Samples collection
Five collection sites were selected in Guandu Watershed (22°50'22.11" S and 43°36'36.70" O) (Queimados, Guandu, Piraí and Macacos rivers and Guandu Dam), and five sites in São João Watershed (22°37'36.60" S and 42°17'54.36" O) (Capivari, Bacaxá and São João rivers, São João river mouth and Juturnaíba Dam). Samples (5–10 L from each site) were collected in sterile bottles, six months apart (January and june/2015). The samples were refrigerated until processed in the laboratory. In addition, in January/2015, 11 tap drinking water samples (5–10 L from each point) were collected in different neighborhoods from Rio de Janeiro city (Centro, Copacabana, Ilha do Governador, Jacarepaguá, Jardim Botânico, Leblon, Realengo, Santa Teresa, Vista Alegre) and from the cities of Unamar and Itaguaí in Rio de Janeiro State.
Antimicrobial Residues Detection
Chemicals and materials
Amoxicillin tryhidrate (AMOX), ampicillin (AMPI), cefaclor (CFCL), cefadroxil (CFDX), cefalexin hydrate (CFLX), cefazolin (CFZL), clarithromycin (CLA), ciprofloxacin hydrochloride (CPF), norfloxacin (NOR), tetracycline hydrochloride (TC) and sulfamethoxazole (SMZ) were chemical reference substances from the Brazilian Pharmacopeial Convention (Santa Maria, RS, Brazil). Azithromycin dehydrate (AZI), roxithromycin (ROX), spiramycin (SPI), oleandomycin (OLE), tilmicosin (TILM), and cefquinome sulphate salt (CFQN) were obtained from Dr. Ehrenstorfer (Augsburg, Germany). Oxytetracycline (OTC), doxycycline hyclate (DC), hydrochloride salts of chlortetracycline (CTC) and demeclocycline (DMC), dapsone (DAP), sulfacetamide (SCT), sulfadimethoxine (SDM), sulfamerazine (SFM), sulfamethazine (SMT), sulphaquinoxaline (SQN), sulfathiazole (STZ), tylosin tartarate (TYL), troleandomycin (TRO), erythromycin (ERY), cephapirin sodium salt (CPPN), ceftiofur (CFTF), cefoperazone (CFPZ), benzylpenicillin sodium salt (PENG), oxacillin sodium salt hydrate (OXA), moxifloxacin (MXF) and ofloxacin (OFX) were supplied from US Pharmacopeial Convention (Rockville, MD, USA). Phenoxymethylpenicillin potassium salt (PENV), cloxacillin sodium salt hydrate (CLOX), dicloxacillin sodium salt hydrate (DCLOX) and nafcillin sodium salt (NAFC) were supplied from WHO Collaborating Centre for Chemical Reference Substances (Stockholm, Sweden). Methacycline (MTC), 4-epioxytetracycline (4-EOTC), 4-epitetracycline (4-ETC) and 4-epichlortetracycline hydrochloride (4-ECTC) were acquired from Acros (Pittsburgh, PA, USA). Ampicillin-d5 (AMPID5) was purchased from Purity Grade Standards (San Francisco, CA, USA). Desacetylcephapirin (DESAC) was supplied from Bristol-Myers Squibb (New York, USA).
Methanol (MeOH) and acetonitrile (ACN) HPLC grade, hydrochloric acid (HCl) and formic acid (FOA) analytical grade were purchased from Merck (Darmstadt, Germany). Sodium hydroxide (NaOH), acetone (ACE) and ascorbic acid (ASA) were purchased from Merck (Darmstadt, Germany). Ethylenediaminetetracetic acid disodium dihydrate (EDTA) was acquired from Calbiochem (Gibbstown, NJ, USA). Ultrapure water was obtained from a Milli-Q purification system (Millipore, Bedford, MA, USA).
Solid-phase extraction (SPE) was performed with 60 mg Oasis® HLB cartridges from Waters Corp. (Milford, MA, USA). Membrane filters of polyvinylidene fluoride (PVDF) with pore size 0.22 µm were purchased from Millipore (Billerica, MA, USA).
