Journal of Chromatography A, 1358 (2014) 277–284

Contents lists available at ScienceDirect

Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma

Highly sensitive analysis of flavonoids by zwitterionic microemulsion electrokinetic chromatography coupled with light-emitting diode-induced fluorescence detection Wan Cao a,1 , Shuai-Shuai Hu a,1 , Xing-Ying Li a , Xiao-Qing Pang a , Jun Cao a,∗ , Li-Hong Ye b , Han-Bin Dai a , Xiao-Juan Liu a , Jian-Hua Da a , Chu Chu c a

College of Material Chemistry and Chemical Engineering, Hangzhou Normal University, Hangzhou 310036, China Integrated Chinese and Western Medicine Hospital of Zhejiang Province, Hangzhou 310003, China c Zhejiang Univ Technol, Coll Pharmaceut Sci, Hangzhou 310014, China b

a r t i c l e

i n f o

Article history: Received 10 April 2014 Received in revised form 18 June 2014 Accepted 24 June 2014 Available online 1 July 2014 Keywords: Flavonoids Fluorescence detector Light-emitting diode Microemulsion electrokinetic chromatography Zwitterionic microemulsion Hawthorn

a b s t r a c t A rapid zwitterionic microemulsion electrokinetic chromatography (ZI-MEEKC) approach coupled with light-emitting-diode-induced fluorescence (LED-IF, 480 nm) detection was proposed for the analysis of flavonoids. In the optimization process, we systematically investigated the separation conditions, including the surfactants, cosurfactants, pH, buffers and fluorescence parameters. It was found that the baseline separation of the seven flavonoids was obtained in less than 5 min with a running buffer consisting of 92.9% (v/v) 5 mM sodium borate, 0.6% (w/v) ZI surfactant, 0.5% (w/v) ethyl acetate and 6.0% (w/v) 1butanol. High sensitivity was obtained by the application of LED-IF detection. The limits of detection for seven flavonoids were in the range of 3.30 × 10−8 to 2.15 × 10−6 mol L−1 without derivatization. Ultimately, the detection method was successfully applied to the analysis of flavonoids in hawthorn plant and food products with satisfactory results. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Capillary electrophoresis (CE) is a powerful analytical approach with significant importance in the food, forensic, environmental and pharmaceutical sciences due to its unique advantages, such as its instrumentation simplicity, high separation efficiency, minimum operation cost and compatibility [1,2]. CE separation depends on the differing migrations of solutes in an electrical field, and electrophoresis is carried out in capillaries filled with a background electrolyte (BGE) [3]. Recently, the coupling of CE with light-emitting-diode-induced fluorescence (LED-IF), a spectroscopic approach used for the determination of molecular structures, flow visualization and the detection of selective species, has received attention. LED-IF possesses the advantages of being small in size, providing a stable output, being lowcost, having a long lifetime and consuming low amounts of energy [4].

∗ Corresponding author. Tel.: +86 571 2886 7909; fax: +86 571 2886 7909. E-mail address: [email protected] (J. Cao). 1 These authors contributed equally to this work. http://dx.doi.org/10.1016/j.chroma.2014.06.081 0021-9673/© 2014 Elsevier B.V. All rights reserved.

Over the past several years, there has been a rapid increase in interest in the study of microemulsions due to their unique physicochemical properties, such as low interfacial tension, thermodynamic stability, optical transparency and strong solubilizing ability [5]. Microemulsion droplets, either water-in-oil or the more common oil-in-water, are generally comprised of nanoscale droplets of oil, water, surfactant and cosurfactant in a certain proportion and can be observed as the swell of micelles. Based on the difference in surfactants, microemulsions are divided into four forms: anionic, nonionic, cationic and zwitterionic (ZI) microemulsion. ZI microemulsions are induced by a three-phase system in which oil surfactant-poor phases coexist with excess water and a ZI surfactant. The ZI surfactant, whose polar hydrophilic heads carries both a positive and a negative charge, has wide applications in many fields [6]. The presence of both cationic and anionic active groups in the same molecule results in the head group hydrophilicity being an intermediate between the nonionic and ionic classes [7]. In addition, it is compatible with other types of surfactant and mild to skin and eyes. Microemulsion electrokinetic chromatography (MEEKC) is a form of CE that utilizes microemulsion droplets as separation media to detect a wide range of analytes. Microemulsions are more

278

W. Cao et al. / J. Chromatogr. A 1358 (2014) 277–284

flexible and can expand better than micelles, providing MEEKC with a higher resolution and a wider separation window than micellar electrokinetic chromatography [8]. In addition, a series of reports have shown the great potential of MEEKC for the separation of lipophilic and hydrophilic substances, such as polyaromatic hydrocarbons [9,10], steroids [11], fat-soluble vitamins [12,14], water-soluble vitamins [13,14], sugars [15] and proteins [16], as well as pharmaceuticals [12,17] and natural products [18,19]. However, no study has been reported on the use of ZI microemulsion as a pseudostationary phase (PSP) for CE. To date, various detectors have been applied in MEEKC analysis, such as diode-array detection (DAD) [20], mass spectrometry (MS) [21], electrochemical detection (ED) [22] and laser-induced fluorescence (LIF) [23,24]. However, it is well known that the use of DAD in this method leads to low sensitivity. Meanwhile, MS requires a salt-free solution, and ED is easily influenced by the pH of the mobile phase and impurities. Although LIF provides high sensitivity, it is usually characterized by high cost, a limited lifetime and high power consumption. To the best of our knowledge, there have been no reports published on the application of LED-IF in MEEKC analysis to date. Therefore, the investigation of microemulsion-LED in CE is highly interesting, particularly for the analysis of complex samples. Crataegus pinnatifida Bge. var. major N.E.Br., also called hawthorn, has been widely used in medicines and food in China and Europe and is thought to promote blood circulation, improve digestion and resolve blood stasis in both traditional and modern medicine [25]. The main chemical components of hawthorn are flavonoids and phenolic acids [26], and pharmacological studies of flavonoids indicate that they possess a number of useful effects, such as hypolipidemic [27], cardiotonic [28] and antioxidative activities [29]. In this study, a ZI-MEEKC method coupled with LED-IF was developed for the rapid simultaneous separation

