ORIGINAL RESEARCH article

Front. Microbiol., 05 December 2019

Sec. Evolutionary and Genomic Microbiology

Volume 10 - 2019 | https://doi.org/10.3389/fmicb.2019.02826

Diverse Fluoroquinolone Resistance Plasmids From Retail Meat E. coli in the United States

  • Office of Research, Center for Veterinary Medicine, U.S. Food and Drug Administration, Laurel, MD, United States

Abstract

Fluoroquinolones are used to treat serious bacterial infections, including those caused by Escherichia coli and Salmonella enterica. The emergence of plasmid-mediated quinolone resistance (PMQR) represent a new challenge to the successful treatment of Gram-negative infections. As part of a long-term strategy to generate a reference database of closed plasmids from antimicrobial resistant foodborne bacteria, we performed long-read sequencing of 11 E. coli isolates from retail meats that were non-susceptible to ciprofloxacin. Each of the isolates had PMQR genes, including qnrA1, qnrS1, and qnrB19. The four qnrB19 genes were carried on two distinct ColE-type plasmids among isolates from pork chop and ground turkey and were identical to plasmids previously identified in Salmonella. Seven other plasmids differed from any other sequences in GenBank and comprised IncF and IncR plasmids that ranged in size from 48 to 180 kb. These plasmids also contained different combinations of resistance genes, including those conferring resistance to beta-lactams, macrolides, sulfonamides, tetracycline, and heavy metals. Although relatively few isolates have PMQR genes, the identification of diverse plasmids in multiple retail meat sources suggests the potential for further spread of fluoroquinolone resistance, including through co-selection. These results highlight the value of long-read sequencing in characterizing antimicrobial resistance genes of public health concern.

Introduction

Fluoroquinolones such as ciprofloxacin are critically important antimicrobials in human medicine. They are used to treat serious bacterial infections caused by Gram-negative and Gram-positive organisms (; ). Thus, fluoroquinolone resistance is a public health issue that can lead to treatment failures and the use of alternative agents with greater side effects.

Fluoroquinolones are bactericidal antimicrobials that disrupt DNA replication in bacterial cells. Fluoroquinolone resistance is often mediated by mutations in the DNA gyrase and topoisomerase genes, with gyrA mutations being the most common mechanism in Gram-negative bacteria (). In recent years, plasmid-mediated quinolone resistance (PMQR) has become more frequent and can spread resistance through horizontal gene transmission. Known genes include qnr, aac(6′)-Ib-cr, qepA, and oqxAB (). There are several types of qnr genes, including qnrA, qnrB, qnrC, qnrD, qnrE, qnrS, and qnrVC, which together have over one hundred named alleles (). The presence of these genes differs by bacterial species, with qnrA, qnrB, and qnrS being most common among Salmonella enterica and Escherichia coli in the United States ().

Although PMQR often results in only low-level fluoroquinolone resistance, this can then result in continued selection for gyrA mutants with even higher levels of resistance (). Transmissible fluoroquinolone resistance is significant since it can lead to the rapid spread of resistance in bacterial species of public health importance. Cross-species and cross-genus transfer of resistance determinants is also possible. For example, qnrB genes are thought to have originated in Citrobacter spp. (), but have now been found in many pathogens, including E. coli, S. enterica, and Klebsiella pneumoniae, among others (; ; ).

The National Antimicrobial Resistance Monitoring System (NARMS) is a One Health surveillance system in the United States that monitors antimicrobial resistance of foodborne pathogens from food animals, retail meats, and humans. Previous NARMS surveillance has found that the prevalence of PMQR genes in the food supply in the United States is low (), although ColE plasmids carrying qnrB19 have been found in Salmonella from swine sources ().

Plasmids containing multiple resistance genes can also co-select for resistance, as the use of one antimicrobial can select for resistance to additional drug classes (). Thus, it is important to identify and characterize resistance plasmids, particularly those conferring fluoroquinolone resistance.

In this paper, we report the use of long-read sequencing to characterize PMQR plasmids among E. coli isolated from retail meats in the United States, showing the diversity of mechanisms and potential co-selection.

Materials and Methods

Bacterial Strains

Escherichia coli strains were isolated from routine NARMS retail meat sampling from 13 states in 2015–2017 (). Antimicrobial susceptibility testing was performed per CLSI methods, with ciprofloxacin non-susceptibility defined per CLSI M100-S29 as MICs ≥ 0.125 μg/mL ().

