Abstract
Salmonella enterica is a foodborne pathogen showing increasing multidrug resistance (MDR). We characterized the antimicrobial resistance (AMR) genotype using microarrays in a panel of 105 nontyphoidal S. enterica isolated from food animals and foodstuff. Nineteen isolates were chosen on the basis of their MDR and virulence for determination of heavy metal susceptibilities and screened by polymerase chain reaction for heavy metal resistance genes. Whole-genome sequencing (WGS) was performed on three isolates carrying clinically important AMR genes and the cdtB toxin gene to detect other heavy metal resistance mechanisms, and conjugation assays were performed to evaluate transfer of AMR/toxin genes with heavy metal resistance genes. AMR genotyping results showed isolates harbored between 1 and 12 mobile AMR genes, with 58% being classified as MDR. The tested subset of isolates showed reduced susceptibility to zinc (78%), copper (68%), silver (63%), arsenic (47%), and tellurite (26%); phenotypes that could be attributed to zitB (n = 32%), pcoA/pcoD (n = 32%), tcrB (n = 16%), arsB (n = 16%), silA/silE (n = 42%), and terF (n = 26%) genes. WGS confirmed the presence of other heavy metal resistance genes such as copA, cusA, and czcD. Isolates often harbored multiple heavy metal resistance genes. Two strains (Sal25 and Sal368) were able to conjugate with Escherichia coli J53 at a relatively high frequency (∼10−4 colony-forming units per recipient). Transformants selected in the presence of copper harbored either an IncHI2 (J53/Sal25 transconjugant) or IncF (J53/Sal368 transconjugant) plasmid with decreased susceptibilities to tellurite, zinc, copper, cobalt, arsenic, lead, mercury, and silver. bla CTX-M-1 and mcr-1 genes were also transferred to one transconjugant, and tet(M) and bla TEM-1 genes to the other. This work shows the presence of a diversity of AMR genes in this zoonotic pathogen, and suggests that heavy metals may contribute to selection of clinically important ones through the food chain, such as the plasmid-mediated colistin resistance gene mcr-1.
Introduction
S
Selective pressure due to substantive use of antibiotics in food-producing animals has been implicated in maintenance of AMR determinants in S. enterica. However, in European Union countries, the use of antibiotics in terrestrial feed animal as growth promoters was banned in 2006. Certain feed additives classified as coccidiostats are permitted. In addition, heavy metals such as copper, cobalt, and zinc are commonly added at high concentrations to animal feed to promote growth and health; silver and copper are appropriate for use as disinfectants, antiseptics, or preservatives, whereas mercury, lead, tellurite, cadmium, and arsenic can be found as contaminants in animal feed (Seiler and Berendonk, 2012); (Commission Regulation 1831/2003/EC, 2003).
There has been suggestion that the use of heavy metals as a feed supplement may promote the spread of antibiotic-resistant bacteria through coselection (Seiler and Berendonk, 2012; Wales and Davies, 2015; Sharma et al., 2016). Indeed, an emerging MDR S. enterica serotype 4,[5],12:i:- harboring heavy metal tolerance determinants has been identified in some European countries (Mourão et al., 2015). Yet, the genetic association of AMR and reduced susceptibility to heavy metal determinants has been less explored. Previously, we described the virulence and AMR phenotypes of S. enterica isolates from livestock, processed food, and ready-to-eat food (Figueiredo et al., 2015a, b).
Therefore, the main goals of this study were to identify the AMR genes by DNA microarrays, test the susceptibility of a subset of MDR isolates to heavy metals, and evaluate the potential of mobile AMR and heavy metal genes to be cotransferred.
