Introduction

Convenient, cheap and popular, ready-to-eat (RTE) food products in Taiwan are found in public markets, train stations, sidewalk stalls, and many other locations. The microbiological quality of RTE foods has gained attention due to the 4,284 cases of food poisoning that were officially reported between 1991 and 2010, with the number of cases increasing yearly1. Accordingly, the literature on the microbiological quality of cold RTE foods (e.g., sandwiches, noodles, and rice balls served at 18 °C or lower) and RTE-related ingredients (e.g., staples, meats, vegetables, and seafood) in Taiwan has grown significantly2,3. After analysing 164 RTE food samples served at 18 °C, Fang et al. reported a 42.7% incidence of psychrotrophic Pseudomonas spp., 75% coliforms, 7.9% E. coli, 49.8% B. cereus and 17.9% S. aureus2. According to a separate study conducted by Wei et al., RTE-related products stored at room temperature had the highest incidence of bacterial contamination, and RTE foods served by street vendors in traditional markets had the highest bacterial counts3. However, a review of extant studies indicates that few efforts have been made to determine antibiotic resistance in RTE foods.

The first case of methicillin resistance in Staphylococcus aureus (MRSA) was reported in Great Britain in 19614. The resistance mechanism has been linked to an alternative penicillin-binding protein (either PBP2a or PBP2′) encoded by mecA and transmitted via the excision and insertion of a SCCmec element5,6. SCCmec elements share two important features: a mec gene complex carrying a mecA homologue, and specific insertion sites with flanking repeat sequences via the ccr gene complex6. Recently, several research teams have reported the potential of coagulase-negative staphylococci (CoNS) for transmitting antibiotic-resistant genes7,8,9,10,11. Tulinski et al. found that CoNS strains isolated from pig farms acted as reservoirs for heterogeneous SCCmec elements9. Kloos et al. have described S. sciuri as a reservoir for a methicillin-resistant gene11, and Ruzauskas et al. have reported the cross-sectional prevalence of methicillin-resistant S. haemolyticus in companion animals7.

It is generally accepted that RTE food products serve as reservoirs for antimicrobial-resistant bacteria, but transmission and resistance mechanisms in Taiwan require further investigation. For this project, we looked at proportions of methicillin-resistant coagulase-negative staphylococci (MRCoNS) found in samples of spring rolls, cold noodles, and fruit platters collected from RTE vendors in the densely inhabited cities of Kaohsiung, Taichung and Taipei, and attempted to determine their antibiotic resistance mechanisms.

Results

MRCoNS characterization

We used dnaJ gene sequencing to identify bacterial species in 14 MRCoNS strains (Table 1). The dominant bacteria was S. sciuri (6/14, 42.9%), including 3 isolates of S. sciuri subsp. rodentium, 2 isolates of S. sciuri subsp. sciuri, and 1 isolate of S. sciuri subsp. carnaticus, followed by S. saprophyticus (4/14, 28.7%), S. haemolyticus (2/14, 14.4%), S. lentus (1/14, 7%), and S. pasteuri (1/14, 7%). The most frequent sources were spring rolls (7/14, 50%), cold noodles (5/14, 35.7%), and fruit platters (2/14, 14.3%). Genetic diversity data as determined by PFGE analysis are shown in Table 1 and Supplementary Fig. S1. Only two S. saprophyticus isolates (TPE-21 and TPE-32, isolated from cold noodles and spring rolls, respectively) belong to pulsotype IX (Table 1). According to antimicrobial susceptibility test results, all isolates were resistant to 1–5 antimicrobials, a list that included oxacillin (14/14, 100%), levofloxacin (2/14, 14%), erythromycin (10/14, 71.4%), tetracycline (9/14, 64.3%), gentamicin (4/14, 28.7%; 1 of the 4 was gentamicin-intermediate), and vancomycin-intermediate (1/14, 7%) (Table 1).

