Introduction

Carbapenem-resistant Enterobacteriaceae have become a global public-health concern because these bacteria cause high mortality and are multidrug-resistant1. This resistance is mainly due to the production of carbapenemases, which hydrolyse carbapenem antibiotics. Carbapenemase genes such as blaNDM are often in transmissible plasmids and thus spread readily among Enterobacteriaceae species2. Currently, carbapenemase-producing Enterobacteriaceae (CPE) are found in both clinical and non-clinical settings, including sewage systems3,4. Alarmingly, CPE have been isolated from imported food stuffs (vegetables and seafood) originating in southeast Asian countries5,6. Despite the potential danger, thorough food surveillance is not performed on retail foods in CPE-endemic regions, with a few exceptions in China7,8.

Previously, we identified genetically diverse Enterobacteriaceae isolates with blaNDM-harbouring plasmids in a hospital of Yangon, Myanmar’s largest city9,10. We then found closely related strains in sewage water outside the hospital10, implying that CPE in clinical settings could reach the external environment. These findings prompted us to investigate the extent of CPE dissemination in local communities. Here, we focused on the spread of CPE in local markets.

Results

Diverse E. coli and K. pneumoniae isolates harbouring bla NDM from market foods

Of the 93 food samples from eight markets, we identified 20 CPE-positive samples (21.5%). Within each food category (meat, seafood, and vegetables), the percentage of CPE-positive samples was 25.0% (8/32), 10.3% (3/29), and 28.1% (9/32), respectively (Table 1). We obtained 27 CPE isolates from the 20 samples, with 16 (59.3%) derived from vegetables. Water spinach exhibited the highest CPE-positive rate (4/8, 50%) and yielded eight CPE isolates. Of the 27 CPE isolates, 20 carried blaNDM, including 13 Escherichia coli, six Klebsiella pneumoniae, and one Serratia marcescens. All but one E. coli, along with four of six K. pneumoniae, possessed blaNDM-5; the remaining carried blaNDM-1 (Fig. 1). Every blaNDM-carrying isolate was resistant to imipenem. Those harbouring blaNDM-5 were also resistant to meropenem, but blaNDM-1 did not confer resistance to this antibiotic. All isolates were resistant to extended-spectrum cephalosporin and oxacephem antibiotics, such as ceftazidime and moxalactam, and the minimum inhibitory concentration for these antibiotics was >64 µg/mL and ≥128 µg/mL, respectively. Additionally, 80% (16/20), 55% (11/20), and 70% (14/20) of isolates were resistant to gentamycin, amikacin, and levofloxacin, respectively.

Table 1 Isolation of CPE from food samples in the markets of Yangon, Myanmar. ECC, E. cloacae complex; Sm, S. marcescens.
Full size table
Figure 1
figure 1

Phenotypic and genetic characteristics of CPE isolates harbouring blaNDM. aEc, Escherichia coli; Kp, Klebsiella pneumoniae; Sm, Serratia marcescens. bFood samples were obtained from southern (S), western (W), eastern (E), and northern (N) districts of Yangon. cGrey cells indicate isolates resistant to the indicated antimicrobials. MIC, minimum inhibitory concentration. dNumbers and alphabets in columns denote variant types. For GyrA and ParC, presence of amino acid substitution in the indicated residues are marked in dark yellow. Genes with nucleotide substitution(s) compared with sequences in the ResFinder database are marked with asterisks. eSouthern blots identified blaNDM-5 gene on chromosomes. fNA, not applicable.

Full size image

Resistant isolates possessed multiple genes encoding aminoglycoside modification enzymes, chromosomal point mutation in quinolone-resistance-determining regions of gyrA and parC, as well as plasmid-mediated quinolone-resistance determinants. Notably, we found the 16S rRNA methylase gene rmtB in nine isolates resistant to gentamycin and amikacin. In contrast, all isolates except one were susceptible to fosfomycin and colistin. The only isolate resistant to the latter antibiotic was classified as Serratia, known to be intrinsically colistin-resistant11. Multilocus sequence typing (MLST) classified 13 E. coli isolates into 10 sequence types (STs) and six K. pneumoniae isolates each into a different ST (Fig. 1). Further evidence of isolate diversity includes blaNDM presence in differently sized plasmids or chromosomes (Fig. 1).

