Development and Validation of Liquid Chromatography-Tandem Mass Spectrometry Methods for Determination of Beta-Lactams , Macrolides , Fluoroquinolones , Sulfonamides and Tetracyclines in Surface and Drinking Water from Rio de Janeiro , Brazil

Two analytical methods for determination of five antibiotics classes in surface water and drinking water samples were developed and validated based on solid phase extraction followed by high-performance liquid chromatography coupled to tandem mass spectrometry. Two distinct chromatographic gradients were used according to the polarity of the different pharmaceuticals. The methods were applied for the quantification of 46 analytes belonging to beta-lactams, macrolides, fluoroquinolones, sulfonamides and tetracyclines classes. Validation results showed recoveries above 75% for the studied analytes in water samples. The method limits of detection calculated for the surface water and drinking water samples were, respectively, from 1 to 12 ng L and from 0.15 to 20 ng L. The method limit of quantification ranged from approximately 3 to 38 ng L for surface water samples and from 0.5 to 64 ng L for drinking water samples. The methods showed to be linear over the range of 25 to 1000 ng L with coefficients of determination greater than 0.94. Amoxicillin, cephalexin and sulfamethoxazole as high as 105 ng L were found in surface water and erythromycin, azithromycin and clarithromycin up to 35 ng L could also be found in surface water. Clarithromycin, cefaclor, oxacillin, sulfamethoxazole and troleandomycin were detected in the lower range up to 10 ng L in drinking water.


Introduction
Antibiotics represent one of the most used class of drugs worldwide 1 and correspond to the largest category of compounds used in human and veterinary medicine, as growth promoters or for therapeutic purposes. 2As regards to antimicrobials used in human medicine, non-prescribed medicines are consumed at home, and prescribed ones are consumed in hospitals and clinics. 3Individuals affected by infectious diseases use specific antibiotics, and after administration, the molecules are absorbed, distributed, metabolized partially, and finally excreted from the body.The metabolism eliminates substances in excess and other xenobiotics via a series of enzymatic biotransformations and converts them into more polar and hydrophilic compounds. 4hese substances were developed to be persistent, keeping its chemical properties, with a therapeutic purpose and after use, about 50 to 90% of a drug dose is excreted and persists in the environment. 5The occurrence of antibiotics in the aquatic environment and drinking water (DW) has raised questions about impacts on the environment and public health.The adverse effects caused by pharmaceutical compounds include aquatic toxicity, development of resistance in pathogenic bacteria, genotoxicity and endocrine disorders. 6,7everal methods have been developed to extract antibiotics in water.Currently, liquid chromatography-tandem mass spectrometry (LC-MS/MS) is the technique most widely used in analysis of drugs in complex environmental samples and have shown to be a sensitive analytical tool that leads to efficient results and lower detection (LOD) and quantification limits (LOQ) (in the range from μg L -1 to ng L -1 ), and generate reliable data in the identification of several molecules. 8LC-MS/MS comprises a separation and a detection technique that provides structural confirmation of the analyzed compounds. 8A good chromatographic separation is advisable in order to reduce matrix effects, which usually results in suppression or, less frequently, signal enhancement. 8To minimize matrix interference, water extracts are generally cleaned up and pre-concentrated by solid-phase extraction (SPE), mainly using HLB cartridges.The use of SPE cartridges may greatly influence the recoveries of target compounds. 8Sample preparation is a crucial step in environmental analysis.It is highly influenced by the physical and chemical analytes properties and by matrices.The main objectives of sample preparation are to extract and concentrate the analytes of interest, removing sample matrix interferences for subsequent chromatographic analysis.The whole analytical procedure typically includes five steps: sampling, sample preparation, chromatographic separation, detection and data analysis.The most important part of the analytical process is sample preparation because it can take more than 80% of the total analysis time. 8ome information about contamination of Brazilian aquatic environment by antibiotics has been published in the form of dissertations and theses, but scientific papers are very scarce.][11][12][13][14] The watershed of Guandu River has a fundamental role for Rio de Janeiro metropolitan region where approximately 12.2 million inhabitants live.This watershed is very important because it is the only option for subsistence and development of the Metropolitan Region of the State of Rio de Janeiro.Its waters supply the second largest metropolitan region of the country, and for several productive sectors, such as the steel, petrochemical, clothing, food and beverage industries, among others, and also as a water body for the collection of domestic and industrial sewage. 15he aim of this study was to develop and validate a methodology to determine the antimicrobial residues of beta-lactams (BL), macrolides (MC), fluoroquinolones (FQ), sulfonamides (SF) and tetracyclines (TC) classes in river surface water (SW) and DW samples in the state of Rio de Janeiro (Brazil).These methods were developed based on US EPA method 1694 16 to determine pharmaceuticals in environmental samples by high performance liquid chromatography-tandem mass spectrometry (HPLC-MS/MS).The methods was applied for quantification of 46 analytes of BL, MC, FQ, SF and TC classes in nine SW and ten DW samples collected in the state of Rio de Janeiro (Figure 1).

