3556 Fuying Du1,2 Shunan Cao1 Ying-Sing Fung2 ∗ 1 Department
of Water Quality Engineering, School of Power and Mechanical Engineering, Wuhan University, Wuhan, China 2 Department of Chemistry, The University of Hong Kong, Hong Kong SAR, China
Received May 24, 2014 Revised August 30, 2014 Accepted August 30, 2014
Electrophoresis 2014, 35, 3556–3563
Research Article
A serial dual-electrode detector based on electrogenerated bromine for capillary electrophoresis A new serial dual-electrode detector for CE has been designed and fabricated for postcolumn reaction detection based on electrogenerated bromine. A coaxial postcolumn reactor was employed to introduce bromide reagent and facilitate the fabrication of upstream generation electrode by simply sputtering Pt film onto the outer surface of the separation capillary. Bromide introduced could be efficiently converted to bromine at this Pt film electrode and subsequently detected by the downstream Pt microdisk detection electrode. Analytes that react with bromine could be determined by the decrease of bromine reduction current at the downstream electrode resulting from the reaction between analytes and bromine. The effects of serial dual-electrode detector working conditions including electrode potentials, bromide flow rate, and bromide concentration on analytical performance were investigated using glutathione (GSH) and glutathione disulfide (GSSG) as test analytes. Under the optimal conditions, detection limits down to 0.16 M for GSH and 0.14 M for GSSG (S/N = 3) as well as linear working ranges of two orders of magnitude for GSH and GSSG were achieved. Furthermore, the separation efficiency obtained by our dual-electrode detector design was greatly improved compared with previous reported design. The developed method has been successfully applied to determine the GSH and GSSG impurity in commercial GSH supplement. Keywords: Capillary electrophoresis / Dual-electrode detector / Electrogenerated bromine / Glutathione / Indirect amperometric detection DOI 10.1002/elps.201400257
1 Introduction Since the introduction of CE in the early 1980s [1], the technique has drawn intensive attention due to its high separation efficiency, low sample and solvent consumption, as well as short analysis time. These features make it especially suitable for the analysis of food [2], environment [3], drugs [4], and biological samples [5]. CE has been hyphenated frequently with UV-Vis absorbance, fluorescence, MS, amperometric detection (AD), etc. Among these detection methods, AD with detection limit in the subnanomolar range (0.1 nM) is one of the most sensitive detection methods coupled with CE separation. Unlike optical detection methods, the miniaturization of the detection electrodes does not compromise the analytical performance of AD, which is an attractive feature for CE system. Moreover, the instrumentation for AD is less expensive than that for fluorescence detection and MS, of
Correspondence: Dr. Fuying Du, Department of Water Quality Engineering, School of Power and Mechanical Engineering, Wuhan University, Wuhan 430072, China E-mail: [email protected]
Abbreviations: AD, amperometric detection; GSH, glutathione; GSSG, glutathione disulfide
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
which sensitivity is comparable to AD. In the past few years, there has been a great deal of research integrating CE with AD [6–9]. In most reported CE-AD systems, one single working electrode is employed either in off-column detection mode [10–14] or in end-column detection mode [15–19]. Recently, the research of CE-AD system employing two working electrodes has started and attracted much more attention due to its flexibility in operation mode and its capability in improving the sensitivity, selectivity, and increasing the applicability of CE-AD through indirect detection scheme. Basically, there are two operation modes for dual-electrode detection: parallel mode and serial mode. Dual-electrode detector in parallel mode can be used to improve the selectivity and sensitivity of the CE-AD. Our group [20] developed a parallel opposite dual-electrode detector controlled at +0.8 and +1.0 V, respectively, to identify polyphenol peaks in red wine. The peak current ratio obtained by the dual-electrode detector was used for identification of the eluted peaks to improve the detection selectivity. Chen et al. [21] fabricated a parallelopposed dual-electrode detector that comprised a gold disk electrode and an on-capillary gold film electrode in a thin-layer
∗ Additional corresponding author: Dr. Fung Ying-Sing, E-mail: [email protected]
www.electrophoresis-journal.com
CE and CEC
Electrophoresis 2014, 35, 3556–3563
geometry. Due to the redox recycling of dopamine between these two electrodes, the sensitivity was enhanced ten times as compared to the conventional single-electrode end-column detection. Lunte’s group [22] also reported a parallel dualelectrode detector operating in both the redox cycling mode and dual-potential mode to improve the detection sensitivity and selectivity. Dual-electrode detector in serial mode has been also constructed and investigated for CE-AD. The initial application of serial dual-electrode detector was reported by Lin et al. [23] for the determination of cysteine and cystine in urine. Cystine was reduced at upstream Au-Hg wire electrode set at –1.0 V and then detected at downstream Au-Hg wire electrode set at +0.15 V. The detection limit was reported to be 100 M for cystine. Zhong et al. [24] improved this detection mode by employing a tubular-wire configuration for serial dual-electrode detector. Due to the high conversion efficiency at the upstream tube electrode, the detection limit was lowered to 0.5 M for cystine. Holland and Lunte [25] implemented the first serial dual-electrode detection after CE separation for indirect detection based on electrogenerated bromine. In this detection strategy, bromine was generated at the upstream electrode and then reduced back to bromide at the downstream electrode. The analytes migrating out of the separation capillary reacted with the electrogenerated bromine to give rise to a decrease in the current detected at the downstream electrode. This approach holds great promise in widening the scope of CE-AD system because a large number of compounds that are difficult to be detected by direct AD can be determined by this method. However, the development of serial dual-electrode detector for CE with high sensitivity and high separation efficiency is not an easy task. First, the arrangement of upstream generation electrode and downstream detection electrode should be carefully designed to obtain high generation and collection efficiency to achieve good detection sensitivity. Second, the way for bromide reagent introduction should be deliberated to reduce the band broadening caused by bromide introduction. In Holland and Lunte’s work, dual electrode in wire– wire configuration was employed in off-column mode after cathodic reservoir. The bromide reagent was introduced into CE system hydrodynamically through a crack on capillary at the cathodic reservoir. Due to the disturbance of the analytical stream by uneven introduction of the bromide reagent at the cathodic reservoir, as well as the back pressure present in the capillary after the decoupling fracture, a serious peak broadening was observed, leading to a poor separation efficiency and hampering the application of this method in the analysis of complex sample matrix. In order to solve this problem, we described a new design for serial dual-electrode detector after CE for indirect AD based on electrogenerated bromine in the present work. Bromide was introduced by a conventional coaxial postcolumn reactor [26] and converted to bromine efficiently at the on-capillary electrode made by sputtering Pt film on the outer surface of separation capillary. The electrogenerated bromine reacted with analytes migrating out of the separation capillary in reaction capillary and detected at a Pt C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
3557
microdisk detection electrode aligned at the outlet of reaction capillary. The analytical performance of the developed serial dual-electrode detector was evaluated by using glutathione (GSH) and glutathione disulfide (GSSG) as test analytes. Experimental parameters governing the analytical performance were investigated and optimized. The results showed that low detection limits for the determination of GSH and GSSG can be achieved. Furthermore, band broadening was greatly reduced by the developed serial dual-electrode detector. The applicability of the developed CE-serial dual-electrode detector for the determination of GSH and GSSG impurity in GSH supplement was demonstrated.