Materials and Methods
Bacterial strains and heavy metal susceptibility
From a total of 258 Salmonella isolated from Portugal during 2011–2012 from ready-to-eat items (hamburgers, chorizos), poultry/swine carcass samples (skin, liver), and feces of poultry, swine, cattle (Figueiredo et al., 2015b), a subset of 105 nontyphoidal S. enterica strains, previously characterized for virulence (Figueiredo et al., 2015a) and antimicrobial susceptibility by the Microscan test (Supplementary Table S1; Supplementary Data are available online at
Based on MDR and virulence features, 15 Salmonella Typhimurium, the most prevalent serotypes associated with human disease, which usually shows high frequency of resistance (EFSA, 2013), and 4 MDR isolates of less common serotypes (Rissen, Hadar, Kentucky, and Derby), were selected for heavy metal susceptibility determination. The reference wild-type strain Salmonella Typhimurium LT2 was included. Minimum inhibitory concentrations (MICs) were determined by microbroth dilution method in an aerobic atmosphere using Mueller-Hinton Broth (BioMérieux, Marcy-l'Étoile, France), supplemented separately with cobalt (CoSO4), cadmium (CdSO4), mercury (HgCl2), arsenic (NaAsO2), lead ([CH3COO]2Pb), silver (AgNO3), and tellurite (K2TeO3) at doubling dilutions from 8 to 0.001 mM (final concentrations), adjusted to pH 7.2. MICs for copper (CuSO4) and zinc (ZnSO4) were determined at doubling dilutions from 36 to 0.25 mM (final concentrations).
All chemicals were acquired from Sigma-Aldrich (Sintra, Portugal). The MIC was determined as the lowest concentration at which there was no visible growth. Considering that there are no recognized breakpoints for heavy metal susceptibilities, the MIC value was obtained by comparing the susceptibility to LT2 reference strain, as described (Fang et al., 2016).
Polymerase chain reaction screening for heavy metal resistance genes and whole-genome sequencing
Polymerase chain reaction (PCR) amplifications to detect heavy metal resistance genes associated with reduced susceptibility to heavy metals for the 19 Salmonella strains and transconjugants (see below) were performed as previously described (Joerger et al., 2010; Araújo et al., 2012; Mourão et al., 2015) for the following genes: copper (pcoA, pcoD, tcrB), zinc (zitB), arsenic (arsB), silver (silA/silE), and tellurite (terF).
Two isolates (Sal25 and Sal368) harboring a large number of AMR genes and showing reduced susceptibility to a number of different heavy metals, for which a genotype was not detected, were selected for whole-genome sequencing (WGS). The cdtB toxin harboring isolate (Sal199), with enhanced virulence, was also selected for WGS. Genomic DNA was prepared as described by Anjum et al. (2007); Nextera DNA library prep kit (Illumina) was used to prepare sequencing libraries; and pair-end sequenced performed using an Illumina Miseq Sequencer.
The sequenced genomes were annotated using RAST (
Conjugation assays
Conjugation assays were performed for three selected strains (Sal25, Sal199, and Sal368) to assess plasmid-mediated transfer of genes conferring reduced heavy metal susceptibility using Escherichia coli J53 strain as a recipient (resistant to sodium azide). In brief, conjugation experiments were performed in triplicate in two independent assays, in trypticase soy broth at a ratio of donor:recipient of 1:10. Selection was performed on MacConkey agar plates supplemented with sodium azide (10 mg/L) and plates separately supplemented with one of each of the nine heavy metals (Supplementary Table S2). The efficiency of conjugation was calculated as the number of transconjugants/donor cells. The susceptibility to heavy metals and antibiotics of the two purified transconjugants from plates containing sodium azide and copper, and E. coli J53, was evaluated, as described above.
In both transconjugants, PCR was used to screen heavy metal resistance and the cotransfer of AMR genes bla CTX-M-1 and mcr-1, and tet(M) and bla TEM-1, as previously described (Araújo et al., 2010, 2012; Joerger et al., 2010; Figueiredo et al., 2015b, 2016; Mourão et al., 2015). Plasmid typing was performed as described by Caratolli et al. (2005).
Results
AMR genotype
AMR genotyping was performed using microarrays, on 105 nontyphoidal S. enterica isolates, which were selected from a larger panel (ST1) to determine the presence of potentially mobile AMR genes. There were generally good correlations between the AMR phenotypes previously observed (Figueiredo et al., 2015a) and the AMR genotype in >90% of isolates for the antimicrobials considered (Supplementary Table S1).
