2738 ˜ Juan Antonio Ocana1 ´ Gonzalez ´ 1,2 Mar´ıa Ramos-Payan 1,3 ´ Rut Fernandez-Torres ´ ´ 1 Manuel Callejon-Moch on 1 ´ ´ Miguel Angel Bello-Lopez 1 Department

of Analytical Chemistry, Universidad de Sevilla, Seville, Spain 2 Department of Analytical Chemistry, Lineberger Cancer Center, University of North Carolina, Chapel Hill, NC, USA 3 Research Centre of Health and Environment (CYSMA), University of Huelva, Huelva, Spain

Received April 4, 2014 Revised June 30, 2014 Accepted July 5, 2014

J. Sep. Sci. 2014, 37, 2738–2744

Research Article

Hollow-fiber liquid-phase microextraction for the direct determination of flumequine in urban wastewaters by flow-injection analysis with terbium-sensitized chemiluminescence A flow-injection analysis chemiluminescence method based on the enhancement effect of the flumequine-Tb(III) complex on the weak native emission of the Ce(IV)-Na2 SO3 system has been developed for the determination of flumequine. The method includes a cleanup and preconcentration stage (750-fold) of the sample by hollow-fiber liquid-phase  microextraction using an Accurel Q 3/2 polypropylene hollow fiber impregnated with 1octanol as the supported liquid membrane. The obtained 50 ␮L acceptor phase was injected in a 1 mM Tb(III) + 4 mM Ce(IV) in 5% v/v H2 SO4 stream and mixed with a 2 mM Na2 SO3 stream before its introduction into the flow cell. The chemiluminescence signal was linear in the 0.3–15 ng/mL range, with detection and quantitation limits of 0.1 and 0.3 ng/mL, respectively. The method allows the selective extraction and determination of flumequine in wastewater samples, using simpler and lower-cost instrumentation and with shorter extraction and analysis times than traditional high-performance liquid chromatography analysis. R

Keywords: Chemiluminescence / Flow injection / Flumequine / Hollow fibers / Wastewater DOI 10.1002/jssc.201400383

1 Introduction Flumequine (FLM) or 9-fluoro-6,7-dihydro-5-methyl-1-oxo1H,5H-benzo[ij]quinolizine-2-carboxylic acid (Fig. 1) is a firstgeneration quinolone antibacterial with a broad spectrum activity against both gram-positive and gram-negative bacteria. As other quinolones, its pharmacological effect is based on the inhibition of bacterial DNA gyrase. FLM is mainly excreted in urine, with an elimination half-time of 6–7 h and two main metabolites: glucuronide conjugate and 7hydroxyflumequine [1]. FLM is frequently used in fish, poultry, and cattle husbandry for the treatment and prevention of diseases. Despite its usefulness, FLM presence in animal tissues and its unexpected introduction in the consumer diet or in natural or wastewaters can lead to the development of bacteria with resistance against this antibiotic. Because of this risk, the use of FLM is regulated in many countries by the establishment of general or specific maximum residue limits in environmental media and in food products. Specifically, EU legislation has ´ Correspondence: Dr. Miguel Angel Bello-Lopez, Department of Analytical Chemistry, Faculty of Chemistry, University of Seville, Seville 41012, Spain E-mail: [email protected] Fax: +34-954557168

Abbreviations: CL, chemiluminescence; FIA, flow-injection analysis; FLM, flumequine; HF-LPME, hollow-fiber based liquid phase microextraction