Carbapenemase-producing E. cloacae complex possesses bla IMI-1

Using the carbapenem-inactivation method (CIM), we identified seven carbapenemase-positive isolates of Enterobacter cloacae complex that are negative for blaNDM, blaKPC, blaIMP, or blaOXA-48 according to a PCR-dipstick test. Instead, whole genome sequencing revealed that they all possessed blaIMI-1, a chromosomally encoded Ambler class A serine β-lactamase gene. In accordance with an earlier report12, blaIMI-1 presence conferred high resistance to imipenem and meropenem, but not to extended-spectrum cephalosporin and oxacephem antibiotics, such as ceftazidime and moxalactam (Fig. 2). These bacteria were also susceptible to gentamycin, minocycline, and levofloxacin. Five isolates out of seven were resistant to fosfomycin, and three were resistant to colistin, a last-resort antibiotic for treating CPE infections.

Figure 2
figure 2

Phenotypic and genetic characteristics of E. cloacae complex isolates harbouring blaIMI-1. aPhylogenetic groups were classified according to Chavda et al.13. bIsolates resistant to the indicated antimicrobials are represented as grey shaded cells. MIC, minimum inhibitory concentration. *Genes with nucleotide substitutions compared with sequences in ResFinder.

Full size image

These six isolates were newly assigned to five STs representing previously undetected populations. We used whole-genome sequencing data of these isolates and those from the GenBank database to generate a core-genome single-nucleotide-polymorphism-based phylogenetic tree of the E. cloacae complex (Fig. 3). This tree grouped the newly assigned isolates into three clusters (G, M, and N)13. Each isolate carried blaIMI-1 between setB and yeiP; the gene was encoded on an integrative mobile element designated as EcloIMEX14. We identified three mobile elements, EcloIMEX-2 (GenBank accession number KR057494.1), -3 (KU870977.1), and -8 (MG547711.1). Of these, EcloIMEX-3 was found in phylogenetically different isolates, F038 and F054V.

Figure 3
figure 3

Phylogenetic analysis based on core-genome SNPs separates market-food-derived IMI-1-producing isolates into groups G, M, and N of the E. cloacae complex. A phylogenetic tree was constructed using the genome of E. cloacae ATCC13047 (GenBank accession number NC_014121.1) as a reference. Genomic sequences of other E. cloacae complex strains were retrieved from NCBI and included in the analysis. The food isolates acquired in this study are in bold; strains belonging to groups G, M, and N are marked in grey.

Full size image

Discussion

This study found diverse CPE in local market foods with higher isolation rates than previously reported (2.4%)8. Vegetables were the most contaminated, exhibiting the highest CPE-positive rate (28.1%) and containing approximately 60% of isolates identified. Following vegetables, meat samples had high isolation rates (25.0%).

Carbapenem-resistant E. coli and K. pneumoniae isolates exhibited multidrug resistance because they harboured blaNDM-1 or blaNDM-5, along with other antimicrobial-resistance determinants. Worryingly, most isolates possessed blaNDM-5, which confers higher resistance against carbapenems than blaNDM-115. Furthermore, we found evidence that E. coli isolates harbouring blaNDM have clinical links; five of the 10 STs identified had been previously detected in a tertiary-care hospital of Yangon10. In contrast, NDM-producing K. pneumoniae isolates belonged to STs that have not been detected thus far from Yangon clinical specimens, although four (ST15, 16, 39, 395) have been found in clinical isolates from other countries16,17. Notably, we identified isolate F069 in chicken and showed that it was resistant to all tested antibiotics except colistin. This strain belongs to ST395, a constituent of worldwide epidemic clonal group 25818.

Isolation of IMI-type carbapenemase is relatively rare. Here, we provided valuable empirical evidence of diverse IMI-producing E. cloacae complexes existing outside hospitals, confirming previous long-term clinical observations that suggested that these bacteria are present in local communities14,19. Moreover, we confirmed the presence of IMI-type carbapenemase in Myanmar, suspected ever since blaIMI-2 was detected in a Dutch individual travelling to the country20.

We linked blaIMI-1 to a putative mobile element EcloIMEX in all isolates containing the gene. Among the three EcloIMEX detected, EcloIMEX-3 was found in genetically different group N (F038) and G (F054V) isolates, suggesting independent acquisition of the element. Thus, EcloIMEX is probably responsible for mobilizing blaIMI-type genes to E. cloacae complex14,19.

We isolated a blaIMI-1-harbouring E. coloacae complex from fish and water spinach, which is usually cultivated in water-rich regions, corroborating previous reports that associated these bacteria with aquatic environments6,21,22. Of note, we isolated two water spinach isolates of the same ST and EcloIMEX type from different districts, suggesting the existence of a shared reservoir of blaIMI-1 dissemination. Other sources are also likely, given the presence of blaIMI-1-harbouring isolates in other vegetables, as well as chicken and mutton, across all four surveyed districts.

