Abstract

The extent and impact of plasmid-mediated AmpC beta-lactamase genes (pAmpCs) prevalence in aquatic environments is poorly understood. The aim of this study was to detect and quantify pAmpCs from the aquatic environment. The following pAmpCs were analysed with clinical TaqMan assays from isolated plasmids: ACC, ACT/MIR, BIL/LAT/CMY, DHA, FOX and MOX/CMY. Quantification was conducted using qPCR and 3D chip-based digital PCR. The results of qPCR yielded 4,875.27 copies/ng DNA and dPCR, 1,640.58 copies/ng (Mann–Whitney U Test, p= 0.868). Redundancy analysis indicated that land coverage explains 90.49% (ANOVA, p= 0.601) of pAmpC variance. There was a correlation between the frequency and quantities of pAmpCs detected in each river and this could be related to anthropogenic influence. Frequencies of detection for pAmpCs were 25/36 for the Crocodile West River and 13/36 for the Marico River. Quantification resulted in higher copy numbers for the Crocodile West River and high copies in only two sites of the Marico River, thus reflecting degrees of anthropogenic influences on both rivers. The presence of these clinically relevant pAmpCs in aquatic systems are cause for concern, considering their potential impact if these genes are harboured by pathogens and become dispersed to human populations.

HIGHLIGHTS

  • Clinically relevant plasmid-mediated AmpC beta-lactamase genes are prevalent in South African aquatic systems.

  • Quantitative PCR and digital PCR are comparable for analysis of AmpC genes.

  • The presence of anthropogenic impact on AmpC genes suggests positively correlated co-factors explaining prevalence.

INTRODUCTION

Aquatic systems serve as a reservoir for the influx and distribution of antibiotic resistance genes (ARGs) (Biyela et al. 2004). Wastewater treatment plants, agricultural activities and urbanisation all contribute to the genetic reservoir from which mutations and transferal between bacterial species of genetic elements occur (Rizzo et al. 2013; Devarajan et al. 2017). Great strides have been made in the field of ARGs analysis in environmental settings, revealing that aquatic ecosystems are reservoirs of ARGs (Amos et al. 2014). Plasmid-mediated AmpC beta-lactamase genes (pAmpCs) are mobile and convey resistance to several antibiotics, including third- and fourth-generation cephalosporins (von Tippelskirch et al. 2018). The World Health Organization (WHO) has recognised this group of antibiotics as critically important antimicrobials (WHO 2017). Global issues regarding the ease of dissemination, high evolutionary rate and clinical impact of pAmpCs have caught the attention of the world (Cantón 2009; Tang et al. 2014).

The cycle of ARGs in clinical settings is well understood, however, ARGs in environmental niches are somewhat understudied (Eckert et al. 2018). ARGs from clinical origins have invaded various environment types and utilise these environments in an almost symbiotic manner to promote the spread and evolution of these genes (Martinez 2009). For example, the bacterial richness of soils allows ARGs to cross to previously unexposed species, or previously susceptible bacteria, thereby inducing further ARG evolution and diversity. Moreover, air, water and animal vectors are means by which ARGs disseminate between environments and human settings (Leonard et al. 2015; Dolejska & Papagiannitsis 2018; Tiedje et al. 2019). The role of pAmpCs in the cycle of ARG dissemination and impact within the environment is generally unexplored. In order to determine the extent of pAmpCs within the environment, quantitative detection methods must be employed (Coertze & Bezuidenhout 2019).

Due to the increased morbidity and mortality rates associated with bacterial infections that are in turn associated with pAmpCs (Tang et al. 2014; Harris 2015), it is imperative that rapid, affordable and reliable detection methods for pAmpCs are available (Reuland et al. 2015). It is undeniable that pAmpCs may elude detection by shortcomings in detection methodology (El-Hady & Adel 2015) or by residing within non-pathogenic organisms (Cantón 2009). In culture-dependent studies, pathogenic organisms are usually the subject of analysis (Rizzo et al. 2013) and therefore non-pathogens harbouring these ARGs may go undetected (Baquero et al. 2008). In mixed bacterial communities, such as environmental samples, pathogens may acquire antibiotic resistance genes as part of mobile genetic elements via horizontal gene transfer from non-pathogens (Martinez 2009). Recent studies have identified pAmpCs in bacterial species from which the plasmid did not originate (Harris 2015). This supports the capability of these genes to disseminate among a variety of bacterial species in an unbiased manner. It would therefore be beneficial to analyse both pathogenic and non-pathogenic DNA in order to understand the genetic potential of an immediate environment, particularly in terms of antibiotic resistance and associated horizontal-gene-transfer and virulence factors. Therefore, analyses of environmental DNA (eDNA) are essential.

The current dynamics regarding AmpC beta-lactamase genes in aquatic environments are largely unknown. In South Africa, there is a lack of knowledge regarding the prevalence of these genes or methods of analysis from environmental samples that are viable considering the limited resources available. Therefore, the objectives of the current study were to (i) use a culture-based method for enriching aerobic heterotrophic bacteria from aquatic systems (anaerobic bacteria were excluded to reduce the number of variables that could affect the outcome of the study) and to isolate plasmid DNA from the mixed population, and (ii) detect and quantify clinical variants of AmpC genes using quantitative PCR and chip-based digital PCR.

METHODS AND MATERIALS

Sampling and enrichment

River water (two 1 L bottles) was sampled from six sites on the Crocodile West and Marico rivers respectively in the North West Province, South Africa (Figure 1). Water was sampled aseptically approximately 1 m from the edge of the river from running water at a depth of approximately 30 cm. Information on the rivers and their respective sampling sites was obtained from a geospatial analysis report by Bezuidenhout et al. (2017). Sites of the Crocodile West River are adjacent to numerous agricultural, mining and urban influences (Table S1). In comparison, the Marico River is marginally affected by anthropogenic influences and mostly surrounded by natural environments with agricultural activities adjacent to some sites and urban influence at Site 6 (Table S1). Due to the inaccessibility of Site 1 of the Marico River and lack of sampling water at Site 1 of the Crocodile West River, these sites were not included in further analysis. Historical studies of water quality on these rivers indicated that on average the Crocodile West River was similar in physical and chemical parameters and higher in microbial water quality measurements (faecal coliforms, Enterococci, Clostridia, and heterotrophic plate counts) than the Marico River (Bezuidenhout et al. 2017).

