Background
The administration of antimicrobial growth promoters in animal husbandry has been prohibited in the EU since 2006 [
1]. As alternatives to the application of antimicrobials, non-antimicrobial substances including heavy metal ions like zinc and copper, prebiotics or probiotics have been suggested to improve animal health and growth rates [
2‐
5]. Zinc is one of the compounds widely used in the pig farming industry to overcome problems during weaning of piglets, including infections caused by pathogenic
E. coli [
6‐
9]. The essential trace elements zinc and copper are both involved in numerous physiological and cellular functions in all organisms [
10‐
12]. Zinc concentrations and resistance are highly regulated through uptake and efflux mechanisms in different organisms [
11,
13]. However, recent studies have suggested that feeding zinc in high concentrations during weaning of piglets increases the proportion of multi-drug resistant
E. coli in the gut of the piglets [
14‐
19]. The enhancement in the spread of antimicrobial resistance by the use of zinc confounds the usefulness of zinc supplementation in piglets and raises the question as to the underlying mechanisms of this observation.
One possible mechanism could be co-selection for both heavy metal/biocide and antimicrobial resistance, either in the form of co- or cross-resistance [
16,
20‐
22]. Cross-resistance occurs as a result of physiological adaptations and affects susceptibility to different compounds, for example through efflux pump regulation or changes in cell wall permeability [
23]. Co-resistance phenomena include changes involving genetic linkage of different genes encoding resistance to different classes of antimicrobials [
20,
24]. A number of different studies have described possible mechanisms for co-selection of antimicrobial and heavy metal (zinc) resistance [
16,
25‐
29]. Physiological coupling, genetic coupling and linked/co-localized resistance genes on mobile genetic elements have been suggested as possible mechanisms of both cross- and co-resistance [
19,
23,
28,
30‐
32]. Zinc dependent beta-lactamases, effects of zinc on ampicillin stability or bacterial conjugation rates, and class 1 integrons (involved in co-selection) proximity to genes coding the efflux pump CzcA have been proposed as mechanisms involved in simultaneous reduction of susceptibility to antimicrobials and zinc/copper [
19,
33‐
37]. Both intrinsic and acquired resistance mechanisms including efflux pumps and cellular detoxification of high concentrations of copper in bacteria have been reported in different studies [
10,
38‐
40]. In addition to zinc, copper has also been suggested to contribute to antibiotic resistance in gram-negative and positive bacteria [
28,
41,
42].
In this study, we tested the hypothesis that the increased antimicrobial resistance of E. coli isolates observed in weaning piglets fed with high zinc concentrations is caused by co-selection via phenotypic zinc tolerance. For this purpose, we used selected isolates [including antimicrobial resistant, multi-drug resistant (MDR), and non-resistant/susceptible (S) E. coli] and screened the level of their phenotypic zinc tolerance by determining a zinc minimal inhibitory concentration. Isolates originated from two, independent zinc-feeding trials of piglets with two different sampling schemes performed by our group over a period of 5 years. In both trials, groups of piglets were administered either high concentrations of zinc (> 2000 ppm) or a background control (50–70 ppm). From both feeding groups, we determined whether feeding of zinc resulted in higher proportions of phenotypically zinc resistant E. coli, and whether multi-drug resistant isolates also tolerated higher zinc concentrations, indicative of a co-selection process. In addition, we also compared phenotypic zinc resistance of these isolates with their phenotypic copper resistance values to determine whether there is a correlation between phenotypic resistance/tolerance to different bivalent ionic metals.
Discussion
During two, independent animal trials, we observed an increase in multi-drug resistant (MDR)
E. coli in isolates of piglets when fed with high concentrations of zinc. One possible explanation for this effect is a co-selection for heavy metal and antimicrobial resistance, as has been previously suggested [
16,
19,
20,
25,
43]. To determine whether there is an association between MDR phenotype and phenotypic zinc tolerance, we screened both MDR and non-MDR (NMDR) isolates for the level of phenotypic zinc tolerance. Out of a total of 210 isolates selected from both zinc supplementation trials, 63 isolates (30%) were multi-drug resistant.