of seven flavonoids (kaempferide, apigenin, quercetin, isovitexin, apigenin 8-C-glucoside, isoquercitrin and hyperoside) without derivatization. The separation of the analytes was obtained using the optimized surfactant, cosurfactant, buffer, pH and fluorescence parameters, and the analytical performance of this method was investigated in terms of selectivity, recovery, linearity and precision. Ultimately, the developed detection method was successfully applied to the analysis of flavonoids in hawthorn plant and food products with satisfactory results. 2. Experimental 2.1. Chemicals and reagents N-dodecyl-N,N-dimethyl-3-ammonio-1-propanesulfonate (DAPS), 3-(N,N-dimethylmyristylammonio)propanesulfonate (MAPS), 3-(N,N-dimethylpalmitylammonio)propanesulfonate (PAPS), sodium dodecyl sulfate (SDS) and sodium borate were purchased from Sigma-Aldrich, Inc. (St. Louis, MO, USA). Brij 35 was obtained from ANPEL Scientific Instrument Co., Ltd. (Shanghai, China). All other chemicals used (chromatography reagents), including 1-butanol, methanol, ethyl acetate, 2-propanol, 1pentanol and cyclohexanol, were provided by Tianjin Siyou Fine Chemical Co. Ltd. (Tianjin, China). Standards of kaempferide, apigenin, quercetin, isovitexin, apigenin 8-C-glucoside, isoquercitrin and hyperoside were purchased from Shanghai Winherb Medical Science and Technology Development Co., Ltd. (Shanghai, China). The purity of each standard was determined to be higher than 98% by normalization of the peak areas detected by MEEKC-LED-IF. The structures of these seven analytes are shown in Fig. 1. Hawthorn plant powder was obtained from local pharmaceutical stores (Hangzhou, China), and hawthorn food products (appetizing hawthorn, iron hawthorn, hawthorn maltose, mini hawthorn and

Fig. 1. Chemical structures of the seven compounds investigated.

W. Cao et al. / J. Chromatogr. A 1358 (2014) 277–284

crataegus) were purchased from a local supermarket (Hangzhou, China). 2.2. Apparatus and MEEKC-LED-IF conditions The experiments were carried out using an Agilent CE instrument (7100, Palo Alto, CA, USA) equipped with an LED-IF detector (max = 480 nm, spectral bandwidth = 25 nm; Picometrics Technology Corporation; Labège, France). The LED was used as its excitation source for fluorescence detection, and the excited fluorescence from the microchannel was detected by a photomultiplier (PM HV) tube after crossing a high pass filter (515 nm). This filter stopped the wavelength before 510 nm and transmitted the wavelength higher than 510 nm until 760 nm. The high-voltage supply for the PM HV was in the range of 630–670 V, and the rise time was set from 0.05 to 0.4 s. An uncoated fused-silica capillary (Yongnian Optic Fiber Factory, Hebei, China) with a 75-␮m id and a total length of 65 cm (44-cm effective length) was employed throughout the study. The virgin capillary was conditioned by flushing with 1 M NaOH for 15 min, 0.1 M NaOH for 15 min, water for 8 min, 10 mM sodium borate for 10 min, 0.1 M NaOH for 5 min and then buffer solution for 5 min. Before each run, the capillary column was rinsed with 0.1 M NaOH for 3 min and buffer for 3 min. A renewed buffer was used after every three runs to ensure good repeatability. Samples were injected into the end of the capillary (anode) with an injection pressure of 50 mbar for 5 s. The capillary was thermostated at 35 ◦ C. All solutions were filtered before entering the instrument. 2.3. Preparation of microemulsion buffer The microemulsion buffer was prepared by weighing 0.5% w/v ethyl acetate, 0.3–1.2% w/v ZI surfactant (DAPS, MAPS, PAPS), 5.5–7.0% w/v cosurfactant (1-butanol, 2-propanol, 1-pentanol, cyclohexanol) and 2.5–10 mM salt solution (sodium borate, disodium hydrogen phosphate, sodium acetate, tris). This mixture was sonicated (100 W, 40 kHz) to form a clear and highly stable microemulsion at room temperature. The buffer pH was then adjusted to the desired value with 1.0 M H3 BO3 or 1.0 M NaOH. The microemulsion buffers were filtered through 0.22-␮m membrane filters before use. 2.4. Preparation of sample solutions A 2.0-g sample of hawthorn plant powder was precisely weighed and subjected to ultrasonication at 100 W (40 kHz) for 60 min after the addition of 20 mL of methanol. The sample solution was then adjusted to the initial volume. After filtering, 1 mL of the filtrate was transferred to a 1.5 mL Eppendorf tube and evaporated to dryness in 80 ◦ C digital dry baths. The residue was then dissolved in 300 ␮L of water and the solution was sonicated for 1 min. The resulting mixture was extracted twice with 600 ␮L of petroleum ether, discarded petroleum ether solution, and the aqueous solution was extracted twice with 300 ␮L of ethyl acetate, then the combined extracts were dried to a residue. The residue was re-dissolved with 200 ␮L of methanol; this was the sample solution. Hawthorn food products (appetizing hawthorn, iron hawthorn, hawthorn maltose, mini hawthorn and crataegus) for analysis were prepared by weighing the appropriate amount of each sample (approximately 2.0 g), grounding it into fine particles and then dissolving the particles in methanol. These food samples were extracted according to the above method and dissolved with 200 ␮L of methanol to get a final volume. All solutions were filtered through 0.22-␮m membrane filters prior to MEEKC-LED-IF analysis.

