Molecular Mechanisms and Epidemiology of Carbapenem-Resistant Escherichia coli Isolated from Chinese Patients During 2002–2017

Introduction

Escherichia coli is one of the most commonly isolated microorganisms in clinical specimens. Multidrug resistance in E. coli has become an upsetting issue observed in humans¹ and has been recognized as a contributor to the dissemination of antibiotic-resistance genes.² Controlling the dissemination of multidrug-resistant (MDR) strains is problematic due to very few new antibiotics available.^3,4 Because of increasing resistance to third-generation cephalosporins, fluoroquinolones and aminoglycosides, carbapenems have gradually become the last resort for life-threatening MDR E. coli infections because of their broad-spectrum antimicrobial agents.^5,6 Nevertheless, with an increasing consumption of carbapenems, the emergence of carbapenems resistant E. coli has become a serious public health concern worldwide.^7,8

The mechanisms of carbapenem resistance are strongly associated with carbapenemase production (acquisition of carbapenemase genes), combination of porin loss with extended-spectrum β-lactamases (ESBLs) and the overexpression of efflux pumps.^9,10 Several studies have reported that acquired carbapenemase isolates might cause hospital outbreaks and become endemic in healthcare settings.^11,12 Globally predominant carbapenemases include KPC, NDM, VIM, IMP, and OXA, which are encoded by bla_KPC, bla_NDM, bla_VIM, bla_IMP, and bla_OXA genes present in both the plasmid and the chromosome.^13,14 In addition, the carbapenemase genes could co-exist with ESBLs and other resistance genes on plasmids, which further limit the treatment options. Moreover, previous studies have reported that the outer membrane porins of E. coli are involved in the MDR phenotype.^15,16 Choi et al have constructed mutants of porins (ompC and ompF mutations) in E. coli and discovered that porins have a distinct role in antibiotic resistance and membrane integrity.¹⁷ With the increase in the prevalence of carbapenem-resistant E. coli strains worldwide,^18,19 longitudinal epidemiological surveillance and mechanisms research on the carbapenem-resistant E. coli are of great clinical significance for the global control and prevention of the distribution and spread of resistance, as well as the guidance on antibacterial treatment. Nonetheless, there is still a lack of data on the long-term evolution of carbapenem-resistant E. coli in China. In the present study, we characterized the epidemiology prevalence and molecular mechanisms of 58 E. coli clinical isolates during large-scale surveillance for carbapenem resistance in the southeast of China.

Materials and Methods

Bacterial Isolates

A total of 5796 E. coli clinical isolates were collected from the First Affiliated Hospital of Wenzhou Medical University (Wenzhou, China) between 2002 and 2017. Identification of all isolates was performed using a VITEK^®2 system (bioMérieux, Marcy-l’Étoile, France). After collection, isolates were stored in 30% glycerol at –80°C. Relevant clinical data were collected from the medical records. We collected the information about isolation date, age, gender, sample, and ward.

Minimum Inhibitory Concentration Determination

MICs of 12 antimicrobial agents, including imipenem, meropenem, ertapenem, ampicillin, ceftriaxone, ceftazidime, ciprofloxacin, levofloxacin, gentamicin, tobramycin, amikacin, and fosfomycin, were determined by the agar dilution method according to the guidelines recommended by the latest Clinical and Laboratory Standards Institute (CLSI).²⁰ Colistin MIC determination was performed with broth microdilution and interpreted by the recommendation of the European Committee on Antimicrobial Susceptibility Testing clinical breakpoints (http://www.eucast.org/). E. coli ATCC 25922 was used as the control strain for antimicrobial susceptibility testing.

Detection of Carbapenemases and Antibiotic Resistance Determinants

The modified carbapenem inactivation method (mCIM) was used to screen isolates for the production of carbapenemases, according to CLSI guidelines. The presence of resistant mechanisms, including carbapenem resistance genes (bla_KPC-2, bla_NDM, bla_IMP, bla_VIM, and bla_OXA-48), ESBLs genes (bla_SHV, bla_TEM, bla_CTX-M-1, and bla_CTX-M-9), outer membrane porins genes (ompC and ompF), fosfomycin resistance genes (fosA3 and fosA) and colistin resistance genes (mcr-1 and mcr-3) were identified by polymerase chain reaction (PCR) and sequencing. Each isolate DNA was extracted from fresh bacterial colonies using a Biospin Bacterial Genomic DNA Extraction kit (Bioer Technology, Hangzhou, China). The primers used for amplification and sequencing were listed in Table S1. Positive amplification products were sent to Shanghai BGI Technology Co. (Shanghai, China) for sequencing. Nucleotide sequences were compared by BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi). The online PROVEAN platform (http://provean.jcvi.org/seq_submit.php) was used to predict alterations in the biological function of the proteins.

