Sensitive analysis of antibiotics via hyphenation of field-amplified sample stacking with reversed-field stacking in microchip micellar electrokinetic chromatography

Journal of Pharmaceutical and Biomedical Analysis 103 (2015) 91–98

Contents lists available at ScienceDirect

Journal of Pharmaceutical and Biomedical Analysis journal homepage: www.elsevier.com/locate/jpba

Sensitive analysis of antibiotics via hyphenation of field-amplified sample stacking with reversed-field stacking in microchip micellar electrokinetic chromatography Minglei Wu a , Fan Gao a , Yi Zhang a , Guan Wang a , Qingjiang Wang a,∗ , Hui Li b,∗∗ a b

Department of Chemistry, East China Normal University, 500 Dongchuan Road, Shanghai 200241, PR China College of Chemistry and Chemical Engineering, Shanghai Jiaotong University, 800 Dongchuan Road, Shanghai 200240, PR China

a r t i c l e

i n f o

Article history: Received 29 August 2014 Received in revised form 30 October 2014 Accepted 2 November 2014 Available online 11 November 2014 Keywords: Microchip micellar electrokinetic chromatography Poly (styrene sulfonic acid) sodium salt Field-amplified sample stacking Reversed-field stacking Antibiotics

a b s t r a c t An on-chip multiple-concentration method combining field-amplified sample stacking (FASS) and reversed-field stacking (RFS) in microchip micellar electrokinetic chromatography (MCMEKC) was developed for the simultaneous analysis of three antibiotics (kanamycin, vancomycin, and gentamycin) using poly (styrene sulfonic acid) sodium salt (PSS) as the pseudostationary phase. Results indicated that the polymeric surfactant PSS provided high stability, unique selectivity, and high efficiency for the separation of these antibiotics as compared to SDS micelles, and the multiple-preconcentration strategy could greatly improve the sensitivity enhancement over those classical CE-LIF methods for antibiotics detection. The stacking and separation mechanism as well as important parameters governing preconcentration and separation have been investigated in order to obtain maximum resolution and sensitivity. Under optimal conditions, three antibiotics were successfully focused and completely separated within <3 min. The limits of detection for kanamycin, vancomycin, and gentamycin were 0.25, 0.20, and 0.80 ␮g/L (S/N = 3), respectively, and the detection sensitivities were improved 259-, 296-, and 308-fold, respectively. The method also gave accurate and reliable results in the analysis of these antibiotics in river water samples. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Antibiotic resistance, which results in the risk for ineffective treatment of infections, has become one of the major human health threats of the 21st century [1]. The factor contributing to the problem of antibiotic resistance is the use of antibiotics in public health and animal husbandry [2]. But widespread abuse of antibiotics, including overprescribing in humans and use in animal feed, accelerates the problem [1,2]. Among antibiotics, vancomycin is an amphoteric glycopeptide antibiotic that is active against grampositive bacteria [3]. However, resistance to vancomycin occurs commonly in Enterococcus species, which are frequently detected and considered to cause human death among hospital-acquired diseases [2]. Both kanamycin and gentamycin are aminoglycoside antibiotics, which have been extensively used in both human and veterinary medicine [4,5]. As a consequence, their residues can

∗ Corresponding author. Tel.: +86 21 54340015. ∗∗ Corresponding author. Tel.: +86 21 54743271. E-mail addresses: [email protected] (Q. Wang), [email protected] (H. Li). http://dx.doi.org/10.1016/j.jpba.2014.11.004 0731-7085/© 2014 Elsevier B.V. All rights reserved.

be found in a variety of food products. The residual amount of kanamycin and gentamycin in the food stuff may lead to antibiotic resistance from the pathogenic bacterial strains, which can endanger the consumer [6]. The most effective measure to reduce the magnitude of these antibiotic resistance is to limit antibiotic use in animal husbandry and avoid the over-prescription [2]. Therefore, efficiently monitoring concentration of these antibiotics is necessary and urgent to assure human health. Liquid chromatography (LC) is the most popular technique for separation and determination of antibiotics [7,8]. In recent decades, capillary electrophoresis (CE) has been accepted as a powerful analytical technique and an important choice to LC in analyzing antibiotics due to its lower consumption of sample and better separation efficiency compared to LC [9]. A number of applications of CE for the separation and determination of antibiotics with high efficiency have been reported [9,10]. Microchip electrophoresis (MCE), as one of the most important applications of micro-fluidics, is increasingly being viewed as a successful alternative to CE. When compared with CE, MCE offers significant benefits such as reduced sample and reagent consumption, rapid operation, and easier miniaturization of the analytical instrumentation [11]. As one of the most sensitive of the variety of detection methods,