Preparation of standard solutions
The stock standard solutions were prepared to obtain a concentration of approximately 1000 µg mL-1. Stock solutions of β-lactams (BL) were prepared in water while those of fluoroquinolones (FQ) in a 0.03 mol L-1 NaOH. Finally, macrolides (MC), sulfonamides (SF) and tetracyclines (TC) solutions were prepared in MeOH. The amount weighed for each standard was calculated considering purity, water content and free acid/basic corrections. The solutions were transferred to microtubes and stored in a freezer at -70°C or below. DMC and AMPID5 were used as internal standards.
Intermediate and working standard solutions were freshly prepared at several concentrations by appropriate dilution of stock standard solutions.
Analytical method
The extraction methodology for antimicrobial residues was based on the standard method from the United States Environmental Protection Agency (US EPA) - Method 1694 71 ,with modifications described by Monteiro et al.72
Samples were previously filtered through filter paper and 0.22 µm PVDF membrane filter. A 50 mL aliquot of each sample was spiked with 100 ng L-1 of the internal standards, acidified to pH 2.5 with HCl, and 2 mL of 25 mg L-1 EDTA stock solution was added. For drinking water samples, 2 mL of 625 mg L-1 ASA was added to reduce any residual chlorine. This solution was applied to an Oasis® HLB cartridge that had been previously conditioned in sequence with 3 mL of MeOH, 3 mL of ultrapure water and 3 mL of ultrapure water acidified to pH 2.5 with HCl. After being washed twice with 2 mL of water, SPE cartridges were vacuum-dried (-35 kPa) for 2 min. Antimicrobials were eluted with three portions of 2 mL MeOH and one portion of 2 mL ACE, using gravity flow only. 4 mL aliquots of the eluate were transferred to two centrifuge tubes and evaporated to dryness with N2 in a temperature up to 47°C. The residues were reconstituted with 1 mL of 0.1% FOA:MeOH (80:20, v/v) for TC and SF analysis and 1 mL of MeOH:H2O (65:35, v/v) for BL, MC and FQ analysis, vortexed for 30 s and filtered through a 0.22 µm polyvinylidene fluoride (PVDF) syringe filter into amber auto-sampler vials.
The chromatographic analysis was performed on a Shimadzu Prominence HPLC (Kyoto, Japan) equipped with a quaternary pump (LC-20AD), a membrane degasser (DGU-20A5), an auto-sampler (SIL-20AC), a column oven (CTO-20AC) and a system controller (CBM-20A) interfaced to a triple quadrupole mass spectrometer (API5000, Applied Biosystems/MDS Sciex, Foster City, CA, USA) with the TurboIonSpray® source. Analyst® V1.4.2 LC/MS control software was used. The analytical column was a Pursuit™ C18 RS (100 mm × 2 mm i.d., 3 µm particle size, 200 Å), with a respective guard column (Varian, Lake Forest, CA, USA). Mobile phases A, B and C were prepared using water, ACN and MeOH, respectively, all with 0.1% FOA. A gradient elution program for TC and SF method was used with a flow rate of 0.15 mL min–1 at 25 ºC and for BL, MC and FQ another gradient elution was used with a flow rate of 0.30 mL min–1 at 35 ºC. The injection volume was 25 µL for both methods. The auto-sampler was set at 4°C. Positive electrospray ionization technique (ESI+) in Multiple Reaction Monitoring (MRM) acquisition mode was used to monitor two ions for each substance.
A six-point calibration set was freshly prepared by spiking varying levels of working standard solutions in ultrapure water. The analytical curves for all analytes in the concentration range from 25 to 1000 ng L-1 were constructed in order to quantify the analytes in samples.
The chromatographic peaks were integrated with the IntelliQuan algorithm of the Analyst® software. A signal-to-noise ratio of the peaks equal or greater than 3:1 for at least 2 transitions was required for detection. Relative retention times and relative abundances between quantification and confirmation MRM transitions in both matrix-fortified standards and samples were used as confirmation criteria according to recommended tolerances in 2002/657/EC Commission Decision, that was in place when this work was carried out 73. Samples were considered contaminated when analytes were detected according to identification criteria by liquid chromatography tandem mass spectrometry and the concentration values exceeded the limits of detection (LOD).