Sequencing and Assembly

Genomic DNA was extracted with DNeasy Blood and Tissue Kits (Qiagen, Valencia, CA, United States) per the manufacturer’s instructions. Whole-genome sequencing was performed on the Pacific Biosciences (PacBio) Sequel Sequencer, as previously described (). Continuous long reads were assembled by the PacBio Hierarchical Genome Assembly Process (HGAP4.8) program. Assembly of the qnrB19 plasmids was done using CLC Genomics Workbench version 10.0.

Annotation and Sequencing Analysis

The closed genomes were annotated by the Prokaryotic Genome Annotation Pipeline (PGAP) version 4.8 (). Resistance genes were identified by the NCBI Pathogen Detection Pipeline by AMRFinder (). BLASTn was also used to compare plasmids identified with those in GenBank. Plasmid typing was determined by PlasmidFinder, comparing with the database to known plasmid types with 95% sequence identity and 60% sequence length (). Multi-locus sequence typing (MLST) was done with assembled sequences using MLST 2.0 (), using the E. coli scheme previously described (). Sequences were submitted to GenBank, with BioSamples in Table 1 and plasmid nucleotide accession numbers listed in Table 2.

TABLE 1

Isolate IDSourceYearStateSTCIP MIC (μg/mL)PMQR mechanismBioSample
N55972Pork chop2015GA100.25qnrB19SAMN12698087
N56338Pork chop2015GA10790.5qnrB19SAMN12698088
N56639Ground beef2015GA5180∗0.5qnrA1SAMN12698089
N62675Ground turkey2015GA3980.25qnrS1SAMN12698090
N16EC0140Pork chop2016OR130.25qnrB19SAMN12698091
N16EC0879Ground turkey2016TN580.25qnrS1SAMN12698092
N17EC0211Ground turkey2017IA5400.12qnrA1SAMN10221061
N17EC0320Ground turkey2017OR5400.12qnrA1SAMN10221115
N17EC0326Ground turkey2017OR100.25qnrB19SAMN10221118
N17EC0616Chicken leg2017CO14850.12qnrS1SAMN10221255
N17EC1164Pork chop2017TX22070.25qnrS1SAMN10221523

Metadata for isolates with PMQR genes.

∗Closest ST type to that of this isolate, as its fumC allele differs from that of ST5180.

TABLE 2

Isolate IDPlasmid typePlasmid lengthAccessionPlasmid resistance genes
P3_N55972ColE3,071CP043760qnrB19
P2_N56338ColE3,071CP043756qnrB19
P1_N56639IncR48,263CP043753qnrA1 aadA2 blaCARB–1mphA floR sul1 (3x) tetA dfrA1
P1_N62675IncR81,916CP043751qnrS1 aadA2 dfrA12 sul3 tetA
P2_N16EC0140ColE3,071CP043749qnrB19
P1_N16EC0879IncF138,918CP043745qnrS1 blaCTX–M–55tetA aac(3)-IId
P1_N17EC0211IncF125,644CP043743qnrA1 blaCARB–2aadA2 (2x) dfrA12 ant(3″)-Ia cmlA1 sul1 sul3 tetA
P2_N17EC0320IncF126,972CP043741qnrA1 blaCARB–2aadA2 (2x) dfrA12 ant(3″)-Ia cmlA1 sul1 sul3 tetA
P4_N17EC0326ColE2,699ROAP02000006qnrB19
P1_N17EC0616IncF179,651CP043737qnrS1 blaTEM–1BstrA strB sul2 dfrA14 tetA
P1_N17EC1164IncR101,987CP043734qnrS1 blaTEM–1BtetA sul3 ant(3″)-Ia cmlA1 aadA2 dfrA12 tet(M)

Characteristics of PMQR plasmids.

Results

From 2015 to 2017, NARMS recovered 3,267 E. coli isolates from retail meat sampling. We performed short-read sequencing on approximately 1,500 of these isolates to identify resistance mechanisms. Most isolates with fluoroquinolone resistance mechanisms had gyrA mutations, comprising 42 isolates. Another 11 isolates lacked gyrase mutations but carried PMQR genes, including qnrA1, qnrS1, and qnrB19. Isolates with these genes were from a variety of sources, including retail chicken, turkey, beef, and pork (Table 1). There were no isolates with both PMQR genes and gyrA mutations.