92

M. Wu et al. / Journal of Pharmaceutical and Biomedical Analysis 103 (2015) 91–98

laser-induced fluorescence (LIF) detection can be used in MCE, making MCE-LIF possible to be a rapid, sensitive, and portable analytical tool for antibiotics. Although a high separation efficiency can be obtained by MCE, the extremely short separation distances in microchip-based separation systems sometimes cause insufficient resolution power. To improve the separation performance, various separation modes have been employed to MCE, such as electrochromatography [12], gel electrophoresis [13], electrokinetic chromatography [14], and so on. Among them, micellar electrokinetic chromatography (MEKC), which is based on the partition equilibrium of analytes between the micellar pseudostationary phase and the surrounding medium, can greatly improve the separation power in MCE [15]. Biogenic compounds such as amino acids, peptides, proteins, biogenic amines, DNA, and estrogens have been reported to be analyzed by microchip MEKC (MCMEKC) with high separation efficiencies [15]. However, other applications of MCMEKC are scarce in comparison with its applications for biogenic compounds, and no report concerning the separation and determination of antibiotics by MCMEKC-LIF exists in the literature. Conventional micelles, such as sodium dodecyl sulfate (SDS), have been extensively used as the pseudostationary phase in MEKC. However, the critical micelle concentration (CMC) of conventional micelles depend on a number of factors, such as ionic strength, the addition of organic solvents, and temperature, leading to the concern of chemical stability of the pseudostationary phase [16]. Moreover, conventional micelles require sufficient surfactant concentrations for maintaining stable structure, and higher concentrations of ionic surfactant result in Joule heating and thereby the reduction of separation efficiency [17]. To overcome these drawbacks, polymeric micelles have been introduced as alternative pseudostationary phase in MEKC [16,17]. These polymeric micelles provide very stable PSP and unique chemical selectivity relative to conventional micelles [16]. In this paper, poly (styrene sulfonic acid) sodium salt (PSS) was first used as a pseudostationary phase for the separation of antibiotics, which provide high stability, unique selectivity, and high-efficiency separations as compared to SDS micelles. Because dosage concentrations of antibiotics are generally quite low, the detection sensitivity is inadequate in the quantitation of low-abundance antibiotics due to the small injection volume of the sample and the short optical path length in MCE. To improve the concentration sensitivity, various on-line sample preconcentration techniques have been developed in MCE, including field-amplified sample stacking (FASS) [18], sweeping [19], transient isotachophoresis [20], isoelectric focusing [21], and so on. However, a single-step stacking method in MCE sometimes cannot meet the low detection limit requirement of low-abundance analytes. The combination of two or more on-line preconcentration techniques is needed to achieve higher signal enhancement. Such multiple-concentration approaches have been reported to afford 3000- to 6000-fold improvements in detection sensitivity [18,19]. The aim of this study was to find a suitable combination of concentration approaches for the detection of low-abundance antibiotics by MCMEKC. After a series of experiments, we combined FASS and reversed-field stacking (RFS) in a multiple-concentration approach for MCMEKC-LIF detection of antibiotics. The applications of FASS to CE have been well described to provide high concentration enhancement [22]. However, the enrichment effect of FASS is greatly weakened in MCE because of its extremely short separation distance as compared to CE. Moreover, the vacant and long sample matrix remaining in the separation channel decreases the effective separation length, thus reducing the separation efficiency of MCE. In this case, the addition of RFS could lengthen the effective separation length, provide a longer time for FASS, and push most of the vacant sample matrix out of the separation channel,

thus improving the separation performance and achieving stronger signal amplification. Herein, we applied an on-line multiple-preconcentration approach combining FASS and RFS to the simultaneous analysis of three antibiotics including kanamycin, vancomycin, and gentamycin by MCMEKC with LIF detection. To the best of our knowledge, this is the first description concerning the polymeric surfactant PSS as the pseudostationary phase in MCMEKC, and the combination of FASS-RFS stacking technique was first employed for determination of varied compounds in MCMEKC. Using this method, three antibiotics could be successfully focused and well separated with high efficiency and sensitivity. This strategy also allows fast and sensitive analysis of these antibiotics in river water samples. 2. Materials and methods 2.1. Apparatus and chemicals All MCE experiments were conducted using a homemade MCE system coupled with an laser-induced fluorescence (LIF) detection device that was created by Shanghai Spectrum Ltd. Co., Zhejiang University and our research group. Briefly, a diode laser (5 mW) was used to generate an excitation beam at 635 nm. The fluorescence signal was spectrally isolated using an edge filter and was subsequently collimated with an achromatic lens before being focused onto the photomultiplier tube. The high voltage power unit variable in the range 0 ± 6 kV was used for on-chip sample injection and zone electrophoretic separation. The amplified current was transferred directly through a 10 k resistor to a 24 bit A/D interface at 10 Hz (Borwin, JMBS Developments, Le Fontanil, France) and stored in a personal computer. Analytical-grade chemicals were used unless otherwise stated. Kanamycin sulphate, gentamycin sulphate, vancomycin hydrochloride, and poly (styrene sulfonic acid) sodium salt (PSS, MW = 70,000) were purchased from Alfa Aesar (Johnson Matthey, Tianjin, China). Poly (ethylene) oxide (MW = 600,000) was purchased from Aldrich (Milwaukee, WI, USA). Polyvinyl pyrrolidone (MW = 40,000) was purchased from Sigma (St. Louis, MO, USA). Sodium tetraborate (Na2 B4 O7 ), sodium hydroxide (NaOH), hydrochloric acid (HCl), and acetonitrile were from Sinopharm Chemical Reagent Co. Ltd (Shanghai, China). Sulfoindocyanine succinimidyl ester (Cy5) was obtained from GE Healthcare Company (Shanghai, China). All solutions were prepared by ultrapure water supplied by Milli-Q water purification system (Millipore, Bedford, MA, USA). 2.2. Background electrolyte and sample preparation Running buffer solution was prepared by dilution of the appropriate amount of Na2 B4 O7 in ultrapure water to a concentration of 100 mM, whereas both sample buffer and derivatization buffer solutions were made by dissolving Na2 B4 O7 with water to 10 mM. Desired pH values of these buffer solutions were obtained with 1 M NaOH or 0.5 M hydrochloric acid. Standard solutions of individual kanamycin, gentamycin, and vancomycin were prepared with a concentration of 10 g/L in ultrapure water and diluted to 0.1 g/L with derivatization buffer. The stock solution Cy5 (1 mM) was made by dissolving in anhydrous acetonitrile. Above solutions were sealed and stored at 4 ◦ C in a refrigerator. 2.3. Derivatization procedure Standard solutions of 10 g/L kanamycin, gentamycin, and vancomycin were diluted to 0.1 g/L with derivatization buffer (10 mM borate solution at pH 8.5). Derivatization of individual compound