Figure 1

Map illustrating the locations of the sampling sites of the Crocodile West River (red circles) and Marico River (blue triangles). The green square in the upper left corner represents the sampling area in the context of South Africa. The map was generated using the ggmap package (version 3.0.0) in R (version 3.6.0).

Figure 1

Map illustrating the locations of the sampling sites of the Crocodile West River (red circles) and Marico River (blue triangles). The green square in the upper left corner represents the sampling area in the context of South Africa. The map was generated using the ggmap package (version 3.0.0) in R (version 3.6.0).

Sampling of the two rivers was conducted over a period of two days (one day for each river). After sampling of each river, samples were kept on ice and analysed within 12 hours after sampling. Three 9 ml Luria–Bertani (LB) broths were inoculated with 1 ml river water from each sampling site. Enrichment of aerobic heterotrophic bacteria was conducted by incubating the inoculated broths at 37 °C while shaking at 220 rpm until an OD600 measurement of approximately 0.6 was reached. This was measured every 30 min (n = 3) using the NanoDrop One Spectrophotometer, following the manufacturer's recommendations (ThermoFisher Scientific, USA). The enrichment of the environmental water samples was followed in an effort to enhance the detection signals of pAmpCs, as suggested by Viršek et al. (2017). It is acknowledged that the one-time sampling is a limitation to this study.

DNA extraction

Plasmid DNA was extracted from each successfully cultured broth using the NucleoSpin Plasmid DNA Isolation kit (Machinery-Nagel, Germany). This was done according to the protocol of the manufacturer. Qubit analysis using a broad-range kit was used to determine the concentrations of extracted plasmids (ThermoFisher Scientific, USA).

Detection of plasmid-mediated AmpC genes

In order to ensure the highest possible comparability across all variables, all the subsequent reagents and PCR quantification platforms were from ThermoFisher Scientific (USA). All PCR reactions (including quantification PCRs) were conducted with the addition of no template controls (one for each reaction and target gene) and positive controls. Positive controls were obtained through PCR screening of environmental samples, purified, and identified using Sanger Sequencing and BLAST (data not shown). Plasmid DNA samples (n = 3 per site) were pooled prior to further analysis. Samples harbouring AmpC genes were determined by using the QuantStudio Quantitative PCR presence/absence protocol. PCR reactions were conducted in triplicate and set up in 20 μl reactions. Each PCR reaction consisted of 20 ng DNA, 1x QuantStudio 3D Digital PCR MasterMix v2, 1x of the appropriate TaqMan assay, and filled to volume with nuclease-free water. The following FAM florescent dye TaqMan gene expression assays were used to amplify the various pAmpCs: Pa04646144_s1 (ACC), Pa04646124_s1 (ACT/MIR), Pa04646135_s1 (BIL/LAT/CMY), Pa04646120_s1 (DHA), Pa04646126_s1 (FOX) and Pa04646156_s1 (MOX/CMY). Probe sequences are not made available by ThermoFisher, thus only product codes are provided. Thermal cycling conditions consisted of a Pre-Read Stage: 60 °C for 30 s, Hold Stage: 95 °C for 10 min, PCR Stage: 40 cycles of 95 °C for 15 s followed by 60 °C for 1 min, and finally a Post-Read Stage at 60 °C for 30 s.

Quantification of plasmid-mediated AmpC genes

Due to the mixed bacterial nature of the DNA samples, optimal gene copy numbers and thus DNA input could not be calculated. Therefore, the DNA input in PCR reactions varied between samples although the results were standardised by using Equation (1): 
formula
(1)
where copies/ng DNA refers to the gene copies of a target gene per nanogram of template DNA. Copies in rxn represents the copies resulting from a qPCR/dPCR platform and calculated to include the entire reaction volume. DNA mass in rxn symbolises the mass of DNA in nanograms that was added to the total reaction volume. The use of this equation made it possible to standardise the AmpC gene copy numbers for the applications of this study.

Quantification using quantitative PCR

Absolute quantification of plasmid-mediated AmpC genes was performed using the QuantStudio 3 Quantitative PCR system in 20 μl reactions. Each reaction consisted of 1x QuantStudio 3D Digital PCR MasterMix v2, 1x of the appropriate TaqMan assay, DNA input (which varied depending on undetectable or oversaturated measurements of previous test runs) and was filled to volume with nuclease-free water. Each sample was independently tested (n = 3). Thermal cycling conditions were comprised of an initial step in which the reaction was kept at 50 °C for 2 min followed by denaturation at 95 °C for 10 min. The cycling steps consisted of 40 cycles of 95 °C for 15 s and 60 °C for 1 min.

DNA used for standard curves were constructed using positive control samples for each target gene containing known copies. The number of copies for each gene was determined using dPCR as described below. At least three repetitions of each gene were subjected to dPCR at various dilutions in order to determine the highest possible accuracy of gene copies. Standard curves were determined by using five-fold dilutions in independent triplicate reactions. Copies ranging between 20,000 and 2 comprised the dilution series. Amplification efficiencies of standard curves were between 90% and 110% (E = 10−1/slope − 1) (Pfaffl 2001) and R2 > 0.97 were considered as reliable comparison trend lines for quantification of unknown gene samples.

Quantification using digital PCR

Absolute quantification of plasmid-mediated AmpC genes was achieved using the QuantStudio 3D Digital PCR System. Each dPCR reaction was prepared to a final volume of 15 μl, which included a 6.67% excess, consisting of 1x QuantStudio 3D Digital PCR MasterMix v2, 1x TaqMan assay, varying mass DNA input and was filled up to volume with nuclease-free water. A final volume of 14 μl was transferred to the digital PCR chip. PCR conditions were set according to the standard ProFlex 2x Flat PCR System PCR Method. No positive control was tested because external references are not necessary for dPCR analyses (Hudecova 2015). Analysis of the 3D PCR results was conducted using the QuantStudio 3D AnalysisSuite Cloud Software. Parameters were set using the Poisson Plus quantification algorithm version 4.4.10 (Majumdar et al. 2017), with a confidence level of 95% and a precision difference of 10% maximum. All samples were quantified with at least three independent repetitions. Samples that did not present within quality parameters were repeated, following adjustment of DNA input quantities.

