°C for 24 h.
The identification of suspected strains were performed by VITEK 2 Compact Automatic Bacterial Identification and Drug Sensitivity Analysis System (bioMerieux, France) using specific Gram-positive bacteria identification cards (bioMerieux, France), followed by further confirmation of 16S rDNA-based polymerase chain reaction (PCR) screen using universal primers 27-F and 1,492-R, and specific primers for E. faecium and E. faecalis (Additional file 1: Table S1).

Antimicrobial susceptibility tests
Isolates confirmed as E. faecium or E. faecalis were subjected to antimicrobial susceptibility testing to determine their resistance to twenty common antibiotics of nine categories (Additional file 1: Table S2) in accordance with the Clinical and Laboratory Standards Institute (CLSI) [22]. Each isolate was inoculated into 1 mL of sterilized normal saline by picking 3 to 5 colonies using cotton swab from an overnight culture on BHI agar to visually match a McFarland turbidity standard of 0.5. The bacteria liquid was evenly coated onto the surface of un-supplemented Mueller-Hinton (Oxoid, UK) agar plates using cotton swab. Each plate was pasted with five different antimicrobial disks (Beijing Tiantan, China), and then incubated at 37°C for 24 h. The diameter of inhibition zones was measured to the nearest millimeter and interpreted according to the CLSI standards and previous studies [22,23]. The E. faecalis reference strain ATCC 29212 was used as the quality control.

Choice of donor and recipient strains
The only VRE strain, E. faecium strain, Efm4, isolated from a FSBM sample, resistant to vancomycin but susceptible to chloramphenicol, was chosen as donor strain. While another E. faecalis strain, Efs2, isolated from a microbial feed additive sample, with opposite vancomycin and chloramphenicol resistant phenotype to Efm4, was chosen as recipient strain. The genomic DNA of Efm4 and Efs2 were extracted using Wizard® Genomic DNA Purification Kit (Promega, USA) according to manufacturer's protocol and further evaluated by PCR for the detection of van and chl genes. (Additional file 1: Table S1).
Amplified PCR products were purified using Wizard® SV Gel and PCR Clean-Up System (Promega, USA). Cleaned fragments were submitted to Sangon Biotech (Beijing, China) and sequenced from both the forward and reverse stands. The sequence was then analyzed using BLAST.
Freeze-dried powder of Efm4 and Efs2 were produced by National Feed Engineering Technology Research Center (Beijing, China), with initial bacteria counts1 0 10 CFU/g, respectively, then stored at − 20°C. Before use, the bacterial activity would be detected by measuring bacterial colony-forming ability after incubation for 24 h at 37°C on BHI agar plates.