279

2.5. Calculations The following equation was used to compute the number of theoretical plates:

 N = 5.54

tR Wh/2

2

where tR is the migration time and Wh/2 is the peak width at half the peak height. The resolution was calculated using the following equation:



R=

2 tR(2) − tR(1)



W1 + W2

where tR is the migration time and W is the peak width. The electroosmotic mobility (EOF ) was calculated by: EOF =

Le Lt tm V

where Le is the length of the capillary to the detector, Lt is the total length of the capillary, tm is the migration time of EOF, and V is the applied voltage. 2.6. Spectroscopic analysis Flavonoids were dissolved in the ZI microemulsion and then sonicated for 10 min, the spectra were recorded by using Lambda 750 UV/VIS/NIR spectrometer (PERKINELMER, Waltham, MA, USA) in the range of the wavelength 250–500 nm. ZI microemulsion was used as blank. The fluorescence emission spectra of flavonoids were recorded with LS55 fluorescence spectrometer (PERKINELMER, Waltham, MA, USA) in the range of the wavelength 400–750 nm with the excitation wavelength 385–399 nm, which were got from UV spectra. The PM HV type was R928 and the voltage was automatic. Then the scan speed was set at 400 nm min−1 and the excitation and emission slits were 10 nm. 3. Results and discussion 3.1. Optimization of surfactant conditions The surfactant is a very important parameter influencing the physicochemical characteristics of a microemulsion, such as its solubility, density and osmotic pressure [30,31]. Additionally, it has a significant impact on the separation efficiency of target analytes in LED-IF detection [4]. Therefore, three ZI surfactants, including DAPS, MAPS and PAPS, were investigated by comparing their analytical performance under the same conditions. In the experiment, it was observed that the increase of the hydrophobic chain length (12 to 16) in ZI surfactants resulted in strengthening of the hydrophobic interactions between the surfactant tails and aqueous phase, which extended the time required to reach a dynamic equilibrium of oil and water. Furthermore, it also implied that the tested ZI surfactants having the same quaternary ammonium cation and sulfonate anion but a different number of methylene units could cause the difference in their microemulsion-forming abilities. Previous studies showed that ZI surfactants could adsorb onto the bare capillary wall in a hemimicellar or bilayer model, which tended to reverse the direction of EOF. A negative voltage was often applied [30]. However, in this work, a positive polarity voltage of 30 kV was chosen instead of the traditional negative voltage. A possible explanation of the difference may be that silanols (with a free counterion) were still present and could not be completely dynamically modified by the adsorbed surfactant; therefore, it was not necessary to reverse the electrode polarity in the microemulsion system. As

280

W. Cao et al. / J. Chromatogr. A 1358 (2014) 277–284

Fig. 2. Schematic of the ZI-MEEKC separation process.

shown in Fig. 2, the microemulsion droplets attempted to migrate towards the anode, but the stronger force of the EOF pushed them towards the cathode end of the capillary and through the detector in MEEKC. One hypothesis to explain this phenomenon may be that the adsorption of DAPS onto the surface may expose the outermost anionic sulfonate group to form a large zeta potential. Another hypothesis was that anions from the BGE adsorbed to the surfactant-adsorbed surface causing a large zeta potential [32]. As presented in Fig. 3, the value of EOF decreased markedly from 6.62 × 10−4 to 6.45 × 10−4 cm2 V−1 s−1 while the migration times of the target compounds increased as the alkyl chain length of the surfactant increased. Moreover, although the separation for peaks 3 and 4 was increasing by MAPS, a decreased separation of peaks 4 and 5 was also obtained in Fig. 3. Based on the above observation, DAPS was chosen as the optimum surfactant species in this study. The surfactant concentration also has a significant influence on the analysis selectivity [30]. In this study, a series of DAPS surfactant concentrations from 0.3% to 1.2% (w/v) were investigated. The resolution of peaks 3 and 4 increased substantially with increasing DAPS concentration from 0.3% to 0.6% (w/v), but at higher

Fig. 3. Impact of surfactant type on flavonoid resolution and selectivity. MEEKC conditions: 0.6% w/v surfactants, 6.0% w/v 1-butanol, 0.5% w/v ethyl acetate and 92.9% v/v 5 mM sodium borate buffer (pH 9.0); capillary length, 65 cm total (44 cm effective length) × 75 ␮m id; temperature, 35 ◦ C; voltage, 30 kV; injection, 50 mbar, 5 s; detection, 480 nm. Surfactants: (A) 0.6% w/v DAPS, (B) 0.6% w/v MAPS, (C) 0.6% w/v PAPS. Analytes: (1) kaempferide, (2) apigenin, (3) quercetin, (4) isovitexin, (5) apigenin 8-C-glucoside, (6) isoquercitrin, (7) hyperoside.

concentration (>0.6% w/v), the resolution of the two analytes decreased slightly (data not shown). This signified that at very high concentration of the pseudostationary phase, undifferentiated partitioning of the analytes can occur. In addition, increasing the concentration of DAPS in the microemulsion system also increased the migration times of analytes due to the increase of the ionic strength of the buffer. Thus, 0.6% (w/v) DAPS was selected for further study. 3.2. Selection of cosurfactant type and concentration The cosurfactant, which is the more hydrophilic organic solvent added to the microemulsion, is an important component controlling the selectivity of the MEEKC system and plays a significant role in affecting the partition and migration of the flavonoids. Based on this factor, the influence of four alcohols (isopropanol, 1-butanol, 1-pentanol and cyclohexanol) on separation efficiency was investigated to gain further insight into the effect of the cosurfactant choice. As shown in Fig. 4, analytes 4–7 were not baseline separated when using isopropanol, and analytes 3–5 overlapped visibly when 1-pentanol was added to the microemulsion. Similarly, microemulsions based on cyclohexanol as cosurfactant also could not produce