2 Materials and methods 2.1 Chemicals and samples L-GSH
and GSSG were purchased from Sigma-Aldrich (St. Louis, MO, USA). Sodium tetraborate and potassium bromide were obtained from BDH Chemicals (Poole, UK). All chemicals were used as received. Stock solutions of GSH and GSSG were made up to 1 mM in DI water and stored at –20°C until use. Standard solutions with desired concentrations for calibration were prepared by diluting the stock solutions with the running buffer before analysis. The running buffer consisted of 20 mM borate buffer with pH adjusted to 8.2. The bromide reagent was prepared by dissolving a desired amount of potassium bromide in the running buffer. Pharmaceutical GSH supplement claimed to have 50 mg GSH per tablet was purchased from a local drug store. A tablet was weighed (0.6035 ± 0.0037 g, n = 5) and then ground thoroughly. A 120 mg portion of the powdered sample was dissolved in 100 mL running buffer, diluted ten times by running buffer, and filtered to remove insoluble cellulose. Prior to analysis, all the solutions were filtered through a 0.2 m hydrophilic polyethersulfone membrane filter (Pall, Washington, NY, USA). 2.2 Instrumentation
A self-assembled reversible high-voltage power supply (Bertan, NY, USA) was used to provide a variable voltage from 0 to 24 kV for CE separation. Fused-silica capillaries (Yongnian, Hebei, China) with the same outer diameter at 365 m and different inner diameters of 25 and 150 m were used as the separation and reaction capillaries, respectively. The potentials for the upstream generation electrode and downstream detection electrode were controlled by a CHI 760d (CH Instruments, Shanghai, China) and a WPI MicroC potentiostat (Sarasota, FL, USA), respectively. Data were collected and processed by Origin 6.1 and N2000 chromatography data system (Zhi Da, Zhejiang, China). A CO2 laser engraver (Vseries, Pinnacle, USA) was used to fabricate channel pattern onto the PMMA chip to accommodate the separation capillary, reaction capillary, and detection electrode. www.electrophoresis-journal.com
3558
F. Du et al.
Electrophoresis 2014, 35, 3556–3563
Figure 1. (A) Schematic diagram showing the serial dualelectrode detection system based on coaxial postcolumn reactor. (B) The enlarged view for the serial dual-electrode configuration.
2.3 Fabrication of serial dual-electrode detector A schematic diagram showing the design of serial dualelectrode detector after CE separation is illustrated in Fig. 1. A coaxial postcolumn reactor as shown in Fig. 1A was employed to introduce bromide reagent hydrodynamically. The dual-electrode detector consisted of an on-capillary Pt film upstream electrode and a Pt microdisk downstream electrode as shown in Fig. 1B. The upstream generation electrode was made by sputtering Pt film onto the outer surface of separation capillary as reported by Baldwin [27]. Before sputtering, the end of the separation capillary was etched in hydrofluoric acid (48%) until its outer diameter was reduced to about 110 m so that it can be inserted into the reaction capillary. Then, the capillary was placed into the sputter chamber (BAL-TEC SCD 005 Sputter Coater, Switzerland) for platinum coating. To avoid undesirable electrogeneration of bromine in the reagent vessel, the outer surface of the separation capillary exposed to solution in the reagent vessel was covered with a layer of nonconductive epoxy. The electrical connection to the on-capillary Pt film electrode was made by cementing copper wire onto the capillary with conductive silver epoxy (ITW Chemtronics) at the edge of PMMA plates as shown in Fig. 1A. The length for the separation capillary inserted into the reaction capillary was fixed at 3.0 mm to make the area of the upstream electrode constant throughout the experiments. In order to make the upstream electrode free from interference from high separation voltage, its corresponding Ag/AgCl reference electrode was also placed in the reagent vessel. The distance between the end of the separation capillary and end of the reaction capillary, which determines the reaction zone of the postcolumn reactor, was kept at 4.0 mm in all experiments. Pt microdisk electrode placed about 20 m away from the outlet of the reaction capillary was employed as the down C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
stream detection electrode. The fabrication method for the Pt microdisk electrode was similar to those previously reported in the literature [28,29]. Platinum wire with 200 m diameter (Goodfellow, UK) was inserted into the capillary (250 m id, 365 m od) and fixed in position by glue. After the glue was cured, the protruded platinum was cut and ground into a flat surface before polishing by the 1.0 and 0.3 m Al2 O3 powder at a polishing cloth until a mirror-shine surface was obtained. Before use, the Pt microdisk electrode was sonicated in DI water and rinsed successfully with ethanol, acetone, and DI water. After the preparation of generation electrode and detection electrode, the separation capillary, reaction capillary, and detection electrode were secured in place as shown in Fig. 1A. A straight guide channel fabricated on the PMMA plate by a CO2 laser engraver was used to accommodate the separation capillary, reaction capillary, and detection electrode. Since outer diameters of the separation capillary, reaction capillary, and detection electrode were all 365 m, they could be readily aligned in the vertical direction. Under the observation of a microscope, the separation capillary integrated with Pt film electrode was carefully inserted into the reaction capillary and the Pt microdisk electrode was aligned at the outlet of the reaction capillary.
2.4 Procedure for conditioning and running of CE The new capillary was preconditioned by 1 M NaOH overnight and rinsed by 0.1 M NaOH and DI water each for 5 min before use. Between each run, the capillary was rinsed by the running buffer for 2 min. Before the CE run, a vacuum was applied to the detection vial to fill the separation capillary with the running buffer, the reagent vessel with the KBr reagent, and the reaction capillary www.electrophoresis-journal.com
Electrophoresis 2014, 35, 3556–3563
with a mixture of running buffer and KBr reagent. The vacuum was applied until the KBr reagent completely filled up the reagent vessel with no observed air bubbles. Then, the reagent vial was placed at a specific height to begin the reagent flow. High voltage (+12 kV) was applied across the separation capillary (40 cm in length) and both the upstream generation electrode and downstream detection electrode were turned on. When a stable baseline was obtained at the detection electrode, samples could then be injected for analysis. The sample was injected into the separation capillary electrokinetically by applying 6 kV for 12 s between the sample vial and detection vial. During sample injection, the reagent vial was kept at the same height as the sample and detection vials to avoid back pressure. After sample injection, a high voltage was applied between the buffer vial and detection vial to carry out CE separation and at the same time, the reagent vial was raised back to its original height to restore the reagent flow in the reaction capillary.
3 Results and discussion 3.1 Preliminary study of serial dual-electrode detector From the detection strategy, it is clear that the advantage of the serial dual-electrode detector is its versatility in determination of a large amount of analytes that are difficult to be determined by other methods. All analytes that can react with bromine could be determined. However, as mentioned above, the implementation of serial dual-electrode detection for CE is not an easy task. The serial-dual electrode detection system should be designed carefully to obtain high generation and collection efficiency for dual-electrode detector to achieve good detection sensitivity, as well as to minimize the band broadening caused by bromide introduction to achieve high separation efficiency. Thus, in this work, we developed a new serial dual-electrode detector based on a coaxial postcolumn reactor as shown in Fig. 1. It can be seen that the coaxial postcolumn reactor can be used not only to introduce the bromide reagent after CE separation, but also to facilitate the construction of upstream generation electrode by simply sputtering Pt film onto the outer surface of the separation capillary. Thus, bromide driven into the reaction capillary by gravity was oxidized to bromine at this Pt film electrode efficiently and reacted with the analytes eluted out from the separation capillary. The decrease of bromine was then collected efficiently at the downstream Pt microdisc electrode. From this design, the band broadening caused by bromide introduction can be reduced and satisfactory separation efficiency can be achieved by optimizing the relative flow rate of bromide reagent and separation running buffer as described in the next section. The applicability of the serial dual-electrode detector based on electrogenerated bromine after CE separation was assessed by using GSH as the test analyte. The electropherogram for the injection of 50 M GSH is shown in Fig. 2. It is C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
CE and CEC
3559
Figure 2. The electropherogram for 50 M GSH. The upper solid line was obtained with the generation electrode on while the lower dash line was obtained with the generation electrode off. Generation electrode: +1.0 V versus Ag/AgCl electrode; detection electrode: +0.2 V versus Ag/AgCl electrode; KBr concentration: 5 mM; height difference between bromide vial and buffer vial: 10 cm; CE conditions: 20 mM borate buffer at pH 8.2, separation voltage: 12 kV, sample injection: 6 kV for 12 s.
