Analysis of combined resistance to oxazolidinones and phenicols among bacteria from dogs fed with raw meat/vegetables and the respective food items

The gene optrA is the first gene that confers resistance to the oxazolidinone tedizolid, a last resort antimicrobial agent in human medicine. In this study we investigated the presence of optrA and the multi-resistance genes poxtA and cfr in enterococci and staphylococci from (i) pet animals known to be fed raw meat and vegetables and (ii) the respective food items. We examined 341 bacterial isolates from cats and dogs, 195 bacterial isolates from supermarket food items and only one E. faecium collected from industrial food in Beijing during 2016. Thirty-five (6.5%) of the 537 isolates, including 31/376 (8.2%) enterococci and 4/161 (2.5%) staphylococci, were positive for optrA, while all isolates were negative for poxtA and cfr. S1-nuclease pulsed-field gel electrophoresis (PFGE) and Southern blotting confirmed that optrA was located in the chromosomal DNA of 19 isolates and on a plasmid in the remaining 16 isolates. Whole genome sequencing revealed several different genetic environments of optrA in plasmid- or chromosome-borne optrA genes. PFGE, multilocus sequence typing (MLST) and/or SNP analysis demonstrated that the optrA-carrying Staphylococcus and Enterococcus isolates were genetically heterogeneous. However, in single cases, groups of related isolates were identified which might suggest a transfer of closely related optrA-positive E. faecalis isolates between food items and dogs.


Results and Discussion
Analysis of resistance mechanisms to oxazolidinones and phenicols. A total of 450 anal and nasal samples were collected from dogs and cats and 187 enterococcal isolates and 154 staphylococcal isolates were recovered from these samples ( Supplementary Fig. 1). An additional 224 samples were collected from vegetables and raw meat products from 28 supermarkets in Beijing, resulting in the isolation of 188 enterococcal isolates and seven Staphylococcus aureus isolates ( Supplementary Fig. 1). From the 35 samples collected from industrial pet food, only one E. faecium isolate was recovered. Resistance to florfenicol was detected in 12.8% of the Enterococcus isolates (24/187; 23 Enterococcus faecalis and a single Enterococcus faecium) and 2.6% of the Staphylococcus isolates (4/154; four S. sciuri) from companion animals. Florfenicol resistance was also identified in 6.4% of the Enterococcus isolates (12/188; nine E. faecalis and three Enterococcus casseliflavus), but in none of the seven S. aureus isolates recovered from supermarket vegetables and raw meat samples. In addition, the single E. faecium recovered from industrial pet food was susceptible to florfenicol. PCR-based amplification of optrA, cfr, and poxtA indicated that 6.5% (35/537) of all enterococci and staphylococci and 87.5% (35/40) of the florfenicol-resistant isolates harbored optrA, including 28 E. faecalis, three E. casseliflavus, and four S. sciuri isolates. All florfenicol-resistant isolates were also positive for the phenicol exporter gene fexA, whereas none of the florfenicol-resistant isolates harbored cfr, cfr(B), cfr(C), poxtA, or fexB ( Table 1). None of the isolates in this study revealed the presence of oxazolidinone resistance-mediating mutations in domain V of the 23S rRNA gene or in the genes coding for ribosomal proteins L3, L4, and L22.
Antimicrobial susceptibility profiles. All optrA-positive enterococci from companion animals originated from dogs, were resistant to florfenicol, chloramphenicol, and erythromycin, and exhibited high MICs to virginiamycin M1. Most of these isolates also exhibited resistance to minocycline (95%), as well as high-level gentamicin resistance (75%). Furthermore, about half of these isolates were resistant to ciprofloxacin. However, all optrA-positive enterococci from companion animals were susceptible to vancomycin, daptomycin, and ampicillin. All optrA-positive enterococci isolated from food were resistant to florfenicol and chloramphenicol. Most of these isolates also showed resistance to minocycline (72.7%), ciprofloxacin (72.7%), exhibited high MICs to virginiamycin M1 (72.7%), and all were susceptible to ampicillin and intermediate to vancomycin (Table 1).