Standard solutions
Stock solutions of 1 mg L -1 were prepared in MeOH for MC, SF and TC, in ultrapure water for BL and in 0.03 mol L -1 NaOH for FQ.Stock solutions of DMC (1 mg L -1 ) and AMPID5 (1 mg L -1 ) were prepared using MeOH and ultrapure water, respectively.All stock solutions were stored at ≤ -70 °C.
DMC was used as internal standard/surrogate for SF and TC quantification.AMPID5 was used as internal standard for BL, MC and FQ.

Sampling and sample preparation
Water samples used for the development and validation An aliquot of 250 mL of each water sample was collected in polypropylene bottles, identified and transported under refrigeration to the laboratory for analysis.DW samples were taken from the tap at the National Institute for Quality Control in Health/Oswaldo Cruz Foundation (Rio de Janeiro, RJ), from the residences in the city of Barra Mansa and from the city of São Gonçalo (Rio de Janeiro).SW samples were collected from some rivers that make up the Guandu system (Guandu and Queimados rivers), which is the main source of the DW supply for the greater metropolitan area of Rio de Janeiro and Parado River in the Lidice District of Rio Claro, in the state of Rio de Janeiro, Brazil.A total of six samples were collected for validation, according to Table 1.Water samples were first filtered using 8 μm paper filters from Whatman (England), followed by 0.22 μm PVDF membrane filters (Millipore, Billerica, MA, USA).Water samples were used for method development experiments.They proved to be blank samples in previous analysis, for this reason they were used for all validation experiments.
Water samples collected for method application The method was applied to analyze nine SW samples collected in June 2016 from Guandu River (Paraíba do Sul, Piraí, Macacos, Queimados, Guandu and Santana rivers, Guandu lagoon mouth, Guandu main dam and adductor to Ribeirão das Lajes River).Figure 1 shows sampling sites position.In addition, ten DW samples were collected in July 2016, from residences in Rio de Janeiro State (Barra Mansa, Belford Roxo, Resende, Rio de Janeiro and Volta Redonda cities).
The sample codes and GPS coordinates are listed in Table 2 referring to the SW samples.The objective of this investigation was to determine the selected antimicrobials residues in surface and drinking water in the state of Rio de Janeiro, Brazil.

SPE procedure
The following procedure was developed based on US EPA method 1694 16 and on a previously published method for TC and SF analysis in river SW. 12 A 50 mL aliquot of each sample (SW and DW) was spiked at 100 ng L -1 (BL, MC, FQ, SF and TC) and spiked with 100 ng L -1 of internal standards/surrogate (DMC and AMPID5).Then, the samples were acidified to pH 2.5 with HCl, and 2 mL of 25 mg L -1 EDTA stock solution was added.For the sample DW, it was added 2 mL of 625 mg L -1 ASA to reduce any residual chlorine that had been added as a disinfectant.This solution was applied to an Oasis HLB ® cartridge previously conditioned with 3 mL of MeOH, 3 mL of ultrapure water and 3 mL of ultrapure water acidified to pH 2.5 with HCl.A manifold vacuum from Alltech (Deerfield, IL, USA) was used for SPE.The samples were percolated at a flow rate of approximately 3 mL min -1 .Cartridges were washed twice with 2 mL of ultrapure water and then dried under vacuum (-35 kPa) for 2 min.Antimicrobials were eluted with three portions of 2 mL methanol 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 under N 2 in a temperature up to 47 °C, using an evaporator with nitrogen flow (Pierce Reacti-Therm IIITM and Pierce Reacti VapTM III, Rockford, IL, USA).The dry residues were reconstituted with 1 mL of 0.1% FOA:MeOH (80:20, v/v) for TC and SF analysis (diluent 1) and 1 mL of MeOH:H 2 O (65:35, v/v) for BL, MC and FQ analysis (diluent 2), vortexed for 30 s and filtered through a 0.22 μm polyvinylidene fluoride (PVDF) syringe filter into amber auto-sampler vials.