found that when the generation electrode is off, no detectable signal for GSH can be observed, indicating that GSH cannot be directly determined by the detection electrode held at +0.2 V. When the generation electrode is turned on, a peak due to bromine depletion by reaction with GSH is observed as expected, demonstrating the feasibility of the serial dualelectrode for the determination of analytes that can react with bromine. Electrode fouling is an important problem when electrode is in continuous use. It can be minimized by cleaning the electrode by cycling the potential between 0 and 1.2 V versus Ag/AgCl electrode for 50 scans to obtain reproducible behavior. Such cycling can oxidize and desorb surface impurities and return the electrode surface to a reproducible state. 3.2 Study of serial dual-electrode detector working conditions Serial dual-electrode detector with good generation efficiency and collection efficiency is important to achieve good sensitivity. Suitable generation and detection potentials should be selected to generate a sufficient amount of bromine and detect the reduction of bromine with high sensitivity. Thus, the effect of potentials applied at both electrodes was investigated. As is shown in Fig. 3A, the oxidation of bromide starts at +0.9 V and the generation current gradually increases with the increase of generation potential. Although maximum generation current was achieved at +1.5 V, solvent decomposition may occur at this potential. Thus, in the present work, +1.0 V was selected as the generation potential to generate sufficient bromine for postcolumn reaction detection and to avoid undesirable interference. In Fig. 3B, it is shown that the www.electrophoresis-journal.com
3560
F. Du et al.
Figure 3. (A) The effect of the generation potential on the current generated. (B) The effect of the detection potential on the detection current when the generation electrode is fixed at 1.0 V. Reference electrode: Ag/AgCl electrode. Height difference between bromide reagent vial and detection vial: 10 cm; KBr concentration: 2 mM; separation voltage: 12 kV; CE running buffer: 20 mM borate at pH 8.2.
bromine reduction current is detected at +0.75 V and reaches a plateau at +0.2 V. Use of a more negative potential could result in the liberation of hydrogen, thus +0.2 V was selected as the detection potential in the present work. The flow rate of bromide reagent can also greatly affect the analytical performance of the dual-electrode detector since it determines the mixing process of electrogenerated bromine and analytes. Mixing is a critical process that determines the yield of postcolumn reaction between bromine and analytes. It is also a potential source of peak broadening that can deteriorate the separation efficiency. Since the bromide reagent was introduced hydrodynamically, the height difference between the reagent and buffer vials determines the hydrostatic pressure, and thus determines bromide reagent flow rate. Therefore, to assess the effect of bromide flow rate on the analytical C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Electrophoresis 2014, 35, 3556–3563
Figure 4. (A) The effect of the height difference between reagent vial and buffer vial on the number of theoretical plates (N; ) and the peak height () of 20 M GSH. (B) The effect of the height difference between reagent vial and buffer vial on the number of theoretical plates (N; ) and the peak height () of 20 M GSSG. Generation electrode: +1.0 V versus Ag/AgCl electrode; detection electrode: +0.2 V versus Ag/AgCl electrode; KBr concentration: 2 mM; CE conditions: 20 mM borate buffer at pH 8.2, separation voltage: 12 kV, sample injection: 6 kV for 12 s.
performance of the system, the effect of the height difference on the separation efficiency and detection sensitivity in terms of the number of theoretical plates (N) and peak height (nA), respectively, was investigated with results shown in Fig. 4. GSH and GSSG, which have been demonstrated to be excellent candidates for this detection strategy [30], were used as the test analytes. The electrochemical reactions involved in both detectors and for analyte detection are as follows: Generation electrode : 2Br − 2e → Br2 GSH + 3Br2 + 3H2 O → GSO3 H + 6HBr GSSG + 5Br2 + 6H2 O → 2GSO3 H + 10HBr Detection electrode : Br2 + 2e → 2Br− www.electrophoresis-journal.com
CE and CEC
Electrophoresis 2014, 35, 3556–3563
3561
Table 1. The effect of bromide concentration on the detection limit and linear working range of GSH and GSSG by the developed CE-serial dual-electrode detector
KBr concentration
GSH
GSSG
(mM)
Detection limita) (M)
Linear working rangeb) (M)
Detection limita) (M)
Linear working rangeb) (M)
1 2 5 10
0.096 0.16 0.54 2.2
0.330 0.550 2100 5200
0.085 0.14 0.46 1.9
0.325 0.440 1.580 4150
Experimental conditions: Generation electrode: +1.0 V versus Ag/AgCl electrode; detection electrode: +0.2 V versus Ag/AgCl electrode; height difference between bromide reagent vial and detection vial: 10 cm; CE conditions: 20 mM borate buffer at pH 8.2, separation voltage: 12 kV, sample injection: 6 kV for 12 s. a) Calculated based on S/N = 3. b) Correlation coefficient >0.995, peak height versus analyte concentration.
It is found that the increase in height difference leads to an increase in the number of theoretical plates for both GSH and GSSG. The lower N at smaller height difference may be due to the band broadening caused by longer residence time in reaction capillary. On the other hand, the use of a larger height difference causes the reduction in peak height for GSH and GSSG, which is due to the dilution of analyte zone migrating out of separation capillary at a high bromide flow rate. As a compromise, 10 cm is chosen as the working height difference and used for the subsequent studies and application described below. It is noteworthy that at this height difference, the separation efficiency (N around 30 000) obtained by our design is significantly higher than the previously reported dual-electrode detector (N estimated at 10002000) [25], making our design more favorable in the analysis of complex sample matrix. Since analytes are determined indirectly by the decrease of the concentration of electrogenerated bromine due to its stoichiometric reaction with analytes, the detection limit and linear working range are greatly dependent on the electrogenerated bromine concentration. We examined the effect of different bromide concentrations on the detection limit and linear working range for GSH and GSSG with results summarized in Table 1. It is found that lowest detection limits for GSH and GSSG were achieved at a bromide concentration of 1 mM, which exhibits the lowest background noise and highest S/N ratio. Both the lower and upper limits of the working range for GSH and GSSG increase with the increase in the bromide concentration. However, the greatest linear working range for GSH and GSSG was obtained at 2 mM bromide concentration. Since GSSG is usually present at a level much lower than GSH in pharmaceutical samples, greater linear working ranges are preferred to achieve simultaneous determination of GSSG and GSH in one CE run. Thus, in this study, 2 mM bromide was selected and used for the subsequent application. Using working conditions described above, the detection limit, linearity, and reproducibility were determined and summarized. The detection limit at S/N = 3 was 0.16 M for GSH and 0.14 M for GSSG. The linear working range for C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
GSH was 0.5–50 M with a correlation coefficient of 0.998 and a slope of 0.64 nA/M. The linear working range for GSSG was 0.4–40 M with a correlation coefficient of 0.997 and a slope of 0.79 nA/M. The reproducibility of the method was evaluated by analyzing the mixture of 20 M GSH and 20 M GSSG on the same day in five replicates (intraday reproducibility) and over 5 consecutive days in duplicates (interday reproducibility). The intraday RSDs for the determination of GSH and GSSG were 5.7 and 6.2%, respectively (n = 5). The interday RSDs for the determination of GSH and GSSG were 6.4 and 6.7%, respectively (n = 5). From these results, it is clear that the developed serial dual-electrode detector for CE based on coaxial postcolumn reactor is suitable for this indirect detection scheme based on electrogenerated bromine. The concentration detection limits for GSH and GSSG obtained by our method are lower than most of CE-based methods, including CE-UV [31, 32] and CE-MS [33]. Compared with CE-LIF, which is the most sensitive CE-based method, the detection limits of our method are several times higher than that of CE-LIF [34–36]. However, CE-LIF is more time consuming since derivatization of GSH and GSSG is usually required due to the lack of fluorophores in their chemical structure. In addition, one important feature of our proposed method is the tunable linear working range for GSH and GSSG achieved by varying the bromide reagent concentration. This makes it more flexible to meet the requirement of detecting different levels of GSH and GSSG in different applications.