In this study, we determined two different classifications of antibiotic resistance. We compared multi-drug resistant (MDR) to non-multi-drug resistant isolates (NMDR) according to the definition of Schwarz et al
. [
44], as well as resistant isolates (R), defined as resistance to at least one antimicrobial agent, and susceptible (S) isolates, defined as not resistant to any antimicrobial agents. For both definitions of antimicrobial resistance, we obtained the same result. Isolates tested in this study are not the whole set of isolates derived from two previous studies. We also did not want to show differences in the number of multi-drug resistant strains. In contrary, we chose almost identical number of strains for this experiment to compare their zinc resistance and whether it correlates with their original MDR phenotype. Therefore, it should not necessarily be a significant difference between the number of MDR isolates from zinc and control-feeding groups as was determined in our previous studies.
When comparing susceptible isolates (S) to isolates harboring at least one (or more) resistances (R), we observed no significant difference (P = 0.107) in their zinc MIC values. In addition, the zinc MIC values for zinc of MDR E. coli and NMDR isolates also showed no significant difference, suggesting that there is no association between antimicrobial resistance and phenotypic zinc tolerance of these isolates.
The observed increase in MDR—
E. coli during the zinc feeding trials is therefore not likely a result of co-selection of zinc and antimicrobial resistance. As proposed by Ciesinski et al. [
18], the increase of multi-drug resistant isolates in swine treated with a high dietary zinc, is likely due to formation of a persistent population of resistant bacteria already present in the gut. Furthermore, we found no difference in zinc tolerance levels of isolates from zinc-treated groups in comparison to the control groups, suggesting that the overall zinc tolerance of
E. coli in the gut of piglets therefore seems not be affected by zinc feeding.
There are no universal interpretative criteria for classification of
E.
coli resistance towards zinc, and studies determining the MIC values for heavy metal ions are scarce [
45‐
48]. In this study, we used a custom-made plate for phenotypic zinc tolerance levels in
E.
coli and which include all inhibitory concentrations mentioned in previous studies in MIC assays [
45].
Despite variations in the testing methods used in previous studies and our study, such as use of agar plates or broth micro-dilution, the medium used for growing bacteria, and the formulation of zinc used in the experiments, the biological upper cut-off of phenotypic zinc tolerance for the most of
E. coli isolates in these studies were around 2–2.5 mM of zinc ion. This is in accordance with the reported MIC of 2.2 mM Zn 2+ for
E. coli TG1 in LB-medium [
47]. The highest reported concentration of zinc (Zn 2+) which could be tolerated by
E. coli isolates in the literature was 5 mM [
48].
To determine whether our findings were similar for other heavy metals, we also compared our isolates for copper tolerance. All tested isolates in our experiment, with two exceptions, had the same MIC values of 1024 µg/ml for copper sulphate (~ 6.4 mM). The highest MIC concentration of copper (Cu 2+) detected for E. coli isolates in prior studies was 10.5 mM. Our results indicated no difference between copper MIC values of MDR and NMDR isolates, suggesting that there is no association between phenotypic antimicrobial resistance and phenotypic copper tolerance of the isolates. Interestingly, we also observed no correlation between the zinc and copper MIC values of the same isolates.
Co-selection for antimicrobial and metal-resistance has been suggested in many studies [
11,
39‐
41]. In many of these studies, co-resistance was not shown, but a co-existence of resistance was reported in the same bacteria. Nevertheless, it is believed that some metal and antimicrobial resistance genes are linked and co-resistance of antimicrobial and heavy metal resistant bacteria have been discussed in several studies as likely to arise through co-selection [
22,
29‐
34].
These studies are mostly on genome level and several of studied antimicrobial and metal ion resistance genes are on plasmids [
16,
32,
43,
49‐
51]. For example, the plasmids of
Salmonella abortus equi were found to co-transfer antimicrobial resistance (ampicillin-resistance) and heavy metal resistance (As, Cr, Cd, Hg) genes in mating experiments with
E.
coli strains.
Salmonella strains cured of the plasmids were found to be sensitive towards ampicillin and heavy metals [
32]. In a genomic transcriptional study, Lee et al. [
30] found up-regulation of the
mdtABC operon after exposure to high levels of zinc which suggested a potential influence of metal stresses on bacterial resistance to antibiotics.