In vitro conjugation: soybean meal fermentation and sample collection
Fresh bacteria liquid cultures of Efm4 and Efs2 were produced by overnight incubation in BHI broth at 37°C, respectively, and sub-cultured before use. The initial concentration was adjusted to~10 8 CFU/mL for both Efm4 and Efs2. FSBM was conducted in aseptic fermentation bags (5 L, provided by Ministry of Agricultural Feed Industry Centre, Beijing, China) using the following ingredients: sterilized soybean meal (SBM), 500 g; glucose, 20 g; sterilized ddH 2 O, 435 mL; yeast, 5 g; bacteria liquid, 50 mL, with the volume ratio was 1:5 (v/v, Efm4 / Efs2) [24,25]. Two control groups with the same ingredients supplemented with 50 mL Efm4 liquid or 50 mL Efs2 liquid, respectively, were set.
All fermentation ingredients except Efm4 were pre-confirmed to be negative for vanA. BHI agar plates containing either 128 mg/L vancomycin or 64 mg/L chloramphenicol were used for detection of Efm4 and Efs2, respectively. BHI agar plates supplemented with vancomycin and chloramphenicol were used for the detection of transconjugants. Five replicates were maintained for each treatment. Efm4, Efs2 and transconjugants were counted through serial plating on selective BHI agar plates at 0, 1, 2, 4, 6, 8, 12, and 24 h, and then daily from day 2 to day 10.
On each selective plate, five clones were randomly selected, cultured, identified by species assessment as described above. Putative transconjugants obtained on double resistant selective plates were further confirmed by PCR amplification on the detection of van and chl genes [26,27].
The conjugant frequency was calculated using the equation described by Liu  In vivo conjugation: animals, feeding and sample collection All procedures used in these experiments were conducted in accordance with the Chinese Guidelines for Animal Welfare and were approved by the China Agricultural University Institutional Animal Care and Use Committee (Beijing, China) [29]. The pig cages, pigpens and feeding appliances were pre-disinfected, then confirmed the absence of enterococci as well as van and chl genes using methods described previously [30]. A total of thirty barrows (Duroc × Landrace × Yorkshire) at the age of 60 d, with initial body weight of 15.38 ± 2.8 kg were individually housed and provided with a non-medicated corn-soybean meal basal diet and sterilized water ad libitum for 10 d. Fresh fecal samples were collected daily to confirm the absence VRE and vanA. After acclimatization, pigs were randomly assigned to one control group (n = 6), fed with the same basal diet previously described; and four treatment groups (n = 6), fed with basal diet supplemented with 10% FSBM, 1% Efm4, 5% Efs2 or a combination of 1% Efm4 + 5% Efs2 for 7 d, respectively.
The estimation of FSBM consumption was about 5 kg/ d, so 5 bags of FSBM were performed daily, from − 1 d to 6 d of the experiment as described above. The fermentation process lasted for 2 d. At this time point, the bacteria concentrate of Efm4 and Efs2 was both~10 9 CFU/g FSBM. Then 5 kg fresh FSBM were uniformly dispersed into 45 kg basal diet to feed FSBM group barrows. Meanwhile, the initial bacteria concentrate of Efm4 in Efm4 group and Efm4 + Efs2 group was~1.0 × 10 8 CFU/g feed, and the initial bacteria concentrate of Efs2 in Efs2 group and Efm4 + Efs2 group was~5.0 × 10 8 CFU/g feed, respectively.
Fresh rectal feces sample of each pig was collected once every 24 h for serial plating and counting of Efm4, Efs2, and transconjugants using selective BHI agar plates. Donor, recipient and transconjugant strains on selective plates and then resistant genes were identified as described above.

Multilocus sequence typing
To investigate the genetic heterogeneity of the donor, recipient and transconjugants, the MLST analysis was performed. 7 housekeeping genes were amplified using the primers and protocol specified by MLST website (https://pubmlst.org/efaecium/ and https://pubmlst.org/ efaecalis/), respectively. Allele and sequence types (STs) assignments were made at the publicly accessible MLST database described above [2,31]. S1-PFGE and southern hybridization S1-PFGE was performed as described previously [2]. Overnight incubated cells of Efm4, Efs2 and transconjugants were embedded in InCert Agarose (Lonza, USA), respectively, followed by digestion with S1 nuclease (TaKaRa, Japan). The DNA restriction fragment were separated in 1% SeaKem Gold® Agarose (Lonza, USA) using a pulse gel electrophoresis apparatus (Bio-Rad, USA). Genomic DNA of Salmonella serovar Braenderup strain H9812, digested with XbaI (TaKaRa, Japan), was used as a molecular standard.
Southern hybridization was performed on the S1-PFGE gel using DIG High Prime DNA Labeling and Detection Starter Kit II (Roche, Germany) with digoxinlabeled DNA probes specific for vanA, followed by manufacturer's protocol.

Plasmid sequencing
The whole genome of Efm4 including vanA-harboring plasmid was sequenced using the Pacific Biosciences RS II (Pacific Biosciences, Menlo Park, CA, USA) sequencing platform. De novo assemblies of Pacbio Reads were performed by SMRT Analysis pipeline v2.3.0 in conjunction with RS_HGAP_Assembly. Three protocols and additional assemblies were performed by minimus 2 from the AMOS package [32]. The open reading frames (ORFs) were identified and annotated with the prodigal.
The vanA-carrying plasmids from transconjugants obtained from FSBM and Efm4 + Efs2 group were partially sequenced (~13,000 bp) to verify their transfer into Efs2. By targeting the regions adjacent vanA, primers (Additional file 2: Supplemental data 1) were designed based on the sequence of vanA-carrying plasmid in Efm4. Multiple alignment of acquire sequence was conducted using SnapGene (Version 4.2.6) against the sequence of vanA-carrying plasmid Efm4.