Fig. 4. Effect of cosurfactant type on the separation of seven analytes using MEEKC. Microemulsion buffers: 0.6% w/v DAPS, 6.0% w/v cosurfactants, 0.5% w/v ethyl acetate and 92.9% v/v 5 mM sodium borate buffer (pH 9.0). Cosurfactant: (A) 6.0% w/v isopropanol, (B) 6.0% w/v 1-butanol, (C) 6.0% w/v 1-pentanol, (D) 6.0% w/v cyclohexanol. The other conditions were the same as in Fig. 3.

W. Cao et al. / J. Chromatogr. A 1358 (2014) 277–284

Fig. 5. Influence of the type of buffer on the separation of seven analytes using MEEKC. Microemulsion buffers: 0.6% w/v DAPS, 6.0% w/v 1-butanol, 0.5% w/v ethyl acetate, 92.9% v/v 5 mM buffers. Buffers: (A) 5 mM sodium borate, (B) 5 mM tris–HCl, (C) 5 mM disodium hydrogen phosphate, (D) 5 mM acetate. The other conditions were the same as in Fig. 3.

a good separation. However, it was obvious that 1-butanol could provide a good separation. Additionally, it was observed that only under heating conditions can 1-pentanol and cyclohexanol be used to form a stable microemulsion. This may be due to that for isopropanol, the log P value (n-octanol/water partition coefficient) was lower (0.25), so it remained mainly in the water with a low tendency to penetrate into the microemulsion droplets. And 1pentanol and cyclohexanol (log P = 1.25 and 2.67), with relatively strong hydrophobicity, could not bridge the oil–water interface efficiently. In addition, in case of 1-butanol, having a log P value of 0.81, it was reported that solvent could penetrate into the microemulsion droplets successfully [33]. Therefore, 1-butanol was chosen as the optimal cosurfactant. To further test how the cosurfactant affected the electrophoretic mobilities, 1-butanol concentrations of 5.5 to 7.0% w/v were investigated. The results suggested that an increase in 1-butanol concentration led to longer migration times and lower theoretical plate numbers, such as the values of analytes 6 and 7 decreased from 94,591 and 90,461 to 71,144 and 68,130. Additionally, the optimum resolution was observed to occur at 6.0% 1-butanol in the solvent, the possible reasons may be that due to the relatively high log P value and the high concentration of 1-butanol (6.0%), the alcohol was present not only in the Stern layer but also inside the microemulsion droplets, acting as a true class I modifier [33,34]. In addition, an increased mass transfer between the microemulsion droplets and the external aqueous phase was mediated by the increase of co-surfactant solvent [34]. However, when the concentration of 1-butanol was beyond 6.5%, longer migration times and lower resolutions (R = 3.38 and 2.47) of analytes 3 and 4 were observed. This phenomenon may be attributed to the increase in the size of the oil droplet as the cosurfactant concentration increased, affecting the charge density on the droplet. Considering these findings, it seemed reasonable to select 6.0% of 1-butanol as the best concentration.

281

analytes. Additionally, all studied analytes co-eluted into a single peak when disodium hydrogen phosphate (pH = 6.0) and acetate buffers (pH = 5.0) were used in MEEKC. A possible reason was that the synergistic influence of attractive electrostatic and hydrophobic interactions led to quite large migration factors of all compounds under acidic condition, resulting in their co-elution. Furthermore, it was clear that the borate solution provided the highest resolution of the target analytes among the studied buffers. A possible explanation of this finding is that the borate ions interacted strongly with the flavonoids tested with vicinal diols on glucosides or adjacent hydroxyl groups to form negatively charged complexes and thus changes in the electrophoretic mobility led to separation with high efficiency [35,36]. As a result, borate was used in the following experiments. The effect of the concentrations of borate on the separation of seven compounds was investigated in the range of 2.5 to 10 mM. The fluorescent response of the analytes improved markedly with increasing borate concentration due to the increase of the ionic strength. As observed, the value of EOF decreased from 6.00 × 10−4 to 5.85 × 10−4 cm2 V−1 s−1 due to the decreased zeta potential. Additionally, a striking improvement in the peak resolution of analytes 3 and 4 was observed when the borate concentration increased from 2.5 to 5 mM (data not shown). However, the resolution of two compounds decreased obviously as the concentration increased from 5 to 10 mM, which may indicate that their partitioning to the microemulsion droplets was reduced, especially in the presence of higher Joule heating in the capillary. It should be noted that an increase in the borate concentrations in the microemulsion buffers resulted in a large increase in the operating current (from 10 to 41 ␮A), which was deleterious to MEEKC experiments. Therefore, 5 mM borate was selected as the appropriate condition. 3.4. Optimization of pH It is known that the pH value can influence the degree of solute ionization, thus affecting the affinity between the analytes and microemulsions. In this study, buffer pH was evaluated over the range of 8.0 to 10.0 using 5 mM borate solution supplemented with 1 M boric acid and 1 M sodium hydroxide. As presented in Fig. 6, the influence of pH on the microemulsion was, interestingly, different from that of the buffer concentrations: It was well known that all studied flavonoids were weak acidic (pKa = 6.37–8.20) to be partially deprotonated in the studied pH range. The trend observed in Fig. 6 indicated that the migration time slightly decreased for compounds 1 and 3 in the pH range of 8.0–9.0, and delayed at pH 9.5, the exceptional phenomenon was possibly related to the ionization process. The molecules of 1 and 3 (pKa1 = 8.2 and 8.45) were to be partially deprotonated at pH 8.5 in ionization supporting