M. Wu et al. / Journal of Pharmaceutical and Biomedical Analysis 103 (2015) 91–98

was performed by mixing diluted solution with 1 mM Cy5 solution 1:1 (v:v) in a 1.5 mL centrifuge tube. The three mixtures reacted in darkness for 6 h at room temperature, and then diluted to required concentrations with sample buffer (10 mM borate solution at pH 9.3) before analysis. 2.4. Spiked river water sample preparation River water samples were obtained from local rivers in Shanghai, China. Before analysis, the water samples were filtered through Millipore membrane filters (nominal pore size 0.45 ␮m). Samples were purified by solid-phase extraction (SPE C18-U, Waters). C18 cartridges were activated with 2 mL of methanol and 2 mL of methanol/water (50:50, v/v), and then 2 mL of sample (pH = 7.5) was passed through the cartridge. Finally, the column was eluted with 4 mL methanol/acetic acid (0.5 M) (50:50). The eluent was lyophilized, and the residue was diluted with 2 mL of 10 mM borate solution and adjusted to pH 8.5. Finally, the diluted sample solutions were derivatized as stated in Section 2.3. 2.5. MCE conditions The glass microchip design used in these experiments consisted of a simple cross channel. The separation channel was 60 mm in length and 45 mm from the injection intersection to the detection point. All other channels had a length of 10 mm measured from the channel intersection. Microchannels were etched to a depth of 25 ␮m and a width of 70 ␮m. Platinum electrodes were inserted into the reservoirs, providing electrical contact from the power supply to the electrolyte solutions. All experiments were running in full filling mode. Before a new microchip was first used, it was washed with 98% H2 SO4 and ultrapure water for 10 min. Subsequently, the channels were flushed with 1 M NaOH for 20 min, ultrapure water for 10 min, and the MCE running buffer for 10 min. Before each injection, the microchannels were washed for 2 min with 1 M NaOH, 1 min with ultrapure water and 2 min with MCE running buffer. 2.6. FASS-RFS-MCMEKC procedure For MCMEKC separation of three antibiotics, 0.5 g/L PSS was added to running buffer as PSP. For FASS, the samples were prepared with an appropriate sample buffer (10 mM borate solution) with a much lower conductivity as compared to running buffer (100 mM borate solution). RFS was performed by applying the reversed voltage to make the sample plug retreat back to the injection cross within an appropriate distance. Under these conditions, the analytes in sample plug were piled up and then separated through MCMEKC. A four-step MCE procedure with the voltage configuration of each step was designed as shown in Fig. 1, which realized the separation and determination of three antibiotics with high efficiency and sensitivity. 3. Results and discussion 3.1. Optimization of derivatization conditions The simultaneous analysis of three antibiotics was carried out by using LIF detection in our work. Since the antibiotics assayed have no native fluorescence, they were derivatized with sulfoindocyanine succinimidyl ester (Cy5). The scheme of Cy5 for labeling primary amine groups-containing analytes can be seen in Fig. 2. To increase the yield of derivative, the derivatization reaction was optimized with respect to the derivatization borate buffer, time, and temperature. The composition of the derivatization borate