LC-MS/MS instrumentation
An LC-MS/MS system consisting in a Shimadzu Prominence HPLC instrument (Shimadzu, Kyoto, Japan) equipped with a solvent delivery pump (LC-20AD), a quaternary gradient kit, a membrane degasser (DGU-20A5), an auto-sampler (SIL-20AC), a column oven (CTO-20AC), a system controller (CBM-20A) interfaced to a triple quadrupole mass spectrometer (API5000, Applied Biosystems/MDS Sciex, Foster City, CA, USA) with a TurboIonSpray ® ESI source was used.Analyst ® V1.4.2 LC/MS software was used for data acquisition.Positive electrospray ionization technique (ESI+) in multiple reaction monitoring (MRM) acquisition mode was used to monitor two ions for each substance.Nitrogen was employed as nebulizer gas (Gas 1, 40 psi), dryer gas (Gas 2, 40 psi), collision-activated dissociation (CAD) gas (6 a.u.) and Curtain™ gas (10 psi).Other parameters selected during automatic tuning were: ion spray potential = 5000 V; source temperature = 500 °C (SF and TC), 550 °C (BL, MC and FQ); entrance potential = 10 V.The column temperature  3. 12,17 Fragmentation studies with beta-lactams and fluoroquinolones for tuning the mass spectrometer were performed with mixed standard solutions at concentrations between 50 and 100 ng mL -1 in MeOH:1% FOA (50:50, v/v).ESI+ in multiple reaction monitoring (MRM) acquisition mode was used to monitor two ions for each substance.MRM experiments for TC analysis in electrospray positiveion mode (ESI+) were described by Spisso et al.; 17 for SF by Monteiro et al.; 12 for MC by Spisso et al. 18 and Costa et al., 19 and the analytical conditions used were listed in Table 3.

Validation
The validation of optimized method was performed according to protocol for EPA 20 approval of new methods for organic and inorganic analytes in wastewater and DW.Validation method was further evaluated in terms of sensitivity, initial precision and recovery (IPR), intermediate precision and linearity.
Sensitivity (method limits of detection (LOD) and method limits of quantification (LOQ)) The method limits of detection (LOD) is calculated using seven replicates of river (SW) and drinking water (DW) samples spiked in concentration of 20 ng L -1 .The LODs were calculated by multiplying the standard deviation from the seven measurements by the Student's t-test value for six degree of freedom at 99% confidence level (3.143).The LOQ were calculated by multiplying 3.18 times the LOD.

Initial precision and recovery (IPR)
The IPR for each compound was determined spiking four replicates at 100 ng L -1 in three samples of water from different origins (DW1, DW2, DW3, SW1, SW2 and SW3).A total of twelve samples of each type (DW and SW) were analyzed.The spiked samples were proceeded by SPE and then analyzed by LC-MS/MS.
The overall recovery was obtained comparing the analyte response in the extract of water samples (SW and DW) post-extraction reconstituted with 1 mL of 100 ng L -1 solutions (BL, MC, FQ, SF, TC, DMC and AMPID5) prepared with respective dilution solvents, 1 mL of diluent 1 and 1 mL of diluent 2, and the theoretical concentration in the final extract assuming 100% SPE recovery.Precision was assessed with respect to repeatability (intraday precision) and intermediate precision.

Linearity
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 the SW and DW samples.