3.3 Applicability of the CE-serial dual-electrode detector for determination of GSH and GSSG in pharmaceutical supplement GSH is one of the most powerful antioxidants in the biological systems and it can help people stay healthy and prevent aging, cancer, heart disease, etc. GSH in pharmaceutical supplement is mainly present in the reduced form. However, it may convert into the oxidized form GSSG during the production. Thus, the simultaneous determination of GSH and its www.electrophoresis-journal.com
3562
F. Du et al.
Electrophoresis 2014, 35, 3556–3563
Table 2. The amount of GSH and GSSG impurity in GSH supplement tablet determined by the developed CE-serial dual-electrode detector
1 2 3 4 5 Mean ± SD
GSHfound mg/tablet
GSHfound /GSHclaimed (%)
GSSGfound mg/tablet
GSSGfound / GSHfound (%)
50.1 48.3 49.4 53.5 52.9 50.8 ± 2.3
100.2 96.6 98.8 107.0 105.8 101.7 ± 4.5
1.13 1.12 1.16 1.31 1.22 1.19 ± 0.08
2.26 2.31 2.35 2.45 2.31 2.34 ± 0.07
4 Concluding remarks A new serial dual-electrode detector after CE separation was designed and established for postcolumn reaction detection based on electrogenerated bromine. With appropriate introduction of bromide and carefully design of serial dualelectrode detector based on a coaxial postcolumn reactor, the serial dual-electrode detection is shown with capability in significantly improving the separation efficiency and obtaining high sensitivity for GSH and GSSG. The method has been successfully applied to the determination of GSH and GSSG in a pharmaceutical supplement sample. Future work will focus on the detection of GSH and GSSG in biological samples such as cells and detection of other analytes that can react with bromine.
Figure 5. The typical electropherogram of the L-glutathione supplement sample. Inset is the enlarged electropherogram for the GSSG peak. Generation electrode: +1.0 V versus Ag/AgCl electrode; detection electrode: +0.2 V versus Ag/AgCl electrode; KBr concentration: 2 mM; height difference between bromide vial and buffer vial: 10 cm; CE conditions: 20 mM borate buffer at pH 8.2, separation voltage: 12 kV, sample injection: 6 kV for 12 s.
The authors would like to acknowledge the support of the Hong Kong Research Grants Council of the Hong Kong Special Administrative Region, China (no. 702210), Seed Fund in Basic Research from the Hong Kong University Research and Conference Grants Committee, Specialized Research Fund for Doctoral Program of Higher Education of China (no. 20130141120049), and Self-topic Fund of Wuhan University (no. 2042014kf0005). The authors have declared no conflict of interest.
impurity GSSH in pharmaceutical supplement is required for the quality control and routine analysis. Although AD is sensitive, direct AD of these compounds, especially GSSG, is very difficult because of their high overpotentials for oxidation [37]. Thus, the developed CE-serial dual-electrode detection based on indirect detection is applied to determine the GSH and GSSG impurity in commercial available supplement. Figure 5 is the typical electropherogram of the supplement sample. Five replicate determinations were made with results shown in Table 2. The amount of active ingredient GSH in pharmaceutical supplement was found to be 50.8 ± 2.3 mg/tablet (n = 5), which is in good agreement with the claimed value on the label. The weight percentage of the impurity GSSG to GSH was 2.34 ± 0.07% (n = 5), indicating that a small portion of GSH converted to GSSG during production process. The recovery experiments were performed by spiking 10 M GSH and 1 M GSSG to the sample. Recoveries for GSH and GSSG are calculated to be 102 and 105%, respectively, indicating the reliability of the developed CE-serial dual-electrode detection method. C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
5 References [1] Jorgenson, J. W., Lukacs, K. D., Anal. Chem. 1981, 53, 1298–1302. [2] Castro-Puyana, M., Garcia-Canas, V., Simo, C., Cifuentes, A., Electrophoresis 2012, 33, 147–167. [3] Pan, Y. L., Chen, F., Zhang, M. Y., Wang, T. Q., Xu, Z. C., Zhang, W., Chu, Q. C., Ye, J. N., Electrophoresis 2013, 34, 1241–1248. [4] Rabanes, H. R., Guidote, A. M., Quirino, J. P., Electrophoresis 2012, 33, 180–195. [5] Alhusban, A. A., Breadmore, M. C., Guijt, R. M., Electrophoresis 2013, 34, 1465–1482. [6] Tang, D. F., Ma, G., Zhang, L. Y., Chen, G., J. Chromatogr. A 2013, 1316, 127–134. [7] Liu, X. Q., Yu, Y. Y., Gu, H., Zhou, T. S., Wang, L. L., Mei, B., Shi, G. Y., Electrophoresis 2013, 34, 935–943. [8] Inoue, J., Kaneta, T., Imasaka, T., Electrophoresis 2012, 33, 2743–2747.
www.electrophoresis-journal.com
Electrophoresis 2014, 35, 3556–3563
[9] Xu, L., Guo, Q. H., Yu, H., Huang, J. S., You, T. Y., Talanta 2012, 97, 462–467. [10] Wallingford, R. A., Ewing, A. G., Anal. Chem. 1987, 59, 1762–1766. [11] Park, S., McGrath, M. J., Smyth, M. R., Diamond, D., Lunte, C. E., Anal. Chem. 1997, 69, 2994–3001. [12] Wang, C. W., Chen, I. C., Anal. Chem. 1992, 64, 2461–2464.
CE and CEC
3563
[24] Zhong, M., Lunte, S. M., Anal. Chem. 1999, 71, 251–255. [25] Holland, L. A., Lunte, S. M., Anal. Chem. 1999, 71, 407–412. [26] Rose, D. J., Jorgenson, J. W., J. Chromatogr. A 1988, 447, 117–131. [27] Voegel, P. D., Zhou, W., Baldwin, R. P., Anal. Chem. 1997, 69, 951–957.
[13] Kok, W. T., Sahin, Y., Anal. Chem. 1993, 65, 2497–2501.
[28] Liu, J., Cao, W., Qiu, H., Sun, X., Yang, X., Wang, E., Clin. Chem. 2002, 48, 1049–1058.
[14] Hu, S., Wang, Z. L., Li, P. B., Cheng, J. K., Anal. Chem. 1997, 69, 264–267.
[29] Liu, Y. M., Li, J., Yang, Y., Du, J., Mei, L., Can. J. Chem. 2012, 90, 180–185.
[15] Huang, X., Zare, R. N., Sloss, S., Ewing, A. G., Anal. Chem. 1991, 63, 189–192.
[30] Zhou, H., Holland, L. A., Liu, P., Analytst 2001, 126, 1252–1256.
[16] Sloss, S., Ewing, A. G., Anal. Chem. 1993, 65, 577–581.
[31] Hoque, M. E., Arnett, S. D., Lunte, C. E., J. Chromatogr. B 2005, 827, 51–57.
[17] Wang, J., Tian, B. M., Chatrathi, M. P., Escarpa, A., Pumera, M., Electrophoresis 2009, 30, 3334–3338. [18] Garcia, M., Alonso-Fernandez, J. R., Escarpa, A., Anal. Chem. 2013, 85, 9116–9125. [19] Chua, C. K., Pumera, M., Electroanalysis 2013, 25, 945–950. [20] Du, F. Y., Fung, Y. S., Electrophoresis 2010, 31, 2192–2199. [21] Chen, D. C., Chang, S. S., Chen, C. H., Anal. Chem. 1999, 71, 3200–3205.
[32] Carru, C., Zinellu, A., Pes, G. M., Marongiu, G., Tadolini, B., Deiana, L., Electrophoresis 2002, 23, 1716–1721. [33] He, T., Quinn, D., Fu, E., Wang, Y. K., J. Chromatogr. B 1999, 727, 43–52. [34] Zhang, J., Hu, Z., Chen, X., Talanta 2005, 65, 986–990. [35] Lavigne, V., Pons, A., Dubourdieu, D., J. Chromatogr. A 2007, 1139, 130–135.
[22] Dorris, M. K., Crick, E. W., Lunte, C. E., Eletrophoresis 2012, 33, 2725–2732.
[36] Qin, J., Ye, N., Yu, L., Liu, D., Fung, Y., Wang, W., Ma, X., Lin, B., Electrophoresis 2005, 26, 1155–1162.
´ L. A., Zare, R. N., J. Chromatogr. A 1994, [23] Lin, B. L., Colon, 680, 263–270.
[37] Safavi, A., Maleki, N., Farjami, E., Mahyari, F. A., Anal. Chem. 2009, 81, 7538–7543.
