Rapid Increase in Carbapenemase-Producing Enterobacteriaceae in Retail Meat Driven by the Spread of the bla_NDM-5-Carrying IncX3 Plasmid in China from 2016 to 2018

The presence and molecular characteristics of carbapenemase-producing Enterobacteriaceae (CPE) among meat products in China were investigated. A total of 110 carbapenem-resistant Enterobacteriaceae (CRE) isolates, including 94 Escherichia coli and 10 Klebsiella pneumoniae isolates, were identified from 105 of 794 (13.2%) samples. The positive rates markedly increased from 2016 (9.4%) to 2018 (22.2%). Only bla_NDM genes were detected; 79.1% of bla_NDM genes were carried by IncX3 plasmids.

ABSTRACT

The presence and molecular characteristics of carbapenemase-producing Enterobacteriaceae (CPE) among meat products in China were investigated. A total of 110 carbapenem-resistant Enterobacteriaceae (CRE) isolates, including 94 Escherichia coli and 10 Klebsiella pneumoniae isolates, were identified from 105 of 794 (13.2%) samples. The positive rates markedly increased from 2016 (9.4%) to 2018 (22.2%). Only bla_NDM genes were detected; 79.1% of bla_NDM genes were carried by IncX3 plasmids. Routine monitoring of carbapenemase-producing Enterobacteriaceae in the animal food supply is highly recommended.

TEXT

In the last decades, there has been a rapid increase of carbapenem-resistant Enterobacteriaceae (CRE) along with the threat of limited treatment options, which are mainly attributed to carbapenemase enzymes, especially, Klebsiella pneumoniae carbapenemase (KPC) and New Delhi metallo-β-lactamase (NDM) (1, 2). Before 2014, the prevalence of CRE remained low in animals and the environment, and the existence of carbapenemase producers in food products was unknown (1, 3). However, in the last few years, CRE have gradually appeared in animals and food (4, 5). In China, multiple studies have demonstrated the increasing occurrence of CRE, mainly, NDM-producing Enterobacteriaceae, in livestock, pets, and the environment. However, little is known about their prevalence or molecular characteristics in meat products (4, 6, 7). In this study, the prevalence and molecular epidemiological features of CRE in retail meat from food markets in Guangzhou were evaluated.

Between 2016 and 2018, 794 fresh retail meat samples (512 pork, 222 chicken, and 60 beef samples) were collected twice a year from farmer’s markets and supermarkets in 7 districts of Guangzhou, China (Table 1; see also Fig. S1 in the supplemental material). Carbapenem-resistant Enterobacteriaceae isolates were selected by MacConkey agar containing 1 μg/ml imipenem. Colonies with different morphologies were selected from each positive sample of carbapenemase genes (bla_NDM, bla_KPC, bla_IMP, bla_VIM, and bla_OXA-48-like) (8, 9), and only different species, identified by matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) or 16S rRNA sequencing, were retained. In total, 110 nonduplicate carbapenemase-producing Enterobacteriaceae isolates were recovered from 105 (13.2%) samples (64 pork, 38 chicken, and 3 beef samples) (Table 1), with results showing that 110 isolates harbored 98 bla_NDM-5, 11 bla_NDM-1, and 1 bla_NDM-7 gene, which was in accordance with studies on farm animals in China (4, 6, 7). The carriage rates of CRE in chicken samples (17.1% [38/222]) were higher than those in pork (12.5% [64/512]) and beef (5.0% [3/60]) samples. According to the annual isolation of pork and chicken samples, we found that the prevalence of CRE among chicken and pork samples increased significantly (P < 0.05) from 2016 to 2018 (9.4% [34/360] in 2016, 11.1% [15/135] in 2017, and 22.2% [53/239] in 2018). The high frequency of CRE in retail meat (13.2% [105/794]) and the increased prevalence in pork and chicken were worrying. Between 2013 and 2017, CRE was reported in pigs, chickens, and cows in China, with various detection rates (6.5% to 61.0%) (6, 7). Additionally, CRE were also detected in chickens and pigs at slaughterhouses (4). Data from the China Antimicrobial Surveillance Network (CHINET) revealed that the rate of carbapenem resistance in K. pneumoniae increased from 9.3% in 2011 to 20.9% in 2017, whereas in Escherichia coli, the rate was maintained at 1.0% to 1.9% over this period (10). Thus, it seemed that CRE detected in retail meat samples might mainly be attributable to contamination from livestock during slaughter but not from human activity and hospital waste.

TABLE 1.