Results and Discussion
Development of the LC-MS/MS method MRM acquisition mode is the most suitable for quantification due to its sensitivity and specificity.Declustering potential (DP), collision energy (CE) and The two most abundant fragment ions were monitored for each compound.For target analytes, the most abundant transition was used for quantification purposes, whereas the second was used to confirm the identity of the substances.
The transition ERY-H 2 O could be monitored, because at pH below 7, ERY is immediately converted into its main degradation product ERY-H 2 O. 21 All compounds showed a good chromatographic peak resolution.
Two chromatographic methods were developed to obtain an increase in substances sensitivity, because the physicochemical properties of the five antimicrobials classes analyzed were different.Both methods were used according to polarity and extraction of different pharmaceuticals.

SPE procedure
Sample preparation is a crucial step in environmental analysis.It is highly influenced by the physical and chemical properties of the studied analytes and the matrices.The main objectives are to concentrate the analytes in the sample, remove matrix interferences and prepare the analyte in the form suitable for subsequent chromatographic analysis.Usually, the sample preparation step includes adjusting the pH of the solution, plus the use of a chelator (EDTA) followed by an extraction procedure, extract treatment and final preparation for the following chromatographic analysis.In most of the methods presented in the literature, Oasis HLB ® cartridge has been used.This cartridge usually works at a neutral pH.][23][24][25][26] EDTA was used as a chelating agent, it is recommended in the analysis of antibiotic residues in environmental samples.A chelating agent was added to water samples, prior to extraction, in order to chelate metals that are found in water, making possible to achieve good extraction efficiencies. 21scorbic acid (ASA) was added to remove residual chlorine in DW, because it can react with some antibiotics, including CPF, CTC, DC, ERY, OTC, sulfamethoxazole and TC. 27,28It is important to do the removal of free chlorine in water samples, because this fact leads to a more precise analysis and resulting in a reliable data, without affecting the stability of the antibiotics in water.
DMC and AMPID5 were used as surrogate standards, they were added to the samples before extraction and were also used for the quantification of the samples.Internal standard/surrogate was therefore added to the sample to compensate the losses originated from both the sample preparation procedure and from matrix effects.
Table 5 presents the comparison between the developed method and the 1694 US EPA method. 16The lower amount of sample, consumables and time show that as a result, the developed method is faster and cheaper than the 1694 US EPA method. 16lidation Drug residues are frequently detected and quantified in aquatic environments.Unreliable analytical data can lead to misinterpretation and wrong decision-making.Therefore, the validation of the analytical method is important to obtain a correct analysis of the possible effects of these compounds on human health, as well as on non-target organisms.Methods developed by laboratories, that is, non-standardized, should be validated.Therefore, the validation of the analytical method is an important step to assure the reliability of the results and hence to enable a correct analysis of the possible effects of these compounds on human health, as well as on non-target organisms.

Sensitivity (LOD and LOQ)
The LOD and LOQ were estimated from the injection of spiked real samples (SW and DW).Results for each matrix are presented in Table 6 (SW) and Table 7 (DW).LODs calculated for SW samples were from 1 to 12 ng L -1 and for DW samples were from 0.15 to 20 ng L -1 .LOQs ranged from approximately 3 to 38 ng L -1 for SW samples and from 0.5 to 64 ng L -1 for DW samples.It is worth mentioning that in the validated method, low LODs and LOQs were achieved for all antibiotics, even though low sample volumes were used for sample preconcentration.By reducing sample volume of complex samples such as river water samples, a decrease in matrix effects may be achieved.

IPR
The achieved recoveries for all target compounds ranged from 49 to 117% and from 50 to 110% for SW and DW samples, respectively (Tables 6 and 7).Good performance with recoveries above 75% among 80% of the 46 analytes for the surface and drinking water sample was achieved.Only PENG samples showed recovery rates below of 40%.This fact can be explained by their instability in water, related to their chemical structure. 21High recoveries obtained for fluoroquinolones can be explained by the retention of these antibiotics in acidic conditions. 21,24][20][21][22][23] In both matrices, the relative standard deviation (RSD) obtained are less than 58% for all analytes for repetitivity and repeatability, which is lower than values reported by   1694 US EPA method 16 and high deviations may have occurred due to the validation of three different water source types.RSD values were acceptable, considering the specifications laid down by European Commission 29 and by Codex Alimentarius Commission. 30he internal standard/surrogate recoveries and standard deviations for antibiotics in both SW and DW water are presented in Tables 6 and 7, respectively.

