J. Sep. Sci. 2007, 30, 2493 – 2500 J. M. Herrero-Martnez et al. 2493

Jos Manuel Herrero-Martnez Original Paper Fadoua Z. Oumada Mart Ross Elisabeth Bosch Determination of flavonoid aglycones in several Clara Rfols food samples by mixed micellar electrokinetic Departament de Qumica chromatography Analtica, Universitat de Barcelona, Barcelona, Spain The application of mixed micellar electrokinetic chromatography to the separation of ten flavonoid aglycones (catechin, epicatechin, naringenin, morin, fisetin, querce- tin, kaempferol, galangin, apigenin, and chrysin) belonging to four different classes (flavanols, flavanones, flavonols, and flavones), and expected to be prominent in commonly consumed foods, has been developed. A micellar system composed of 25 mM SDS and 25 mM sodium cholate buffered at pH 7.0 provided a simultaneous separation of all compounds in less than 20 min. The procedure could be easily adapted to the determination of some flavonoids from each of these classes in real complex samples (propolis, Ginkgo biloba, etc.). The LODs of these compounds were in the range of 1.2–4 lg/mL, and the peak area and migration time repeatabilities were below 6.0 and 3.1%, respectively.

Keywords: Flavonoids / Food analysis / MEKC / Mixed micelles / Received: March 26, 2007; revised: April 18, 2007; accepted: April 22, 2007 DOI 10.1002/jssc.200700124

1 Introduction ual flavonoid glycosides in foods would be desirable, however, most of the reference compounds are not com- Flavonoids are polyphenolic natural products found pre- mercially available. Additionally, differences in flavo- dominantly in plants, comprising an important part of noid glycosides composition from one plant to another our daily diet [1, 2]. These compounds have structures are present, despite these compounds derive only from a based on 2-phenylbenzopyrone (Fig. 1), and differ in their few aglycones. In order to facilitate their study, the con- pattern of hydroxylation, methylation, and glycosila- version of glycosides into aglycones by acid hydrolysis tion, in the degree of unsaturation and in the type and has been described as a practical method for the quanti- position of sugar links [2]. Flavonoids occur in free state tative determination of flavonoids [8–10]. or as aglycones and glycosides. They have been mainly Analysis of flavonoids has been accomplished usually studied due to their antioxidant and free radical scaveng- by HPLC [11, 12]. Nevertheless, most of the chromato- ing activity [3, 4]. Their increasing interest lies in their graphic methods reported in the literature for a simulta- broad pharmacological activities (e.g., antimicrobial, neous measure of the different classes of prominent food spasmolytic, antiallergic, anti-inflammatory, antiviral, flavonoids as their aglycones [13–17] require long anal- anticarcinogenic) [5–7]. Given their wide range of biolog- ysis times, even some works [9, 10, 14] have required two ical functions, there is a growing concern in isolating different chromatographic procedures to separate these and separating the flavonoids from natural sources. The compounds. number of flavonoids present in medicinal plants and Among the modern separation techniques, CE has food products is large and their analysis becomes com- won increasing acclaim for its extremely high efficiency, plex owing to their existence mostly in O-glycosidic small sample volume, high speed, and good resolution. forms. Although a quantitative determination of individ- CE has been applied to the separation of flavonoids since 1991 [18–40]. Nevertheless, many works [25, 31–37] have Correspondence: Dr. Jos Manuel Herrero-Martnez, Departa- focused on the potential of this technique in the meas- ment de Qumica Analtica, Universitat de Barcelona, Av. Diago- urement of the common flavonoids (single or a few sub- nal 647, Barcelona 08028, Spain classes) present in a particular food, and they have paid a E-mail: [email protected], [email protected] little attention to the potential in the characterization of Fax: +34-934021233 the different classes of flavonoids in several matrix sam- Abbreviations: BHT, butylated hydroxytoluene; SC, sodium ples. The analysis of flavonoid glycosides has been cholate addressed by using CZE [19, 21–23, 26] or MEKC [18, 20, i 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com 2494 J. M. Herrero-Martnez et al. J. Sep. Sci. 2007, 30, 2493 – 2500

Figure 1. Molecular structures of studied fla- vonoid aglycones.