In vivo conjugation
No pigs were colonized by vancomycin or chloramphenicol-resistant enterococci prior to the study (day−10 to − 1, data not shown). Besides, during the test period, no VRE strain was detected in the fecal samples collected from the control group fed with basil diet.
The presence vanA and catA1 in transconjugants was confirmed by further PCR bothly in vitro and in vivo experiments.

Molecular typing of probiotic enterococci isolates
The MLST analysis identified that the ST of Efm4, Efs2 and transconjugants was ST1014, ST69 and ST69, respectively (Table 1). S1-PFGE and southern hybridization S1-PFGE analysis exhibited one visible plasmid bands (~140 kb) in the donor strain Efm4 and another bands (~60 kb) in the recipient strain Efs2. It could be observed clearly that the transconjugants harbored two visible plasmids from both Efm4 and Efs2 (Fig. 3).
Southern hybridization confirmed that vanA hybridized on the visible plasmid of Efm4 with a size of~140 kb, which was subsequently named pQZ67 in this study.

Analysis of the plasmid pQZ67
The complete DNA sequence of plasmid pQZ67 was obtained by whole-genome sequencing (Fig. 4). 142,988-bp pQZ67 consisted of 38,000 amino acid residues and containing 185 potential edcoding sequences (CDSs), with a G + C content of 34.03%. The vanRSHAX gene cluster was located between two insertion sequences which were IS21and IS256. It was a Tn1546-like compound transposon, which played a crucial role in horizontal transmission and might assist in the horizontal transfer of vanA in Enterococcaceae.
Multiple alignment showing the vanA-carrying fragments (~13,780 bp) from the FSBM as well as Efm4 + Efs2 group were identical to the corresponding region of