Major discrepancies were observed in only five isolates: Salmonella Typhimurium Sal188 harboring tet(E) but being susceptible to tetracycline; Salmonella Virchow Sal193 harboring bla MOX/bla CMY but being ampicillin susceptible; Salmonella Enteritidis Sal226 harboring bla TEM but being ampicillin susceptible; and another (Sal289) harboring bla OXA-2-like but being ampicillin susceptible. Isolate Sal329 harbored sul2 but did not harbor any trimethoprim resistance genes, and so its susceptibility to cotrimoxazole (Trimethoprim/Sulfametoxazole) was expected; this was seen in several isolates. Macrolide (ereA; mphA) and streptomycin (aadA1-like; aadA2-like; strA/B) resistance genes were detected in several isolates but could not be correlated with a phenotype as these antibiotics were not tested. Also, presence of the plasmid-mediated quinolone resistance (PMQR) gene (qnrB) did not result in phenotypic resistance.
AMR genotyping detected on average four AMR genes per isolate, with 31 different AMR genes present in these isolates. Fifty-eight percent of the isolates were genotypically MDR, and the most common genes were as follows: sul1 and sul2 (sulfonamide resistance); tet(B) and tet(G) (tetracycline resistance); strB (streptomycin resistance); aadA2 (aminoglycoside resistance); bla TEM-like and bla PSE-1-like (β-lactam resistance); and floR (chloramphenicol resistance) (Supplementary Table S1). Four genes, strB, bla TEM-like, sul2, and tet(B), were present together in 27 Salmonella Typhimurium isolates. Seventeen Salmonella Typhimurium isolates harbored bla PSE-1, floR, aadA2, sul1, and tet(G), which is characteristic of SGI-1.
The integrase gene intI1 was detected in 34 isolates (32.1%), of which 26 harbored aadA2 and sul1; 4 were positive for aadA1, cmlA1, tet(A), and sul3; and 2 carried aadA1, cmlA1, dfrA12, and sul3. The class 1 integrase was also present in a Salmonella Enteritidis isolate that harbored the aadA1, aadA2, cmlA1, tet(A), and sul3 and the bla SHV-like extended-spectrum β-lactamase (ESBL) gene.
Susceptibility to heavy metals
To explore the susceptibility to heavy metals of selected isolates that may potentially be more hazardous to human health, a subset of 19 isolates representing the diversity of AMR pheno- and genotypes were selected. The majority of these isolates were collected from ready-to-eat or processed food (Table 1). Most isolates were MDR, but a Salmonella Typhimurium with increased virulence due to presence of a mobile cdtB gene, with the pltA and pltB genes, encoding a tripartite toxin that causes cellular distension of mammalian host cells (Spanó et al., 2008; Figueiredo et al., 2015a), was also included. Therefore, the susceptibilities to nine heavy metals were determined in this subset. Table 1 shows the MIC to heavy metals where differences with respect to the LT2 reference strain are highlighted. There were no apparent differences between the Typhimurium and non-Typhimurium serotypes.
Minimum Inhibitory Concentrations to Heavy Metals and Resistance Genes of Salmonella enterica Isolates and Transconjugants
Shaded gray boxes indicate MIC values higher than LT2, and presence or possible presence of heavy metal resistance genes.
Shaded dark grey boxes indicate MIC values of transconjugants more than two doublings higher than J53 recipient strain.
Transconjugant T1 derived from conjugation with Sal25.
Transconjugant T2 derived from conjugation with Sal368.
MIC, minimum inhibitory concentration; RTE, ready-to-eat food; PF, processed food; CS, carcass sample.
These isolates and LT2 were screened by PCR for the presence of some heavy metal resistance genes (Table 1). The results indicated that all isolates harboring a heavy metal resistance gene showed reduced susceptibility phenotypically to the corresponding heavy metal, with respect to the LT2 control. Conversely, not all isolates that showed reduced susceptibility to a heavy metal, with respect to LT2, harbored the corresponding resistance gene from our panel. Tellurite was an exception, which showed 100% correlation between pheno- and genotype.
Characterization of isolates with reduced heavy metal susceptibility by WGS
Two MDR isolates were selected for further characterization by WGS because they harbored many AMR genes and showed reduced heavy metal susceptibility to almost all heavy metals tested (Table 1); the cdtB toxin harboring strain was selected for WGS to determine if heavy metal resistance genes might be linked with virulence on mobile elements (Table 2). All three isolates harbored a plethora of AMR genes that matched the microarray data (Table 2), including presence of the tet(M) gene in Sal368. The mcr-1 gene, as previously reported (Figueiredo et al., 2015b, 2016), was the only additional gene detected by WGS, in comparison with microarray results.
