Detection of Antibiotic Resistance Genes in Source and Drinking Water Samples from a First Nations Community in Canada

Water samples were collected from a First Nations community in the Island Lake Region of Manitoba, which is located about 500 km from Winnipeg, Manitoba, Canada. The community has an on-reserve population of approximately 4,000 and a total registered population of about 4,500, with a median age of ∼20 years. This is a fly-in community, which is accessible from Winnipeg by two flights per day during summer (with no road access during summer) and by ice road during winter. The community has a water treatment plant, and just over 300 homes in the community are served by piped tap water. The majority of the houses that do not have piped tap water from the water treatment plant get their water delivered by a water truck, which fills up at the water treatment plant. During the time of sample collection for this study, there was only one operational water truck for the community. Water is stored in cisterns (water storage tanks used in approximately 150 homes) with most cisterns located underground. Families in homes without running water use small plastic buckets to obtain their drinking water from a community standpipe after which the water is stored (often in open buckets lined with a garbage bag) in the home. It is also quite common to use the lake water for washing and cleaning purposes. In addition, families sometimes utilize lake water as drinking water or collect rain water for general washing and cleaning purposes.

Water samples were collected between 21 and 24 July 2014 from various sites within the community (Table 1) using standard methods (SM) (13): SM 9060A (sample bottle pretreatment) and SM 9060B (preservation and storage). The samples were collected over a period of only 3 days due to the challenges associated with the access to the community; however, we tried to overcome this limitation by collecting samples from multiple households and sources. During water collection, the Hach Chlorine Pocket Colorimeter II (VWR, Mississauga, ON, Canada) was used to determine free residual and total residual chlorine following the adapted U.S. EPA DPD Method 8021 (14). For all other analyses, samples were transported in coolers to Winnipeg by air on the same day of collection when possible or on the next morning after being stored overnight in a refrigerator. In addition, water collected from a tap located on the University of Manitoba grounds and a sample collected from the Red River flowing through the University of Manitoba grounds were used as control samples. Total coliform and E. coli counts were immediately processed upon receipt of the samples in our laboratories in Winnipeg using the “Standard Methods for the Examination of Water and Wastewater” of the American Public Health Association/American Water Works Association/Water Environment Federation as outlined in SM 9222 (13). Briefly, 100 ml of water sample was filtered through a sterile polyethersulfone membrane (0.22-μm pore size, 47-nm diameter; Mo Bio Laboratories, Carlsbad, CA, USA), and the filter paper was placed on agar plates containing Brilliance E. coli and coliform medium (Fisher Scientific, Ottawa, ON, Canada) and incubated at 35°C for 24 h. Some samples required dilution as the bacterial counts were too high. All samples were plated within 24 h of collection, and a control sample was included every time samples were plated. Bacterial counts in water samples were calculated as follows: E. coli CFU/milliliter = number of purple colonies/volume of filtered sample (100 ml) × dilution factor; total coliform bacteria CFU/milliliter = number of purple + pink colonies/volume of filtered sample (100 ml) × dilution factor. We were unable to plate biological replicates for the samples due to the limited number of samples we were able to collect and bring to our laboratories in Winnipeg in a timely fashion.

DNA was extracted from water samples immediately upon arrival in our laboratories in Winnipeg by filtering 100 ml of water samples (R. Li, H. M. Tun, M. Jahan, A. Farenhorst, A. Kumar, W. G. D. Fernando, and E. Khafipour, submitted for publication) through sterile polyethersulfone membranes (0.22-μm pore size; Mo Bio Laboratories, Carlsbad, CA, USA). Filters were subjected to DNA extraction using the PowerWater DNA isolation kit (Mo Bio Laboratories), following the manufacturer's instructions. The DNA concentration was quantified using a NanoDrop 2000 spectrophotometer (Thermo Scientific, Waltham, MA, USA). The DNA samples were normalized, and quality was verified by PCR amplification of the 16S rRNA gene using universal primers (Table 2) as described previously (15). Amplicons were verified by agarose gel electrophoresis. DNA samples were stored at −80°C until further analyses. Sterile water was used as an extraction control. The yield of DNA by this method ranged from 5 ng/μl to 275 ng/μl (100 μl in total), which was sufficient for the library construction as well as the quantitative PCR (qPCR) analysis described below.