Table 1 Antibiotic resistance and pulsotypes identified in CoNS isolated from the 270 RTE food samples.

mecA was detected in all 14 oxacillin-resistant isolates (14/14, 100%) (Table 1). Among the 10 erythromycin-resistant isolates, S. saprophyticus KHH-2, S. sciuri subsp. sciuri TXG-24, and S. sciuri subsp. rodentium TXG-28 carried both ermA and ermC genes, while S. haemolyticus TXG-25, S. sciuri subsp. sciuri TXG-15, and S. sciuri subsp. rodentium TPE-18 only carried the ermC gene. No erm genes were detected in the other 4 erythromycin-resistant isolates. Among the 9 tetracycline-resistant isolates, S. sciuri subsp. sciuri TXG-15 harboured 3 tetracycline-resistant genes, while S. haemolyticus TXG-25 and S. sciuri subsp. rodentium TXG-28 and TPE-18 contained tet(K). Both tet(M) and tet(K) were found in S. saprophyticus KHH-20, and tet(M) and tet(O) were found in S. lentus TXG-26. We also observed aac(6′)Ie-aph(2″)Ia in all gentamicin-resistant isolates (4/4, 100%) (Table 1). Staphylococcal super-antigenic genes encoded staphylococcal enterotoxins (SEs), an SE-related toxin, and toxic shock syndrome toxin-1 (TSST-1). Among the 14 isolates, 12 (85.7%) carried one or more staphylococcal super-antigenic genes, but they were not detected in S. haemolyticus TXG-25 or S. saprophyticus TPE-21. Among enterotoxins and enterotoxin-like proteins, the most prevalent genes were sec (3/14, 21.4%), selk (5/14, 35.7%) and seln (5/14, 35.7%). The sed and selp genes were not detected in any isolates (Table 1).

Genetic analysis of the mecASs gene complex and SCCmecTXG24

Gene analysis results indicate the presence of two mecA homologues (mecA and mecASs) in S. sciuri subsp. sciuri TXG-24. Genomic structure analysis data for the mecASs region are shown in Fig. 1. The mecASs gene complex of TXG-24 is closely related to S. sciuri subsp. sciuri ATCC29062 (GenBank accession number AB547234.1), with the exception of a downstream 4-gene cobalt ABC transporter homologue. The mecASs region of S. sciuri subsp. rodentium ATCC700061 (AB547235.1) shares a high degree of similarity with TXG-24, except for the upstream ugpQ of two hypothetical protein genes, the ABC transporter gene, and the amino acid/polyamine/organocation (APC) family transporter gene.

Figure 1: Genomic structure of the mecASs complex in S. sciuri subsp. sciuri TXG-24.
figure 1

Homologous regions are in pink. Arrows indicate genes and their directions. Most sequences are closely related to the sequences of S. sciuri subsp. sciuri (ATCC29062) and S. sciuri subsp. rodentium (ATCC700061).

The SCCmec element of S. sciuri subsp. sciuri TXG-24 has a complex genomic structure that contains a class A mec gene complex (IS431-mecA-mecR1-mecI), an IS1216 mobile element carrying tet(S), partial DNA recombinase with methyltransferase, a heavy metal-resistant gene complex, and a ccr gene complex (Fig. 2). The mec gene complex of SCCmecTXG24 is closely related to S. sciuri subsp. carnaticus GVGS2 (HG515014) and S. pseudintermedius KM241 (AM904731), except for two hypothetical protein genes and a truncated mecR2 gene. The SCCmecTXG24 region containing partial DNA recombinase with methyltransferase is highly similar to the comparative region of Streptococcus suis SC84 (FM252031.1), except for a truncated apt gene. Compared to S. capitis CR01 (KF049201), the heavy metal-resistant gene complex is associated with the absence of two cadmium-resistant genes (cadD and cadB). The proximal left boundary of SCCmec consists of the ccr gene complex, the putative helicase gene, and some hypothetical protein genes that are associated with comparative regions in S. sciuri subsp. carnaticus GVGS2 and S. pseudintermedius KM241.

Figure 2: Data from a genomic analysis of SCCmec in S. sciuri subsp. sciuri TXG-24, compared to data for SCCmec in S. sciuri subsp. carnaticus GVGS2, S. pseudintermedius KM241, S. capitis CR01, and a partial sequence in the integrative and conjugative element (ICE) of Streptococcus suis SC84.
figure 2

Colours indicate various homologous regions in bacterial isolates.