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.electrophoresis-journal.com
of Water Quality Engineering, School of Power and Mechanical Engineering, Wuhan University, Wuhan, China 2 Department of Chemistry, The University of Hong Kong, Hong Kong SAR, China
Received May 24, 2014 Revised August 30, 2014 Accepted August 30, 2014
Electrophoresis 2014, 35, 3556–3563
Research Article
A serial dual-electrode detector based on electrogenerated bromine for capillary electrophoresis A new serial dual-electrode detector for CE has been designed and fabricated for postcolumn reaction detection based on electrogenerated bromine. A coaxial postcolumn reactor was employed to introduce bromide reagent and facilitate the fabrication of upstream generation electrode by simply sputtering Pt film onto the outer surface of the separation capillary. Bromide introduced could be efficiently converted to bromine at this Pt film electrode and subsequently detected by the downstream Pt microdisk detection electrode. Analytes that react with bromine could be determined by the decrease of bromine reduction current at the downstream electrode resulting from the reaction between analytes and bromine. The effects of serial dual-electrode detector working conditions including electrode potentials, bromide flow rate, and bromide concentration on analytical performance were investigated using glutathione (GSH) and glutathione disulfide (GSSG) as test analytes. Under the optimal conditions, detection limits down to 0.16 M for GSH and 0.14 M for GSSG (S/N = 3) as well as linear working ranges of two orders of magnitude for GSH and GSSG were achieved. Furthermore, the separation efficiency obtained by our dual-electrode detector design was greatly improved compared with previous reported design. The developed method has been successfully applied to determine the GSH and GSSG impurity in commercial GSH supplement. Keywords: Capillary electrophoresis / Dual-electrode detector / Electrogenerated bromine / Glutathione / Indirect amperometric detection DOI 10.1002/elps.201400257
1 Introduction Since the introduction of CE in the early 1980s [1], the technique has drawn intensive attention due to its high separation efficiency, low sample and solvent consumption, as well as short analysis time. These features make it especially suitable for the analysis of food [2], environment [3], drugs [4], and biological samples [5]. CE has been hyphenated frequently with UV-Vis absorbance, fluorescence, MS, amperometric detection (AD), etc. Among these detection methods, AD with detection limit in the subnanomolar range (0.1 nM) is one of the most sensitive detection methods coupled with CE separation. Unlike optical detection methods, the miniaturization of the detection electrodes does not compromise the analytical performance of AD, which is an attractive feature for CE system. Moreover, the instrumentation for AD is less expensive than that for fluorescence detection and MS, of
Correspondence: Dr. Fuying Du, Department of Water Quality Engineering, School of Power and Mechanical Engineering, Wuhan University, Wuhan 430072, China E-mail: [email protected]
Abbreviations: AD, amperometric detection; GSH, glutathione; GSSG, glutathione disulfide
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
which sensitivity is comparable to AD. In the past few years, there has been a great deal of research integrating CE with AD [6–9]. In most reported CE-AD systems, one single working electrode is employed either in off-column detection mode [10–14] or in end-column detection mode [15–19]. Recently, the research of CE-AD system employing two working electrodes has started and attracted much more attention due to its flexibility in operation mode and its capability in improving the sensitivity, selectivity, and increasing the applicability of CE-AD through indirect detection scheme. Basically, there are two operation modes for dual-electrode detection: parallel mode and serial mode. Dual-electrode detector in parallel mode can be used to improve the selectivity and sensitivity of the CE-AD. Our group [20] developed a parallel opposite dual-electrode detector controlled at +0.8 and +1.0 V, respectively, to identify polyphenol peaks in red wine. The peak current ratio obtained by the dual-electrode detector was used for identification of the eluted peaks to improve the detection selectivity. Chen et al. [21] fabricated a parallelopposed dual-electrode detector that comprised a gold disk electrode and an on-capillary gold film electrode in a thin-layer
∗ Additional corresponding author: Dr. Fung Ying-Sing, E-mail: [email protected]
www.electrophoresis-journal.com
CE and CEC
Electrophoresis 2014, 35, 3556–3563
geometry. Due to the redox recycling of dopamine between these two electrodes, the sensitivity was enhanced ten times as compared to the conventional single-electrode end-column detection. Lunte’s group [22] also reported a parallel dualelectrode detector operating in both the redox cycling mode and dual-potential mode to improve the detection sensitivity and selectivity. Dual-electrode detector in serial mode has been also constructed and investigated for CE-AD. The initial application of serial dual-electrode detector was reported by Lin et al. [23] for the determination of cysteine and cystine in urine. Cystine was reduced at upstream Au-Hg wire electrode set at –1.0 V and then detected at downstream Au-Hg wire electrode set at +0.15 V. The detection limit was reported to be 100 M for cystine. Zhong et al. [24] improved this detection mode by employing a tubular-wire configuration for serial dual-electrode detector. Due to the high conversion efficiency at the upstream tube electrode, the detection limit was lowered to 0.5 M for cystine. Holland and Lunte [25] implemented the first serial dual-electrode detection after CE separation for indirect detection based on electrogenerated bromine. In this detection strategy, bromine was generated at the upstream electrode and then reduced back to bromide at the downstream electrode. The analytes migrating out of the separation capillary reacted with the electrogenerated bromine to give rise to a decrease in the current detected at the downstream electrode. This approach holds great promise in widening the scope of CE-AD system because a large number of compounds that are difficult to be detected by direct AD can be determined by this method. However, the development of serial dual-electrode detector for CE with high sensitivity and high separation efficiency is not an easy task. First, the arrangement of upstream generation electrode and downstream detection electrode should be carefully designed to obtain high generation and collection efficiency to achieve good detection sensitivity. Second, the way for bromide reagent introduction should be deliberated to reduce the band broadening caused by bromide introduction. In Holland and Lunte’s work, dual electrode in wire– wire configuration was employed in off-column mode after cathodic reservoir. The bromide reagent was introduced into CE system hydrodynamically through a crack on capillary at the cathodic reservoir. Due to the disturbance of the analytical stream by uneven introduction of the bromide reagent at the cathodic reservoir, as well as the back pressure present in the capillary after the decoupling fracture, a serious peak broadening was observed, leading to a poor separation efficiency and hampering the application of this method in the analysis of complex sample matrix. In order to solve this problem, we described a new design for serial dual-electrode detector after CE for indirect AD based on electrogenerated bromine in the present work. Bromide was introduced by a conventional coaxial postcolumn reactor [26] and converted to bromine efficiently at the on-capillary electrode made by sputtering Pt film on the outer surface of separation capillary. The electrogenerated bromine reacted with analytes migrating out of the separation capillary in reaction capillary and detected at a Pt C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
3557
microdisk detection electrode aligned at the outlet of reaction capillary. The analytical performance of the developed serial dual-electrode detector was evaluated by using glutathione (GSH) and glutathione disulfide (GSSG) as test analytes. Experimental parameters governing the analytical performance were investigated and optimized. The results showed that low detection limits for the determination of GSH and GSSG can be achieved. Furthermore, band broadening was greatly reduced by the developed serial dual-electrode detector. The applicability of the developed CE-serial dual-electrode detector for the determination of GSH and GSSG impurity in GSH supplement was demonstrated.