93

buffer was optimized by changing the pH (7.5–9.5) and the concentration of borate solution (5–50 mM). The maximum LIF signal was obtained when increasing pH values up to 8.5 and then the signal decreased, which can be ascribed to the hydrolytic degradation of the Cy5 at pH over 8.5 [23]. Therefore, pH 8.5 was selected as optimum. In the labeling reaction, 10 mM derivatization borate buffer was found to achieve maximum analytical signal. Further increases in derivatization buffer concentration led to the decreases in the signal intensity, which can be attributed to the formation of negatively charged polyol-borate complexes [24]. The concentration of this negatively complex increases with rising borate concentration and results in an electrostatic repulsion with Cy5 during the derivatization reaction and thereby the reduction of reaction yield. Time and temperature have a significant influence in the reaction efficiency, and the maximum reaction yield was obtained at room temperature with 6 h of reaction time. These conditions were used for all subsequent experiments. 3.2. FASS-RFS-MCMEKC mechanism In this study, a combination of FASS-RFS-MCMEKC strategy was employed to realize the highly efficient and sensitive analysis of kanamycin, vancomycin, and gentamycin. It should be noted that gentamycin consists of a mixture of closely related and structurally similar components like C1 , C1a , C2 , and C2a [25], and the proposed method was used to detect the total amount of this antibiotic. Schematic mechanism of this method is shown in Fig. 1. At the preloading stage, the low-conductivity sample solution was introduced to the sample waste reservoir from the sample reservoir through the injection cross by adjusting the voltages in each of the reservoirs (Fig. 1A). As shown in Fig. 1B, a gated injection method was then applied by changing the voltage configuration during the loading step, which introduced the sample plug into the separation channel. Under these conditions, FASS occurred in the separation channel (the dark gray zone in Fig. 1B) because of the difference in concentration between running buffer and sample buffer. In FASS, the samples were prepared in an appropriate sample buffer with a much lower concentration as compared to MCE running buffer. However, microchip-based separation systems have short separation distances that are 10–100 times shorter than those of standard capillary-based systems, limiting the concentration enhancement. Moreover, after the FASS concentration at the sample buffer/running buffer boundary, the vacant and long sample matrix remaining in the separation channel significantly decreases the effective separation length, thus reducing the separation efficiency. Therefore, RFS was performed to solve these problems by applying the reversed voltage (Fig. 1C). Under the RFS conditions, the concentrated sample plug in L4 retreated back to the intersection within an appropriate distance, and FASS happened again due to the presence of the boundary between two buffers with different concentrations. Hence, a prolonged time was obtained for boundary enrichment, thus enhancing the enrichment effect of FASS. Meanwhile, most of the long and vacant sample matrix remaining in the separation channel (L4) was removed and pumped into L3, thus lengthening the effective separation distance and leading to further concentration enhancement. When all analytes were concentrated, separation occurred (Fig. 1D). Its mechanism was based on MCMEKC using PSS as the pseudostationary phase. PSS is a commercially available and water-soluble anionic polymeric micelle. Because no CMC of such a polymeric micelle is required to maintain the structure of the polymer, detection interferences and Joule heating can be reduced or eliminated. Thus, the drawbacks of conventional micelles (e.g., SDS) for separation of antibiotics would not occur as PSS used as the pseudostationary phase. PSS has been reported as a pseudostationary phase for separation of neutral compounds in capillary

94

M. Wu et al. / Journal of Pharmaceutical and Biomedical Analysis 103 (2015) 91–98

(A)

SR

(B)

(1.35kv)

SR

L1 BR

SW

BR

(1.90kv)

(0.65kv)

(0.65kv)

L4

L3

L2

L4 BW

(Ground)

SR (1.35kv)

L1

SW L2

(D)

(0.30kv)

L1

(1.30kv)

L3

SR

(C)

(1.05kv)

SW (Ground)

L3

L2

L4 BW

(Ground)

L1 BR

SW

BR

(0.30kv)

(1.30kv)

(1.90kv)

L3

L2

L4 BW

(1.00kv)

BW

(Ground)

Fig. 1. Schematic diagram of the sample loading, on-line multiple-preconcentration, and MCMEKC separation of antibiotics: (A) preloading, (B) FASS, (C) RFS, (D) MCMEKC separation. The clear zone represents the running buffer, the light gray zone represents the sample matrix, the dark gray zone represents the concentrated sample by FASS before using RFS, and the black zone represents the area of the concentrated sample after using RFS. SR = sample reservoir; BR = buffer reservoir; SW = sample waste reservoir; BW = buffer waste reservoir. L1-4 represents four channels of microchips. Arrows indicate bulk flow directions. The voltages (kv) applied to the reservoirs at each step are indicated.

electrokinetic chromatography [26]. Analyte hydrophobicity was found to have a major influence on the migration behavior of neutral compounds. However, for separation of Cy5 derivatives (Fig. 2), both the electrostatic and hydrophobic interactions with PSS contributed to the difference in the migration times of all analytes. The magnitude of these interactions can be directly controlled by adjusting the PSS concentration added in running buffer to obtain the best separation. 3.3. Optimization of MCMEKC separation conditions To obtain the best separation among the peaks of labeled antibiotics and the unreacted Cy5, different electrophoretic conditions were tested using microchip zone electrophoresis and MCMEKC. Initially, microchip zone electrophoresis studies were carried out by using borate solution as running buffer and by changing the pH (8.8–9.8) and the running buffer concentration (10–75 mM). The best resolution of all analytes was achieved using 70 mM borate solution at pH 9.3 (Fig. 3A). However, as can be seen in Fig. 3A, this simple microchip zone electrophoresis mode failed to solve overlapping problems even under optimized separation conditions. The addition of organic solvent, such as methanol, ethanol, and acetonitrile, was also studied and was found ineffective in resolving this problem. Finally, MCMEKC was assayed to improve the separation performance. As the most common used surfactant in MEKC, SDS was added to the running buffer at different concentrations (5–25 mM). However, the addition of SDS had no effect in improving separation efficiency. Moreover, higher SDS concentrations (>15 mM) led to baseline undulation and reduction of separation efficiency due to Joule heating. Since polymeric micelles addresses many of the problems associated with conventional micelles, we tested some polymeric micelles, including poly (ethylene) oxide,