Prevalence of carbapenemase-producing Enterobacteriaceae isolates by origin

Year(s)	Sample type	No. of samples	No. of carbapenemase-producing isolates	No. and type of carbapenemase gene(s)
2016	Pork	285	27 (23 E. coli, 3 K. pneumoniae, 1 E. aerogenes)	26 blaNDM-5, 1 blaNDM-1
	Chicken	75	7 E. coli	7 blaNDM-5
	Beef	36	3 (1 E. coli, 2 K. pneumoniae)	3 blaNDM-5
	Total	396	37 (31 E. coli, 5 K. pneumoniae, 1 E. aerogenes)	36 blaNDM-5, 1 blaNDM-1
2017	Pork	78	8 E. coli	8 blaNDM-5
	Chicken	57	7 E. coli	6 blaNDM-5, 1 blaNDM-1
	Beef	24	0	0
	Total	159	15 E. coli	14 blaNDM-5, 1 blaNDM-1
2018	Pork	149	30 (24 E. coli, 4 K. pneumoniae, 1 E. aerogenes, 1 Proteus mirabilis)	25 blaNDM-5, 5 blaNDM-1
	Chicken	90	28 (24 E. coli, 1 K. pneumoniae, 3 P. mirabilis)	23 blaNDM-5, 4 blaNDM-1, 1 blaNDM-7
	Total	239	58 (48 E. coli, 5 K. pneumoniae, 1 E. aerogenes, 4 P. mirabilis)	48 blaNDM-5, 9 blaNDM-1, 1 blaNDM-7
2016–2018	Pork	512	65 (55 E. coli, 7 K. pneumoniae, 2 E. aerogenes, 1 P. mirabilis)	59 blaNDM-5, 6 blaNDM-1
	Chicken	222	42 (38 E. coli, 1 K. pneumoniae, 3 P. mirabilis)	36 blaNDM-5, 5 blaNDM-1, 1 blaNDM-7
	Beef	60	3 (1 E. coli, 2 K. pneumoniae)	3 blaNDM-5
	Total	794	110 (94 E. coli, 10 K. pneumoniae, 2 E. aerogenes, 4 P. mirabilis)	98 blaNDM-5, 11 blaNDM-1, 1 blaNDM-7

According to the CLSI guidelines (11) and epidemiological cutoff (ECOFF) values or clinical breakpoints of EUCAST (http://www.eucast.org), all 110 isolates showed resistance to ampicillin, cefotaxime, ceftazidime, and cefoxitin (see Fig. S2). Resistance genes, including mcr-1 to mcr-8, rmtB, fosA3, floR, and bla_CTX-M, were screened by PCR and sequencing, as described previously (see Table S1). Twenty-one (19.1%) isolates coproduced CTX-M enzymes, and 94 (85.5%) carried the floR gene (see Table S2). In addition, 20, 16, and 3 isolates showed resistance to colistin, fosfomycin, and amikacin, respectively, while 16, 14, and 3 were positive for mcr-1, fosA3, and rmtB, respectively; no other mcr genes were detected. Interestingly, the mcr-1 positive rates of the 110 isolates decreased significantly (P < 0.05) from 2016 (24.3% [9/37]) to 2017 (13.3% [2/15]) and to 2018 (8.6% [5/58]). Furthermore, 103 isolates were sensitive to tigecycline (Table S2).

The 94 NDM-producing E. coli isolates belong to 51 distinct sequence types (STs). ST10 (n = 12) and ST48 (n = 13) were the most prevalent STs, followed by ST7111 (n = 4) (Fig. 1 and Table S2). A previous report pointed out the dominance of bla_NDM-positive E. coli ST48 clones in one pig farm in China (7). ST10, ST48, ST117, ST2732, ST5229, ST746, ST119, and ST93 (Table S2) were detected in multiple years. Pulsed-field gel electrophoresis (PFGE) analysis was performed on the same ST clone. In total, 10 and 8 different PFGE patterns were observed among 12 ST10 strains and 13 ST48 strains, respectively (see Fig. S3a and b). Among the 10 K. pneumoniae isolates carrying bla_NDM, 8 STs were detected, with ST1 (n = 3) being dominant, and had indistinguishable PFGE patterns (Fig. S3c). Thus, nonclonal dissemination played a vital role in the spread of NDM-producing Enterobacteriaceae strains from retail meat in China despite the occasional clonal spread. Most of the characterized STs were different from those of human sources in China. Yet, there were clones shared by isolates from humans and retail meat, such as E. coli ST10, ST167, ST410, ST641, ST744, and ST5229 and K. pneumoniae ST1, ST11, ST35, and ST45 (12–14).

FIG 1.