 C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

established maximum residue limits values for unmetabolized FLM residuum, since this is the main microbiologically active compound found in most tissues [2]. There is extensive literature concerning the analytical determination of fluoroquinolones (including FLM) in environmental samples and in animal tissues/food products. The determination of these antibiotics in environmental or wastewaters is usually based on LC separation with MS detection, such Q-TOF [3, 4], triple quadrupole [5, 6], ion trap [7], and triple quadrupole linear ion trap [8]. With respect to animal tissues and food products, published methods are also usually based on LC separation. This way, FLM (among other quinolones) has been determined using UV detection in milk [9, 10] and bovine liver and porcine kidney [11], fluorimetric detection in eggs [12, 13], muscle and milk [12], and aquatic products [14], while MS detection has been applied to honey [15], bovine and lamb feed [16], eggs [17], chicken liver [18], shrimp [19], and pig tissues [20]. Finally, other techniques as CE for milk [21], chicken [22] and varied foodstuff samples [23], and TLC for milk samples [24] have been applied for the determination of FLM. Chemiluminescence (CL) methods have been successfully applied to the determination of fluoroquinolone antibiotics in different matrices (pharmaceuticals, biological fluids, animal tissues, etc.) by coupling to a flow-injection system [25–27] or other analytical systems as HPLC [28–30] or CE [31]. Analytical methods based on CL detection combined with flow-injection analysis (FIA–CL) have some advantages over HPLC, mainly with respect to analysis time (since carrier www.jss-journal.com

Sample Preparation

J. Sep. Sci. 2014, 37, 2738–2744

2739

2 Materials and methods 2.1 Chemicals and reagents All chemicals were of analytical reagent grade. All solutions and dilutions were prepared with ultrapure water from a Milli-Q Plus water purification system (Millipore, Billerica, MA, USA). Dihexyl ether, 1-octanol, FLM, and TbCl3 ·6H2 O were purchased from Fluka-Sigma-Aldrich (Madrid, Spain) and the rest of chemical reagents were obtained from Merck  (Darmstadt, Germany). The Accurel Q 3/2 polypropylene hollow fibers (600 ␮m id, 200 ␮m wall thickness, and 0.2 ␮m pore size) were purchased from Membrana (Wuppertal, Germany). Methanolic stock solution of 200 ␮g/mL FLM was prepared and stored at 4⬚C; working solutions of FLM were prepared daily by dilution with water from this stock solution. Both Ce(IV) and Tb(III) in 0.5% v/v H2 SO4 solution and Na2 SO3 solution were prepared daily. R

Figure 1. Chemical structure of FLM.

flows are higher than those usually found for mobile phases) and cost (due to of the use of aqueous solvents instead of organic solvents, peristaltic pumps instead of high-pressure pumps, etc.). Many quinolones can enhance the weak native CL signal obtained from the redox reaction between Ce(IV) and sulfite. The observed CL emission is based on an energy transfer between the excited products of the Ce(IV)-sulfite reaction and the quinolone molecule. However, this signal can be further enhanced by the presence of a trivalent lanthanide ion as Tb(III) and the formation of a stable lanthanide– quinolone complex. In this case, the CL signal is based on an energy transfer between the excited quinolone and metal cation [32, 33]. FLM forms complexes with Tb(III) [34–36], so the described CL mechanism is a potential method for its determination. FIA–CL has been applied to the determination of both inorganic and organic pollutants in wastewaters, taking advantage of the sensitivity of the technique, simple instrumentation, and short measurement times [37]. Nevertheless, the absence of analyte matrix separation in FIA–CL makes this technique easily susceptible to interference effects (as quenching effects on the CL emission) in the analysis of complex matrices as urban wastewaters. The application of hollow-fiber LPME (HF-LPME) in the pretreatment stage prior to their CL determination, could be an interesting option, since this procedure can achieve an effective cleanup for several kinds of samples and it also leads to a preconcentration of the analyte, increasing the sensitivity of the measurements [38–40]. This way, HF-LPME has been applied to the determination of eight fluoroquinolones (including FLM) in different environmental water samples by HPLC with diode-array and fluorescence detection [41]. The objective of this work, on the other hand, is the development of an HF-LPME method for the selective extraction of FLM (even in the presence of other quinolone-based antibiotics), in order to significantly reduce the required analysis time and to allow the use of simpler and lower cost instrumentation (FIA–CL instead of HPLC). The proposed procedure has been applied to the direct determination of this pharmaceutical in wastewaters.  C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