Overall, our results demonstrate widespread dissemination of blaIMI-1-harbouring E. coloacae complex in Yangon. These organisms are susceptible to several common antibiotics, meaning that multiple treatment options are available in the event of infection such as reported previously23,24. However, they may sporadically become more resistant to drugs through acquiring multidrug-resistance plasmids in local communities. We therefore strongly recommend vigorous countermeasures to prevent pathogen spread through daily consumption of food stuffs in regions heavily contaminated by CPE and other multidrug-resistant microbes.

In conclusion, this study provided evidence that local market foods in Yangon, Myanmar frequently harboured diverse CPE isolates, some with highly drug-resistant phenotypes. We note two important limitations of our study: the small sample size and the possible existence of multiple sources for CPE spread. These factors prevented us from tracking exact CPE dissemination profiles in market food samples. Thus, further study is necessary to fully understand the extent of CPE dissemination in this region and to identify the major source(s) of their spread. Such knowledge will enable the development of effective countermeasures against contamination of food with CPE. Nevertheless, our research highlights the immediate need to take corrective action against the current situation of CPE contamination in food.

Methods

Isolation of carbapenemase-producing organisms from market food

Food sampling and bacterial isolation were conducted as previously described25. From November to December 2017, chicken, pork, beef, fish, clam, and assorted vegetables were collected from eight street stalls in Yangon. Samples were placed in an ice box and transported to the laboratory at the Department of Medical Research, Yangon, where they were either examined immediately or stored at 4 °C until further analysis. For sample processing, 25 g of flesh was mixed with 225 mL buffered peptone water in a stomacher bag and incubated at 35 °C for 18 ± 2 h. Subsequently, a loopful of enrichment culture was streaked onto CHROMagar ECC (CHROMagar, Paris, France), supplemented with 0.25 μg/mL meropenem and 70 μg/mL ZnSO426, then cultured at 35 °C for 24 ± 2 h.

Bacterial identification, antimicrobial-susceptibility tests, and PCR for carbapenemase genes

Bacteria were identified with matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI Biotyper; Bruker Daltonics, Germany). Antimicrobial susceptibility was measured using a MicroScan WalkAway 96 Plus (Beckman Coulter, Brea, CA, USA) and dry plate (Eiken Chemical, Tokyo, Japan), then classified based on the Clinical and Laboratory Standards Institute Guidelines (M100-S24). For colistin, resistance was defined according to the European Committee on Antimicrobial Susceptibility Testing criteria. Carbapenem-resistant Enterobacteriaceae isolates were screened for carbapenemase production using CIM27. In addition, PCR-dipstick chromatography was performed to detect four major carbapenemase genes (blaNDM, blaKPC, blaIMP, and blaOXA-48)28.

Whole-genome sequencing and plasmid analysis

Bacterial isolates were cultured overnight in brain-heart infusion broth (BD Bacto, Franklin Lakes, NJ, USA) containing 0.25 μg/mL meropenem. Genomic DNA was then prepared using a PowerSoil DNA isolation kit (Qiagen, Valencia, CA, USA). Library preparation was performed as described previously9 or using KAPA Hyper Plus Kits (KAPA Biosystems, Woburn, MA, USA). Paired-end sequencing with 250 bp reads was performed in MiSeq (Illumina, San Diego, CA, USA). Reads were then de novo-assembled using CLC Genomics Workbench 11.0.1 (CLC Bio, Aarhus, Denmark) for subsequent analyses.

Multilocus sequence typing for E. coli and K. pneumoniae were performed using EnteroBase (http://enterobase.warwick.ac.uk/) and the Institute Pasteur’s MLST database (http://bigsdb.pasteur.fr/klebsiella/klebsiella.html), respectively. Existing methods29 were followed for MLST of the E. cloacae complex. Resistance-gene identification was performed in ResFinder 2.130. Evolutionary relatedness of isolates was assessed using CSI Phylogeny 1.431, and iTOL was used to construct the phylogenetic tree32. To classify EcloIMEX, local BLAST searches were performed on sequences around blaIMI-1, which contain setB and yeiP genes.

To determine the blaNDM-harbouring plasmid size, pulsed-field electrophoresis (PFGE) plugs prepared from isolates were treated with S1 nuclease (Takara Bio, Shiga, Japan) and subjected to PFGE in the CHEF Mapper XA system (Bio-Rad, Hercules, CA, USA). Separated DNA was then transferred to a nylon membrane and probed with a digoxigenin-labelled (Roche Diagnostics, Basel, Switzerland), blaNDM-specific DNA probe for Southern hybridization.