Analysis of insertion sequence element carrying the tet(S) tetracycline-resistant gene

The tet(S)-carrying IS1216 mobile element was found at the 3′ end of ΔmecR2 (Fig. 2). According to our sequence analysis, orf25-orf26-orf27-tet(S) had a high degree of similarity with both the Lactococcus lactis subsp. lactis pK214 plasmid (GenBank accession number X92946) and Streptococcus dysgalactiae subsp. equisimilis NTUH_1743 (EF682209) (Fig. 3). Comparisons of IS1216 regions revealed exceptionally high degrees of shared identity (99.4% and 99.6%) with the L. lactis sp. lactis pK214 plasmid, but much lower degrees of shared identity (69.1% and 76.5%) with S. dysgalactiae subsp. equisimilis NTUH_1743 due to a truncated gene. The ΔtnpA gene was only found downstream of orf25 in L. lactis sp. lactis pK214.

Figure 3: Data from a genetic analysis of the tetracycline-resistant tet(S) gene complex inserted in SCCmecTXG24 and compared to the complexes Lactococcus lactis subsp. lactis plasmid pK214 and Streptococcus dysgalactiae subsp. equisimilis NTUH_1743.
figure 3

Homologous regions are shaded in gray. Black arrow, transposase. Corresponding regions are shadowed.

ccr gene phylogenetic trees

SCCmec is a genetic element that encodes methicillin resistance and that carries a unique site-specific recombinase (the ccr gene) in charge of SCCmec element integration and excision6,12. For the present study, we identified a ccr gene complex in S. sciuri subsp. sciuri TXG-24. Lengths of ccrA and ccrB were 1350 and 1629 bp, respectively. Phylogenetic trees for the ccrA and ccrB sequences (23 each) are shown in Fig. 4a and b. ccrA matching identity was 84.5% to ccrA5 in S. pseudintermedius KM241 (GenBank accession number AM904731). ccrB matching identity was 92.1% to ccrB3 in S. pseudintermedius AI16 (LN864705.1).

Figure 4: Phylogenetic trees for the cassette chromosome recombinase (ccr) gene.
figure 4

(a) 23 ccrA genes. (b) 23 ccrB genes. Trees were generated using neighbour-joining MEGA7 software. Numbers next to nodes indicate confidence levels, expressed as percentages of occurrence over 2,000 bootstrap samples. Scale bar indicates evolutionary distance.

SCCmecTXG24 boundaries

To investigate SCCmecTXG24 boundaries, we aligned the left and right boundaries of SCCmec types I–VII with the SCCmec element of S. sciuri subsp. carnaticus GVGS2 (Fig. 5). SCCmecTXG24 integration occurred at almost the same nucleotide position at the 3′ end of the orfX gene as the SCCmec complex of S. sciuri subsp. carnaticus GVGS2 and S. pseudintermedius KM241, with both sharing identical direct repeats (DR) at their left and right boundaries. However, nucleotide positions in the other SCCmec types were different from that of SCCmecTXG24, and the inverted repeats (IR) of each SCCmec type were variant.

Figure 5: SCCmec boundaries.
figure 5

Left and right boundaries of SCCmec types I to VII and the SCCmec element of S. sciuri subsp. carnaticus GVGS2 were aligned with SCCmecTXG24. Black arrows indicate direct repeats (DRs). Gray arrows indicate inverted repeats (IRs) of SCCmec elements.