Linearity
The linearity was evaluated with matrix-matched analytical curve at six concentration levels.The results showed good linearity over the range of 25 to 1000 ng L -1 with coefficient of determination (R²) greater than 0.97 for SW and greater than 0.94 for DW.

Method application
The method was applied to the analysis of nine SW and ten DW samples.According to the results, showed in Table 8, compounds were found present in eight out of nine SW samples.Antibiotics were not detected only in the adductor to the Ribeirão das Lajes SW samples.The results showed levels of AMOX, CFLX and SMZ as higher as 105 ng L -1 .Also, concentrations of ERY, AZI, CLA up to 35 ng L -1 could be found in the river water.
CLA, CFCL, OXA, SMZ and TRO were detected in the lower range up to 10 ng L -1 in DW water (Table 9).
Figure 2 shows the MRM chromatograms of a contaminated SW samples with a maximum concentration of AMOX, CFLX, SMZ, ERY, AZI and CLA.The other antibiotics analyzed were below the method limits of detection (LOD).
Many compounds have been found worldwide in several different types of water.A recent review described that among 22 pharmaceuticals detected in SW around the world, about 13 are common in Brazil and other countries, being the most commonly detected antibiotics. 9Studies conducted by Locatelli et al. 10 and Monteiro et al. 11 in rivers located in São Paulo, Brazil, showed that NOR, AMOX, CFLX, CPF, SMZ, TC, trimethoprim, OTC and florfenicol were determined with a concentration between 2.2 and 484 ng L -1 and in a river in Rio de Janeiro, Brazil, OTC and SMZ were detected in concentration between 44.1 and 467 ng L -1 , respectively. 12In SW samples from Dilúvio Creek in Porto Alegre, Brazil, SMZ, CPF, NOR and AZI were detected between 15.7 and 572 ng L -1 . 13he non-detection of TCs in water samples may be due to their strong adsorption on organic matter and, although TCs are very soluble in water and are weakly adsorbed by biomass, mechanisms like metal complexation likely played a significant role in the sorption of TCs into solids. 11,14Similarly, FQs were not found in water, possibly because these molecules are strongly adsorbed by sediment, especially when the concentration of Ca 2+ and Mg 2+ is high. 14

Conclusions
The analytical method developed and validated, based on SPE followed by LC-MS/MS analysis, for the simultaneous extraction of beta-lactams, macrolides, fluoroquinolones, sulfonamides and tetracyclines classes, showed good performance.The achieved recoveries for all target compounds ranged from 49 to 117% and from 50 to 110% for SW and DW samples, respectively.In both matrices, the obtained RSD are less than 58% for all analytes for repetitivity and repeatability, which is lower than values reported by 1694 US EPA method.As a result, a fast and cost-effective method was developed.
The developed and validated method in this study was applied to evaluate the occurrence of compounds in SW and DW from Rio de Janeiro.The results showed that several compounds are occasionally present at high levels, indicating that the evaluated rivers receive uncontrolled loads of wastewater of different sources and/or that these compounds are not efficiently removed in the wastewater treatment plant.The results highlight the worries related to the presence of these compounds in the environment, because of their possible ecotoxicological effects on non-target organisms and on human health arising from the food chain via the water cycle.

Figure 1 .
Figure 1.Map of the state of Rio de Janeiro with the sampling locations.

Table 1 .
Sample collection for validation

Table 2 .
Sampling site details and GPS coordinates (surface sample (SW) samples)

Table 3 .
Gradient elution programs for sulfonamides (SF), tetracyclines (TC), beta-lactams (BL), macrolides (MC), fluoroquinolones (FQ) cell exit potential (CXP) values for MC, SF, TC, BL and FQ precursor/product ion pairs obtained in MRM mode are shown in Table4.For BL and FQ, only protonated molecules [M + H] + were observed and selected as precursor ions, and no adducts were noted.
a A, B, C: mobile phases with 0.1% FOA prepared using H 2 O, ACN and MeOH, respectively.collision

Table 5 .
16mparisons between the developed method and the 1694 US EPA16

Table 6 .
Performance data for pharmaceuticals in surface water (SW) (cont.)

Table 7 .
Performance data for pharmaceuticals in drinking water (DW) (cont.)