In a recent genomic study by Pal et al
. [
28], a total of 2522 fully sequenced bacterial genomes and 4582 plasmids were analyzed. The authors concluded from their large-scale study that plasmids have only a limited potential for horizontal transfer of biocides and metals resistance by co-selection.
Prior studies have also tested isolates at both the genomic and phenotypic levels. One such study showed co-regulation of resistance to heavy metals and carbapenems through the CzcR–CzcS system in
Pseudomonas aeruginosa strain PT5. In that study, it was shown that a mutation in the CzcS sensor protein found in zinc and imipenem resistant isolates led to efflux pump CzcCBA overexpression and down-regulation of the OprD porin resulting in a co-selection for both increased zinc and carbapenem resistance [
52]. In a series of retrospective studies screening
E. faecium isolated from different species, it was found that
tcrB (transfer copper resistance) and
ermB (transfer macrolide resistance) genes were present on the same conjugative plasmid. However, the data did not demonstrate a co-selection between these two phenomena and the strong correlation between copper and macrolide resistance was found only in pig isolates. In addition, while the prevalence of macrolide resistance in isolates decreased during the years covered in the study, the prevalence of copper resistance among pig
E. faecium isolates remained more or less the same. The authors argued that the reduction in the antimicrobial usage during this period lead to a decrease in antimicrobial resistance, whereas in the same time period the use of copper derivatives remained unchanged. Therefore, they concluded that copper exposure might not alone be sufficient to induce antimicrobial resistance and a strong selective pressure of macrolide administration should be present to select the antimicrobial resistance [
40,
53,
54].
There are few experimental studies available evaluating the induction of antimicrobial resistance following metal exposure. Peltier et al
. [
55] investigated antimicrobial resistance to ciprofloxacin, oxytetracycline, and tylosin in zinc-activated sludge bioreactors. Zinc application alone did not affect zinc and antimicrobial resistance to ciprofloxacin and oxytetracycline. Increased antimicrobial resistance could be the result of co-exposure of zinc and antimicrobial agents. Berg et al
. [
56] found that strains isolated from soil treated with copper for 21 months were more resistant to both copper and indirectly antimicrobials compared to control plots.
In contrast to the above-mentioned studies in which co-selection was the subject of discussion, there are also studies reporting counter-selection of heavy metal and antimicrobial resistance [
57]. Hölzel et al
. [
26] found that while exposure to zinc and copper increased the rate of β-lactam resistance in
E. coli, the presence of mercury was associated with a lower rate of antimicrobial resistance.
Conclusions
In summary, our results do not indicate a co-selection process of antimicrobial resistance and higher zinc tolerance in the MDR isolates of our feeding trials. An increase of E. coli more tolerant to zinc due to the feeding of high zinc concentrations as an explanation for the increase of multi-drug resistant isolates via co-selection can therefore be excluded. This seems to be also true for copper tolerance levels. These results would appear to argue against a co-selection mechanism for drug-resistance after zinc supplementation, since we did not find an association between antimicrobial resistance and phenotypic zinc/copper tolerance for the same isolates. We also found that zinc exposure did not have an effect on either zinc or copper phenotypic tolerance of the isolates.
An explanation for an increase in MDR isolates from piglets with high zinc dietary feeding in our previous studies could be that resistant bacteria to antimicrobial agents are more persistent to stresses such as zinc or copper exposure. Ciesinski et al. have argued that the increase in multi-drug resistant E. coli populations is associated with persistence of the resistant population under the influence of high dietary zinc, while in that study the total number of E. coli population had been decreased.
Another explanation might be that in the zinc-fed groups, zinc activates genes involved in metal ion resistance to deal with the metal ion load, and which might also be involved in antibiotic resistance, but this is a transient phenotypic zinc resistance. In accordance to this argumentation, Peltier et al
. also found that zinc exposure increases resistance to antibiotics but had a minimal effect on zinc resistance [
55]. In addition, the duration of experiments, co-exposure to both metal and antimicrobial agents and concentration of the substances could play role in either in vivo or in vitro-resistance studies.
Interpretation and analysis of resistance data based only on genetic data should be made carefully, a combination of both genetic and phenotypic resistance determinations is required, and it will also be important to show whether resistance could be developed in non-resistant isolates. The result of these types of studies could have implications for the prophylactic use of zinc in the field, i.e. pigs daily fed zinc to prevent infections.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.