2 Materials and methods 2.1 Chemicals and samples L-GSH
and GSSG were purchased from Sigma-Aldrich (St. Louis, MO, USA). Sodium tetraborate and potassium bromide were obtained from BDH Chemicals (Poole, UK). All chemicals were used as received. Stock solutions of GSH and GSSG were made up to 1 mM in DI water and stored at –20°C until use. Standard solutions with desired concentrations for calibration were prepared by diluting the stock solutions with the running buffer before analysis. The running buffer consisted of 20 mM borate buffer with pH adjusted to 8.2. The bromide reagent was prepared by dissolving a desired amount of potassium bromide in the running buffer. Pharmaceutical GSH supplement claimed to have 50 mg GSH per tablet was purchased from a local drug store. A tablet was weighed (0.6035 ± 0.0037 g, n = 5) and then ground thoroughly. A 120 mg portion of the powdered sample was dissolved in 100 mL running buffer, diluted ten times by running buffer, and filtered to remove insoluble cellulose. Prior to analysis, all the solutions were filtered through a 0.2 m hydrophilic polyethersulfone membrane filter (Pall, Washington, NY, USA). 2.2 Instrumentation
A self-assembled reversible high-voltage power supply (Bertan, NY, USA) was used to provide a variable voltage from 0 to 24 kV for CE separation. Fused-silica capillaries (Yongnian, Hebei, China) with the same outer diameter at 365 m and different inner diameters of 25 and 150 m were used as the separation and reaction capillaries, respectively. The potentials for the upstream generation electrode and downstream detection electrode were controlled by a CHI 760d (CH Instruments, Shanghai, China) and a WPI MicroC potentiostat (Sarasota, FL, USA), respectively. Data were collected and processed by Origin 6.1 and N2000 chromatography data system (Zhi Da, Zhejiang, China). A CO2 laser engraver (Vseries, Pinnacle, USA) was used to fabricate channel pattern onto the PMMA chip to accommodate the separation capillary, reaction capillary, and detection electrode. www.electrophoresis-journal.com
3558
F. Du et al.
Electrophoresis 2014, 35, 3556–3563
Figure 1. (A) Schematic diagram showing the serial dualelectrode detection system based on coaxial postcolumn reactor. (B) The enlarged view for the serial dual-electrode configuration.
2.3 Fabrication of serial dual-electrode detector A schematic diagram showing the design of serial dualelectrode detector after CE separation is illustrated in Fig. 1. A coaxial postcolumn reactor as shown in Fig. 1A was employed to introduce bromide reagent hydrodynamically. The dual-electrode detector consisted of an on-capillary Pt film upstream electrode and a Pt microdisk downstream electrode as shown in Fig. 1B. The upstream generation electrode was made by sputtering Pt film onto the outer surface of separation capillary as reported by Baldwin [27]. Before sputtering, the end of the separation capillary was etched in hydrofluoric acid (48%) until its outer diameter was reduced to about 110 m so that it can be inserted into the reaction capillary. Then, the capillary was placed into the sputter chamber (BAL-TEC SCD 005 Sputter Coater, Switzerland) for platinum coating. To avoid undesirable electrogeneration of bromine in the reagent vessel, the outer surface of the separation capillary exposed to solution in the reagent vessel was covered with a layer of nonconductive epoxy. The electrical connection to the on-capillary Pt film electrode was made by cementing copper wire onto the capillary with conductive silver epoxy (ITW Chemtronics) at the edge of PMMA plates as shown in Fig. 1A. The length for the separation capillary inserted into the reaction capillary was fixed at 3.0 mm to make the area of the upstream electrode constant throughout the experiments. In order to make the upstream electrode free from interference from high separation voltage, its corresponding Ag/AgCl reference electrode was also placed in the reagent vessel. The distance between the end of the separation capillary and end of the reaction capillary, which determines the reaction zone of the postcolumn reactor, was kept at 4.0 mm in all experiments. Pt microdisk electrode placed about 20 m away from the outlet of the reaction capillary was employed as the down C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
stream detection electrode. The fabrication method for the Pt microdisk electrode was similar to those previously reported in the literature [28,29]. Platinum wire with 200 m diameter (Goodfellow, UK) was inserted into the capillary (250 m id, 365 m od) and fixed in position by glue. After the glue was cured, the protruded platinum was cut and ground into a flat surface before polishing by the 1.0 and 0.3 m Al2 O3 powder at a polishing cloth until a mirror-shine surface was obtained. Before use, the Pt microdisk electrode was sonicated in DI water and rinsed successfully with ethanol, acetone, and DI water. After the preparation of generation electrode and detection electrode, the separation capillary, reaction capillary, and detection electrode were secured in place as shown in Fig. 1A. A straight guide channel fabricated on the PMMA plate by a CO2 laser engraver was used to accommodate the separation capillary, reaction capillary, and detection electrode. Since outer diameters of the separation capillary, reaction capillary, and detection electrode were all 365 m, they could be readily aligned in the vertical direction. Under the observation of a microscope, the separation capillary integrated with Pt film electrode was carefully inserted into the reaction capillary and the Pt microdisk electrode was aligned at the outlet of the reaction capillary.
2.4 Procedure for conditioning and running of CE The new capillary was preconditioned by 1 M NaOH overnight and rinsed by 0.1 M NaOH and DI water each for 5 min before use. Between each run, the capillary was rinsed by the running buffer for 2 min. Before the CE run, a vacuum was applied to the detection vial to fill the separation capillary with the running buffer, the reagent vessel with the KBr reagent, and the reaction capillary www.electrophoresis-journal.com
Electrophoresis 2014, 35, 3556–3563
with a mixture of running buffer and KBr reagent. The vacuum was applied until the KBr reagent completely filled up the reagent vessel with no observed air bubbles. Then, the reagent vial was placed at a specific height to begin the reagent flow. High voltage (+12 kV) was applied across the separation capillary (40 cm in length) and both the upstream generation electrode and downstream detection electrode were turned on. When a stable baseline was obtained at the detection electrode, samples could then be injected for analysis. The sample was injected into the separation capillary electrokinetically by applying 6 kV for 12 s between the sample vial and detection vial. During sample injection, the reagent vial was kept at the same height as the sample and detection vials to avoid back pressure. After sample injection, a high voltage was applied between the buffer vial and detection vial to carry out CE separation and at the same time, the reagent vial was raised back to its original height to restore the reagent flow in the reaction capillary.
3 Results and discussion 3.1 Preliminary study of serial dual-electrode detector From the detection strategy, it is clear that the advantage of the serial dual-electrode detector is its versatility in determination of a large amount of analytes that are difficult to be determined by other methods. All analytes that can react with bromine could be determined. However, as mentioned above, the implementation of serial dual-electrode detection for CE is not an easy task. The serial-dual electrode detection system should be designed carefully to obtain high generation and collection efficiency for dual-electrode detector to achieve good detection sensitivity, as well as to minimize the band broadening caused by bromide introduction to achieve high separation efficiency. Thus, in this work, we developed a new serial dual-electrode detector based on a coaxial postcolumn reactor as shown in Fig. 1. It can be seen that the coaxial postcolumn reactor can be used not only to introduce the bromide reagent after CE separation, but also to facilitate the construction of upstream generation electrode by simply sputtering Pt film onto the outer surface of the separation capillary. Thus, bromide driven into the reaction capillary by gravity was oxidized to bromine at this Pt film electrode efficiently and reacted with the analytes eluted out from the separation capillary. The decrease of bromine was then collected efficiently at the downstream Pt microdisc electrode. From this design, the band broadening caused by bromide introduction can be reduced and satisfactory separation efficiency can be achieved by optimizing the relative flow rate of bromide reagent and separation running buffer as described in the next section. The applicability of the serial dual-electrode detector based on electrogenerated bromine after CE separation was assessed by using GSH as the test analyte. The electropherogram for the injection of 50 M GSH is shown in Fig. 2. It is C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
CE and CEC
3559
Figure 2. The electropherogram for 50 M GSH. The upper solid line was obtained with the generation electrode on while the lower dash line was obtained with the generation electrode off. Generation electrode: +1.0 V versus Ag/AgCl electrode; detection electrode: +0.2 V versus Ag/AgCl electrode; KBr concentration: 5 mM; height difference between bromide vial and buffer vial: 10 cm; CE conditions: 20 mM borate buffer at pH 8.2, separation voltage: 12 kV, sample injection: 6 kV for 12 s.