The microbial compositions of the water samples were determined by PCR amplification of the V4 region of the 16S rRNA gene using modified F515/R806 primers (16) as described before (17). For each sample, the PCR was performed in duplicate and contained 1.0 μl of prenormalized DNA, 1.0 μl of each forward and reverse primer (10 μM), 12 μl of high-performance liquid chromatography (HPLC)-grade water (Fisher Scientific, Ottawa, ON, Canada), and 10 μl of 5 Prime HotMasterMix (5 Prime, Inc., Gaithersburg, MD, USA). Reactions consisted of an initial denaturing step at 94°C for 3 min, followed by 35 amplification cycles at 94°C for 45 s, 50°C for 60 s, and 72°C for 90 s, with a final extension step at 72°C for 10 min in an Eppendorf Mastercycler pro (Eppendorf, Hamburg, Germany). PCR products were purified using a ZR-96 DNA cleanup kit (Zymo Research, Irvine, CA, USA) to remove primers, deoxynucleoside triphosphates (dNTPs), and reaction components. The V4 library was then generated by pooling 200 ng of each sample, quantified by PicoGreen double-stranded DNA (dsDNA) (Invitrogen, Burlington, ON, Canada). This was followed by multiple dilution steps using prechilled hybridization buffer (HT1; Illumina, San Diego, CA, USA) to bring the pooled amplicons to a final concentration of 5 pM, as determined with a Qubit 2.0 Fluorometer (Life Technologies, Burlington, ON, Canada). Finally, 15% of the PhiX control library was spiked into the amplicon pool to improve the unbalanced and biased base composition, a common characteristic of low-diversity 16S rRNA libraries. Customized sequencing primers for read1 (5′-TATGGTAATTGTGTGCCAGCMGCCGCGGTAA-3′), read2 (5′-AGTCAGTCAGCCGGACTACHVGGGTWTCTAAT-3′), and the index read (5′-ATTAGAWACCCBDGTAGTCCGGCTGACTGACT-3′) were synthesized and purified by polyacrylamide gel electrophoresis (Integrated DNA Technologies, Coralville, IA, USA) and added to the MiSeq reagent kit v2 (300-cycle; Illumina). The 150 paired-end sequencing reaction was performed on a MiSeq Illumina platform at the Gut Microbiome Laboratory (Department of Animal Science, University of Manitoba, Winnipeg, MB, Canada).

The FLASH assembler (18) was used to merge overlapping paired-end Illumina fastq files, and all sequences with mismatches in the overlapping region were discarded. The output of the fastq file was then analyzed by downstream computational QIIME pipelines (19). Assembled reads were demultiplexed according to the barcode sequences and exposed to additional quality filters so that reads with ambiguous calls and those with Phred quality scores (Q scores) of <20 were discarded. Chimeric reads were filtered using UCHIME (20), and sequences were assigned to operational taxonomic units (OTU) using the QIIME implementation of UCLUST (21) at a 97% pairwise identity threshold. Taxonomies were classified to the representative sequence of each OTU using the Ribosomal Database Project (RDP) classifier (22) and aligned with the Greengenes Core reference database (23) using PyNAST algorithms (24).