3.3. Effect of buffer type and concentration MEEKC separations are usually performed using low-ionicstrength buffers, such as 5–10 mM borate or phosphate, as they can generate a sufficient EOF and relatively low current [9]. In this work, four different buffer systems (borate, tris, disodium hydrogen phosphate and acetate) in the microemulsions were investigated. As depicted in Fig. 5, the tris system (pH = 8.0) provided good peak-to-peak resolutions for analytes 1–5, but analytes 6 and 7 were not baseline separated. This might be ascribed to the dissociation of most analytes and the complex hydrogen bonding and electrostatic interactions between microemulsion droplets and

Fig. 6. Influence of pH on the separation of seven analytes using MEEKC. Microemulsion buffers: 0.6% w/v DAPS, 6.0% w/v 1-butanol, 0.5% w/v ethyl acetate, 92.9% v/v 5 mM sodium borate. pH: (A) 8.0, (B) 8.5, (C) 9.0, (D) 9.5, (E) 10.0. The other conditions were the same as in Fig. 3.

282

W. Cao et al. / J. Chromatogr. A 1358 (2014) 277–284

solvents like water and methanol [37], and further deprotonated at 9.5 (pKa2 = 9.5 and 9.3) [37], which resulted in the differences in the migration time. For analytes 6 and 7, the analysis time became gradually short as pH raised from 8.0 to 9.5. The possible reason was that the resistance of compounds 6 and 7 migrating to anode increased with the increases of the degree of solute ionization. But analytes 2, 4 and 5, lacking the 3-OH group in the structures, were insensitive to the change of pH values, thus the analysis times of these solutes were kept almost unchanged in this range. At pH 10.0, the larger volume of modifier (NaOH) added into buffer resulted in peaks of all analytes considerably broader, and peaks 5 and 6 overlapped visibly. In addition, the fluorescence intensity of all compounds became gradually lower when the pH increased from 8.0 to 10.0. This may be due to that the dissociation degree of the solutes increased with the increase of pH values, and more molecules can be incorporated into the ZI microemulsion, which led to a very large bathochromic shift [38]. Considering that pH 9.0 gave the optimum separation efficiency and the minimum migration time, pH 9.0 was considered to be the best value for the running buffer.

Fig. 7. Fluorescent emission spectrum of analytes in the presence of ZI microemulsion.

3.5. Fluorescence parameter

C4 O···HOC3 may permit the excited-state proton transfer (ESPT) process [39]. As shown in Fig. 7, analytes 1 and 3 (em = 553 and 556 nm) at the same concentration had stronger fluorescence intensities than analytes 2, 4 and 5 (em = 490, 512 and 502 nm) with the absence of OH in the C3 position, this was mainly due to the solubilization of both C4 O and C3 OH in the ZI microemulsion which could enhanced the ESPT process [39]. Accordingly, it was also obtained that analytes 6 and 7 with the glycosides in C3 position caused a decrease in fluorescence intensity. A possible reason may be that the presence of glycosides in the molecules may interfere with the formation of hydrogen bonds between C4 O and HOC5 and disturb the ␲-electron conjugated system of analytes 6 and 7. Therefore, it was worth noting that the interaction between analytes and ZI microemulsion and the structures of flavonoids had exerted a remarkable effect on the fluorescence intensity.

The rise time function is generally used to filter the signal, and its value has a significant influence on the nature of the observed electropherogram. It could be seen that an increase in rise time led to a decrease in fluorescence intensity, but when a low rise time (0.05 s) was selected, less smoothing took place (data not shown). Based on the results, a 0.1 s rise time was adopted in this study. The PM HV supply field is used to set the high voltage on the photomultiplier electrodes, and the influence of the applied voltage in the range of 610 to 670 V on the peak intensity was investigated using the standard microemulsion conditions described above at pH 9.0. The findings indicated that higher applied voltage could increase both the fluorescence intensity and the background noise (data not shown). Considering that an excessively high PM HV value could damage the photomultiplier, a voltage of 650 V, which is allowed by the instrument, was finally used.

3.7. Validation of the method

3.6. Structure-fluorescence relationships

The validation of method, including the linearity, limit of detection (LOD), precision and recovery, was performed using the optimized experimental conditions (0.6% DAPS, 0.5% ethyl acetate, 6.0% 1-butanol and 92.9% 5 mM sodium borate, pH 9.0); 30 kV as the applied voltage, 35 ◦ C as the capillary temperature, a 0.1 s rise time and a 650 V PM supply). The linearities of the peak areas and analyte concentrations were constructed in Table 1 using a series of standard solutions over the range of 1.67–133.60 to 107.67–10767.00 (×10−7 mol L−1 ). The LODs, corresponding to a signal-to-noise ratio (S/N) of 3, were evaluated for the seven flavonoids by progressive dilution and calculated in the range of 0.33 to 21.67 (×10−7 mol L−1 ) and the sensitivity of LED-IF was about 8 to 40 times than the UV for most analytes. Moreover, the limits of quantification (LOQs), corresponding to an S/N of 10,

The interaction of flavonoids with ZI microemulsion was further investigated by the fluorescence method. The emission spectra of the seven analytes at 480–700 nm (Fig. 7) when excited at 385–399 nm (data not shown) can be attributed to the proton transfer tautomer fluorescence (PTRF) band [39]. When analytes were solubilized in the ZI microemulsion, their relative fluorescence intensity was increased markedly with a large red shift compared to the solutes in methanol (data not shown). The enhancement of the PTRF of flavonoids by ZI microemulsion can be ascribed to the interference of the intramolecular hydrogen bonds in the flavonoid molecules. It was known that the H-bond between C4 O···HOC5 can improve the non-radiative deactivation while that between