21, 24, 27, 40]. However, analysis of flavonoid aglycones In this work, we developed an analytical method capa- by these techniques has not been studied to the same ble of separating several classes of flavonoid aglycones extent, and only few works on CZE [38] and MEKC [20, (flavanols, flavanones, flavonols, and flavones) com- 28–30, 40] analyses on the flavonoid aglycones have monly present in food samples by mixed MEKC. Within been published. Some of these MEKC procedures have each class, several flavonoids were chosen according to employed organic solvents [28–30, 40] in addition to the their biological interest and abundance in commonly surfactant (usually SDS), as a common additive in order consumed foods. The optimized method was successfully to enhance the separation and selectivity of different applied to the analysis of complex extracts of natural classes of flavonoids in MEKC. Compounds such as flavo- and food products containing several flavonoid agly- noids strongly interact with micelles and consequently cones from each of these classes. selectivity may be varied by modifying the micellar phase. Mixing surfactants with different structural and polar properties to generate mixed micelles has the 2 Materials and methods potential to provide variable micellar phases. Thus, by 2.1 Instrumentation tailoring the micellar environment, solute–micelle interactions can be manipulated to generate the desired MEKC experiments were carried out with a Beckman selectivity [41–44]. Recently, we have reported the suc- instrument P/ACE 5500 (Palo Alto, CA, USA) equipped cessful application of mixed micellar systems to the sep- with a diode array spectrophotometric detector. The sep- aration of procyanidins in food samples [45, 46]. arations were performed at 258C on an uncoated fused- i 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com J. Sep. Sci. 2007, 30, 2493 – 2500 Electrodriven Separation 2495 silica capillary (47.0 cm650 lmid6375 lm od) ether layer was separated and evaporated to dryness obtained from Polymicro Technologies (Phoenix, USA). using a rotavapor. The dry residue was dissolved in (0.5 mL) methanol and injected into the CE system. The extraction protocol followed for orange (pulp and 2.2 Reagents and solutions peel) was based on the work of Careri et al. [48]. About All chemicals were of analytical reagent grade unless oth- 3.5 g of orange material was extracted with 4 mL of erwise noted. Sodium dihydrogenphosphate and hydro- methanol, 1 mL of 12 M hydrochloric acid, and 12 mg of genphosphate, hydrochloric acid, sodium hydroxide, BHT, as an antioxidant, in order to obtain a mixture con- diethyl ether, and butylated hydroxytoluene (BHT) were sisting of 1.5 M HCl in a methanol–water solution from Merck (Darmstadt, Germany). SDS from Merck and (50:50, v/v) containing 1500 mg/L BHT. This mixture was sodium cholate (SC) from Fluka (Buchs, Switzerland) mixed and refluxed at 908C for 60 min. After cooling, were used as surfactants. HPLC-grade methanol and ACN methanol was added to a final volume of 10 mL. The were from Merck. All solutions were prepared with mean recovery of flavanones was in the 98 l 2% range, deionized water (Milli-Q deionizer, Millipore, Bedford, while quercetin yielded a recovery higher than 85%. MA, USA). (+)-Catechin hydrate, (–)-epicatechin, querce- All sample solutions were kept at –208C and protected tin dihydrate, kaempferol, and apigenin were obtained from light until their use. from Fluka. Morin, fisetin, chrysin, galangin, and narin- genin were purchased from Sigma (St Louis, MO, USA). 2.4 MEKC procedure Stock solutions (1 mg/mL) of catechin, epicatechin, morin, fisetin, quercetin, kaempferol, and galangin were Before first use, a new capillary was conditioned at 258Cas prepared in methanol. Apigenin and chrysin stock solu- follows: 10 min with 1 M NaOH, 10 min with 0.1 M NaOH, tions were prepared in a methanol–ACN mixture (70:30, 10 min with water, and finally 30 min with the running v/v). All stock solutions were stored at –208C until their buffer. Between runs, the capillary was rinsed with 0.1 M use and protected from daylight. Prior to injection, work- NaOH (1 min), water (1 min), and running buffer (3 min). ing solutions were daily prepared by dilution of stock sol- At the end of each working session, the capillary was utions with methanol. rinsed with deionized water for 10 min. Standards and samples were injected hydrodynamically at 0.5 psi (1 psi = 6894.76 Pa) for 5 s. Before injection, all solutions 2.3 Sample preparation were filtered through a 0.45 lm pore size nylon filter Wine and orange samples were purchased at a local (Whatman, Maidstone, Kent, UK). Detection was carried supermarket, while propolis and Ginkgo biloba were out at 214, 270, 295, and 350 nm. Unless otherwise indi- obtained from a local pharmacy. Different extraction cated, the applied voltage was 20 kV of positive polarity. procedures were carried out according to the matrix The electrophoretic mobility le and the resolution fac- sample. Propolis samples were treated according to the tor Rs were calculated using the following expressions, protocol reported by Cao et al. [37]. About 2 g of raw prop- respectively: olis was extracted with 15 mL of methanol for 10 min in  LdLt 1 1 an ultrasonic bath. After centrifugation at 3000 rpm6 l ¼ e V t t 15 min, the resultant supernatant solution was put in a m eof volumetric flask of 50 mL. The extraction procedure was 2ðtm2 tm1 Þ Rs ¼ repeated three times. The total extracted solutions were ðw1 þ w2Þ diluted with methanol until 50 mL. where tm and tm are the migration times of two solutes, The extraction conditions for the acid hydrolysis of G. 1 2 w1 and w2 are the corresponding peak widths of the two biloba leaves were taken from Sticher et al. [47]. Briefly, adjacent peaks, teof is the migration time for an dried and pulverized Ginkgo leaves (approximately 4 g) uncharged solute (i.e., methanol), tm the migration time were refluxed with 70 mL of methanol and 10 mL of 25% of the analyte, Lt is the total length of the capillary, Ld is hydrochloric acid at 908C for 60 min. After cooling, the the effective length of the capillary to the detector, and V mixture was filtered through a filter paper (Whatman is the applied voltage. Mobility and resolution measure- no. 1) and transferred into a 100 mL volumetric flask and ments were done by triplicate. made up to the volume with methanol. This solution was directly injected into the CE system. The reported recov- eries of the flavonols quercetin and kaempferol were 3 Results and discussion 100.6 and 101.5%, respectively. 3.1 Optimization MEKC studies Wine sample preparation was similar to that proposed by Rodrguez-Delgado et al. [32]. Wine (5 mL) was The initial CE conditions used for the separation of flavo- extracted with diethyl ether (5 mL), and the resulting noid aglycones (Fig. 1) were based on the previous meth- i 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com 2496 J. M. Herrero-Martnez et al. J. Sep. Sci. 2007, 30, 2493 – 2500