polyvinyl pyrrolidone, and PSS as the pseudostationary phase, and found that these additives provided better stability than SDS. However, only PSS achieved baseline separation for all analytes. As compared with SDS, PSS had several advantages including high stability, unique selectivity, and high-efficiency separations for the MCE analysis of antibiotics. Therefore, PSS was selected as the pseudostationary phase in MCMEKC. To determine the impact of the amounts of this polymeric micelle on the effect of separation efficiency in MCMEKC, a series of electrophoretic separation with PSS at different concentrations (0–0.75 g/L) were tested using no concentration method. From Fig. 3A–C, it can be observed that the resolution increased with the addition of PSS to the running buffer. Further increases in the amount of PSS in separation buffer (Fig. 3D) was found detrimental to the MCMEKC separation efficiency. As a result of the experiment, three antibiotics were classified to have best resolution with 0.5 g/L PSS (Fig. 3C). Accordingly, 0.5 g/L PSS was used for the MCMEKC-LIF experiment. As shown in Fig. 4A, the enrichment effect of the multipleconcentration technique is detrimental to the separation efficiency in MCMEKC. To enhance the separation performance of MCMEKC, other alternatives such as the pH and the concentration of borate solution in running buffer were investigated. From experimental results, it was found that the pH value of running buffer solution did not markedly affect separation efficiency, but running buffer concentration did. As can be observed from Fig. 4A–C, resolution was improved with the increase in buffer concentration and the perfect separation was achieved when the buffer concentration was increased to 100 mM (Fig. 4C). For optimal resolution and stability, 100 mM borate solution at pH 9.3 (containing 0.5 g/L PSS) was selected as running buffer solution in MCMEKC (Fig. 4C).

Fig. 2. Scheme of the labeling reaction with Cy5.

M. Wu et al. / Journal of Pharmaceutical and Biomedical Analysis 103 (2015) 91–98

95

Fig. 3. Effect of the concentration of PSS on separation efficiency in MCMEKC without any concentration step. The concentration of PSS added in running buffer: (A) 0 g/L, (B) 0.25 g/L, (C) 0.50 g/L, (D) 0.75 g/L. The concentrations of kanamycin, vancomycin, and gentamycin were 100, 80, and 200 ␮g/L, respectively. The running buffer was 70 mM borate solution at pH 9.3, and the samples were prepared with the same borate solution in running buffer. The sample injection time was 2 s. Peak identification: 1, kanamycin; 2, the excess of Cy5; 3, vancomycin; 4, gentamycin.

3.4. Optimization of on-line preconcentration conditions 3.4.1. Effect of sample buffer concentration During the on-line preconcentration process, the addition of RFS step could provide a longer time for FASS by pushing back the sample plug. Thus, all the preconcentration process was accompanied by FASS. In FASS, the samples should be prepared with an appropriate sample buffer with a much lower conductivity as

compared to MCE running buffer. Sample stacking occurs at the boundary between a high-electric-field sample zone and a lowelectric-field background solution zone. The concentration ratio of running buffer to sample buffer affects the distribution of the electric field across the channel and thereby influences the result of signal enhancement. Theoretically, higher ratios result in stronger fluorescent signals of analytes. However, a laminar flow caused by the mismatch of electroosmotic velocity inside the column results

Fig. 4. Effect of the running buffer concentration on separation efficiency using FASS-RFS-MCMEKC technique. The concentration of borate buffer solution in running buffer: (A) 70 mM, (B) 85 mM, (C) 100 mM. The concentrations of kanamycin, vancomycin, and gentamycin were 80, 80, and 120 ␮g/L, respectively. The running buffer contained 0.5 g/L PSS. The samples were prepared with a 10-fold-diluted borate solution in running buffer, and all buffer pH values were 9.3. The sample injection time was 10 s, and the reversed-polarity time was 8 s. Other conditions and peak labels are the same as those in Fig. 3.

96

M. Wu et al. / Journal of Pharmaceutical and Biomedical Analysis 103 (2015) 91–98

Fig. 5. Effect of injection time and reversed-polarity time on peak intensity and separation efficiency. Injection time and reversed-polarity time: (A) 5 s, 3 s; (B) 10 s, 8 s; (C) 15 s, 13 s; (C) 20 s, 18 s. The concentrations of kanamycin, vancomycin, and gentamycin were 8, 9, and 16 ␮g/L, respectively. Other conditions are the same as those in Fig. 4C.

in the reduction of stacking efficiency. In addition, higher concentration ratios of the two buffers may lead to a larger laminar flow [27]. Therefore, the increase in concentration ratio of the running buffer to the sample buffer should be balanced against the negative effect of the laminar flow. Under optimal MCMEKC separation conditions (0.5 g/L PSS and 100 mM borate solution at pH 9.3 in running buffer), sample buffer concentration was investigated to obtain the strongest signal amplification. Different concentrations of borate solution (5–75 mM) in sample buffer were studied on the basis of the observed signal intensity of vancomycin. As shown in Fig. S1, the peak height of vancomycin was maximized at sample buffer concentration of 10 mM (Fig. S1B). Higher concentrations of sample buffer resulted in deterioration of the peak and band broadening of signal (Fig. S1C and D). Thus, 10 mM borate solution (pH 9.3) was selected as the optimum sample buffer. 3.4.2. Effect of injection time and reversed-polarity time Because the signal intensities of the developed MCE method depended on both the sample volume and enrichment effect, the injection time and the reversed-polarity time became important factors that influenced signal amplification. During the injection step, the injection time controlled the sample volume introduced into the separation channel for preconcentration. To obtain stronger signal amplification, the reversed polarity step using RFS technique was performed by pushing back the sample plug to provide a longer time for FASS. Moreover, most of the long and vacant sample matrix was simultaneously pushed out of the separation channel by using this reversed polarity step, thus further improving the concentration enhancement. However, with a longer reversed-polarity time, the analytes would potentially be pushed out of the separation channel with the vacant sample matrix, leading to the reduction of signal intensities. Due to these facts, the reversed-polarity time was designed 2 s shorter than the injection time, making the sample plug move to the intersection within an appropriate distance to avoid pushing out the concentrated samples. As it can be seen (Fig. 5A–D), although the increases in injection time and reversed-polarity time could strengthen the signal intensities of all analytes, the time interval between peaks was increasingly shorter, resulting in the decreases in separation efficiency (Fig. 5C and D). Therefore, the increases in injection time must be balanced against the reduction of separation efficiency.