Table 1 Linear regression data, precision, repeatability, limits of detection (LODs) and limits of quantitation (LOQs) of the investigated compounds. Analytes

1 2 3 4 5 6 7

Calibration curve

y = 919299x − 0.0256 y = 64893x − 0.0450 y = 625260x + 0.0405 y = 344883x − 0.1078 y = 321936x − 0.2059 y = 7085.6x + 0.0095 y = 8413.8x + 0.0029

Linear range (×10−7 mol L−1 )

1.67–133.60 9.25–740.00 4.97–397.60 5.78–462.40 5.78–462.40 107.67–10767.00 107.67–8613.60

Precision (RSD %)

Repeatability

LODs (×10−7 mol L−1 )

LOQs (×10−7 mol L−1 )

Intra-day (n = 6)

Inter-day (n = 6)

(RSD%) n = 5

Time

Area

Time

Area

Time

Area

UV

LED-IF

1.27 1.33 1.31 1.46 1.46 1.33 1.39

1.04 2.52 1.24 2.97 2.75 2.11 2.87

3.24 3.33 3.68 3.49 3.59 3.37 3.51

4.14 3.33 3.46 3.84 4.11 3.38 5.04

0.89 1.18 0.75 1.23 1.05 1.56 1.33

3.01 2.53 3.18 2.56 2.94 2.34 3.11

13.32 14.80 19.87 18.50 13.88 8.61 8.82

0.33 1.85 0.99 1.16 1.03 21.53 21.67

1.67 5.25 4.97 5.78 5.26 67.67 67.34

W. Cao et al. / J. Chromatogr. A 1358 (2014) 277–284

283

Fig. 8. Electropherogram of seven analytes by three MEEKCs. MEEKC conditions: 0.6% w/v surfactants, 6.0% w/v 1-butanol, 0.5% w/v ethyl acetate and 92.9% v/v 5 mM sodium borate buffer (pH 9.0). Microemulsions: (A) DAPS microemulsion, (B) SDS microemulsion, (C) Brij 35 microemulsion. The other conditions were the same as in Fig. 3.

were also estimated from 1.67 to 67.67 (×10−7 mol L−1 ). As presented in Table 1, the repeatability, which was conducted in parallel five times to assess the extraction of real samples, was obtained as the relative standard deviation (RSD). The RSDs of peak areas and migration times were based on six replicated injections for intra-day. And the values of analytes for migration times were 1.27–1.46% (intra-day, n = 6) and 0.75–1.56% (repeatability, n = 5). For the peak areas, the RSD values were 1.04–2.97% (intra-day, n = 6) and 2.34–3.18% (repeatability, n = 5). Inter-day precision of analysis time and peak areas across three days of analysis was lower than 3.68% and 5.04%, Therefore, the MEEKC-LEDIF method indeed provided better separation efficiency and selectivity for the determination of the studied analytes.

Fig. 9. Typical electropherograms of samples under the optimum conditions. (A) Hawthorn, (B) iron hawthorn. Separation conditions were the same as in Fig. 3.

separated from the mixture, except for analytes 2 and 4. Therefore, the above observations confirmed that the DPAS microemulsion system was much more effective in separating flavonoids than the SDS and Brij 35 microemulsion system, probably reflecting the more complex physicochemical properties of ZI microemulsion.

3.8. Comparison among the ZI, anionic and nonionic MEEKC separations of flavonoids

3.9. Sample analysis

Under optimum conditions, ZI, anionic and nonionic MEEKC were compared in order to evaluate the separation selectivity of flavonoids. The experiment was performed under different surfactants employed in the microemulsion system with the other parameters were kept fixed. Fig. 8 showed the electropherogram of seven flavonoids by different microemulsions. First, it was found that separation times of all target analytes were extended by using anionic and nonionic MEEKC when compared with ZI MEEKC. Furthermore, the migration order of anionic and nonionic MEEKC differed from that of ZI MEEKC. For SDS-MEEKC system, analytes 4–7 with glucosides moved forward compared with ZI microemulsion, and a significant decrease of all compounds in separation selectivity was clearly observed in Fig. 8B, especially for peaks 5, 6 and 7. It was also worthy to note that the running current for SDS-MEEKC was twice as high as that of DPAS-MEEKC, so it led to longer analysis times as well as poorer resolution. When the nonionic Brij 35-MEEKC system was employed to test the separation performance, we observed that most compounds could be

The developed MEEKC method was used for the determination of seven flavonoids in hawthorn plant extract and food products (appetizing hawthorn, iron hawthorn, hawthorn maltose, mini hawthorn and crataegus) under the optimized conditions. In this work, the methanol extracts were filtered and injected directly onto the capillary column. Additionally, we used the fluorescence spectra and migration times of standards and the recoveries of samples to further identify substances. A typical electropherogram of the extractions was shown in Fig. 9. It can be seen that the total flavonoid contents varied markedly among the six species, with that of hawthorn maltose and crataegus being less than those of the other four species (Table 2). Most interesting was the results for analyte 6 and 7, which were the highest in concentration and accounted for more than 70% of the flavonoids. Additionally, the recoveries were calculated by adding an appropriate amount of standards to the samples in which the contents of seven flavonoids had been investigated. The recoveries are listed in Table 2 and were within 85–103%. As observed, most target components were

Table 2 Quantitative analytical results of hawthorn samples in MEEKC. Analyte

1 2 3 4 5 6 7 a b

Content (×10−7 mol L−1 ) Hawthorn

Appetizing hawthorn

Iron hawthorn

Hawthorn maltose

Mini hawthorn

Cratogus

2.35 23.89 tra 12.98 16.02 184.18 162.95

tr 8.63 tr 6.61 7.86 77.16 99.95

ndb 12.48 6.07 6.90 nd 82.80 102.33

1.69 nd tr tr 6.09 68.21 88.95

2.05 10.36 tr 5.82 nd 75.32 100.59

1.79 nd tr tr 5.31 76.48 67.69

Trace. Not detected.