Data analysis

Normalisation of the datasets was calculated using the Shapiro–Wilk normality test. Normally distributed data were considered for P-values larger than the alpha value (0.05). The Mann–Whitney U Test was used to determine the statistical significance between the medians of the two quantification methods for both rivers, as well as the rivers individually. This was also performed to determine significance between the target genes for each river and between methods. Analysis of the various AmpC targets for each river was performed using the Kruskal–Wallis H Test. Post-hoc analysis was conducted using Dunn's Test. Redundancy analysis (RDA) was used to determine if variations of the AmpC target genes (considered as species data) can be explained by land coverage surrounding the sampling sites (Agriculture, Mines, Urban, Natural, Water, Wetlands, Plantations, Erosion, Bare Ground and WWTPs; Table S1). This was done using the rda function from the vegan package (version 2.5–6) (Oksanen et al. 2019). Land coverage data was obtained from Bezuidenhout et al. (2017). The analysis was based on prevalence (absence/presence) of the AmpC genes and not their copy numbers. Monte Carlo simulations, 999 permutations, were conducted to test the significance of the observed variations using the R base anova function. Statistical significance over all tests was recognised for P-values < 0.05. Statistical analyses were conducted in the R statistical programming language (version 3.6.0) (R Core Team 2013). Relevant graphs were constructed using the ggplot2 package (version 3.1.1) (Wickham 2016).

RESULTS

Plasmid-mediated AmpC beta-lactamases were detected in all sites of the Crocodile West River and from five out of six sites sampled in the Marico River (Table 1). The results for each site and target genes obtained by both the quantitative PCR and digital PCR methods are available in the Supplementary Material as Figure S1.

Table 1

Average copy numbers of pAmpCs for each AmpC group and sampling sites of each river and quantification method used

Crocodile West River
Marico River
n (6)qPCRdPCRMediann (6)qPCRdPCRMedian
Target genes Copies/ng DNA Copies/ng DNA 
ACC 4.67 6.48 1.07 23.02 35.35 0.00 
ACT/MIR 4,296.79 3,475.68 2,195.74 621.02 461.57 1.58 
BIL/LAT/CMY 3,476.58 3,128.55 424.18 1,221.84 1,062.19 0.00 
DHA 132.98 145.70 1.98 983.54 1,011.14 0.00 
FOX 346.58 396.04 79.75 1,860.12 882.80 148.35 
MOX/CMY 4,192.30 2,710.50 2,486.10 41,343.81 6,370.93 0.00 
Total 25/36    13/36    
Sampling sites Copies/ng DNA Copies/ng DNA 
Site 2 3,933.81 3,389.19 71.24 5,049.92 1,816.49 633.88 
Site 3 1,062.44 842.07 87.19 11.12 10.99 0.00 
Site 4 2,045.01 2,596.57 1,285.32 136.70 115.08 0.00 
Site 5 2,719.58 1,666.42 748.08 40,836.68 7,849.36 4,344.83 
Site 6 2,447.02 1,135.65 79.76 18.93 32.05 0.00 
Site 7 242.04 233.04 1.07 0.00 0.00 0.00 
Total 25/36    14/36    
Crocodile West River
Marico River
n (6)qPCRdPCRMediann (6)qPCRdPCRMedian
Target genes Copies/ng DNA Copies/ng DNA 
ACC 4.67 6.48 1.07 23.02 35.35 0.00 
ACT/MIR 4,296.79 3,475.68 2,195.74 621.02 461.57 1.58 
BIL/LAT/CMY 3,476.58 3,128.55 424.18 1,221.84 1,062.19 0.00 
DHA 132.98 145.70 1.98 983.54 1,011.14 0.00 
FOX 346.58 396.04 79.75 1,860.12 882.80 148.35 
MOX/CMY 4,192.30 2,710.50 2,486.10 41,343.81 6,370.93 0.00 
Total 25/36    13/36    
Sampling sites Copies/ng DNA Copies/ng DNA 
Site 2 3,933.81 3,389.19 71.24 5,049.92 1,816.49 633.88 
Site 3 1,062.44 842.07 87.19 11.12 10.99 0.00 
Site 4 2,045.01 2,596.57 1,285.32 136.70 115.08 0.00 
Site 5 2,719.58 1,666.42 748.08 40,836.68 7,849.36 4,344.83 
Site 6 2,447.02 1,135.65 79.76 18.93 32.05 0.00 
Site 7 242.04 233.04 1.07 0.00 0.00 0.00 
Total 25/36    14/36    

The median of both methods for sites and AmpC groups are also shown.

Comparison of qPCR and dPCR

The two quantification methods resulted in varying AmpC gene copies; qPCR (4,875.27) and dPCR (1,640.58) copies/ng DNA values. However, there was no significance between the medians for these methods (Mann–Whitney U Test, p= 0.868). Moreover, individual analysis of the Crocodile West River, qPCR (2,074.98 copies/ng DNA) and dPCR (1,643.82 copies/ng DNA), and Marico River, qPCR (7,675.56 copies/ng DNA) and dPCR (1,637.33 copies/ng DNA), also demonstrated that there was no significant difference between the two methods (Kruskal–Wallis H Test, p= 0.981 and p= 0.857 respectively).

Each target gene, individually for each river, was evaluated to determine if there was a significant difference for the measured copies using each method. Results for both the Crocodile West River and Marico River indicated no significance for the specific target gene levels (Mann–Whitney U Test, p > 0.05), irrespective of the method used. Statistically the methods were comparable and therefore the median of both methods were used for further investigations.

AmpC enriched gene quantification

Among the six Crocodile West River sites, the following AmpC gene groups and their detection frequencies were observed, in descending order: ACT/MIR (5/6), BIL/LAT/CMY (5/6), MOX/CMY (5/6), FOX (4/6), ACC (3/6) and DHA (3/6) (Table 1). In the Marico River, fewer of these genes were detected: FOX (4/6), ACT/MIR (3/6), BIL/LAT/CMY (2/6), MOX/CMY (2/6), ACC (1/6) and DHA (1/6) (Table 1). Overall, considering both rivers, the most prevalent AmpC gene groups were FOX & ACT/MIR (8), followed by MOX/CMY & BIL/LAT/CMY (7), ACC (4) and DHA (4). At site numbers 3, 4 and 5 of the Crocodile West River and sites 2 and 5 of the Marico River the highest abundance of target AmpC gene was detected (Table 1). Sites with the lowest abundance of pAmpCs were sites 4 and 6 of the Marico River in which only one target gene was detected at each (FOX and ACC respectively). None of the targeted pAmpCs were detected at Site 7 of the Marico River. The lowest abundances of AmpC gene groups detected in the Crocodile West River were at sites 6 (ACT/MIR, BIL/LAT/CMY and MOX/CMY) and 7 (ACC, DHA and MOX/CMY).