Fig. 6. Signal enhancement of the multiple concentration in MCMEKC: (A) signal intensity without concentration. Running buffer was the same as that in Fig. 4C. The sample was prepared with 100 mM borate solution at pH 9.3, and the sample injection time was 2 s. (B) Signal intensity with FASS. Running buffer and sample buffer were the same as those in Fig. 4C, and the sample injection time was 2 s. (C) Signal intensity with a combination of FASS and RFS. Running buffer and sample buffer were the same as those in Fig. 4C. The sample injection time was 10 s, and the reversed-polarity time was 8 s. The concentrations of kanamycin, vancomycin, and gentamycin in (C) were 3.4, 1.5, and 7 ␮g/L, respectively, and the sample concentration in (C) were 1/10 of that in (B) and 1/100 of that in (A). Other conditions are the same as those in Fig. 4C.

After comprehensively considering every factor, 10 s and 8 s were selected as the optimum injection time and reversed-polarity time, respectively (Fig. 5B). 3.5. Method validation The linearity, precision (RSD), limit of detection (LOD), and sensitivity-enhancement results were determined to assess performance and reliability of the proposed method. Table 1 gives the equations for the calibration curves obtained by plotting the peak height against analyte concentration in the linear range, the intraday precision obtained by repeating the analysis three times, the LODs calculated as the minimum analyte concentration providing signals 3 times the background noise, and the sensitivity enhancement factors of all analytes using the multipleconcentration method compared to when no concentration was performed. The linearity was evaluated by analyzing the mixture of three Cy5-labeled antibiotics at different concentrations (each antibiotic at eight different concentrations), which was satisfactory in the linear range studied with coefficients of determination from 0.9975 to 0.9982. The intraday precision of the method, expressed as the RSD, was examined by three consecutive injections of the mixture of three antibiotics (containing 5.2 ␮g/L kanamycin, 3.8 ␮g/L vancomycin, and 10.5 ␮g/L gentamycin). The RSDs of the migration time and peak height were in the range of 1.4–1.5% and 1.7–3.4%, respectively, indicating good reproducibility and high precision. The LODs were 0.25, 0.20, and 0.80 ␮g/L for kanamycin, vancomycin, and gentamycin, respectively. Fig. 6 displays the electrophoretic profiles for the different concentration steps. The standard electropherograms shown in Fig. 6B and C were obtained by 10 and 100-fold dilution of the sample used in Fig. 6A, respectively. The electropherogram of a typical gated injection method using no concentration step is shown in Fig. 6A. For FASS (Fig. 6B), all analytes were clearly enriched. For the combined FASS-RFS stacking (Fig. 6C), detection sensitivities of all analytes increased considerably. Compared with conventional gated injection using no concentration step (Fig. 6A), the sensitivity

M. Wu et al. / Journal of Pharmaceutical and Biomedical Analysis 103 (2015) 91–98

97

Table 1 Analytical performance and enrichment factor results of the multiple-concentration MCE method. Compound

Linearity range (␮g/L)

Linear regressiona

Correlation of determination (R2 )

LOD (␮g/L)

RSD% (n = 3)b MT/peak height

Enrichment factor

Kanamycin Vancomycin Gentamycin

2–60 1–40 4–110

y = 42.650x − 8.6659 y = 48.469x + 7.4664 y = 14.119x + 4.3327

0.9982 0.9975 0.9981

0.25 0.20 0.80

1.5/1.7 1.4/2.3 1.5/3.4

259 296 308

a

In the regression equation, the x value was the concentration of analytes (␮g/L), the y value was the peak height (mv). MT, migration time. Concentrations of kanamycin, vancomycin, and gentamycin were 5.2, 3.8, and 10.5 ␮g/L, respectively. Other conditions are the same as those in Fig. 4C. b

enhancement factors in peak heights for kanamycin, vancomycin, and gentamycin using the combined FASS-RFS stacking were 259, 296, and 308, respectively. Moreover, the sensitivity of antibiotics obtained via the combination of FASS-RFS stacking technique (Fig. 6C) was improved by 18- to 35-fold over the single FASS stacking technique (Fig. 6B), which testifies to the stronger enrichment effect of the combination of on-line preconcentration approaches. Because little has been reported on the simultaneous detection of these three antibiotics, the proposed method was compared with CE-LIF by comparing the detection sensitivity of kanamycin [28,29]. The sensitivity of kanamycin detection obtained via FASS-RFSMCMEKC was improved by 10- to 30-fold over those of previously reported CE-LIF methods [28,29].