Recovery (%) 85 90 93 98 101 97 103

284

W. Cao et al. / J. Chromatogr. A 1358 (2014) 277–284

successfully tested in a short time (5.0 min), showing no strong matrix interference. 4. Conclusions In this study, a MEEKC-LED-IF method using a ZI microemulsion with a ZI surfactant as the pseudo-stationary phase was developed for the analysis of flavonoids. The important factors known to influence separation efficiency, such as surfactant, cosurfactant and pH, were systematically investigated. The results showed the unique advantage of the modified PSP in terms of resolution and selectivity, and most flavonoids were completely separated within 5 min without derivatization. This proposed method was further applied to the analysis of active compounds in hawthorn plant and food products with satisfactory results. We recommend this proposed method because of the following advantages: (a) microemulsions are more flexible and can expand better than micelles, providing higher resolution and a wider separation window, (b) the use of LED-IF is less expensive than a laser source and (c) good sample separation can be obtained in a short time. Acknowledgments This study was supported by General Program of National Natural Science Foundation of China (No. 81274065), Research on Public Welfare Technology Application Projects of Zhejiang Province (No. 2014C37069), Changjiang Scholars and Innovative Research Team in Chinese University (IRT 1231), Scientific Research Foundation of Hangzhou Normal University (2011QDL33), Young and middle-aged academic leaders of Hangzhou (2013-45), and the new-shoot Talents Program of Zhejiang Province (2013R421044, 2014R421019). References [1] S. Fanali, Editorial on “Stereoselective determination of drugs and metabolites in body fluids, tissues and microsomal preparations by capillary electrophoresis (2000–2010)” by Jitka Caslavska and Wolfgang Thormann, J. Chromatogr. A 1218 (2011) 588–601. [2] Q. Zhang, Y.X. Du, J.Q. Chen, G.F. Xu, T. Yu, X.Y. Hua, J.J. Zhang, Investigation of chondroitin sulfate D and chondroitin sulfate E as novel chiral selectors in capillary electrophoresis, Anal. Bioanal. Chem. 406 (2014) 1557–1566. [3] L. Suntornsuk, Recent advances of capillary electrophoresis in pharmaceutical analysis, Anal. Bioanal. Chem. 398 (2010) 29–52. [4] J. Xu, Y. Xiong, S.H. Chen, Y.F. Guan, Light emitting diode induced fluorescence detector, Prog. Chem. 21 (2009) 1325–1334. [5] A.K. Ganguli, A. Ganguly, S. Vaidya, Microemulsion-based synthesis of nanocrystalline materials, Chem. Soc. Rev. 39 (2010) 474–485. [6] D.W. Tondo, E.C. Leopoldino, B.S. Souza, G.A. Micke, A.C.O. Costa, H.D. Fiedler, C.A. Bunton, F. Nome, Synthesis of a New Zwitterionic surfactant containing an imidazolium ring. Evaluating the chameleon-like behavior of zwitterionic micelles, Langmuir 26 (2010) 15754–15760. [7] Z.Q. Li, L. Zhang, Z.C. Xu, D.D. Liu, X.W. Song, X.L. Cao, L. Zhang, S. Zhao, Effect of zwitterionic surfactants on wetting of quartz surfaces, Colloids Surf., A: Physicochem. Eng. Aspects 430 (2013) 110–116. [8] R. Ryan, S. Donegan, J. Power, E. McEvoy, K. Altria, Recent advances in the methodology, optimisation and application of MEEKC, Electrophoresis 30 (2009) 65–82. [9] K.A. Kahle, J.P. Foley, Two-chiral-component microemulsion electrokinetic chromatography-chiral surfactant and chiral oil: Part 1. Dibutyl tartrate, Electrophoresis 28 (2007) 1723–1734. [10] K.A. Kahle, J.P. Foley, Chiral microemulsion electrokinetic chromatography: effect of cosurfactant identity on enantioselectivity, methylene selectivity, resolution, and other chromatographic figures of merit, Electrophoresis 27 (2006) 4321–4333. [11] X.J. Ni, M.J. Yu, Y.H. Cao, Microstructure of microemulsion modified with ionic liquids in microemulsion electrokinetic chromatography and analysis of seven corticosteroids, Electrophoresis 34 (2013) 2568–2576. ˛ J. Paradziej-Łukowicz, M. Taciak, B. [12] I. Oledzka, P. Kowalski, A. Baluch, T. Baczek, Pastuszewska, Quantification of the level of fat-soluble vitamins in feed based on the novel microemulsion electrokinetic chromatography (MEEKC) method, J. Sci. Food Agric. 94 (2014) 544–551.