Effect of Efflux Pump Mechanism on Carbapenem-Resistance in E. coli

Carbonyl cyanide m-chlorophenylhydrazone (CCCP) is an energy uncoupler, has been identified as a compound to reverse MDR in E. coli over-expressing efflux pumps.²¹ CCCP (Sigma, St Louis, MO) was used to measure the activity of efflux pumps in carbapenem-resistant E. coli isolates. The change in the MICs of carbapenems was determined by the agar dilution method in the absence or presence of 10 μg/mL CCCP. A phenotype for positive efflux was defined as a ≧4-fold reduction of the carbapenem MIC in the presence of CCCP.²²

Molecular Epidemiology Analysis

MLST analyses of the carbapenem-resistant isolates were carried out by amplifying eight housekeeping genes (dinB, icdA, pabB, polB, putP, trpA, trpB, and uidA). Sequence types were assigned by querying against the database available at the Institut Pasteur’s E. coli MLST website (http://bigsdb.web.pasteur.fr/ecoli/ecoli.html).

To further identify potential clonal spread, PFGE was performed using a CHEF-Mapper XA PFGE system (Bio-Rad, Hercules, CA). Briefly, genomic DNA was extracted from all tested isolates, followed by Xba I restriction enzyme (Takara Bio, Inc., Kusatsu, Japan) digestion. Electrophoresis was then performed under the following conditions: temperature, 14°C; voltage, 6 V/cm; pulse angle 120°; and pulse duration, 2.16–54.17 s for 18.5 hrs. The universal standard strain Salmonella enterica serotype H9812 was used as a molecular marker.²³ Band patterns were analyzed and interpreted according to the criteria proposed by Tenover et al.²⁴

Results

Bacterial Strains and Antimicrobial Susceptibility Testing

A total of 58 (1.0%) carbapenem-resistant E. coli isolates were identified with carbapenems (including imipenem, meropenem, and ertapenem), MICs ranging from 2 μg/mL to ≧16 μg/mL. Carbapenem-resistant E. coli isolates at our hospital were first detected in 2012; after that, the resistance rate has increased from 0.85% to 1.85% as was detected in 2017 (Table 1). Table 2 summarized the patient characteristics and species distribution. Overall, the carbapenem-resistant organisms were mainly from urine samples (31.0%, 18/58), followed by blood (27.6%, 16/58) and drainage (19.0%, 11/58). There were more isolates from males than females (62.1% vs 37.9%, respectively). Isolates were cultured from patients aged 19 to 91 years (average age 62.5 years). The majority of the isolates were from patients in the intensive care unit (ICU) (31.0%, 18/58), Hepatobiliary Surgery (10.3%, 6/58). The antimicrobial resistance profiles of the 58 carbapenem-resistant isolates were summarized in Table 3. According to the results of antimicrobial susceptibility testing, all 58 isolates showed higher resistance rates to cephalosporins, fluoroquinolones, and aminoglycosides. Thereinto, 55 (94.8%) isolates were resistant to fluoroquinolones, including levofloxacin and ciprofloxacin; 49 (84.5%) isolates were resistant to aminoglycosides, including gentamicin, tobramycin and amikacin. Furthermore, 58 (100%) isolates were resistant to ampicillin, while 18 (31.0%) isolates were resistant to fosfomycin and 2 (3.4%) to colistin.

Table 1 Carbapenems Susceptibility of E. coli Clinical Isolates

Table 2 Patient’s Clinical Data and Characteristics of Analyzed Strains

Table 3 Minimum Inhibitory Concentrations (MICs) of 58 Carbapenem-Resistant E. coli Isolates

Prevalence of β-Lactamase Genes

Forty-three carbapenem-resistant E. coli isolates produced carbapenemases (Figure S1 and Table S2). The prevalence rates of bla_KPC-2 and bla_NDM in carbapenem-resistant isolates were 22.4% and 51.7%, respectively (Figure 1), while bla_IMP, bla_VIM and bla_OXA-48 were not detected. In addition, the number of isolates harbored bla_NDM-1 or bla_NDM-5 were 3 (5.2%) and 27 (46.6%), respectively. Moreover, the most prevalent CTX-M-type among analyzed strains was bla_CTX-M-1 (75.9%, 44/58), followed by bla_CTX-M-9 (65.5%, 38/58). In general, the CTX-M-type β-lactamase genes accounted for 96.6% (56/58). In addition to bla_CTX-M genes, bla_TEM-1 genes were also identified in 58.6% of tested strains, bla_SHV was not detected.