Table 2 Recoveries of FASS-RFS-MCMEKC for the analysis of antibiotics in the river water samples.a Antibiotics

Added amount (ppb)

Recovery%

Kanamycin

25.0 4.0 2.0

95.7 104.7 104.2

Vancomycin

25.0 3.0 2.0

96.7 93.8 97.5

Gentamycin

66.7 13.3 4

98.8 96.3 107.4

a

Experiment conditions are the same as those in Fig. 4C.

3.6. Application to real-world sample matrices

4. Conclusions

To evaluate the usefulness of the method developed, the river water was collected and analyzed for trace antibiotics. The samples were prepared according to procedures in Section 2.4. Preliminary tests revealed the absence of antibiotics, so the viability of the FASSRFS-MCMEKC method was tested with river water spiked with antibiotics. In order to avoid the suppression of sensitivity of FASSRFS-MEKC due to the presence of salts in the matrix, we analyzed river water spiked with antibiotics by desalting the samples with a C18 SPE column. Electropherograms for the blank (A) and spiked (B) river water sample are presented in Fig. 7. Because the migration times of the analytes varied in the real sample analysis, the peaks were verified by a standard addition method. Under optimized conditions, river water spiked with a mixture of the three antibiotics gave recoveries of 93.8–107.4% (Table 2). These results demonstrated that real water sample matrices had virtually no effect on the performance of the method proposed.

We have presented a sensitive method on the hyphenation of FASS and RFS in MCMEKC-LIF for the simultaneous detection of antibiotics. Using this method, three antibiotics could be simultaneously focused and separated within <3 min. The polymeric surfactant PSS was first employed as the pseudostationary phase in MCMEKC and provided high stability, unique selectivity, and high efficiency for the separation of antibiotics. Under optimal conditions, the strategy allowed for the determination of these antibiotics at very low concentrations (0.20–0.80 ␮g/L) and afforded 259- to 308-fold improvements in peak height. Moreover, the method was applied to the determination of these antibiotics in river water samples with a satisfactory recovery. Compared with conventional detection methods for antibiotics, this multipleconcentration MCMEKC method provides many merits, including its miniaturization, high sensitivity, low sample consumption, and rapid operation. Acknowledgements This work was financially supported by the Program New Century Excellent Talents in University (Grant NCET-08-0191) and the National Program on Development of Scientific Instruments and Equipment (Grant 2011YQ150072). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jpba.2014.11.004. References

Fig. 7. Electropherograms for blank river water sample (A) and sample spiked with kanamycin and vancomycin, 25.0 ppb; and gentamycin, 66.7 ppb (B). Other conditions are the same as those in Fig. 4C.

[1] H. Rho, B. Shin, O. Lee, Y.-H. Choi, J. Rho, J. Lee, Antibiotic resistance profile of bacterial isolates from animal farming aquatic environments and meats in a peri-urban community in Daejeon, Korea, Environ. Monit. 14 (2012) 1616–1621. [2] D.C. Hooper, A. DeMaria, B.M. Limbago, T.F. O’Brien, B. McCaughey, Antibiotic resistance: how serious is the problem, and what can be done? Clin. Chem. 58 (2012) 1182–1186. [3] G. Ye, X. Cai, B. Wang, Z. Zhou, X. Yu, W. Wang, J. Zhang, Y. Wang, J. Dong, Y. Jiang, Simultaneous determination of vancomycin and ceftazidime in cerebrospinal

98

[4]

[5]

[6] [7]

[8]

[9]

[10]

[11] [12]

[13]

[14] [15] [16] [17]