To further characterize the isolates with PMQR genes, we performed long-read sequencing using Pacific Biosciences technology. From this sequencing, we obtained closed, circular chromosomes and plasmids from each isolate.

Four of the isolates possessed qnrB19 genes, which were found on small plasmids as expected. Long-read sequencing is not optimal for plasmids under 10 kb, so plasmids were closed using short-read sequencing data. We identified two distinct ColE-type plasmids of approximately 3 kb (Table 2) containing these genes. Interestingly, one of the isolates was from ground turkey, and had a different plasmid from the other three isolates, which were from pork chop. The four isolates were genetically distinct, comprising three different E. coli sequence types (STs) (Table 1). The two isolates that were ST10 were from different sources and not within 50 single-nucleotide polymorphisms (SNPs) of each other in the NCBI Isolates Browser (). The two plasmids we found were identical to those identified in swine sources of Salmonella in the United States () and have also been found in E. coli and Salmonella in South America ().

Three isolates had plasmids containing qnrA genes, with two isolates from ground turkey and one from ground beef. The ground turkey isolates were both identified as ST540 from in silico MLST (), and were 15 SNPs away from each other according to the NCBI Isolates Browser (). These two isolates had nearly identical PMQR plasmids of approximately 126 kb each, indicating likely clonal spread of this strain and its plasmid. These IncF plasmids had limited homology to known plasmids, and contained genes conferring resistance to beta-lactams, aminoglycosides, phenicols, sulfonamides, and tetracycline (Table 2). A graphical representation of one of these plasmids is shown in Figure 1. Interestingly, this plasmid also contains the iroN, iroBCDE, and sitABCD genes, which are involved in iron uptake and may contribute to virulence (; ). Furthermore, the plasmid contains the copB gene, which exports copper and confers copper resistance (), as well as the mer operon, which encodes mercury resistance (). Thus, the presence of copper or mercury could co-select for fluoroquinolone resistance.

FIGURE 1

The qnrA plasmid from the ground beef isolate comprised a 48-kb IncR replicon (Figure 2) with limited similarity to known plasmids. This plasmid had additional genes conferring resistance to beta-lactams, macrolides, phenicols, and sulfonamides. Since fluoroquinolones, beta-lactams, and macrolides are some of the most important antimicrobials used to treat serious Gram-negative infections, potential transfer of this plasmid to other pathogens could compromise the effectiveness of multiple potential therapies. The plasmid also contained the vapBC toxin–antitoxin system, which plays a role in greater plasmid stability () that may help with plasmid persistence. This toxin–antitoxin system was also present in the IncF plasmids containing qnrA.

FIGURE 2

Four isolates had qnrS on unique IncF and IncR plasmids, with sizes from 81 to 180 kb (Table 2). These four isolates included those from retail pork, chicken, and turkey. One of these was an IncF plasmid from N16EC0879 that also had blaCTX–M–55, tetA, and aac(3)-IId (Figure 3). The presence of blaCTX–M–55 is noteworthy, as this is an extended-spectrum beta-lactamase gene that confers resistance to cephalosporins. Since cephalosporins such as ceftriaxone are used in human medicine, the combined transfer of both blaCTX–M–55 and qnrS1 on one plasmid could compromise therapy to multiple drug classes. While all four resistance genes were within a 15-kb fragment on the plasmid, only qnrS1 and aac(3)-IId were on an insertion sequence together, an IS2 element. This plasmid also contained the iron uptake genes sitABCD and iucAC (), in addition to macAB, which may contribute to virulence (; ). This plasmid also had the copB copper resistance gene, as described in the IncF plasmids with qnrA. The two isolates with qnrS on IncR plasmids also had the sil operon genes, which confer silver resistance (Table 2; ).

FIGURE 3

Only a subset of the PMQR plasmids had identifiable conjugal transfer genes, such as those in the tra locus (). In fact, only the IncF plasmids had known conjugal transfer genes, indicating that the other plasmids may not be self-transmissible, or may require helper plasmids for successful spread through conjugation.