All optrA-positive staphylococci originated from dogs and were identified as S. sciuri. They were resistant to florfenicol, chloramphenicol and oxacillin. Most of these isolates also exhibited resistance to erythromycin (60%) and high MICs to tiamulin (60%), but all were susceptible to vancomycin, minocycline, daptomycin, gentamicin, and rifampicin (Supplementary Table 1). The presence of resistant bacteria did not differ significantly between the samples from pets and supermarket-derived food items, except for those showing resistance to minocycline and erythromycin (Supplementary Table 2). A previous study from Denmark identified three optrA-positive enterococci isolated from food products that had a similar resistance pattern as the isolates in the current study, showing resistance to ciprofloxacin, erythromycin, aminoglycosides, and tetracycline 24 .
The MICs of linezolid and tedizolid of the optrA-positive staphylococci and enterococci were in the range of 2-16 µg/ml and 0.25-4 µg/ml, respectively. Among the 35 isolates, 30 were classified as non-susceptible to linezolid (4-16 µg/ml), with 22 of them being also classified as non-susceptible to tedizolid (≥1 µg/ml). The percentage of non-susceptibility to linezolid among enterococcal isolates from pets in the present study (8.56%, 16/187) is relatively high compared with previous studies conducted in Canada (0.40%), Korea (0.33%), and China (2.03%) [25][26][27] . Linezolid use in animals is banned in China, and phenicol antibiotics are rarely used at the China Agricultural University Veterinary Teaching Hospital for the treatment of dogs.
Location and characteristics of optrA. S1-PFGE and Southern blotting revealed that optrA was located in the chromosomal DNA of 12 E. faecalis isolates and on a plasmid in the remaining 16 isolates. In the three E. casseliflavus isolates and the four S. sciuri isolates, optrA was located in the chromosomal DNA ( Supplementary  Fig. 2). Table 2 shows that all isolates with linezolid MICs of 2 µg/ml have chromosomal optrA genes and all five isolates with tedizolid MICs of 0.25 µg/ml and seven of the eight isolates with tedizolid MICs of 0.5 µg/ml have chromosomal optrA genes. This observation may suggest that a chromosomal location of optrA may be associated with lower oxazolidinone MICs. However, the OptrA variant present in the respective isolates also needs to be taken into account.
Comparison of the deduced OptrA amino acid sequences of the 35 isolates with the OptrA of E. faecalis E349 (designated as the wild-type) revealed the presence of the wild-type OptrA and ten OptrA variants, all of which differed in at least one amino acid position from the wild-type (Table 2), and all these variants also differed distinctly from the variant OptrA E35048 found in Italy 28 . OptrA variants D, DD, DP, KD, RDK and KDP have previously been detected in large screening studies on enterococci of human and animal origin 20,29,30 . Here, we report four new variants: DM, RD, YDD and DNDM. The variants DM and RD were detected in samples from dogs and food, whereas the variants YDD and DNDM were detected only in three S. sciuri isolates from nasal samples of dogs. Based on the associated MIC values for linezolid and tedizolid, it is possible that the different OptrA variants have an impact on the relative oxazolidinone susceptibility of the corresponding isolates (Table 2).   www.nature.com/scientificreports www.nature.com/scientificreports/ Genetic environment of plasmid-borne optrA. Sixteen E. faecalis isolates harbored optrA on plasmids ranging in size between 60-100 kb. Most of these plasmids carried the same core erm(A)-optrA-fexA resistance gene cluster (Fig. 1). Furthermore, the insertion sequence IS1216E was found either downstream and/or upstream of optrA in all plasmids except that of isolate 131AC, in similar optrA environments as previously reported in plasmids pSS85, pSS92, pSS79, and pFX13 29,31 . Among the plasmids with two copies of IS1216E bracketing the optrA-carrying central region, these copies were located in opposite orientations on the plasmids from isolates 8-3, 8-2, 114AC and 82AC, but in the same orientation in the other seven plasmids. When located in the same orientation, recombination between the two copies of IS1216E may form minicircles that contain the bracketed region plus one copy of IS1216E 29 , which could accelerate the transmission of optrA. Inverse PCR confirmed that minicircles of isolates 190AC, 192NC and 182NS were detectable, which contained the central region and one intact IS1216E (data not shown). This observation indicated that the IS1216E-flanked segment was unstable in these plasmids. Only one copy of IS1216E, located in the region downstream of optrA, was identified on the plasmids of isolates 75AC, 61NC and 207AE.