found that when the generation electrode is off, no detectable signal for GSH can be observed, indicating that GSH cannot be directly determined by the detection electrode held at +0.2 V. When the generation electrode is turned on, a peak due to bromine depletion by reaction with GSH is observed as expected, demonstrating the feasibility of the serial dualelectrode for the determination of analytes that can react with bromine. Electrode fouling is an important problem when electrode is in continuous use. It can be minimized by cleaning the electrode by cycling the potential between 0 and 1.2 V versus Ag/AgCl electrode for 50 scans to obtain reproducible behavior. Such cycling can oxidize and desorb surface impurities and return the electrode surface to a reproducible state. 3.2 Study of serial dual-electrode detector working conditions Serial dual-electrode detector with good generation efficiency and collection efficiency is important to achieve good sensitivity. Suitable generation and detection potentials should be selected to generate a sufficient amount of bromine and detect the reduction of bromine with high sensitivity. Thus, the effect of potentials applied at both electrodes was investigated. As is shown in Fig. 3A, the oxidation of bromide starts at +0.9 V and the generation current gradually increases with the increase of generation potential. Although maximum generation current was achieved at +1.5 V, solvent decomposition may occur at this potential. Thus, in the present work, +1.0 V was selected as the generation potential to generate sufficient bromine for postcolumn reaction detection and to avoid undesirable interference. In Fig. 3B, it is shown that the www.electrophoresis-journal.com
3560
F. Du et al.
Figure 3. (A) The effect of the generation potential on the current generated. (B) The effect of the detection potential on the detection current when the generation electrode is fixed at 1.0 V. Reference electrode: Ag/AgCl electrode. Height difference between bromide reagent vial and detection vial: 10 cm; KBr concentration: 2 mM; separation voltage: 12 kV; CE running buffer: 20 mM borate at pH 8.2.
bromine reduction current is detected at +0.75 V and reaches a plateau at +0.2 V. Use of a more negative potential could result in the liberation of hydrogen, thus +0.2 V was selected as the detection potential in the present work. The flow rate of bromide reagent can also greatly affect the analytical performance of the dual-electrode detector since it determines the mixing process of electrogenerated bromine and analytes. Mixing is a critical process that determines the yield of postcolumn reaction between bromine and analytes. It is also a potential source of peak broadening that can deteriorate the separation efficiency. Since the bromide reagent was introduced hydrodynamically, the height difference between the reagent and buffer vials determines the hydrostatic pressure, and thus determines bromide reagent flow rate. Therefore, to assess the effect of bromide flow rate on the analytical C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Electrophoresis 2014, 35, 3556–3563
Figure 4. (A) The effect of the height difference between reagent vial and buffer vial on the number of theoretical plates (N; ) and the peak height () of 20 M GSH. (B) The effect of the height difference between reagent vial and buffer vial on the number of theoretical plates (N; ) and the peak height () of 20 M GSSG. Generation electrode: +1.0 V versus Ag/AgCl electrode; detection electrode: +0.2 V versus Ag/AgCl electrode; KBr concentration: 2 mM; CE conditions: 20 mM borate buffer at pH 8.2, separation voltage: 12 kV, sample injection: 6 kV for 12 s.
performance of the system, the effect of the height difference on the separation efficiency and detection sensitivity in terms of the number of theoretical plates (N) and peak height (nA), respectively, was investigated with results shown in Fig. 4. GSH and GSSG, which have been demonstrated to be excellent candidates for this detection strategy [30], were used as the test analytes. The electrochemical reactions involved in both detectors and for analyte detection are as follows: Generation electrode : 2Br − 2e → Br2 GSH + 3Br2 + 3H2 O → GSO3 H + 6HBr GSSG + 5Br2 + 6H2 O → 2GSO3 H + 10HBr Detection electrode : Br2 + 2e → 2Br− www.electrophoresis-journal.com
CE and CEC
Electrophoresis 2014, 35, 3556–3563
3561
Table 1. The effect of bromide concentration on the detection limit and linear working range of GSH and GSSG by the developed CE-serial dual-electrode detector
KBr concentration
GSH
GSSG
(mM)
Detection limita) (M)
Linear working rangeb) (M)
Detection limita) (M)
Linear working rangeb) (M)
1 2 5 10
0.096 0.16 0.54 2.2
0.330 0.550 2100 5200
0.085 0.14 0.46 1.9
0.325 0.440 1.580 4150
Experimental conditions: Generation electrode: +1.0 V versus Ag/AgCl electrode; detection electrode: +0.2 V versus Ag/AgCl electrode; height difference between bromide reagent vial and detection vial: 10 cm; CE conditions: 20 mM borate buffer at pH 8.2, separation voltage: 12 kV, sample injection: 6 kV for 12 s. a) Calculated based on S/N = 3. b) Correlation coefficient >0.995, peak height versus analyte concentration.
It is found that the increase in height difference leads to an increase in the number of theoretical plates for both GSH and GSSG. The lower N at smaller height difference may be due to the band broadening caused by longer residence time in reaction capillary. On the other hand, the use of a larger height difference causes the reduction in peak height for GSH and GSSG, which is due to the dilution of analyte zone migrating out of separation capillary at a high bromide flow rate. As a compromise, 10 cm is chosen as the working height difference and used for the subsequent studies and application described below. It is noteworthy that at this height difference, the separation efficiency (N around 30 000) obtained by our design is significantly higher than the previously reported dual-electrode detector (N estimated at 10002000) [25], making our design more favorable in the analysis of complex sample matrix. Since analytes are determined indirectly by the decrease of the concentration of electrogenerated bromine due to its stoichiometric reaction with analytes, the detection limit and linear working range are greatly dependent on the electrogenerated bromine concentration. We examined the effect of different bromide concentrations on the detection limit and linear working range for GSH and GSSG with results summarized in Table 1. It is found that lowest detection limits for GSH and GSSG were achieved at a bromide concentration of 1 mM, which exhibits the lowest background noise and highest S/N ratio. Both the lower and upper limits of the working range for GSH and GSSG increase with the increase in the bromide concentration. However, the greatest linear working range for GSH and GSSG was obtained at 2 mM bromide concentration. Since GSSG is usually present at a level much lower than GSH in pharmaceutical samples, greater linear working ranges are preferred to achieve simultaneous determination of GSSG and GSH in one CE run. Thus, in this study, 2 mM bromide was selected and used for the subsequent application. Using working conditions described above, the detection limit, linearity, and reproducibility were determined and summarized. The detection limit at S/N = 3 was 0.16 M for GSH and 0.14 M for GSSG. The linear working range for C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
GSH was 0.5–50 M with a correlation coefficient of 0.998 and a slope of 0.64 nA/M. The linear working range for GSSG was 0.4–40 M with a correlation coefficient of 0.997 and a slope of 0.79 nA/M. The reproducibility of the method was evaluated by analyzing the mixture of 20 M GSH and 20 M GSSG on the same day in five replicates (intraday reproducibility) and over 5 consecutive days in duplicates (interday reproducibility). The intraday RSDs for the determination of GSH and GSSG were 5.7 and 6.2%, respectively (n = 5). The interday RSDs for the determination of GSH and GSSG were 6.4 and 6.7%, respectively (n = 5). From these results, it is clear that the developed serial dual-electrode detector for CE based on coaxial postcolumn reactor is suitable for this indirect detection scheme based on electrogenerated bromine. The concentration detection limits for GSH and GSSG obtained by our method are lower than most of CE-based methods, including CE-UV [31, 32] and CE-MS [33]. Compared with CE-LIF, which is the most sensitive CE-based method, the detection limits of our method are several times higher than that of CE-LIF [34–36]. However, CE-LIF is more time consuming since derivatization of GSH and GSSG is usually required due to the lack of fluorophores in their chemical structure. In addition, one important feature of our proposed method is the tunable linear working range for GSH and GSSG achieved by varying the bromide reagent concentration. This makes it more flexible to meet the requirement of detecting different levels of GSH and GSSG in different applications.