Absolute quantification of resistance genes in water samples was carried out for four different genes, namely, ampC, vanA, tet(A), and mecA, using the StepOnePlus real-time PCR system (Life Technologies Inc., Burlington, ON, Canada), following a previously described method with slight modifications (25). DNA samples were diluted to a concentration of 0.724 ng/μl, and 2.68 μl of the diluted DNA (to a total of 1.94 ng of DNA per reaction) was used in a total reaction volume of 8 μl which contained 9 μM respective primers and 2× SsoFast EvaGreen Supermix (Bio-Rad Canada, Mississauga, ON, Canada). Primers used in the study are listed in Table 2. Reactions were carried out in triplicate. The copy number of each gene per nanogram of DNA was calculated by creating a standard curve for each gene. Briefly, DNA was isolated from bacterial cells that contained the target genes using the DNA extraction kit (Biobasic, Markham, ON, Canada). The bacterial strains used were clinical isolates E. coli 99270 and E. coli ER2925 for ampC and tet(A), respectively, a methicillin-resistant Staphylococcus aureus clinical isolate HA-MRSA 100697 for mecA, and Enterococcus faecium ATCC 13048 for vanA. For the creation of the standard curve, we first calculated the copy number of each gene per nanogram of DNA in bacterial strains used as positive controls with the following formula (http://www6.appliedbiosystems.com/support/tutorials/pdf/quant_pcr.pdf): m = n/1.069 × 10⁻²¹ (g/bp), where n is genome size in base pairs and m is mass in grams.

Genome sizes used for each of the strains are as follows: 4,640,000 bp for E. coli; 2,839,469 bp for methicillin-resistant S. aureus; and 3,218,031 bp for E. faecium. The sizes of the genomes represent published sequences of these organisms, and while they may not represent the exact sizes of the genomes of the clinical isolates used in our study, we suspect that the sizes are fairly comparable to those used in our calculations. The standard curve was created by using a 10-fold dilution series for the target gene copy number starting with 200,000 copies to 20 copies. The copy number per nanogram of DNA for each sample was then calculated using the slope of the standard curve. Although there are other known methods that either calculate the relative quantification of the 16S rRNA gene copy number or absolute quantification of the target gene per volume of the water sample (26–28), we used calibration curves derived from different antibiotic resistance reference bacteria for absolute quantification. We prefer this method, which has been used previously (25), because our samples were likely to contain uncharacterized bacteria and also contain bacteria with different genetic backgrounds (e.g., multicopy plasmids), and, therefore, we believe our method of quantifying resistance genes normalizing against the amount of DNA circumvents the possible problems associated with mismatches in the 16S qPCR primers.

A multiplex PCR was carried out for the detection of five different β-lactamase genes, namely, SHV-type, TEM-type, CTX-M-type, OXA-1, and CMY-2-type. A list of primers used in the study is provided in Table 2. PCR was performed with Q5 high-fidelity DNA polymerase (New England BioLabs, Whitby, ON, Canada) using the amplification cycle that consisted of the following steps: 1 cycle at 95°C (15 min); 30 cycles at 94°C for 30 s, 63.5°C for 90 s, and 72°C for 90 s; and a final extension step at 72°C for 7 min. Final concentrations of the primers used were as follows: SHV and CMY-2, 0.3 μM each; TEM, 0.1 μM; CTX, 0.4 μM; and OXA-1, 0.2 μM. PCRs were validated using the following strains as positive controls: Klebsiella pneumoniae N09-00080 for SHV, TEM, CTX, OXA-1, and CMY-2; K. pneumoniae N09-0431 for KPC-2; E. coli 12-123T for OXA-48; E. coli A44413 for GES-5; E. coli 10469T for NDM; Pseudomonas aeruginosa C-10 for VIM-2; and A. baumannii C-3 for IMP-4.

Detection of six different carbapenemase-encoding genes, namely, KPC, IMP, VIM, NDM, GES-5, and OXA-48, was carried out using separate PCRs. All genes, except IMP, were detected using a multiplex reaction, while IMP detection was carried out in a separate reaction. Primers for this reaction are listed in Table 2. Final concentrations of the primers used in the reaction were 0.2 μM for each except for OXA-48 (0.1 μM). The PCR consisted for the following steps: 1 amplification cycle at 95°C (15 min); 30 cycles at 94°C for 30 s, 60°C for 90 s, and 72°C for 60 s; and a final extension step at 72°C for 7 min. PCR products were resolved on a 1.3% agarose gel.

The sequencing data were deposited into the Sequence Read Archive (SRA) of NCBI (http://www.ncbi.nlm.nih.gov/sra) and can be accessed via accession number SRR3189861.