The Crocodile West River resulted in an average of 1,859.40 copies/ng DNA of AmpC beta-lactamase gene copies and the Marico River 4,656.45 copies/ng DNA. The difference in copies between the two rivers was significant (Mann–Whitney U Test, p< 0.000). The average copies and medians of AmpC genes for each river, target gene and sampling sites are represented in Table 1.

Crocodile West River: AmpC gene copy number

The highest average AmpC copy numbers (for individual genes) were obtained for ACT/MIR (3,886.23 copies/ng DNA), followed by MOX/CMY (3,451.40 copies/ng DNA), BIL/LAT/CMY (3,302.56 copies/ng DNA), FOX (371.31 copies/ng DNA), DHA (139.34 copies/ng DNA) and ACC (5.57 copies/ng DNA). Differences between target groups were significant (Kruskal–Wallis H Test, p < 0.018). It was calculated that the copy numbers for ACT/MIR and MOX/CMY was significant toward ACC and DHA (Dunn's Test, p < 0.05).

The highest average AmpC copy numbers (including all genes) for each sampling site was Site 2 (3,661.50 copies/ng DNA), followed by Site 4 (2,320.79 copies/ng DNA), Site 5 (2,193.00 copies/ng DNA), Site 6 (1,791.33 copies/ng DNA), Site 3 (952.26 copies/ng DNA) and Site 7 (237.54 copies/ng DNA). Differences between sampling sites were not significant (Kruskal–Wallis H Test, p = 0.400).

Marico River: AmpC gene copy number

The highest average AmpC copy numbers (for individual genes) were obtained for MOX/CMY (23,857.37 copies/ng DNA), followed by FOX (1,371.46 copies/ng DNA), BIL/LAT/CMY (1,142.01 copies/ng DNA), DHA (997.34 copies/ng DNA), ACT/MIR (541.30 copies/ng DNA) and ACC (29.18 copies/ng DNA). Differences between target groups were not significant (Kruskal–Wallis H Test, p = 0.763).

The highest average AmpC copies (including all genes) for each sampling site was Site 5 (24,343.02 copies/ng DNA), followed by Site 2 (3,433.21 copies/ng DNA), Site 4 (125.89 copies/ng DNA), Site 6 (25.49 copies/ng DNA), Site 3 (11.06 copies/ng DNA) and Site 7 (0.00 copies/ng DNA). Differences between sampling sites were significant (Kruskal–Wallis H Test, p = 0.003). It was calculated that the average AmpC copies obtained at sites 2 and 5 were significant toward sites 3, 4, 6 and 7 (Dunn's Test, p < 0.05).

AmpC target gene relationships with land coverage

In order to determine if land coverage explains variance in the prevalence of AmpC genes, RDA was performed, and the resulting ordination plot is illustrated in Figure 2. Variance between AmpC target genes was explained by 90.49% of the first two axes with all variables, although this variation was not significant (F= 0.952, p = 0.601). The first axis explained 45.15% (F = 23.74, p= 0.590) of variations and the second axis 19.70% (F = 10.36, p= 0.881). The lack of significance could imply unknown variables affecting variance in AmpC target genes.

Figure 2

Redundancy analysis (RDA) plot illustrating variance of AmpC target genes (thick blue arrows and blue text) explained by land coverage (thin red arrows and red text) surrounding the sampling sites of both rivers.

Figure 2

Redundancy analysis (RDA) plot illustrating variance of AmpC target genes (thick blue arrows and blue text) explained by land coverage (thin red arrows and red text) surrounding the sampling sites of both rivers.

DISCUSSION

The aim of this study was to detect and quantify plasmid-mediated AmpC beta-lactamase genes from DNA obtained from water samples from two river systems using two quantification methods. Major groups of clinically relevant pAmpCs were detected throughout both the Crocodile West River and Marico River. This is cause for concern regarding the potential impact on the environment and surrounding human settings.

Comparison between quantification methods

There were no significant differences between the results produced by dPCR and qPCR methods. The dPCR method has operational advantages over qPCR in that quantification mostly does not require a standard curve (Jones et al. 2016), is able to detect low gene copy numbers (Bosman et al. 2015) and is less prone to inhibition (Powell & Babady 2018). However, the dynamic range of dPCR is limited to four log units (Alikian et al. 2017), while qPCR extends the range to seven log units (Powell & Babady 2018). When using dPCR, it is recommended to conduct analysis of a sample in which the number of copies can preliminarily be estimated so that the number of copies fall within the dynamic range of dPCR (Hudecova 2015). This is not always possible with environmental DNA since the nature of the samples is unpredictable (Blaya et al. 2016). Therefore, for the application of the present study, qPCR was preferred. The larger dynamic range allowed more room for the quantification of unpredictable copy numbers, which is the case in most environmental DNA samples.

A scenario in this study in which the limited dynamic range of dPCR affected the results was quantification of the MOX/CMY gene at Site 5 of the Marico River (229,786.3 copies/ng DNA for qPCR versus 32,887.4 copies/ng DNA for dPCR). The addition of unknown high copy numbers of DNA to a dPCR chip could have entailed that multiple pAmpCs were contained within a single dPCR chip partition, which is why saturated samples may not have been accurate and resulted in underestimation of gene copies. Similar observations were made by Alikian et al. (2017), who focused on the quantification of the clinical BCR-ABL1 gene, and Blaya et al. (2016), who quantified environmental bacteria, Phytophthora nicotianae. These two studies both found that at higher concentrations, the two methods were not comparable. This problem can be alleviated by optimising each sample in replicate dilutions to find the ideal quantity of DNA to analyse in a dPCR chip (Blaya et al. 2016). However, concerning large numbers of samples it would be impractical and financially unfeasible to accomplish.

Environmental implications of AmpC genes

The Crocodile West River and Marico River were chosen for analysis due to the differences in anthropogenic impact on the two rivers. A report on South African rivers by Bezuidenhout et al. (2017) specified types of human exposure by land coverage surrounding the sampling sites (Table S1). Generally, sampling sites of the Crocodile West River are more impacted by mining, WWTPs and industrial activities, whereas the Marico River sites are less impacted by such activities and more by agriculture.