Discussion
To the best of our knowledge, this is the first time of the observation on the transfer of vanA of probiotic origin occurred in the process of FSBM and in the digestive tract of growing pigs. Over the past few decades, with the extensive emergence of antibiotic-resistant strains, probiotics are increasingly used in human and food animals as suitable replacements of antibiotics [33,34]. As traditional lactic acid bacteria, E. faecium and E. faecalis have been wildly used in foods, drugs, dietary supplements, microbial feed and feed additives [35]. However, some prospective studies showed that these bacteria can serve as an accumulator of antibiotic resistance genes (ARGs) potentially providing them to pathogens, raising the problem of pathogen resistance [36,37]. For the past two decades, the production and consumption of microbial products associated with the increase in intensive livestock farming, and the safety of these products is important for food safety. Previous studies have shown concern on probiotics applied in feed containing transferable ARGs [38,39]. Using filter mating experiments, Nawaz et al. [40] found that ermB and tetM from lactobacillus strains could be successfully transferred to E. faecalis, with conjugation frequency ranged from 10 − 5 to 10 − 6 . Moreover, Ma et al. [41] used Lactobacillus delbrueckii strain harboring mcrC as donor strain, and E. faecalis ATCC 29212 as recipient strain, resulting in a conjugation frequency as high as 2.2 × 10 − 2 . Similarly, our study found out that, during the production of FSBM, the vanA could be transferred easily at the early stage (detected only 1 h after the fermentation began), and the transconjungations increased steadily during the whole fermentation process of FSBM. Nevertheless, the conjugation frequency ranged from~10 − 3 tõ 10 − 4 , which was about 10 to 100 fold higher than primary studies [42,43]. It might be because that FSBM provided a suitable environment for conjugations to happen, where three necessary conditions were need: full contact, absolute stationary and sufficient nutrition [44,45].
Most of the previous experiments of resistance genes transfer in vivo were performed on specific pathogen   [46] revealed that vanA could transfer from pig derived E. faecium to human derived E. faecalis in germ-free mice intestines, but only 2 colonies of transconjugants were detected after administration donor and receiver strains intragastrically for 14 d. Moreover, the transfer of vanA-harboring plasmid from poultry and pig derived strain to human fecal E. faecium strains in digestive tract of germ-free mice was observed by Dahl et al. [47], indicating that even transient colonization strains might provide a significant reservoir to transfer resistance genes to permanent commensal bacteria. A study by Lester et al. [48] demonstrated that conjugation could occur in the human gastrointestinal tract. In this study, three of six volunteers who ingested~10 9 CFU vanA-harboring E. faecium bacteria in 250 mL milk had transconjugant bacteria in their feces at a mean conjugation frequency of~10 − 7 transconjugants / recipient. This was in line with the observations in our results, in particular for the Efm4 + Efs2 group, as the  It was noteworthy that transconjugants in the FSBM passed through the digestive tract of growing pigs successfully and could be detected after 3 d of feeding. Moreover, this cycle was 2 d sooner than feeding Efm4 and Efs2 directly. In addition, in commercial production facility, the SBM fermentation process usually lasted only for 2 to 5 d before further processing, and the inoculation amount of bacteria was much more than we used in the current study [49]. This emphasized the possibility that the resistance genes harbored by commercial probiotics used in feed and feed additives can be disseminated in digestive tracts of animal, which may cause further infections under extreme conditions.
Many researchers explored the transferability of resistance genes from farm to fork [50,51]. Nawaz et al. [40] analyzed the antibiotic resistance in lactic acid bacteria from retail fermented foods in Xi'an, China. Not only did they identify tetS gene from Lactobacillus brevis and Lactobacillus kefiri for the first time, but they also observe the ermB gene from L. fermentum and L. salivarius. Furthermore, the tetM gene from L. plantarum and L. brevis was also observed to be successfully transferred to E. faecalis by filter mating, with a conjugation frequency as~10 − 6 to~10 − 5 . In addition, He et al. [52] analyzed the oxazolidinone / phenicol resistance gene optrA from 17 non-related E. faecalis isolates from human and animal origin, and the IS1216E elements on the plasmid were also found to contribute the most in the dissemination of the optrA gene. The IS 1216E elements were also detected on pQZ67 plasmid in our study, demonstrating the likelihood of in vitro and in vivo transfer of vanA between enterococci and other Gram-positive bacteria from food or feed to animal or human via food chain in vitro and in vivo.
In this study, the diversity of STs showed that the commercial probiotic enterococci strains were from multiple sources. Additionally, it should be noted that the sequence type of VRE strain Efm4 was ST1014 (Fig.  3). ST1014 was evolved out of ST78, which was the dominant clone complex (CC) in most cities in China, causing the prevalence and spread of VRE [53]. Furthermore, another reported vancomycin resistant E. faecium strain with the same ST1014 was isolated in a hospital in Shandong Province, China, in 2013 [53]. Although there was no evidence showing the direct link between that isolate and Efm4, the possible affiliation between ST1014 and ST78 (CC17) still rang the alarms of the safety of probiotic enterococci applied in feed and food.
Ideally, probiotics used in food and feed production should harbor none of the transferable resistance genes and should be sensitive to pathogen related antibiotics [54]. European Food Safety Authority (EFSA) recommended that bacterial strains harboring virulence factors or transferable ARGs should not be used in animal feeds, probiotic and fermented foods for human [55,56]. The transmission of ARGs in digestive tracts is a major health concern related to the probiotic application. Unfortunately, until now, ARGs are not included in the standard screening assays before production and application in foods and feeds. Most of the fermentation starter probiotic strains that used for food and animal feed processing were lacking of thorough and rigorous assessments. The donor strain Efm4 used in this study was isolated from a retail FSBM product, revealing the risk of dissemination multiple resistant plasmids was not only through the ingestion of food animals, but also via further feed processing like SBM fermentation. In addition to vanA, the macrolide-resistant gene erm and two aminoglycoside resistant genes aadK and aphA were also identified in pQZ67 (Fig. 4), explained the resistant of Efm4 to erythromycin, kitasamycin and gentamicin. This also indicated that pQZ67 was a multidrug resistance plasmid. In the past decade, the consumption of FSBM in Asian countries has increased sharply accompanied with the rapid development of animal husbandry [57]. To date, although there has been no report of VRE or other antibiotic-resistance probiotics in FSBM and in the intestines of growing pigs, the potential risk of horizontal transfer of resistance genes is still hanging in cliff, because the consumption is huge while the surveillance is lacked.

Conclusion
In conclusion, our study illustrated that vanA could successfully transfer among enterococci during the fermentation process of soybean meal and also in the digestive tract of growing pigs. Thus, it is suggested that when considering an Enterococcus strain as a starter for probiotics, each specific strain should be carefully evaluated to determine the presence of all known virulence factors.

Additional files
Additional file 1: Table S1. Primers used for PCR and DNA sequencing in this study and size of the PCR-targeted products. Table S2