Discussion

We report detailed sequence data on fluoroquinolone non-susceptible E. coli from retail meats. This work shows the value of long-read sequencing in de novo characterization of AMR plasmids. Using only short-read sequencing data makes it difficult to accurately identify plasmids or fully characterize them (). Using short-read sequencing data alone we have accurately identified resistance genes in E. coli (), but not which are co-located on plasmids. In addition, assemblies from short-read data have difficulty identifying multiple copies of the same gene. For instance, some plasmids in this study had multiple copies of aadA and sul genes, but short-read data assemblies typically only identify single copies of resistance genes (; ). Understanding the full plasmid structure also helps uncover potential co-selection, including to heavy metals and other biocides. These details are important in assessing the nature of resistant microbial hazards in food and other sources. Fluoroquinolone use is relatively low in food animal production in the United States, but most PMQR plasmids also had genes conferring resistance to tetracycline, which is the highest selling antimicrobial for food animals in the United States (). This means that tetracycline use in food animals could result in continued selection for fluoroquinolone resistance in E. coli.

Interestingly, E. coli from all retail meat food types had PMQR genes. This contrasts with prior findings of swine as the major contributor to PMQR in retail meat Salmonella (). Most plasmids in this study had minimal homology to known plasmids. This demonstrates the importance of increased sequencing of plasmids even in well-studied bacteria such as E. coli, since completely new plasmids are still being discovered.

One interesting finding was that some E. coli had the same qnrB19 plasmids as those in Salmonella from swine and retail pork from NARMS sampling (). This reflects a likely transmission of plasmids between E. coli and Salmonella, including in non-swine sources. As observed in Salmonella, these isolates were diverse, indicating little serotype-specific barriers to transmission. These plasmids are small and likely not self-transmissible due to the lack of conjugation genes; each isolate had at least one additional plasmid, indicating a potential method for qnrB19 plasmid transmission.

Bacteria from this study were generic E. coli unlikely to cause foodborne disease in humans, so the direct risk of these bacteria impacting human health or treatment is low. Furthermore, none of the isolates were within 50 SNPs of any human isolates in the NCBI Isolates Browser. Nevertheless, as a source of resistance of human health concern, these bacteria could transfer resistance plasmids to pathogenic E. coli or to other genera. The tracking and reporting of PMQR in these bacteria is essential for a One Health strategy to identify emerging public health threats, and is enhanced by long-read sequencing for de novo characterization of novel plasmids.

Disclaimer

The views expressed in this article are those of the authors and do not necessarily reflect the official policy of the Department of Health and Human Services, the U.S. Food and Drug Administration, or the U.S. Government. Reference to any commercial materials, equipment, or process does not in any way constitute approval, endorsement, or recommendation by the U.S. Food and Drug Administration.

Statements

Data availability statement

The datasets generated for this study can be found in the NCBI BioSamples SAMN12587179, SAMN12587180, SAMN12587181, SAMN12587182, SAMN12587183, SAMN12587184, SAMN102 21061, SAMN10221115, SAMN10221118, SAMN10221255, and SAMN10221523.

Author contributions

GT conceived and coordinated the study. GT and PM wrote the manuscript. CL performed the sequencing and did the sequencing analysis. C-HH did the sequencing analysis and made the figures. SB-J performed the antimicrobial susceptibility testing. All authors contributed to finalizing the manuscript.

Funding

This work was funded by the U.S. Food and Drug Administration as part of routine work.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

References

Summary

Keywords

PacBio, antimicrobial resistance, Escherichia coli, plasmids, fluoroquinolones

Citation

Tyson GH, Li C, Hsu C-H, Bodeis-Jones S and McDermott PF (2019) Diverse Fluoroquinolone Resistance Plasmids From Retail Meat E. coli in the United States. Front. Microbiol. 10:2826. doi: 10.3389/fmicb.2019.02826

Received

04 September 2019

Accepted

21 November 2019

Published

05 December 2019

Volume

10 - 2019

Edited by

Ludmila Chistoserdova, University of Washington, United States

Reviewed by

Liang Li, The Lundquist Institute, United States; Xiang-Dang Du, Henan Agricultural University, China

Updates

Copyright

*Correspondence: Gregory H. Tyson,

This article was submitted to Evolutionary and Genomic Microbiology, a section of the journal Frontiers in Microbiology

Disclaimer

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.

Outline

Figures

Cite article

Copy to clipboard


Export citation file


Share article

Article metrics