Figure 1 Antibiotic resistant mechanisms detected in the E. coli strains that were sequenced as part of this study. (A) Modified carbapenem inactivation method (mCIM) for phenotypic detection of carbapenemase production; (B) outer membrane porins genes; (C) carbapenem resistance genes; (D) β-lactam resistance genes; (E) carbonyl cyanide m-chlorophenylhydrazone (CCCP) was used to measure the activity of efflux pumps in carbapenem-resistant E. coli isolates. Black squares represent positive, gray squares represent negative.

Detection of Mutations in ompC and ompF

Mutations in ompC and ompF genes were detected in carbapenem-resistant isolates, including amino acid substitutions and deletions. Deleterious mutations of OmpC and OmpF occurred in 22 and 21 isolates, respectively. Moreover, four carbapenem-resistant isolates had mutations in both OmpC and OmpF. The majority of mutation sites in ompC were D192G followed by Q104_F141del. Notably, several large deletions (S75_V127del, W83_D135del and Q88_D135del) of an amino acid sequence encoded by the ompF gene were also detected. Amino acid substitutions in ompC and ompF were considered deleterious by PROVEAN (Tables 4 and 5).

Table 4 Mutations in Carbapenem-Resistant E. coli Isolates

Table 5 Analysis of Mutations in ompC and ompF

Phenotypic Detection of the Efflux Pump Overexpression

The effect of efflux pumps on the antibiotic resistance profiles of isolates was examined using the efflux pump inhibitor CCCP. When exposed to 10 μg/mL CCCP, none of the isolates showed a ≧4-fold decrease in the carbapenem MIC, suggesting that the antibiotic resistance was not associated with overexpression of efflux pump in this study.

Epidemiological Characterization

MLST analysis assigned the 58 carbapenem-resistant isolates into 19 different sequence types (STs) (Figure 2). ST8 was the predominant ST, accounting for 29.3% (17/58), followed by ST19 (12.1%, 7/58) and ST692 (12.1%, 7/58). Moreover, there were two novel STs (labelled as “New” in Figure 2; currently not registered in the MLST database). PFGE analysis revealed that homology among the resistant isolates was low and sporadic, suggesting a very low likelihood of clonal spread (Figure 2).

Figure 2 PFGE profiles of Xba I-digested chromosomal DNAs of carbapenem-resistant E. coli isolates. Relatedness was analyzed using QualityOne software (Bio-Rad Laboratories, USA). The phylogenetic tree was generated using UPGMA clustering. A genetic similarity index scale is indicated by the vertical line.

Discussion

Carbapenems are extensively applied in clinical settings for the therapeutic management of MDR Gram-negative bacterial infections due to their broad spectrum of antimicrobial activity.²⁵ Yet, several surveillance programs have reported a highly increasing carbapenem resistance, making clinical treatment more challenging.^26,27 In the current study, 58 of 5796 E. coli isolates exhibited an increasing carbapenem-resistant rate from 2002 to 2017. The relatively higher incidence revealed that the ongoing surveillance is urgently warranted in China.

From the clinical perspective, there have been reports of transmission of E. coli in the ICU,^28,29 and clinicians should be vigilant about the potential presence of this species. Our study also confirmed that carbapenem-resistant strains were most commonly isolated from patients aged >65 years who were treated in the ICU. The KPC-type enzyme was first reported in Klebsiella pneumoniae from the southern United States in 2001³⁰ and now endemic all over the world.^31,32 In China, dissemination of KPC-producing Enterobacteriaceae spp. has been confirmed in Shandong, Zhejiang, Taiwan, and other provinces.^33–35 KPC-2 was the most important in K. pneumoniae, whereas NDM-1 was the most important in E. coli. Notably, in previous studies in China, a few strains of E. coli with KPC-2 were detected.³⁶ However, in our study, 22.4% (13/58) of the strains were detected with KPC-2. This finding suggested that more attention should be paid to the spread of KPC-2 in this region. The IMP and VIM genes were reported in several regions, OXA-48 was more common in Europe but had not been found in our study.³⁷ New Delhi metallo-β-lactamase (NDM), which was first reported in Sweden in 2009 in a patient who developed an infection while travelling in India,³⁸ could confer resistance to most β-lactams. Over the recent years, a high prevalence of NDM-1 has been observed in China and India.^39,40 In addition, the rapid global spread of NDM-producing isolates via MDR plasmids has led many into thinking that common infections with such strains may soon be untreatable.⁴¹ Selective pressure caused by increased use of antibiotics may drive the evolution of NDM-1, thus resulting in the emergence of its variants. In the current study, the emergence of NDM-5 reflected a new prevalence since 2017. M154L amino acid substitution in NDM-5 was the most common substitution in all NDMs variants,⁴² responsible for increased carbapenemase activity. Moreover, NDM-5 has an extra V88L substitution; the emergence of V88L may contribute to lower catalytic activity on imipenem and meropenem.^43,44 Although NDM-5 made anti-infective treatment more difficult,⁴⁵ the lower hydrolytic activity of imipenem and meropenem implied these were still the first choice for MDR E. coli isolates. Our study indicated an increased number of carbapenemases-producing E. coli isolates over the last few years. It also revealed the high incidence of bla_NDM since it was first discovered at the hospital between 2015 and 2017. Interestingly, our results revealed that NDM-5 may even replace the NDM-1 in carbapenem-resistant E. coli isolates from 2017 in China. To date, several studies showed that bla_NDM-5 was carried by IncX3 plasmids in China,^46,47 India,⁴⁸ Denmark⁴⁹ and Australia.⁵⁰ The fact that IncX-type plasmids have been shown to be conjugatable in most studies could explain the rapid spread of bla_NDM-5-carrying isolates. Therefore, it is imperative that feasible and effective measures are taken immediately.