3.3 Applicability of the CE-serial dual-electrode detector for determination of GSH and GSSG in pharmaceutical supplement GSH is one of the most powerful antioxidants in the biological systems and it can help people stay healthy and prevent aging, cancer, heart disease, etc. GSH in pharmaceutical supplement is mainly present in the reduced form. However, it may convert into the oxidized form GSSG during the production. Thus, the simultaneous determination of GSH and its www.electrophoresis-journal.com
3562
F. Du et al.
Electrophoresis 2014, 35, 3556–3563
Table 2. The amount of GSH and GSSG impurity in GSH supplement tablet determined by the developed CE-serial dual-electrode detector
1 2 3 4 5 Mean ± SD
GSHfound mg/tablet
GSHfound /GSHclaimed (%)
GSSGfound mg/tablet
GSSGfound / GSHfound (%)
50.1 48.3 49.4 53.5 52.9 50.8 ± 2.3
100.2 96.6 98.8 107.0 105.8 101.7 ± 4.5
1.13 1.12 1.16 1.31 1.22 1.19 ± 0.08
2.26 2.31 2.35 2.45 2.31 2.34 ± 0.07
4 Concluding remarks A new serial dual-electrode detector after CE separation was designed and established for postcolumn reaction detection based on electrogenerated bromine. With appropriate introduction of bromide and carefully design of serial dualelectrode detector based on a coaxial postcolumn reactor, the serial dual-electrode detection is shown with capability in significantly improving the separation efficiency and obtaining high sensitivity for GSH and GSSG. The method has been successfully applied to the determination of GSH and GSSG in a pharmaceutical supplement sample. Future work will focus on the detection of GSH and GSSG in biological samples such as cells and detection of other analytes that can react with bromine.
Figure 5. The typical electropherogram of the L-glutathione supplement sample. Inset is the enlarged electropherogram for the GSSG peak. Generation electrode: +1.0 V versus Ag/AgCl electrode; detection electrode: +0.2 V versus Ag/AgCl electrode; KBr concentration: 2 mM; height difference between bromide vial and buffer vial: 10 cm; CE conditions: 20 mM borate buffer at pH 8.2, separation voltage: 12 kV, sample injection: 6 kV for 12 s.
The authors would like to acknowledge the support of the Hong Kong Research Grants Council of the Hong Kong Special Administrative Region, China (no. 702210), Seed Fund in Basic Research from the Hong Kong University Research and Conference Grants Committee, Specialized Research Fund for Doctoral Program of Higher Education of China (no. 20130141120049), and Self-topic Fund of Wuhan University (no. 2042014kf0005). The authors have declared no conflict of interest.
impurity GSSH in pharmaceutical supplement is required for the quality control and routine analysis. Although AD is sensitive, direct AD of these compounds, especially GSSG, is very difficult because of their high overpotentials for oxidation [37]. Thus, the developed CE-serial dual-electrode detection based on indirect detection is applied to determine the GSH and GSSG impurity in commercial available supplement. Figure 5 is the typical electropherogram of the supplement sample. Five replicate determinations were made with results shown in Table 2. The amount of active ingredient GSH in pharmaceutical supplement was found to be 50.8 ± 2.3 mg/tablet (n = 5), which is in good agreement with the claimed value on the label. The weight percentage of the impurity GSSG to GSH was 2.34 ± 0.07% (n = 5), indicating that a small portion of GSH converted to GSSG during production process. The recovery experiments were performed by spiking 10 M GSH and 1 M GSSG to the sample. Recoveries for GSH and GSSG are calculated to be 102 and 105%, respectively, indicating the reliability of the developed CE-serial dual-electrode detection method. C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
5 References [1] Jorgenson, J. W., Lukacs, K. D., Anal. Chem. 1981, 53, 1298–1302. [2] Castro-Puyana, M., Garcia-Canas, V., Simo, C., Cifuentes, A., Electrophoresis 2012, 33, 147–167. [3] Pan, Y. L., Chen, F., Zhang, M. Y., Wang, T. Q., Xu, Z. C., Zhang, W., Chu, Q. C., Ye, J. N., Electrophoresis 2013, 34, 1241–1248. [4] Rabanes, H. R., Guidote, A. M., Quirino, J. P., Electrophoresis 2012, 33, 180–195. [5] Alhusban, A. A., Breadmore, M. C., Guijt, R. M., Electrophoresis 2013, 34, 1465–1482. [6] Tang, D. F., Ma, G., Zhang, L. Y., Chen, G., J. Chromatogr. A 2013, 1316, 127–134. [7] Liu, X. Q., Yu, Y. Y., Gu, H., Zhou, T. S., Wang, L. L., Mei, B., Shi, G. Y., Electrophoresis 2013, 34, 935–943. [8] Inoue, J., Kaneta, T., Imasaka, T., Electrophoresis 2012, 33, 2743–2747.
www.electrophoresis-journal.com
Electrophoresis 2014, 35, 3556–3563
[9] Xu, L., Guo, Q. H., Yu, H., Huang, J. S., You, T. Y., Talanta 2012, 97, 462–467. [10] Wallingford, R. A., Ewing, A. G., Anal. Chem. 1987, 59, 1762–1766. [11] Park, S., McGrath, M. J., Smyth, M. R., Diamond, D., Lunte, C. E., Anal. Chem. 1997, 69, 2994–3001. [12] Wang, C. W., Chen, I. C., Anal. Chem. 1992, 64, 2461–2464.
CE and CEC
3563
[24] Zhong, M., Lunte, S. M., Anal. Chem. 1999, 71, 251–255. [25] Holland, L. A., Lunte, S. M., Anal. Chem. 1999, 71, 407–412. [26] Rose, D. J., Jorgenson, J. W., J. Chromatogr. A 1988, 447, 117–131. [27] Voegel, P. D., Zhou, W., Baldwin, R. P., Anal. Chem. 1997, 69, 951–957.
[13] Kok, W. T., Sahin, Y., Anal. Chem. 1993, 65, 2497–2501.
[28] Liu, J., Cao, W., Qiu, H., Sun, X., Yang, X., Wang, E., Clin. Chem. 2002, 48, 1049–1058.
[14] Hu, S., Wang, Z. L., Li, P. B., Cheng, J. K., Anal. Chem. 1997, 69, 264–267.
[29] Liu, Y. M., Li, J., Yang, Y., Du, J., Mei, L., Can. J. Chem. 2012, 90, 180–185.
[15] Huang, X., Zare, R. N., Sloss, S., Ewing, A. G., Anal. Chem. 1991, 63, 189–192.
[30] Zhou, H., Holland, L. A., Liu, P., Analytst 2001, 126, 1252–1256.
[16] Sloss, S., Ewing, A. G., Anal. Chem. 1993, 65, 577–581.
[31] Hoque, M. E., Arnett, S. D., Lunte, C. E., J. Chromatogr. B 2005, 827, 51–57.
[17] Wang, J., Tian, B. M., Chatrathi, M. P., Escarpa, A., Pumera, M., Electrophoresis 2009, 30, 3334–3338. [18] Garcia, M., Alonso-Fernandez, J. R., Escarpa, A., Anal. Chem. 2013, 85, 9116–9125. [19] Chua, C. K., Pumera, M., Electroanalysis 2013, 25, 945–950. [20] Du, F. Y., Fung, Y. S., Electrophoresis 2010, 31, 2192–2199. [21] Chen, D. C., Chang, S. S., Chen, C. H., Anal. Chem. 1999, 71, 3200–3205.
[32] Carru, C., Zinellu, A., Pes, G. M., Marongiu, G., Tadolini, B., Deiana, L., Electrophoresis 2002, 23, 1716–1721. [33] He, T., Quinn, D., Fu, E., Wang, Y. K., J. Chromatogr. B 1999, 727, 43–52. [34] Zhang, J., Hu, Z., Chen, X., Talanta 2005, 65, 986–990. [35] Lavigne, V., Pons, A., Dubourdieu, D., J. Chromatogr. A 2007, 1139, 130–135.
[22] Dorris, M. K., Crick, E. W., Lunte, C. E., Eletrophoresis 2012, 33, 2725–2732.
[36] Qin, J., Ye, N., Yu, L., Liu, D., Fung, Y., Wang, W., Ma, X., Lin, B., Electrophoresis 2005, 26, 1155–1162.
´ L. A., Zare, R. N., J. Chromatogr. A 1994, [23] Lin, B. L., Colon, 680, 263–270.
[37] Safavi, A., Maleki, N., Farjami, E., Mahyari, F. A., Anal. Chem. 2009, 81, 7538–7543.
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.electrophoresis-journal.com