According to redundancy analysis of the variations in AmpC genes observed, land coverage explains 90.49% of pAmpC variation (Figure 2). However, since this result is not significant (p= 0.601), land coverage may not be an accurate estimation of ARGs in the environment, but rather an indication. Land coverages, such as urban areas, agricultural settings and WWTPs, are known to be sources of ARGs, antibiotic resistant bacteria (ARB) and selection components (i.e. antibiotics) (Karkman et al. 2019), although these elements are carried by mobile waters to downstream locations and therefore are difficult to link to a specific location or type of land coverage. This explains why land coverages can explain the presence of pAmpCs, although the types of pollution elements and ARB in water co-detected with pAmpCs may provide a more accurate explanation of pAmpC prevalence.

Karkman et al. (2019) demonstrated that ARGs in anthropogenic environments are more likely to be explained by faecal pollution rather than a selective pressure such as antibiotics in the environment. This was potentially reflected by the frequency of pAmpCs detected in the Crocodile West River (25/36), which was higher compared with the Marico River (13/36). The higher anthropogenic exposure of the Crocodile West River could have led to the higher occurrence of pAmpCs, especially considering the high exposure to WWTPs (n = 32). This aspect was investigated by Zeng et al. (2019) where it was established that higher levels of pollution, specifically faecal pollution, lead to higher abundance of ARGs when compared with more pristine environments. The same is also true for the quantities of pAmpCs observed at each site of both rivers. In the Crocodile West River, a majority of sites resulted in relatively high copies of pAmpCs, compared with the Marico River (Mann–Whitney U Test, p< 0.000), whereas in the Marico River only sites 2 and 5 yielded significantly high copies (Table 1, Kruskal–Wallis H Test, p < 0.05).

According to Jacoby (2009), ESKAPE pathogens (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa and Enterobacter spp.) have been observed to harbour, as well as be the origin of, AmpC genes. Escherichia coli and Klebsiella pneumonia are ESKAPE pathogens and known for association with faecal wastewater (Dolejska et al. 2011). In this study, the ACT/MIR and MOX/CMY group was frequently detected in the Crocodile West River (5/6) and yielded significantly higher copy numbers compared with the other genes (ACC and DHA). Similarly, high copies of MOX/CMY were observed in the Marico River (Table 1). A study conducted by Ragupathi et al. (2019) isolated E. coli and K. pneumonia from a clinical environment with the goal of isolating various plasmid variants and to determine resistance constituents. According to the aforementioned study by Ragupathi et al. (2019), the CMY-2, CMY-42 and CMY-65 genes, as well as CMY-4 and CMY-6 genes (members of MOX/CMY and BIL/LAT/CMY) were found on a variety of plasmids carried by E. coli and K. pneumonia respectively. This included the Inc plasmids which are known for carrying diverse groups of ARGs, including beta-lactamase genes, quinolone and aminoglycoside resistance genes, that originate from clinical samples (Carattoli 2009; Ogbolu et al. 2013).

It is known that WWTPs are not always able to remove ARB and ARGs from wastewater (Paul et al. 2018). For example, E. coli and K. pneumonia (known for harbouring pAmpCs) have often been detected in treated wastewater effluent from a WWTP in the Czech Republic (Dolejska et al. 2011). Moreover, numerous other faecal pathogens are also known to carry pAmpCs (Jacoby 2009). If wastewater treatment processes are mismanaged, not optimally used or neglected, these bacteria may enter the environment while harbouring clinically relevant pAmpCs. Both rivers in this study are impacted by WWTPs and its effluent may affect downstream populations (Bezuidenhout et al. 2017). It is frightening that pAmpCs detected from clinical cases have also been observed in the aquatic environments of this study. The pAmpCs groups quantified in this study, such as MOX/CMY, BIL/LAT/CMY or ACT/MIR, are frequently found in pathogenic organisms. For example, the CMY-2 gene is the most commonly detected pAmpC in clinical Enterobacteriaceae (Pietsch et al. 2018), and was also frequently detected in this study at higher levels compared with other AmpC genes. Furthermore, ACT/MIR (Wu et al. 2018), DHA (Hsieh et al. 2015) and ACC (Hasman et al. 2005) have all been detected in clinical isolates and are known to convey resistance to several beta-lactam antibiotics in clinical cases. Outside the boundaries of enriched samples, direct correlation between copy numbers of pAmpCs and number of bacterial carriers in the aquatic environment is alarming.

The various possible pollution sources and their effluent constituents may explain the prevalence of these genes in the river systems. Even the Marico River, which is less anthropogenically impacted, is still exposed to various agricultural activities, which could be promoting the selection of antibiotic resistant bacteria and ARGs (Manyi-Loh et al. 2018; Karkman et al. 2019). In a South African context, ampC genes have been detected in both agricultural and environmental settings (Ekwanzala et al. 2018). This indicates the possibility of transferring these genes between agriculture and the environment. A possible dispersion mechanism of pAmpCs to human settings and subsequent consumption is through irrigation and produce. Njage & Buys (2017) established that E. coli harbouring extended-spectrum beta-lactamase genes (ESBLs) and AmpC genes (associated with pAmpCs detected in this study) were detected in irrigation water for lettuce. After the lettuce was harvested and processed, E. coli still conveying resistance to beta-lactams was detected on the produce. Considering that fresh produce could be globally distributed, it highlights the danger of how easily these genes could be transferred from the environment to human settings.

Another method in which pAmpCs may cross from the environment to humans is direct exposure. A study by Leonard et al. (2015) argued that Escherichia coli harbouring ESBLs could be transferred from ocean water to humans. This was based on the levels of E. coli in ocean water and volumes of ingested water during recreational activities. Considering that the pAmpCs detected in this study originated from river waters, it is likely that bacteria harbouring these genes could be transferred to humans. Both the Crocodile West River and Marico River flow through urban, rural and agricultural settings. It was observed during sampling that these waters are used for consumption, food-preparation, bathing, religious practices, irrigation and livestock consumption (Bezuidenhout et al. 2017), therefore it is possible that these pAmpCs may be transferred to humans. Moreover, river water being used for drinking water, after treatment, ARGs can enter the drinking water system (Rodriguez-Mozaz et al. 2015). This observation has previously been made through isolation of E. coli harbouring IncF plasmids from drinking water sources in taps and wells (Lyimo et al. 2016). Furthermore, it is concerning that pAmpCs have been detected on Inc-type plasmids in clinical settings (Lorme et al. 2018). The presence of pAmpCs and their mobile vectors in both aquatic and clinical settings suggests that the environment might serve as a distribution hub of pAmpCs. This highlights the need for further investigation and environmental surveillance regarding pAmpCs and ARGs in general. In addition, it is also important to establish relationships between pAmpCs and pollution elements in aquatic systems.