Minimum spanning tree based on multilocus sequence typing of the 95 bla_NDM-positive E. coli isolates. Each ST is displayed as a circle, and the size of the circle is proportional to the number of the strains belonging to each type.

Conjugation and transformation experiments were performed on 52 NDM producers collected in 2016 to 2017. All 52 bla_NDM plasmids were successfully transferred to recipients (E. coli C600^str or DH5α) by conjugation or transformation. PCR-based replicon typing was performed on transconjugants/transformants as previously described (15–17). The replicon types included IncX3 (n = 50), F2:A−:B− (n = 1), and untypeable (n = 1) (Table 1). For the 58 NDM-producing isolates collected in 2018, S1-PFGE and hybridization analyses confirmed that bla_NDM genes from 37 isolates were located on ∼50-kb plasmids and the others were located on 60- to 270-kb plasmids or chromosome. The ∼50-kb plasmids carrying bla_NDM were confirmed to be IncX3 type by using primers covering both the backbone of IncX3 plasmid and bla_NDM genes. Totally, among the 110 NDM producers, 87 (79.1%) bla_NDM genes were carried by IncX3 plasmids (Table S2), which was consistent with previous reports that the IncX3 plasmid was a major vector driving the transmission of bla_NDM (7, 12, 13, 18). IncX3 plasmid was first associated with the dissemination of bla_NDM-1 among clinical CRE strains (19). Nowadays, it has disseminated in 23 countries, mostly in Asian countries, and was harbored by diverse Enterobacteriaceae species (see Fig. S5) (13). The wide distribution of bla_NDM-carrying IncX3 plasmids might be explained by their ability to transfer among different bacterial species at a wide range of temperatures (30°C to 37°C) (13). However, further studies are needed to investigate the underlying mechanism and driving force.

Genomic DNA of 22 bla_NDM-positive isolates was extracted and subjected to sequencing using an Illumina HiSeq 2500 platform. To obtain the complete sequence of 9 IncX3 plasmids and analyze genetic backgrounds of other bla_NDM gens carried by IncX3, we designed customized primers (see Table S3) based on plasmid pNDM5_IncX3 (GenBank accession number KU761328.1), a 46,161-bp IncX3 plasmid from K. pneumoniae (20). BLAST analysis indicated that 9 IncX3 plasmids carrying bla_NDM-5 were highly similar (≥99%) to pNDM5_IncX3. Of note, the sequences of pHNYX638 (MK033577) showed 100% similarity and query coverage with pNDM5_IncX3 (see Fig. S4). pHNHZB05, pHNYX658, pHNHZ11, pHNYX644, pHNHZ18, and pHNYX667 were almost identical to pNDM5_IncX3, with only 1 to 5 nucleotide differences in plasmid backbones (pHNHZ18) or mobile modules. A total of 7 genetic contexts (type I to type VII) were found in 87 bla_NDM-carrying IncX3 plasmids, which were due to insertions, truncation, and/or deletions of mobile elements (Fig. 2; Table S2). Type I was the most common structure in this study (81.6% [71/87]) and the GenBank database, encoded by IncX3 plasmids from Enterobacteriaceae of human, pig, and chicken sources in China and other countries (Fig. 2; Table S2 and Fig. S5). Type III was found in 3 bla_NDM-1-carrying K. pneumoniae and Enterobacter aerogenes isolates and was identical to pCRENT-193_2 (CP024814) from Enterobacteriaceae of an inpatient in South Korea. Types II, IV, V, VI, and VII were all novel genetic structures found in this study and were unique in the GenBank database.

FIG 2.

Genetic contexts of bla_NDM in IncX3 plasmids. Regions of ≥99.0% nucleotide sequence identity are shaded gray.

A dramatic increase of CRE in retail meat samples was observed, driven by the prevailing bla_NDM-5-carrying IncX3 plasmid. The CRE and NDM-producing plasmids in retail meat might originate from livestock and endanger public health via direct contact, the food chain, and the environment. It is likely that bla_NDM-5-carrying IncX3 might further spread via international travel and trade, as observed in extended-spectrum β-lactamase (ESBL)-producing Enterobacteriaceae (21). From the one health perspective, more extensive surveillance and development of multisectoral strategies to control its dissemination are urgently needed.

Statistical analysis.

To compare the prevalence of CRE and mcr-1, a χ² test was performed using SPSS version 23.0. A P value of ≤0.05 was considered statistically significant.

Accession number(s).

The complete IncX3 plasmid sequences carrying bla_NDM-5 obtained in this study have been deposited in the GenBank database under the following accession numbers: MK033578, MK033584, MK033580, MK033583, MK033581, MK033579, MK033577, and MK088486.