Antimicrobial Resistance and Heavy Metal Resistance Genes Identified from Polymerase Chain Reaction and Annotation of Whole-Genome Sequencing in Selected Salmonella Isolates
A number of additional heavy metal resistance genes (Table 2) to those found by PCR were detected by WGS (Table 1). For Sal25 and Sal368, which had been positive by PCR for terF, WGS indicated the presence of the entire tellurite resistance operon. Also, the entire arsenic resistance operon and the mercury resistance operon were detected. Both isolates showed decreased susceptibility to mercury, but Sal199, which did not harbor this operon, was susceptible to mercury. A number of copper resistance genes were also detected in these isolates, including the copA gene, which was present in all isolates. Since only two isolates showed reduced susceptibility to copper, further work is required to understand which gene(s) may be responsible for the phenotype. The smtA gene associated with lead resistance and czc genes associated with zinc resistance were detected in all three isolates, and may account for the decreased lead and zinc susceptibilities. However, as czc genes have also been associated with cadmium resistance, absence of this phenotype in Sal199 and Sal368 suggests a role for other mechanisms (Table 1).
Plasmids of incompatibility types IncHI2 and IncQ1 were detected in Sal25, an IncF plasmid was detected in Sal368, but no plasmid Inc-type was found in Sal199 (Table 2).
Transfer of heavy metal and AMR determinants
Conjugation assays were performed to assess whether AMR and heavy metal resistance genes were colocated on plasmids and could be selected in the presence of heavy metals, after plasmid transfer. The three Salmonella Typhimurium isolates for which WGS was available were tested. No transconjugants were obtained for Sal199, which correlated with the absence of any Inc-types from WGS data. For Sal25 and Sal368, transconjugants were obtained, and the plasmids were transferred at ∼10−4 colony-forming units (CFUs) per recipient CFU for both isolates, which were selected in the presence of different heavy metals (Supplementary Table S2).
A representative E. coli J53 transconjugant from Sal25 (T1) and Sal368 (T2) was purified from copper-selective plates. PCR plasmid typing showed that T1 carried an IncHI2 plasmid and T2 harbored an IncF plasmid, in both cases matching the PlasmidFinder data from the parent strain. The heavy metal susceptibility of both transconjugants was determined and compared with the recipient strain (E. coli J53) to assess for heavy metal resistance gene transfer. Transconjugants T1 and T2 showed reduced susceptibility to copper, zinc, arsenic, silver, tellurite, cobalt, lead, and mercury compared with the recipient E. coli J53 (Table 1). Presence of selected heavy metal resistance genes in the transconjugants was confirmed by PCR and matched results obtained with the donor strains. The decreased susceptibilities of isolates T1 and T2 to several antibiotics, and PCR, confirmed that AMR genes bla CTX-M-1 and mcr-1 for T1, and tet(M) and bla TEM-1 for T2 were cotransferred with heavy metal resistance genes on the same plasmids (Table 1).
Discussion
Infections with S. enterica by consumption of contaminated animal-derived food products are one of the most important causes of foodborne diseases worldwide (EFSA, 2013). Our knowledge on the impact of using heavy metals in animal husbandry and coselection of AMR genes is scarce (Seiler and Berendonk, 2012).
To assess the potential for dissemination of AMR genes by heavy metal selection in S. enterica of animal origin, the study was performed in a stepwise manner. In the first instance, the prevalence of AMR genes was assessed in a collection of 105 isolates, to accurately define their genotype. It showed that the isolates harbored AMR genes known to be present on mobile elements, such as plasmids and SGI1 (Boyd et al., 2002; Hopkins et al., 2007; Anjum et al., 2011; Kirchner et al., 2013; Szmolka et al., 2012). Presence of the tetM gene, which is rarely reported in S. enterica (Aminov et al., 2001; Scaria et al., 2010), was detected by microarray and confirmed by WGS.