2.2 Wastewater samples Wastewater samples were obtained from “Guadalquivir”ALJARAFESA Wastewater Treatment Plant (Palomares del R´ıo, Seville, Spain). This plant receives mainly urban wastewaters, with a capacity of 100 000 inhabitants and a discharged flow of 12 433 313 m3 /year (2008 data). From this plant, four types of samples were collected, corresponding to the different treatment stages: (a) influent (raw water), (b) after the primary sedimentation tank, (c) after the aeration tank, (d) effluent (treated water after anaerobic digestion). Once collected, these samples were filtered, successively, through a GDU1 glass fiber filter (10–1 ␮m, Whatman, Maidstone, UK) and a Pall Nylaflo TM nylon membrane filter (0.45 ␮m, Pall Corporation, Ann Arbor, MI, USA). Finally, samples were adjusted to pH 2.0 with HCl and stored at 4⬚C for no more than a week prior to their analysis.

2.3 Instrumentation CL system used in this work is shown in Fig. 2. Flow rate was achieved with a Minipuls 3 peristaltic pump (Gilson, Middleton, WI) with polytetrafluoroethylene tubing connectors (0.8 mm id) and two carrier-stream channels. A Rheodyne injector with a 20 ␮L loop allowed the introduction of the sample in the flow-stream, using an HPLC syringe. The obtained CL signal was measured by a ChemLab CL2 Chemiluminescence Detector model (Camspec, Cambridge, UK) with a 60 ␮L, 5 mm path length quartz flow cell. Before their introduction into the measurement cell, carrier streams were mixed using a Y-shaped element. The obtained CL FIA-grams were acquired and integrated using the Clarity Lite software (DataApex, Prague, Czech Republic). The HPLC system used for the optimization of the HF-LPME procedure consisted of a LaChrom VWR-Hitachi (Barcelona, Spain) with a L-2130 quaternary pump, a www.jss-journal.com

2740

J. Sep. Sci. 2014, 37, 2738–2744

J. A. Ocana-Gonz alez et al. ˜ ´

Figure 2. Scheme of the FIA–CL system.

Rheodyne manual injection valve Model 7725i fitted with  a 20 ␮L sample loop, and a LichroCART75-4 Purosphere STAR RP-18e 3 ␮m (75 × 4.0 mm id) column (VWR, Darmstadt, Germany). R

2.4 Supported liquid membrane preparation and extraction procedure

+ 4 mM Ce(IV) prepared in 5% v/v H2 SO4 . This stream was mixed with a stream of 2 mM Na2 SO3 (channel B), with a 1.0 mL/min total working flow. The analytical signal was calculated as the area of the obtained peak.

3 Results and discussion

 R

An Accurel Q 3/2 polypropylene hollow fiber was cut into 27 cm pieces, washed with acetone for 30 s in an ultrasonic bath in order to eliminate all residues and dried at room temperature. The dried fiber was soaked in 1-octanol for 10 s, and rinsed with water in the ultrasonic bath for 30 s in order to remove the excess of organic solvent. Once the organic liquid membrane was formed, the lumen of the fiber piece was filled with 50 ␮L of acceptor phase (pH 12.5 aqueous solution) using an HPLC syringe. The opened ends were sealed by the application of a hot sol dering tool and covered with plastic film (Parafilm , Pechiney Plastic Packaging Company, Chicago, IL, USA). FLM extraction was carried out by the introduction of the prepared hollow fiber in 50 mL of sample solution (20% w/v Na2 SO4 , with a pH adjusted to 3.7). The sample was stirred in a 50 mL glass beaker for 1 h with an ANS-00/1 Science Basic Solutions magnetic stirrer (Rub´ı, Barcelona, Spain) at 600 rpm. Once the extraction time was reached, the acceptor phase was injected into the CL system without further manipulation using an HPLC syringe. R

2.5 CL determination Using the FIA system described previously, 50 ␮L of acceptor phase was injected into a stream (channel A) of 1 mM Tb(III)  C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