SCCmec typing and mecASs detection in 14 MRCoNS strains

To investigate SCCmec distribution in Taiwan, we analysed 14 strains of MRCoNS from the 270 RTE food samples. Four SCCmec types (IV, V, VIII and TXG-24) were identified in 9 strains (9/14, 64.3%); the other 5 were non-typeable (Table 2). The dominant form was SCCmec type VIII (3/9, 33.3%), found in S. sciuri subsp. carnaticus KHH-57 and TPE-33, and in S. lentus TXG-26. SCCmec type IV (2/9, 22.2%) was found in S. pasteuri TPE-12 and S. saprophyticus TPE-32. SCCmec type V (2/9, 22.2%) was found in S. haemolyticus KHH-11 and S. sciuri subsp. rodentium TXG-28. SCCmecTXG24 (2/9, 22.3%) was found in S. sciuri subsp. sciuri TXG-24 and S. sciuri subsp. rodentium TPE-18. The intrinsic mecASs gene was present in 8 of the 14 MRCoNS strains (57.1%).

Table 2 SCCmec types and mecASs in 14 MRCoNS.

Discussion

In their study of five types of RTE food products in Taiwan, Fang et al. reported 75%, 49.8%, 42.7%, 17.9% and 7.9% contamination rates for coliform, Bacillus cereus, Pseudomonas spp., S. aureus and E. coli, respectively, in food products stored at 18 °C2. S. aureus was found in 26.1% of all ham samples, 21.4% of all seafood samples, 15.4% of other meat samples, and 13.6% of all vegetable samples. A separate study conducted in southern Taiwan found a 9.5% incidence of S. aureus contamination in RTE food products purchased from warehouse stores, 12.7% from traditional markets, and 19.0% from supermarkets3. The two research teams reported the presence of different pathogens in RTE food, but did not address antimicrobial susceptibility or resistance pattern tendencies. For the present study, we isolated 14 MRCoNS strains that were resistant to at least one antibiotic, and identified the dominant sources as spring rolls filled with salad ingredients and stewed ground pork wrapped in thin pastry dough, both prepared by glove-wearing vendors (Table 1). The fillings and pastry cracks are likely bacteria reservoirs13. The second most common source was cold noodles mixed with some kind of sauce, with bacterial proliferation likely due to the relatively higher pH value of the sauce or improper storage temperature2. Bacterial contamination of fruit platters (the third most common source) was likely due to the improper cleaning of knives. Regardless of actual cause or transmission route, the data indicate that RTE food contamination is a likely avenue for transmitting antibiotic-resistant genes and food-borne diseases14,15,16.

Determining genetic relationships in bacterial isolates is an important task for monitoring the spread of bacteria. In one study conducted in Turkey, genetic diversity data for 154 multi-drug-resistant strains of S. aureus found in 1,070 RTE food samples suggested multiple routes for various isolates17. In the present study, only two S. saprophyticus isolates (TPE-21 and TPE-32, both from Taipei city) shared the same pulsotype, indicating genetic diversity in our RTE food samples (Supplementary Fig. S1).

Staphylococcal enterotoxin (SE) contaminated food have been reported in foodborne illness18. Fijałkowski et al. reported that the prevalence of toxin genes in 75 different staphylococcal isolates from 41 food samples in Poland19. The most prevalent SE genes were sei (27/75, 36%), followed by seln (24/75, 32%) and sed (23/75, 31%). Chiang et al. reported that 109 (74.1%) S. aureus isolates contained one or more SE genes in Taiwan20. The most detected SE genes were sei (45/147, 30.6%), followed by sea (42/147, 28.6%) and seb (30/147, 20.4%). These studies and our finding revealed that SEs production of staphylococcal isolates may be associated with food poisoning19,20. Our study found that the dominant SE genes were selk (5/14, 35.7%) and seln (5/14, 35.7%), followed by sec (3/14, 21.4%) (Table 1). Further studies are warranted to determine the importance of SEs-producing CoNS in RTE food.

To date, the mecA gene has been found in multiple homologues, including mecA1 (mecA1, mecASs and mecASv)21,22, mecASf22, and mecC (formerly mecALGA251)23. Although mecASs (from S. sciuri) and mecASv (from S. vitulinus) share 80% and 91% identities with mecA, respectively, neither gene is associated with oxacillin resistance22. mecASf (from S. fleurettii), which belongs to the class A mec complex, shares 99% identity with the mecA gene, suggesting that S. fleurettii may be the ancestor of the SCCmec element in MRSA22. We found a close relationship between the TXG-24 mecASs gene complex and a comparative region of S. sciuri subsp. sciuri ATCC29062 (GenBank accession number AB547234.1) that is not associated with oxacillin resistance (Fig. 1).