S. sciuri
Moreover, four groups of plasmid-bearing isolates -72AC and 37AC; 100AC and 114AC; 190AC, 192NC and 182NS as well as 8-2 and 8-3 -were identified. Within each group, the isolates shared the same or closely related MLST types, had similar PFGE patterns, shared the same resistance pattern, and carried closely related optrA gene regions on similar-sized plasmids ( Table 1, Fig. 1). Their close genetic relationships were also confirmed by SNP analysis (Supplementary Fig. 3). All these characteristics suggested that the respective isolates are related.
Genetic environment of chromosomal optrA. The regions flanking optrA in the chromosomal DNA ( Fig. 2) differed distinctly from those on plasmids. The putative transcriptional regulator araC was located immediately upstream of optrA in seven E. faecalis and three E. casseliflavus isolates. The transposon Tn558, which includes the genes fexA, orf138, and three transposase genes, was detected in six isolates whereas the transposon Tn554, which comprises the genes met, erm(A), spc and three transposase genes, was found in five isolates (Fig. 2).   www.nature.com/scientificreports www.nature.com/scientificreports/ The upstream regions of the complete transposons Tn558 and Tn554 in E. faecalis are closely related (Fig. 2). Moreover, the genomic regions flanking optrA are similar to those of E. faecalis strains SS85 and E016 reported in previous studies 29,31 .
The chromosomal optrA regions of the four S. sciuri isolates 31, 207NS, 210NS, and 200NS were very similar and contained the core araC-optrA segment and Tn558 32 . The region of isolate 31 differed from the others by the absence of the topoisomerase gene top and three reading frames for hypothetical proteins (Fig. 3). The genetic relationships of the E. casseliflavus and S. sciuri isolates were also investigated by SNP analysis ( Supplementary  Fig. 4).
Relatedness of the isolates. PFGE analysis of the 28 E. faecalis isolates revealed 16 patterns (Table 1, Supplementary Fig. 5), whereas each of the three E. casseliflavus isolates exhibited a distinct pattern and the four S. sciuri isolates showed three patterns (Table 1, Supplementary Fig. 6), suggesting a high diversity among the optrA-positive isolates (Table 1). Moreover, the results of genetic relationships between 28 E. faecalis isolates by MLST analysis and SNP analysis were consistent. Nevertheless, closely related PFGE patterns were observed for two E. faecalis isolates from nasal samples of dogs (121NS, 233NC) and one E. faecalis isolate from beef (3-8), for two E. faecalis isolates from anal and nasal samples of dogs (52AC, 203NC) and one isolate (22-4) from a chicken wing, as well as for one isolate (207AE) from an anal sample of a dog and two E. faecalis isolates (11-7, 11-8) from egg and beef samples from supermarkets. However, in two of the three paired groups of isolates, differences in either the location of the optrA gene and/or the resistance patterns were observed. Nevertheless, one group of paired isolates, E. faecalis 52AC and 203NC from two unrelated dogs and E. faecalis 22-4 from a chicken wing revealed the chromosomal location of the optrA gene, had the same MLST type and optrA genetic environment,  Table 1). ST773 (1-7-10-1-1-84-1) is a single locus variant of ST22 (1-7-10-1-1-10-1). (2019) 9:15500 | https://doi.org/10.1038/s41598-019-51918-y www.nature.com/scientificreports www.nature.com/scientificreports/ and showed the same resistance patterns. Moreover, the observation that closely related E. faecalis isolates 8-2 and 8-3, 11-7 and 11-8 with the same MLST types (Table 1, Fig. 4) and resistance patterns and an optrA plasmid of similar size and similar optrA genetic environment (Table 