3.1 Optimization of the HF-LPME extraction The optimization and evaluation of experimental conditions for the HF-LPME extraction (donor and acceptor pH, organic solvent, concentration of salting-out agent, extraction time and stirring speed) were carried out using a previously described HPLC determination with fluorescence detection [41]. In a series of preliminary studies, both dihexyl ether and 1octanol were tested as the organic solvent for the supported membrane formation; the best results were achieved for the latter, so 1-octanol was selected as organic solvent for the rest of experiments. The influence of salting-out agents (NaCl and Na2 SO4 ) in the donor phase on the extraction was also tested, and the best results were achieved in presence of 20% Na2 SO4 . With respect to stirring speed, it was found that >600 rpm speeds lead to the apparition of vortexes in the donor solution, so this speed was selected for further studies. Under these conditions, extraction experiments were carried out using acceptor phase pH in the 10.0–13.0 range and donor phase pH in the 2.0–7.0 range (Fig. 3). Obtained results shown that FLM was selectively extracted from a mixture with other fluoroquinolones using a pH 3.7 donor phase and pH 12.5 acceptor phase, so these conditions were selected for the optimized procedure. Finally, the influence of extraction time was studied. It was found that FLM recovery reached a www.jss-journal.com

J. Sep. Sci. 2014, 37, 2738–2744

Figure 3. Effect of donor phase pH, acceptor phase pH, and extraction time in the HF-LPME of FLM.

maximum value after 1 h, so this extraction time was selected for further experiments (Fig. 3).

3.2 Optimization of the CL determination Preliminary experiments showed that the native CL emission of the Ce(IV)-sulfite-Tb(III) system was nearly negligible, but the presence of FLM led to a notable increase in the obtained signals. Thus, the following experimental parameters were optimized: reagent concentrations (Ce(IV), sulfite and Tb(III)), acidity of the Ce(IV)/Tb(III) solution and flow rate.  C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Sample Preparation

2741

Figure 4. Influence of Ce(IV), Na2 SO3 , and Tb(III) concentrations on the CL signal of a 5 mg/mL FLM solution.

The Ce(IV)/Tb(III) solution was prepared in acid medium in order to obtain a stable solution of these cations. A minimum of 0.25% v/v H2 SO4 was required for the complete dissolution of both salts, but >0.5% v/v H2 SO4 concentrations led to a decrease in the CL signal. The influence of Ce(IV), Tb(III), and sulfite concentrations was studied in the 1–6, 0.1–1.8, and 1–5 mM ranges, respectively. The results are presented in Fig. 4. With these results, the following experimental conditions were selected as optimized procedure: 4 mM Ce(IV), 1 mM Tb(III), and 2 mM sulfite. www.jss-journal.com

2742

J. Sep. Sci. 2014, 37, 2738–2744

J. A. Ocana-Gonz alez et al. ˜ ´

Finally, the effect of flow rate on the CL intensity was studied; it was found that CL signals were stable for flow rates in the 0.8–4.0 mL/min range. So, in order to minimize the reagent consumption, a flow rate of 1 mL/min was chosen as suitable for the determination of FLM.

3.3 Validation of the optimized method Linearity of the CL signal versus FLM concentration was studied by application of the external calibration method. This way, the HF-LPME method was applied to triplicates of 50 mL FLM standards; it was found that the analytical signal (peak area) showed a linear relationship with donor phase FLM concentration in the 0.3–15 ng/mL range, with a correlation coefficient r ࣙ 0.999 and detection and quantitation limits of 0.1 and 0.3 ng/mL, respectively. The obtained signals showed a 750-fold increase (enrichment factor) over a theoretical maximum of 1000 with respect to the measurements obtained without the HF-LPME stage, which is equivalent to an extraction efficiency of 75% of the total analyte from the donor phase. The intraday repeatability, expressed as the RSD of quintuplicate measurements of 3.0, 7.5, and 10.0 ng/mL FLM standards in the same day was found to be 2.3%. Interday precision, expressed as RSD of quintuplicate measurements of 3.0, 7.5, and 10.0 ng/mL FLM standards during a two week period was found to be 4.0%. The influence of the presence of other quinolones in the analytical signal was studied by preparing mixtures of these potential interferences and FLM with a 20:1 concentration relation. It was found that the presence of marbofloxacin, norfloxacin, ciprofloxacin, danofloxacin, gatifloxacin, and grepafloxacin led to no significant changes in the obtained signals (20 min for HPLC versus 20 min for HPLC to