[13] M.S. Aurora-Prado, C.A. Silva, M.F.M. Tavares, K.D. Altria, Rapid determination of water-soluble and fat-soluble vitamins in commercial formulations by MEEKC, Chromatographia 72 (2010) 687–694. [14] E.P. Svidritskii, E.B. Pashkova, A.V. Pirogov, O.A. Shpigun, Simultaneous determination of fat- and water-soluble vitamins by microemulsion electrokinetic chromatography, J. Anal. Chem. 65 (2010) 287–292. [15] I. Miksik, J. Gabriel, Z. Deyl, Microemulsion electrokinetic chromatography of diphenylhydrazones of dicarbonyl sugars, J. Chromatogr. A 772 (1997) 297–303. [16] A.K. Bytzek, C.G. Hartinger, Capillary electrophoretic methods in the development of metal-based therapeutics and diagnostics: new methodology and applications, Electrophoresis 33 (2012) 622–634. [17] R. Ryan, K. Altria, E. McEvoy, S. Donegan, J. Power, A review of developments in the methodology and application of microemulsion electrokinetic chromatography, Electrophoresis 34 (2013) 159–177. [18] J. Cao, H.B. Qu, Y.Y. Cheng, Separation of flavonoids and phenolic acids in complex natural products by microemulsion electrokinetic chromatography using surfactant-coated and carboxylic single-wall carbon nanotubes as additives, Electrophoresis 31 (2010) 1689–1696. [19] M. Unger, Capillary electrophoresis of natural products: current applications and recent advances, Planta Med. 75 (2009) 735–745. [20] E. Liotta, R. Gottardo, C. Seri, C. Rimondo, I. Miksik, G. Serpelloni, F. Tagliaro, Rapid analysis of caffeine in “smart drugs” and “energy drinks” by microemulsion electrokinetic chromatography (MEEKC), Forensic Sci. Int. 220 (2012) 279–283. [21] A.K. Bytzek, M.R. Reithofer, M. Galanski, M. Groessl, B.K. Keppler, C.G. Hartinger, The first example of MEEKC-ICP-MS coupling and its application for the analysis of anticancer platinum complexes, Electrophoresis 31 (2010) 1144–1150. [22] L.S. Yu, X.Q. Xu, L. Huang, J.M. Ling, G.N. Chen, Separation and determination of flavonoids using microemulsion EKC with electrochemical detection, Electrophoresis 29 (2008) 726–733. [23] C.P. Satori, E.A. Arriaga, Describing autophagy via analysis of individual organelles by capillary electrophoresis with laser induced fluorescence detection, Anal. Chem. 85 (2013) 11391–11400. [24] C.P. Satori, V. Kostal, E.A. Arriaga, Individual organelle pH determinations of magnetically enriched endocytic organelles via laser-induced fluorescence detection, Anal. Chem. 83 (2011) 7331–7339. [25] C.Y. Chu, M.J. Lee, C.L. Liao, W.L. Lin, Y.F. Yin, T.H. Tseng, Inhibitory effect of hot-water extract from dried fruit of Crataegus pinnatifida on low-density lipoprotein (LDL) oxidation in cell and cell-free systems, J. Agric. Food Chem. 51 (2003) 7583–7588. [26] Q. Chang, Z. Zuo, F. Harrison, M.S.S. Chow, Hawthorn, J. Clin. Pharmacol. 42 (2002) 605–612. [27] T. Jurikova, J. Sochor, Polyphenolic profile and biological activity of Chinese hawthorn (Crataegus pinnatifida BUNGE) fruits, Molecules 17 (2012) 14490–14509. [28] B. Yang, P. Liu, Composition and health effects of phenolic compounds in hawthorn (Crataegus spp.) of different origins, J. Sci. Food Agric. 92 (2012) 1578–1590. [29] A. Sokol-Letowska, J. Oszmianski, A. Wojdylo, Antioxidant activity of the phenolic compounds of hawthorn, pine and skullcap, Food Chem. 103 (2007) 853–859. [30] A.H. Rageh, U. Pyell, Imidazolium-based ionic liquid-type surfactant as pseudostationary phase in micellar electrokinetic chromatography of highly hydrophilic urinary nucleosides, J. Chromatogr. A 1316 (2013) 135–146. [31] J. Cao, P. Li, J. Chen, T. Tan, H.B. Dai, Enhanced separation of Compound Xueshuantong capsule using functionalized carbon nanotubes with cationic surfactant solutions in MEEKC, Electrophoresis 34 (2013) 324–330. [32] Q. Guan, S.D. Noblitt, C.S. Henry, Electrophoretic separations in poly(dimethylsiloxane) microchips using mixtures of ionic, nonionic and zwitterionic surfactants, Electrophoresis 33 (2012) 2875–2883. [33] S.H. Hansen, C. Gabel-Jensen, S. Pedersen-Bjergaard, Comparison of microemulsion electrokinetic chromatography and solvent-modified micellar electrokinetic chromatography, J. Sep. Sci. 24 (2001) 643–650. [34] S.H. Hansen, C. Gabel-Jensen, D.T.M. El-Sherbiny, S. Pedersen-Bjergaard, Microemulsion electrokinetic chromatography or solvent-modified micellar electrokinetic chromatography? TrAC, Trends Anal. Chem. 20 (2001) 614–619. [35] J. Cao, B. Li, Y.X. Chang, P. Li, Direct on-line analysis of neutral analytes by dual sweeping via complexation and organic solvent field enhancement in nonionic MEKC, Electrophoresis 30 (2009) 1372–1379. [36] J. Cao, L. Yi, P. Li, Y.X. Chang, On-line concentration of neutral analytes by complexation and acetonitrile sweeping in nonionic microemulsion electrokinetic chromatography with direct ultraviolet detection, J. Chromatogr. A 1216 (2009) 5608–5613. [37] V. Stepanic, K.G. Troˇselj, B. Lucic, Z. Markovic, D. Amic, Bond dissociation free energy as a general parameter for flavonoid radical scavenging activity, Food Chem. 141 (2013) 1562–1570. [38] A. Mezzetti, S. Protti, C. Lapouge, J.P. Cornard, Protic equilibria as the key factor of quercetin emission in solution. Relevance to biochemical and analytical studies, Phys. Chem. Chem. Phys. 13 (2011) 6858–6864. [39] W. Liu, R. Guo, Interaction of flavonoid, quercetin with organized molecular assemblies of nonionic surfactant, Colloids Surf., A: Physicochem. Eng. Aspects 274 (2006) 192–199.