The results indicated the importance of genotyping as there was good correlation between geno- and phenotyping for >90% of isolates, although five did not show an AMR phenotype (Figueiredo et al., 2015b). This may indicate a potential for some genes to be expressed under different conditions such as the natural environment, as seen for OXA-48 carbapenemases (Poirel et al., 2012); the PMQR genes (Redgrave et al., 2014); and for a Morganella morganii isolate, which harbored the NDM-1 enzyme but remained susceptible in vitro to meropenem (Kumarasamy et al., 2010). Conversely, genotypic detection by microarray in the absence of a phenotype may indicate absence of a fully functional gene.
In the next step, a subset of 19 S. enterica isolates selected on the basis of their MDR profiles and virulence characteristics were tested for susceptibility against nine heavy metals. The results showed reduction in susceptibility to the majority of heavy metals tested with respect to the LT2 reference; many were confirmed by PCR and WGS. There are currently no MIC breakpoints described for heavy metals, and our results using the LT2 reference were comparable with others for zinc (Medardus et al., 2014) and silver (Mourão et al., 2015), but 1.75-fold less susceptible than that previously reported for copper (Aarestrup and Hasman, 2004).
Genes for mercury were detected by PCR and WGS in two isolates, which correlated with their reduced susceptibility phenotype, as also reported by Mastrorilli et al. (2018); we expect that the reduced susceptibility phenotype for Sal23 will also correlate with presence of the mer operon. There were instances where we could not definitively allocate a gene to a phenotype, for example, for cadmium, indicating that there may be other genes or mechanisms responsible, which requires further work. It is notable that in two isolates an association was found between pcoA/pcoD (copper resistance) and sil genes, which has previously been reported in S. enterica 4,[5],12:i:- (Mourão et al., 2015).
Finally, conjugation assays were performed using a heavy metal, and not an antibiotic, as the selective marker. The results showed cotransferability of heavy metal and AMR genes to the recipient E. coli. In one transconjugant, a plasmid with the ESBL determinant bla CTX-M-1 and the colistin resistance gene mcr-1 was present, and in the other a conjugative plasmid harbored both bla TEM-1 and tet(M). The presence and transfer of genes harboring resistance to high-priority critically important antibiotic classes (third-generation cephalosporins, monobactams, and colistin) with heavy metal resistance genes such as zitB, arsB, silE, and terF are noteworthy.
The finding of a plasmid IncHI2 carrying both bla CTX-M-1 and mcr-1 genes was reported in isolates from cattle in France and in Tunisian chickens of French origin (Grami et al., 2016), but their association with heavy metal resistance genes is not known. Previous work from our group has detected only the mcr-1 gene in conjugative plasmids from MDR nontyphoidal S. enterica; the mcr-1 plasmids did not harbor any additional AMR and heavy metal resistance genes (Anjum et al., 2016; Duggett et al., 2017).
It is also relevant that the heavy metal resistance genes bearing plasmids from this study were transferable at relatively high frequency. Since the isolates are of animal origin, these results suggest a potential for these bacteria to grow in the presence of heavy metals at concentrations that may be used in animal production as additives or decontaminants. For example, the MIC for silver for a number of isolates was 0.125 mM, a concentration that is above the suggested antimicrobial activity of 0.06 mM for silver (Araújo et al., 2012).
In summary, this study shows the presence of a wide diversity of AMR genes among S. enterica isolates. It also shows that reduced susceptibility to heavy metals may occur in MDR S. enterica, which is cotransferable with AMR genes. It is notable that some of these isolates were from ready-to-eat samples, with no necessity of cook, and can consequently infect humans. Future work will address factors that may be involved in dissemination of MDR plasmids with heavy metal resistance genes in the natural environment and through the food chain.
Footnotes
Acknowledgments
We thank Dr. Ana Henriques and Dr. Rui Sereno from ControlVet, Tondela, Portugal for isolation and identification of Salmonella sp. isolates. We also thank Dr. Jacoby from Lahey Clinic for E. coli J53. R.F. was supported by grant SFRH/BD/78833/2011 and N.M. by the grant from SFRH/BPD/45815/2008 from Fundação para a Ciência e Tecnologia. This work was supported financially by the grant SFRH/BPD/45815/2008 from Fundação para a Ciência e Tecnologia, Lisboa, Portugal. Work undertaken at the Animal and Plant Health Agency was funded by the UK government Department for Environment, Food and Rural Affairs.
Disclosure Statement
No competing financial interests exist.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