The CoNS-acquired mecA gene, which has been the focus of multiple studies, is a likely reservoir for transmitting antibiotic-resistant genes7,8,9,10,11. Of the 14 MRCoNS strains that we tested, the most prevalent was S. sciuri—a widespread Staphylococcus species among animals and humans. Reported in a wide range of food products, this bacteria has been described as a reservoir for the methicillin-resistant gene11,24,25. S. sciuri was first described by Kloos et al. and originally isolated from both human and animal skins26. According to one study, ccr genetic diversity in methicillin-susceptible S. sciuri may be useful for capturing the mecA gene and assembling the SCCmec element27. Two research teams have shown that S. saprophyticus, S. haemolyticus and S. lentus in RTE foods are likely routes for antibiotic-resistant gene transmission in Poland28,29. Specifically, Podkowik et al. reported that 40% (17/42) of the CoNS strains they examined were resistant to 4 or more antibiotics, especially 15 isolates (36%) harbouring the mecA gene28. Chajecka-Wierzchowska et al. found that 56.9% (33/58) of the CoNS strains they tested were resistant to at least one antibiotic, with 24 isolates (41.3%) harbouring the mecA gene29. In Taiwan, SCCmec types IV and V have been described as prevalent in community-associated MRSA; these same SCCmec types were also found in the MRCoNS strains we analysed for the present study (Table 2)30,31. Combined, these data indicate that MRCoNS strains can serve as reservoirs for transmitting the SCCmec element to and from MRSA.

Many ccr gene complexes have been identified in MRSA23,32,33. Multiple ccr variants have been found in CoNS strains—for example, ccrA5B3 in S. pseudintermedius KM241 and both ccrA5B13 and ccrA5B9 in S. sciuri27,34. Different compositions of the ccr gene complex may be due to dissimilarities in recognised insertion sites35,36. We found similar ccrA boundaries sequences in S. sciuri subsp. sciuri TXG-24, S. pseudintermedius KM241, and S. sciuri subsp. carnaticus GVGS2, suggesting that SCCmec is easily transmitted across these and perhaps other Staphylococcus species (Figs 4a and 5).

In summary, we found that CoNS strains in contaminated RTE food samples collected in three Taiwanese cities were resistant to multiple types of antibiotics; it is likely that the associated antibiotic-resistant genes can be easily transmitted to other food products, to the homes of consumers, and to hospitals and other clinics. Since S. sciuri carries diverse ccr genes that are globally distributed, further research is called for to determine or refute its role as a reservoir for antibiotic-resistant gene transmission.

Methods

Sample collection and microbiological analysis

A total of 270 food samples (90 spring rolls, 90 cold noodle bowls and 90 fruit platters) were collected between June and November of 2014. All samples were randomly procured and transported to our laboratory in their original packaging, either within 1 h at the original temperature (Kaohsiung and Taichung samples) or 2 h refrigerated at 4 °C (Taipei samples).

For each sample, 10 g were homogenised using a stomacher sample blender, and enriched in brain-heart infusion broth (BD Biosciences) overnight at 37 °C. Single loopfuls of each bacterial suspension were plated on mannitol salt agar. Single colonies were placed on Muller-Hinton agar with 2% NaCl and 4 μg/ml oxacillin. Bacterial identification was performed by dnaJ gene sequencing as previously described37.

Antimicrobial susceptibility testing

Antimicrobial susceptibility testing was performed using standard agar dilution methods according to Clinical and Laboratory Standards Institute guidelines38. Minimum inhibitory concentration (MIC) was defined as the lowest concentration of antibiotic preventing bacterial growth after 16–20 h of incubation at 37 °C. The following antimicrobial agents were tested: erythromycin, gentamicin, levofloxacin, oxacillin, tetracycline and vancomycin.