ESBL-producing E. coli showed higher health risks related to hospital-acquired infections compared to non-ESBL-producing isolates.⁵¹ CTX-M β-lactamases are the most widespread types of ESBLs, which have been identified in the mid-2000s in clinical E. coli isolates.⁵² In this study, 96.6% of ESBL genes were classified as bla_CTX-M. Several reports have indicated that the transfer of CTX-M mobile plasmids could be frequently accompanied by the acquisition of fosfomycin resistance genes.^53,54 In our study, 17 carbapenem-resistant strains harboring CTX-M plasmids were positive for the fosA3 gene. Colistin resistance represents another health concern. Two colistin-resistant E. coli strains detected in our study carried mcr-1 gene. Moreover, co-harboring of bla_NDM-1, fosA3, and mcr-1 were detected in DC-3737, like a reservoir, which posed serious concern on public health.

It has been reported that resistance to carbapenems could be mediated by non-specific outer membrane porins OmpC and OmpF in E. coli.¹⁷ In the current study, the deleterious mutations were detected in 39 isolates, whereas OmpC and OmpF alteration occurred in 22 and 21 isolates, respectively. Mutation prediction showed that the amino acid substitutions in ompC, such as D192G might be the key factor driving resistance to carbapenems, while amino acid deletions could make an important impact in ompF mutations. The mutations in OmpF and OmpC were the important mechanisms contributing to the elevated MICs to carbapenems.

All of the isolates (100%, 58/58) were ertapenem non-susceptible; however, the abundance of imipenem-resistant strains was relatively smaller, promising the suitability of imipenem as the choice of treatment for infections caused by ertapenem-non-susceptible E. coli isolates. Furthermore, the alteration of outer membrane porins combined with carbapenemase production were found in 39 isolates, which further decreased the sensitivity of imipenem and meropenem. Otherwise, it is worth noting that the carbapenem resistance mechanism of DC-38 still remains unclear, and needs to be further researched in the future.

So far, few studies have reported the effect of efflux pump on carbapenems resistance in Enterobacter spp.^55–57 The current study showed that the efflux pump inhibitor CCCP was not able to restore the susceptibility of carbapenem-resistant E. coli, indicating that efflux pump was not involved in the carbapenem resistance in our study.

Our analysis showed that the majority of carbapenem-resistant clinical E. coli isolates showed different PFGE patterns, suggesting that they were genetically unrelated. The results of MLST demonstrated that these carbapenem-resistant isolates were polyclonal without a clonal dissemination. We speculated that carbapenem-resistant E. coli isolates might originate from different lineages and sources, instead of expansion of a single clonal lineage, which is in line with previous reports.⁵⁸ Among them, ST8 was the main clone type (29.3%, 17/58). Interestingly, 76.9% (10/13) KPC-2-producing E. coli isolates belonged to ST8 in our study, indicated that a high prevalence of bla_KPC-2 was linked with ST8. We hypothesized that ST8 had a better ability to capture or accumulate bla_KPC compared with the other types. Furthermore, both STs ST19 and ST692 were present in association with the bla_NDM-5 gene, which was firstly reported to be linked with NDM-5- producing isolates.

In summary, we described the resistance mechanisms and the molecular epidemiology of carbapenem-resistant E. coli isolates at the First Affiliated Hospital of Wenzhou Medical University between 2002 and 2017. To best of our knowledge, this is the first report on the long duration and large scale of carbapenem-resistant E. coli isolates in China. Due to the limited treatment options, the rising resistance rate has further exacerbated the threat to public health. The prevalence of variant bla_NDM-5 represents a new threat. Moreover, ESBLs genes have shown to have a significant role in the carbapenem-resistant E. coli isolates, among which, CTX-M-type ESBLs were prevalent. As carbapenems are becoming ever more used as an effective therapeutic option, monitoring programs are urgently required to prevent the emergence and further spread of its resistance.