Bacterial Community Composition
Water samples were filtered through 0.22 µm cellulose acetate membranes (Millipore, USA) under aseptic conditions. DNA was extracted from the filters using the PowerWater DNA Isolation Kit (Qiagen Science, USA). For the preparation of the amplicon library, the DNA was quantified using a Qubit 2.0 Fluorometer (ThermoFisher Scientific, USA), and samples were diluted to achieve the concentration of 5 ng µL-1 per sample. The V4 hypervariable region of the 16S rRNA gene was amplified by PCR using the primers 16Sf (5 '- GTGCCAGCMGCCGCGGTAA-3') and 16Sr (5'- GGACTACHVGGGTWTCTAAT-3') with the appropriate barcodes and adapters74. PCR products were purified using the ChargeSwitch™ PCR Clean-Up Kit (ThermoFisher Scientific, USA). Each individual sample library was diluted to 4 nM and then pooled and paired-end sequenced on a MiSeq system (Illumina Inc. USA), using the 500 cycles MiSeq Reagent v2 Kit.
Quality analysis of raw reads was carried out with FastQC software75 and the filtering of sequences with an average quality equal to or greater than 20 was performed by the PRINSEQ program76. Data analysis was performed using QIIME (Quantitative Insights Into Microbial Ecology) 1.9.177. The data were compared with the SILVA Ribosomal RNA database (non-redundant) 132 release75 with a maximum e-value of 1e-5, and a minimum identity of 99%, which generated a table with taxonomic groups. Statistical analyzes such as Alpha and Beta diversity were calculated using the MicrobiomeAnalyst web platform (https://www.microbiomeanalyst.ca/)78,79. The diversity of the bacterial communities was assessed using the Chao1 index and Simpson's index calculated for OTUs with the evolutionary distance of 0.01 (or 99% 16S rRNA gene sequence similarity). Principal coordinate analysis (PCoA) between bacterial communities present in the Guandu, São João watersheds and in drinking water was built using the Jaccard method with PERMANOVA and using the bacterial OTUs.
Plasmidome
Plasmid DNA (pDNA) was extracted from the filters by alkaline lysis using the Plasmid Mini Kit (Qiagen Science, USA) according to the manufacturer's protocol. pDNA was precipitated with isopropanol and washed with 70% ethanol. To remove possible traces of genomic DNA the precipitate was treated with ATP-dependent Plasmid Safe DNase (Epicentre, USA) according to manufacturer's instructions50. The pDNA was quantified using a Qubit 2.0 Fluorometer (ThermoFisher ScientificTM, USA) according to the manufacturer's manual.
A pDNA sequencing library was prepared using the Nextera XT DNA Library Prep Kit (Illumina Inc. USA) following the manufacturer's recommendations. Paired-end sequencing was performed with the 600 cycles Miseq Reagent Kit v.3 on the MiSeq platform (Illumina Inc. USA). Sequence quality checks were performed with the FastQC software75 and sequence filtering with an average quality of 20 or higher was performed by PRINSEQ80. Contigs were assembled with SPAdes version 3.1381,82 using the metaSPAdes pipeline.
The sequences were analyzed by the MG-RAST (Meta Genome - Rapid Annotation using Subsystem Technology) platform83, where the annotation can be viewed in several different categories, including subsystems. A subsystem can be understood as a set of functional roles that implement a certain biological or structural process84. The subsystems are classified into hierarchical levels, so that level 1 includes general catabolic and anabolic functions (for example, DNA metabolism), and levels 2 and 3 contain more specific pathways, such as antimicrobial resistance and other compounds85.
In addition, the sequences were compared against the Comprehensive Antibiotic Resistance Database (CARD) database86,87 with DIAMOND88. Only alignments with an e-value < 1e5, coverage > 60% and amino acid identity > 30% were considered89.