CONCLUSIONS

In this study all major groups of clinically relevant plasmid-mediated AmpC beta-lactamase genes were detected from environmental aquatic sources. Detection of these genes are cause for concern considering the prevalence of potential pathogens with which these genes may be associated. The types of anthropogenic exposure on both the Crocodile West River and Marico River serve as influx points of ARB, pAmpCs and possible selective pressure constituents for further ARG selection within the environment. Moreover, these rivers serve as reservoirs and distribution systems of pAmpCs since bacteria harbouring these genes may reach human settings by dissemination through water usage in agriculture and urban activities. Further investigation of environmental DNA is required in order to establish which environmental and/or anthropogenic factors are associated with the prevalence of pAmpCs in aquatic systems.

ACKNOWLEDGEMENTS

We would like to thank the National Research Foundation of South Africa (Grant No. 93621) and the Water Research Commission of South Africa, contact no. K5/2347//3 for aiding in funding this research. We would also like to thank Linda Coertze for proof reading and language corrections of the article.

FUNDING

This work is based on research supported in part by the National Research Foundation of South Africa (Grant No. 93621, 109xxxx) and the Water Research Commission of South Africa, contract no. K5/2347//3. The views expressed are those of the authors and not of the funding entities.

SUPPLEMENTARY MATERIAL

The Supplementary Material for this paper is available online at https://dx.doi.org/10.2166/ws.2020.085.