1, Figs 1 and 2) were found among different food items originating from the same supermarket, points towards a possible cross-contamination at this supermarket. Several pairs of canine E. faecalis (100AC and 114AC; 190AC and 192NC; 72AC and 37AC) were identified which shared the same MLST types (except for 100AC and 114AC which share 6/7 housekeeping gene alleles), and were closely related in their PFGE patterns, their resistance patterns and the location of the optrA gene (Table 1, Figs 1  and 4). Since all samples originated from individual dogs, the occurrence of such related optrA-positive isolates may be explained by the contact between the dogs (and/or their excrements). In this regard, contacts between the dogs in the waiting area of the China Agricultural University Veterinary Teaching Hospital need also be taken into account. Even though most closely related E. faecalis isolates from dogs and food products were detected in only one case, it suggests that transmission of optrA-positive E. faecalis between contaminated food products and dogs may occur.  Schematic representation of the genetic environment of chromosomally located optrA in the twelve E. faecalis isolates investigated in this study. Shaded areas represent regions of >99% nucleotide sequence identity. Arrows indicate the positions and orientations of the genes. White arrows represent genes coding for hypothetical proteins, red arrows those coding for antimicrobial resistance genes, green arrows those coding for transposases, and blue arrows those coding for other known functions. The information in the brackets includes the location of gene optrA (c, chromosomal DNA), PFGE pattern, and the MLST type (both according to Table 1). Genes with known functions coded for a putative NADH oxidase (nox), the septation protein SpoVG (spoVG), a helicase (hel), a topoisomerase (top), a folylpolyglutamate synthase (folC), a valine-tRNA ligase (valS), a thiol peroxidase (tpx), a redox-sensing transcriptional repressor (rex), and a thiamine-phosphate kinase (thil). The radC gene codes for a DNA repair protein and represents the preferential chromosomal integration site for Tn558 and Tn554.

WGS accession numbers. The WGS of all isolates investigated in this study have been deposited at
www.nature.com/scientificreports www.nature.com/scientificreports/ conclusion This is -to the best of our knowledge -the first large-scale screening study for the presence of optrA, poxtA and cfr in enterococci and staphylococci isolated from companion animals and supermarket-derived vegetable and meat samples. The wide distribution of optrA in bacteria of these samples is of great concern for public health, especially when these bacteria are involved human infections. It is noteworthy that some E. faecalis isolates from food products and/or companion animals are closely related in their molecular and phenotypic characteristics, highlighting the possibility of bacterial spread between fresh foods from supermarkets and dogs. Because of their frequent and close contact with companion animals, we assume that bacteria -including oxazolidinone-resistant ones -can be exchanged between humans and companion animals in either direction. Further surveillance and control efforts are needed to reduce optrA-positive bacteria in both companion animals and food products at supermarkets.
All data generated or analysed during this study are included in this article (and its Supplementary  Information Files).

ethics statement. Sampling procedures were approved by the China Agricultural University Laboratory
Animal Welfare and Animal Experimental Ethical Inspection Committee, and were carried out in accordance with the Committee's approved guidelines.