Pulsed-field gel electrophoresis (PFGE)

PFGE typing of SmaI-digested DNA (New England BioLabs, Ipswich, MA) was performed as previously described39. Electricity (200 volts) was applied for 20 h at 13 °C, with pulse durations ranging from 5.3 to 34.9 sec at 6 V/cm. Dice similarity indices40 were used to construct pulsotype relationship dendrograms using an unweighted pair group method with arithmetic means. Pulsotypes exhibiting 85% similarity were assigned to the same clusters.

PCR detection of antibiotic-resistant genes and staphylococcal enterotoxin genes

PCR was used to detect the presence of the following antibiotic-resistant genes: gentamicin (aac(6′)Ie-aph(2″)Ia), oxacillin (mecA), vancomycin (vanA, vanB), erythromycin (ermA, ermB, ermC), and tetracycline (tet(M), tet(O), tet(K)). Primer sets were selected based on a previous study41. The presence of staphylococcal enterotoxin genes, sea, seb, sec, sed, see, seg, seh, sei, selj, selk, sell, selm, seln, selo, selp, selq, selr and tst1, were determined by PCR using primer sets from a previous study42.

Identification of SCCmecTXG24 and the mecASs gene complex

Genomic DNA from S. sciuri subsp. sciuri TXG-24 was extracted manually. Total DNA was subjected to quality control using agarose gel electrophoresis and quantified by Qubit (Invitrogen, Thermo Fisher Scientific, Waltham, MA). The S. sciuri subsp. sciuri TXG-24 genome was sequenced using massively parallel sequencing Illumina (San Diego, CA). Two DNA libraries were constructed: a paired-end library with a 500 bp insert, and a mate-pair library with a 5 kb insert. Both libraries were sequenced with the HiSeq2500 ultra-high-throughput sequencing system (Illumina, San Diego, CA) (PE125 strategy). Library construction and sequencing was performed at Beijing Novogene Bioinformatics Technology Co., Ltd. An in-house quality control program was used for both paired-end and mate-pair reads. Illumina PCR adapter reads and low quality reads were filtered and assembled with SOAPdenovo43,44 to generate scaffolds. All reads were used for subsequent gap closures. SCCmecTXG24 and mecASs gene complex nucleotide sequences from S. sciuri subsp. sciuri TXG-24 were added to GenBank (accession numbers KX774481 and KX774480, respectively).

Phylogenetic tree analysis

The ccrA and ccrB genes identified in this work were compared with 22 publicly available Staphylococcus spp. sequences: S. aureus strains JCSC6943, JCSC6945, COL, NCTC10442, CA05, JH1, JH9, MRSA252, Mu3, Mu50, MW2, N315, 85/2082, BK20781, C10682 and HDE288 (GenBank accession numbers AB505628.1, AB505630.1, CP000046, AB033763.2, AB063172.2, CP000736, CP000703, BX571856, AP009324, BA000017, BA000033, BA000018, AB037671.1, FJ670542.1, FJ390057.1, and AF411935.3, respectively); S. pseudintermedius strains KM241 and AI16 (AM904731 and LN864705.1); S. haemolyticus H9 (EU934095); S. saprophyticus subsp. saprophyticus TSU33 (AB353724.1); S. sciuri MCS24 (AB587080.1); and S. sciuri subsp. carnaticus GVGS2 (HG515014). Phylogenetic trees were analysed by MEGA7 using the neighbour-joining method; tree topologies were estimated using bootstrap analyses with 2,000 replicates to achieve confidence intervals as indicated on each tree node45. Identities shown after each ccr gene were aligned and calculated using DNAman (Lynnon Biosoft, Quebec).

SCCmec type determination and mecASs gene detection

SCCmec types were determined by mec and ccr gene complexes as described in our previous study39. SCCmecTXG24 was determined by the class A mec complex and ccr gene (ccrA5B3) (Supplementary Table S1), and mecAs was determined by mecAs-F and mecAs-R (Supplementary Table S1).

Additional Information

How to cite this article: Yang, T.-Y. et al. mecA-related structure in methicillin-resistant coagulase-negative staphylococci from street food in Taiwan. Sci. Rep. 7, 42205; doi: 10.1038/srep42205 (2017).

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.