REFERENCES

REFERENCES
Alikian
M.
,
Whale
A. S.
,
Akiki
S.
,
Piechocki
K.
,
Torrado
C.
,
Myint
T.
,
Cowen
S.
,
Griffiths
M.
,
Reid
A. G.
,
Apperley
J.
,
White
H.
,
Huggett
J. F.
&
Foroni
L.
2017
RT-qPCR and RT-digital PCR: a comparison of different platforms for the evaluation of residual disease in chronic myeloid leukemia
.
Clinical Chemistry
63
(
2
),
525
531
.
https://doi.org/10.1373/clinchem.2016.262824
.
Amos
G. C. A.
,
Zhang
L.
,
Hawkey
P. M.
,
Gaze
W. H.
&
Wellington
E. M.
2014
Functional metagenomic analysis reveals rivers are a reservoir for diverse antibiotic resistance genes
.
Veterinary Microbiology
171
(
3–4
),
441
447
.
https://doi.org/10.1016/j.vetmic.2014.02.017
.
Baquero
F.
,
Martínez
J.-L.
&
Cantón
R.
2008
Antibiotics and antibiotic resistance in water environments
.
Current Opinion in Biotechnology
19
(
3
),
260
265
.
http://dx.doi.org/10.1016/j.copbio.2008.05.006
.
Bezuidenhout
C. C.
,
Mienie
C.
,
De Klerk
T.
&
Molale
L.
2017
Geospatial Analysis of Microbial Community Structure and Antimicrobial Resistance Analysis in the Management of Natural Streams and Selected Wetlands
.
Water Research Commission
,
Pretoria, South Africa
.
Biyela
P. T.
,
Lin
J.
&
Bezuidenhout
C. C.
2004
The role of aquatic ecosystems as reservoirs of antibiotic resistant bacteria and antibiotic resistance genes
.
Water Science and Technology
50
(
1
),
45
50
.
https://doi.10.2166/wst.2004.0014
.
Blaya
J.
,
Lloret
E.
,
Santisima-Trinidad
A. B.
,
Ros
M.
&
Pascual
J. A.
2016
Molecular methods (digital PCR and real-time PCR) for the quantification of low copy DNA of Phytophthora nicotianae in environmental samples
.
Pest Management Science
72
(
4
),
747
753
.
https://dx.doi.org/10.1002/ps.4048
.
Bosman
K. J.
,
Nijhuis
M.
,
van Ham
P. M.
,
Wensing
A. M. J.
,
Vervisch
K.
,
Vandekerckhove
L.
&
De Spiegelaere
W.
2015
Comparison of digital PCR platforms and semi-nested qPCR as a tool to determine the size of the HIV reservoir
.
Scientific Reports
5
,
13811
.
https://doi.org/10.1038/srep13811
.
Cantón
R.
2009
Antibiotic resistance genes from the environment: a perspective through newly identified antibiotic resistance mechanisms in the clinical setting
.
Clinical Microbiology and Infection
15
(
Supplement 1
),
20
25
.
https://doi.org/10.1111/j.1469-0691.2008.02679.x
.
Carattoli
A.
2009
Resistance plasmid families in Enterobacteriaceae
.
Antimicrobial Agents and Chemotherapy
53
(
6
),
2227
2238
.
https://doi.org/10.1128/aac.01707-08
.
Coertze
R. D.
&
Bezuidenhout
C. C.
2019
Global distribution and current research of AmpC beta-lactamase genes in aquatic environments: a systematic review
.
Environmental Pollution
252
,
1633
1642
.
https://doi.org/10.1016/j.envpol.2019.06.106
.
Devarajan
N.
,
Köhler
T.
,
Sivalingam
P.
,
van Delden
C.
,
Mulaji
C. K.
,
Mpiana
P. T.
,
Ibelings
B. W.
&
Poté
J.
2017
Antibiotic resistant Pseudomonas spp. in the aquatic environment: a prevalence study under tropical and temperate climate conditions
.
Water Research
115
,
256
265
.
https://doi.org/10.1016/j.watres.2017.02.058
.
Dolejska
M.
&
Papagiannitsis
C. C.
2018
Plasmid-mediated resistance is going wild
.
Plasmid
99
,
99
111
.
https://doi.org/10.1016/j.plasmid.2018.09.010
.
Dolejska
M.
,
Frolkova
P.
,
Florek
M.
,
Jamborova
I.
,
Purgertova
M.
,
Kutilova
I.
,
Cizek
A.
,
Guenther
S.
&
Literak
I.
2011
CTX-M-15-producing Escherichia coli clone B2-O25b-ST131 and Klebsiella spp. isolates in municipal wastewater treatment plant effluents
.
Journal of Antimicrobial Chemotherapy
66
(
12
),
2784
2790
.
https://doi.10.1093/jac/dkr363
.
Eckert
E. M.
,
Di Cesare
A.
,
Coci
M.
&
Corno
G.
2018
Persistence of antibiotic resistance genes in large subalpine lakes: the role of anthropogenic pollution and ecological interactions
.
Hydrobiologia
824
(
1
),
93
108
.
https://doi.10.1007/s10750-017-3480-0
.
Ekwanzala
M. D.
,
Dewar
J. B.
,
Kamika
I.
&
Momba
M. N. B.
2018
Systematic review in South Africa reveals antibiotic resistance genes shared between clinical and environmental settings
.
Infection and Drug Resistance
11
,
1907
1920
.
https://doi.10.2147/IDR.S170715
.
El-Hady
S. A.
&
Adel
L. A.
2015
Occurrence and detection of AmpC β-lactamases among Enterobacteriaceae isolates from patients at Ain Shams University Hospital
.
Egyptian Journal of Medical Human Genetics
16
(
3
),
239
244
.
http://dx.doi.org/10.1016/j.ejmhg.2015.03.001
.
Harris
P. N. A.
2015
Clinical management of infections caused by Enterobacteriaceae that express extended-spectrum β-lactamase and AmpC enzymes
.
Seminars in Respiratory and Critical Care Medicine
36
(
1
),
56
73
.
https://doi.org/10.1055/s-0034-1398387
.
Hasman
H.
,
Mevius
D.
,
Veldman
K.
,
Olesen
I.
&
Aarestrup
F. M.
2005
β-Lactamases among extended-spectrum β-lactamase (ESBL)-resistant Salmonella from poultry, poultry products and human patients in The Netherlands
.
Journal of Antimicrobial Chemotherapy
56
(
1
),
115
121
.
https://doi.10.1093/jac/dki190
.
Hsieh
W.-S.
,
Wang
N.-Y.
,
Feng
J.-A.
,
Weng
L.-C.
&
Wu
H.-H.
2015
Identification of DHA-23, a novel plasmid-mediated and inducible AmpC beta-lactamase from Enterobacteriaceae in Northern Taiwan
.
Frontiers in Microbiology
6
,
436
.
https://doi.10.3389/fmicb.2015.00436
.
Hudecova
I.
2015
Digital PCR analysis of circulating nucleic acids
.
Clinical Biochemistry
48
(
15
),
948
956
.
https://doi.org/10.1016/j.clinbiochem.2015.03.015
.
Jacoby
G. A.
2009
AmpC β-lactamases
.
Clinical Microbiology Reviews
22
(
1
),
161
182
.
http://dx.doi.org/10.1128/CMR.00036-08
.
Jones
G. M.
,
Busby
E.
,
Garson
J. A.
,
Grant
P. R.
,
Nastouli
E.
,
Devonshire
A. S.
&
Whale
A. S.
2016
Digital PCR dynamic range is approaching that of real-time quantitative PCR
.
Biomolecular Detection and Quantification
10
,
31
33
.
https://doi.org/10.1016/j.bdq.2016.10.001
.
Karkman
A.
,
Pärnänen
K.
&
Larsson
D. G. J.
2019
Fecal pollution can explain antibiotic resistance gene abundances in anthropogenically impacted environments
.
Nature Communications
10
(
1
),
80
.
https://doi.10.1038/s41467-018-07992-3
.
Leonard
A. F. C.
,
Zhang
L.
,
Balfour
A. J.
,
Garside
R.
&
Gaze
W. H.
2015
Human recreational exposure to antibiotic resistant bacteria in coastal bathing waters
.
Environment International
82
,
92
100
.
https://doi.org/10.1016/j.envint.2015.02.013
.
Lorme
F.
,
Maataoui
N.
,
Rondinaud
E.
,
Esposito-Farèse
M.
,
Clermont
O.
,
Ruppe
E.
,
Arlet
G.
,
Genel
N.
,
the VOYAG-R Study Group, Matheron
S.
,
Andremont
A.
&
Armand-Lefevre
L.
2018