The results are summarized in Table 3. While we did not find any coliforms (including E. coli) in samples collected from the water treatment plant, we detected both total coliforms and E. coli from all other samples with the exception of one tap water sample (T3). Health Canada's “Guidelines for Canadian Drinking Water Quality” require a total coliform and E. coli count of 0/100 ml for treated water to be deemed safe for drinking (www.healthcanada.gc.ca/waterquality). Substantial counts (150 to ∼700 CFU/ml) were observed for various samples collected from the lake, including the area near the inlet of the water treatment plant (LBS, LPS, and LBC). We also found high numbers of total coliforms (420 to 5,000 CFU/ml) and E. coli (390 to 2,000 CFU/ml) in samples from the community fountain. Water samples collected from various sources within houses (tap, cisterns, and buckets) also showed the presence of both total coliforms and E. coli. Two of the bucket samples collected from homes without running water, B1 and B2, showed extremely high numbers of total coliforms (14,600 and 1,100 CFU/ml, respectively) and E. coli (2,300 and 1,090 CFU/ml, respectively), and piped water samples from households showed the presence of 1 to 2 CFU/ml bacteria. Field blanks did not show the presence of any bacteria.

None of the water samples from the community had more than 0.2 ppm of residual free chlorine (minimum concentration recommended by the World Health Organization) except for the posttreatment water directly collected in the treatment plant (P1 and P2) (Table 3). Similarly, total chlorine levels were the highest in the samples collected from the plant.

Taxonomic classification of clustered OTU revealed the presence of 46 bacterial phyla. Abundant phyla (>1% of the population) included Proteobacteria, Bacteroidetes, Actinobacteria, Cyanobacteria, Planctomycetes, Firmicutes, Verrucomicrobia, and Acidobacteria, whereas Armatimonadetes, TM6, Chlorobi, Chloroflexi, Chlamydiae, and Gemmatimonadetes were in medium abundance (between 0.1 and 1% of the population), and the remaining 32 phyla were in low abundance (<0.1% of the population) (Fig. 1). The results for the community analysis showed that samples from the lake, irrespective of the location they were collected from, had a very similar community profile. The same was true for the two posttreatment samples from the treatment plant. Water samples collected from households or from fountains showed a profile that was distinct from that for the posttreatment samples from the treatment plant. Interestingly, we observed a significantly large proportion of Proteobacteria in household/fountain samples compared to that in the samples from the treatment plant (Fig. 1).

Results for the presence of antibiotic resistance genes detected using qPCR are shown in Fig. 2. ampC was detected in one tap water sample (T4), three cisterns (C1, C2, and C4), two bucket samples (B1 and B4), one community fountain sample (BS2), and all lake samples except LPS2 (Fig. 2A). The most abundant resistance gene in water samples was tet(A), which was present in all samples, including the tap water samples, except for the posttreatment samples collected from the water treatment plant (Fig. 2B). mecA was detected in six different lake water samples (LBS1, LPS2, LPS3, LPS4, LBC2, and LBC3), two bucket samples (B3 and B4), and one cistern sample (C3) (Fig. 2C). We did not detect vanA in any of our samples.

Results for the detection of β-lactamase and carbapenemase genes using PCR are shown in Table 4. Among the β-lactamase genes, we detected the presence of TEM in most of the samples except T2 and T3 (tap water), C3 and C4 (cisterns), B1 and B2 (buckets), and LPS1 and LPS2 (plant source lake water). The SHV gene was found only in RBC1, and CTX-M was found only in two bucket samples (B3 and B4). We did not detect the OXA-1 or CMY-2 genes. As for the carbapenemase genes, we found the VIM-like carbapenemase gene in 9 out of 25 water samples tested from the community, including all eight lake water samples tested: OXA-48 and GES were found in four water samples each (OXA-48 in C1, C2, LBC3, and RBC1 and GES in C1, B2, B4, and LBC3) and NDM in two (B2 and B4) water samples. It is worth mentioning that in this study, we detected the presence of antibiotic resistance genes and not their expression.