M. Wu et al. / Journal of Pharmaceutical and Biomedical Analysis 103 (2015) 91–98 fluid in craniotomy patients by high-performance liquid chromatography, J. Pharm. Biomed. Anal. 48 (2008) 860–865. W. Sheng, L. Yang, J. Wang, Y. Zhang, S. Wang, Development of an enzymelinked immunosorbent assay for the detection of gentamycin residues in animal-derived foods, LWT – Food Sci. Technol. 50 (2013) 204–209. F. Li, Y. Guo, X. Sun, X. Wang, Aptasensor based on thionine, graphenepolyaniline composite film, and gold nanoparticles for kanamycin detection, Eur. Food Res. Technol. 239 (2014) 227–236. D.K. Byarugaba, R. Kisame, S. Olet, Multi-drug resistance in commensal bacteria of food of animal origin in Uganda, Afr. J. Microbiol. Res. 5 (2011) 1539–1548. R. Cazorla-Reyes, R. Romero-Gonzalez, A. Garrido Frenich, M.A. Rodriguez Maresca, J.L. Martinez Vidal, Simultaneous analysis of antibiotics in biological samples by ultra high performance liquid chromatography–tandem mass spectrometry, J. Pharm. Biomed. Anal. 89 (2014) 203–212. A.V. Herrera-Herrera, J. Hernandez-Borges, T.M. Borges-Miquel, M. Angel Rodriguez-Delgado, Dispersive liquid–liquid microextraction combined with ultra-high performance liquid chromatography for the simultaneous determination of 25 sulfonamide and quinolone antibiotics in water samples, J. Pharm. Biomed. Anal. 75 (2013) 130–137. E. Dominguez-Vega, V. Perez-Fernandez, A. Luis Crego, M. Angeles Garcia, M. Luisa Marina, Recent advances in CE analysis of antibiotics and its use as chiral selectors, Electrophoresis 35 (2014) 28–49. S. Ge, W. Tang, R. Han, Y. Zhu, Q. Wang, P. He, Y. Fang, Sensitive analysis of aminoglycoside antibiotics via hyphenation of transient moving substitution boundary with field-enhanced sample injection in capillary electrophoresis, J. Chromatogr. A 1295 (2013) 128–135. F. Kitagawa, K. Otsuka, Recent progress in microchip electrophoresis-mass spectrometry, J. Pharm. Biomed. Anal. 55 (2011) 668–678. C. Hackl, R. Beyreiss, D. Geissler, S. Jezierski, D. Belder, Rapid prototyping of electrochromatography chips for improved two-photon excited fluorescence detection, Anal. Chem. 86 (2014) 3773–3779. M. Del Mar Barrios-Romero, A.G. Crevillen, J.C. Diez-Masa, Development of an SDS-gel electrophoresis method on SU-8 microchips for protein separation with LIF detection: application to the analysis of whey proteins, J. Sep. Sci. 36 (2013) 2530–2537. V. Kasicka, Recent developments in capillary and microchip electroseparations of peptides (2011–2013), Electrophoresis 35 (2014) 69–95. F. Kitagawa, K. Otsuka, Micellar eletrokinetic chromatography on microchips, J. Sep. Sci. 31 (2008) 794–802. C.P. Palmer, Recent progress in the use of ionic polymers as pseudostationary phases for electrokinetic chromatography, Electrophoresis 30 (2009) 163–168. B. Wang, X. Ni, M. Yu, Y. Cao, Polymeric micelle as the pseudostationary phase in electrokinetic chromatography, J. Chromatogr. A 1245 (2012) 190–198.

[18] Z.-F. Wang, S. Cheng, S.-L. Ge, H. Wang, Q.-J. Wang, P.-G. He, Y.-Z. Fang, Ultrasensitive detection of bacteria by microchip electrophoresis based on multiple-concentration approaches combining chitosan sweeping, fieldamplified sample stacking, and reversed-field stacking, Anal. Chem. 84 (2012) 1687–1694. [19] Q. Pan, M. Zhao, S. Liu, Combination of on-chip field amplification and bovine serum albumin sweeping for ultrasensitive detection of green fluorescent protein, Anal. Chem. 81 (2009) 5333–5341. [20] Z. Mala, P. Gebauer, P. Bocek, Recent progress in analytical capillary isotachophoresis, Electrophoresis 34 (2013) 19–28. [21] M. Kinoshita, Y. Nakatsuji, S. Suzuki, T. Hayakawa, K. Kakehi, Quality assurance of monoclonal antibody pharmaceuticals based on their charge variants using microchip isoelectric focusing method, J. Chromatogr. A 1309 (2013) 76–83. [22] J. Li, Y. Huang, L. Huang, L. Ye, Z. Zhou, G. Xiang, L. Xu, Determination of imatinib mesylate and related compounds by field amplified sample stacking with large volume sample injection capillary electrophoresis, J. Pharm. Biomed. Anal. 70 (2012) 26–31. [23] E. Orejuela, M. Silva, Rapid and sensitive determination of phosphoruscontaining amino acid herbicides in soil samples by capillary zone electrophoresis with diode laser-induced fluorescence detection, Electrophoresis 26 (2005) 4478–4485. [24] P.A. Hoffstetter-Kuhn, S.E. Gassmann, et al., Influence of borate complexation on the electrophoretic behavior of carbohydrates in capillary electrophoresis, Anal. Chem. 63 (1991) 1541–1547. [25] Y. Bijleveld, T. de Haan, J. Toersche, S. Jorjani, J. van der Lee, F. Groenendaal, P. Dijk, A. van Heijst, A.W.D. Gavilanes, R. de Jonge, K.P. Dijkman, H. van Straaten, M. Rijken, I. Zonnenberg, F. Cools, D. Nuytemans, R. Mathot, A simple quantitative method analysing amikacin, gentamicin, and vancomycin levels in human newborn plasma using ion-pair liquid chromatography/tandem mass spectrometry and its applicability to a clinical study, J. Chromatogr. B 951 (2014) 110–118. [26] B. Pranaitytë, A. Padarauskas, Poly (sodium-4-styrenesulfonate) as the pseudostationary phase in capillary electrokinetic chromatography, Chemija 15 (2004) 45–49. [27] D.S. Burgi, R.L. Chien, Optimization in sample stacking for high-performance capillary electrophoresis, Anal. Chem. 63 (1991) 2042–2047. [28] Y.-F. Lin, Y.-C. Wang, S.Y. Chang, Capillary electrophoresis of aminoglycosides with argon-ion laser-induced fluorescence detection, J. Chromatogr. A 1188 (2008) 331–333. [29] C.-Z. Yu, Y.-Z. He, G.-N. Fu, H.-Y. Xie, W.-E. Gan, Determination of kanamycin A, amikacin and tobramycin residues in milk by capillary zone electrophoresis with post-column derivatization and laser-induced fluorescence detection, J. Chromatogr. B 877 (2009) 333–338.