Acquisition of plasmid-mediated cephalosporinase producing Enterobacteriaceae after a travel to the tropics
.
PLoS ONE
13
(
12
),
e0206909
.
https://doi.10.1371/journal.pone.0206909
.
Lyimo
B.
,
Buza
J.
,
Subbiah
M.
,
Temba
S.
,
Kipasika
H.
,
Smith
W.
&
Call
D. R.
2016
IncF plasmids are commonly carried by antibiotic resistant Escherichia coli isolated from drinking water sources in northern Tanzania
.
International Journal of Microbiology
2016
,
3103672
.
https://doi.10.1155/2016/3103672
.
Majumdar
N.
,
Banerjee
S.
,
Pallas
M.
,
Wessel
T.
&
Hegerich
P.
2017
Poisson Plus quantification for digital PCR systems
.
Scientific Reports
7
(
1
),
9617
.
https://doi.org/10.1038/s41598-017-09183-4
.
Manyi-Loh
C.
,
Mamphweli
S.
,
Meyer
E.
&
Okoh
A.
2018
Antibiotic use in agriculture and its consequential resistance in environmental sources: potential public health implications
.
Molecules
23
(
4
),
795
.
https://doi.10.3390/molecules23040795
.
Martinez
J. L.
2009
Environmental pollution by antibiotics and by antibiotic resistance determinants
.
Environmental Pollution
157
(
11
),
2893
2902
.
https://doi.org/10.1016/j.envpol.2009.05.051
.
Ogbolu
D. O.
,
Daini
O. A.
,
Ogunledun
A.
,
Terry Alli
O. A.
&
Webber
M. A.
2013
Dissemination of IncF plasmids carrying beta-lactamase genes in Gram-negative bacteria from Nigerian hospitals
.
The Journal of Infection in Developing Countries
7
(
5
),
382
390
.
https://doi.10.3855/jidc.2613
.
Oksanen
J.
,
Blanchet
F. G.
,
Friendly
M.
,
Kindt
R.
,
Legendre
P.
,
McGlinn
D.
,
Minchin
P. R.
,
O'Hara
R. B.
,
Simpson
G. L.
,
Solymos
P.
,
Stevens
M. H. H.
,
Szoecs
E.
&
Wagner
H.
2019
vegan: Community Ecology Package
. .
Paul
C.
,
Bayrychenko
Z.
,
Junier
T.
,
Filippidou
S.
,
Beck
K.
,
Bueche
M.
,
Greub
G.
,
Bürgmann
H.
&
Junier
P.
2018
Dissemination of antibiotic resistance genes associated with the sporobiota in sediments impacted by wastewater
.
PeerJ
6
,
e4989
.
https://doi.10.7717/peerj.4989
.
Pfaffl
M. W.
2001
A new mathematical model for relative quantification in real-time RT–PCR
.
Nucleic Acids Research
29
(
9
),
e45
.
https://doi.org/10.1093/nar/29.9.e45
.
Pietsch
M.
,
Irrgang
A.
,
Roschanski
N.
,
Brenner Michael
G.
,
Hamprecht
A.
,
Rieber
H.
,
Käsbohrer
A.
,
Schwarz
S.
,
Rösler
U.
,
Kreienbrock
L.
,
Pfeifer
Y.
,
Fuchs
S.
,
Werner
G.
,
Bühling
A.
,
Domurath
B.
,
von Salviati-Claudius
C.
,
Wendt
C.
,
Valenza
G.
,
Wahl
H. G.
,
Lu
H. M.
,
Laube
H.
,
Hering
J.
,
Hellkamp
J.
,
Arvand
M.
,
Kresken
M.
,
Podschun
R.
,
Schneider
S.
,
Tobisch
S.
,
Witt
M.
,
Eckmanns
T.
,
Hachmann
U.
&
Bührlen
U.
,
RESET Study Group
2018
Whole genome analyses of CMY-2-producing Escherichia coli isolates from humans, animals and food in Germany
.
BMC Genomics
19
(
1
),
601
.
https://doi.org/10.1186/s12864-018-4976-3
.
Powell
E. A.
&
Babady
N. E.
2018
Digital PCR in the clinical microbiology laboratory: another tool on the molecular horizon
.
Clinical Microbiology Newsletter
40
(
4
),
27
32
.
https://doi.org/10.1016/j.clinmicnews.2018.01.005
.
Ragupathi
N. K. D.
,
Bakthavatchalam
Y. D.
,
Mathur
P.
,
Pragasam
A. K.
,
Walia
K.
,
Ohri
V. C.
&
Veeraraghavan
B.
2019
Plasmid profiles among some ESKAPE pathogens in a tertiary care centre in South India
.
Indian Journal of Medical Research
149
(
2
),
222
231
.
https://doi.org/10.4103/ijmr.IJMR_2098_17
.
R Core Team
2013
R: A Language and Environment for Statistical Computing
.
R Foundation for Statistical Computing
,
Vienna
,
Austria
. .
Reuland
E. A.
,
Halaby
T.
,
Hays
J. P.
,
de Jongh
D. M. C.
,
Snetselaar
H. D. R.
,
van Keulen
M.
,
Elders
P. J. M.
,
Savelkoul
P. H. M.
,
Vandenbroucke-Grauls
C. M. J. E.
&
al Naiemi
N.
2015
Plasmid-mediated AmpC: prevalence in community-acquired isolates in Amsterdam, the Netherlands, and risk factors for carriage
.
PLoS ONE
10
(
1
),
e0113033
.
https://doi.org/10.1371/journal.pone.0113033
.
Rizzo
L.
,
Manaia
C.
,
Merlin
C.
,
Schwartz
T.
,
Dagot
C.
,
Ploy
M. C.
,
Michael
I.
&
Fatta-Kassinos
D.
2013
Urban wastewater treatment plants as hotspots for antibiotic resistant bacteria and genes spread into the environment: a review
.
Science of The Total Environment
447
,
345
360
.
http://dx.doi.org/10.1016/j.scitotenv.2013.01.032
.
Rodriguez-Mozaz
S.
,
Chamorro
S.
,
Marti
E.
,
Huerta
B.
,
Gros
M.
,
Sànchez-Melsió
A.
,
Borrego
C. M.
,
Barceló
D.
&
Balcázar
J. L.
2015
Occurrence of antibiotics and antibiotic resistance genes in hospital and urban wastewaters and their impact on the receiving river
.
Water Research
69
,
234
242
.
https://doi.org/10.1016/j.watres.2014.11.021
.
Tang
S. S.
,
Apisarnthanarak
A.
&
Hsu
L. Y.
2014
Mechanisms of β-lactam antimicrobial resistance and epidemiology of major community- and healthcare-associated multidrug-resistant bacteria
.
Advanced Drug Delivery Reviews
78
,
3
13
.
http://dx.doi.org/10.1016/j.addr.2014.08.003
.
Tiedje
J. M.
,
Wang
F.
,
Manaia
C. M.
,
Virta
M.
,
Sheng
H.
,
Ma
L.
,
Zhang
T.
&
Topp
E.
2019
Antibiotic resistance genes in the human-impacted environment: a One Health perspective
.
Pedosphere
29
(
3
),
273
282
.
https://doi.org/10.1016/S1002-0160(18)60062-1
.
Viršek
M. K.
,
Lovšin
M. N.
,
Koren
Š.
,
Kržan
A.
&
Peterlin
M.
2017
Microplastics as a vector for the transport of the bacterial fish pathogen species Aeromonas salmonicida
.
Marine Pollution Bulletin
125
(
1–2
),
301
309
.
https://doi.org/10.1016/j.marpolbul.2017.08.024
.
von Tippelskirch
P.
,
Gölz
G.
,
Projahn
M.
,
Daehre
K.
,
Friese
A.
,
Roesler
U.
,
Alter
T.
&
Orquera
S.
2018
Prevalence and quantitative analysis of ESBL and AmpC beta-lactamase producing Enterobacteriaceae in broiler chicken during slaughter in Germany
.
International Journal of Food Microbiology
281
,
82
89
.
https://doi.org/10.1016/j.ijfoodmicro.2018.05.022
.
WHO
2017
Critically Important Antimicrobials for Human Medicine – 5th Revision
.
World Health Organization
,
Geneva, Switzerland
.
Wickham
H.
2016
ggplot2: Elegant Graphics for Data Analysis
.
Springer-Verlag
,
New York, USA
.
Wu
C.
,
Lin
C.
,
Zhu
X.
,
Liu
H.
,
Zhou
W.
,
Lu
J.
,
Zhu
L.
,
Bao
Q.
,
Cheng
C.
&
Hu
Y.
2018
The beta-lactamase gene profile and a plasmid-carrying multiple heavy metal resistance genes of Enterobacter cloacae
.
International Journal of Genomics
2018
,
4989602
.
https://doi.10.1155/2018/4989602
.
Zeng
J.
,
Pan
Y.
,
Yang
J.
,
Hou
M.
,
Zeng
Z.
&
Xiong
W.
2019
Metagenomic insights into the distribution of antibiotic resistome between the gut-associated environments and the pristine environments
.
Environment International
126
,
346
354
.
https://doi.org/10.1016/j.envint.2019.02.052
.

Supplementary data