Bacterial isolation and detection of resistance genes and mutations. Anal and nasal swab samples were collected from healthy pets (239 dogs and 11 cats) at the Veterinary Teaching Hospital of the China Agricultural University in Beijing. The fact that the dogs and cats included in this study were fed with raw meat, vegetables, or industrial food was assessed by a questionnaire, in which also the type of food items and the supermarket from which it was purchased were requested. Samples of bacteria from the surfaces of vegetables and meat purchased from the aforementioned supermarkets were collected by submerging the foodstuffs in buffered peptone water. Samples of bacteria from industrial food were collected by submerging the foodstuffs in buffered peptone water immediately after the pet food package was opened. Samples were plated on BHI medium and incubated at 37 °C for 24 h. Bacterial species identification was conducted by matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry analysis and 16S rRNA gene sequencing. Phenicol-resistant bacteria were selected on brain heart infusion agar plates containing 8 µg/ml florfenicol. All florfenicol-resistant isolates were screened for the presence of cfr/cfr(B)/cfr(C), fexA, fexB, optrA, poxtA, and mutations in 23S rRNA or in the genes coding for ribosomal proteins L3, L4, and L22 accounting for oxazolidinone resistance by PCR and sequence analysis using previously described primers 12 Table 1. Shaded areas represent regions of >99% nucleotide sequence identity. White arrows represent genes coding for hypothetical proteins, red arrows those coding for antimicrobial resistance genes, green arrows those coding for transposases, and blue arrows those coding for other known functions. Genes with known functions coded for a putative NADH oxidase (nox), the septation protein SpoVG (spoVG), a helicase (hel), a topoisomerase (top), a C4-dicarboxylate ABC transporter permease (dct), a folylpolyglutamate synthase (folC), a valine-tRNA ligase (valS), a thiol peroxidase (tpx), a redoxsensing transcriptional repressor (rex), and a thiamine-phosphate kinase (thil), proteins involved in mercury resistance (merC and merD), and a multidrug ABC transporter (mdlB-1 and mdlB-2). The radC gene codes for a DNA repair protein and represents the preferential chromosomal integration site for Tn558.

Molecular analyses.
The optrA-carrying isolates were subjected to SmaI macrorestriction analysis with subsequent PFGE. PFGE results were analyzed using BioNumerics (version 5.1; Applied Maths, Austin, TX, USA). The definition of a PFGE cluster was based on a similarity cutoff of 80%. The location of optrA was analyzed by S1-nuclease PFGE and Southern blot analysis 19 . MLST of the optrA-positive E. faecalis isolates was done by searching the assembled contigs of the recognized chromosomal DNA sequences, and by specific PCR assays if necessary. The primers used for this approach were those indicated at http://efaecalis.mlst.net/: Sequence types (STs) and corresponding MLST gene allele profiles were entered into BioNumerics (Applied Maths, Belgium). Genome sequencing analysis. DNA was extracted from all optrA-positive isolates using a TIANamp Bacteria DNA Kit (Tiangen Biotech Co., Beijing, China) according to the manufacturer's instructions. Potential DNA degradation and contamination were analysed by electrophoresis of aliquots in 1% agarose gels. The DNA purity was checked by using the NanoDrop UV-Vis Spectrophotometer (Thermo Fisher Scientific, Shanghai, China). DNA libraries were prepared using the KAPA Hyper Prep Kit (Roche, Basel, Switzerland) and sequenced on an Illumina HiSeq X Ten platform with 150-bp paired-end reads by Berry Genomics Company (Beijing, China). The draft genomes were assembled using SPAdes version 3.9.0 (http://cab.spbu.ru/software/spades/).
The draft genomes were used for the analysis of the genetic environment of optrA on plasmids of different sizes or in the chromosomal DNA of the Enterococcus and Staphylococcus isolates. To determine whether recombination between two IS1216E elements can result in the formation of optrA-carrying minicircles in isolates that carried the gene optrA on a plasmid, inverse PCR assays were conducted using outward primers (primer 1: 5′-CCAGCACCCTGAACCATTCT-3′; primer 2: 5′-CCGGTGTCCTCTTTGTCAGG-3′; annealing temperature 58 °C). A draft assembly of the sequences was generated using CLC Genomics Workbench 9 (CLC Bio, Aarhus, Denmark), the assembly algorithm of which uses de Bruijn graphs 38 . All contigs with an average coverage of >100-fold were screened for the presence of optrA using BLAST analysis and LRE-Finder 23 . The regions flanking optrA-carrying contigs were identified by primer walking 19 . Sequence analysis was conducted using ORF Finder