In general, the electrophoretic mobility of flavonoid aglycones increases in absolute value with increase in the mole fraction of SDS in the mixture. However, the electrophoretic mobility of flavonol morin shows an opposite trend. Also, catechin and epicatechin present a much lower retention than the rest of the aglycones. This different behavior can be explained by the ioniza- tion degree of the compounds. Under working pH condi-

tions (pH 7), catechin and epicatechin (pKa1 l9 [51]) remain almost completely uncharged. The rest of the

aglycones, except morin, have pKa1 l7 [51, 52] and thus

remain partially ionized. Morin has a pKa1 = 5.06 [51] and therefore it is almost completely ionized at pH 7. The interaction between the ionized aglycones and the nega- tive charge on the external surface of SDS micelles pro- duced an electrostatic repulsion, larger in the case of the fully ionized morin. These interactions are much lower for SC because its negative charges are oriented inside Figure 2. Effect of the mole fraction of SDS on the electro- the micelle. The electrostatic repulsion forces to morin phoretic mobility of analytes using a mixed SC–SDS system are significantly increased as SDS/SC concentration ratio at the total surfactant concentration of 50 mM. Running buf- increases, since it produces an increase in the total nega- fer, 50 mM phosphate at pH 7. Symbols: Cat (g), Ec (f), Mor tive charge of the external surface of the micelle. As a (h), Nar (6), Fis (*), Quer (G), Kmp (F), Apg (H), Chry (+), consequence of this, the morin peak was located ahead and Gal (0). of epicatechin at high contents of SDS. As shown in Fig. 2, the use of a BGE containing 50 mM odologies reported in the literature [45, 46], and modi- SC showed a significant comigration of solutes due to fied to allow an accurate quantitation of these com- minor differences (or coincidence) in their electropho- pounds in several food samples. A 50 mM phosphate buf- retic mobility values. This translated into a relatively fer at pH 7 containing different ratios of SDS and SC small migration time window of all analytes and, conse- micelles (with 50 mM as the total surfactant concentra- quently, a significant loss of resolution (data not shown). tion) was investigated. The use of SC in combination with With the incorporation of SDS into the SC system, some SDS is interesting due to the type and properties of the of the peak pairs that were poorly resolved with pure SC, micelles that each one forms. SDS forms a spherical such as morin and quercetin, were satisfactorily micelle, while SC forms helical micelles. The SDS micelle resolved. Nevertheless, at 100% SDS, a coelution between consists of a hydrophobic inner core and a charged exter- the chrysin and galangin peaks (the most hydrophobic nal shell, while SC micelle is the reverse, the hydropho- compounds) was observed. bic region faces the aqueous solution while the hydro- In order to establish the analytical conditions that pro- philic moieties turn inward. Therefore, a mixture of SC– vide an adequate resolution of all analytes in a short

SDS will form a complex micellar system in which micel- analysis time, the resolution (Rs) between consecutive lar properties (such as hydrophobicity of the micellar “critical peaks” was evaluated. As can be seen in Fig. 2, interior, and the propensity to solubilize the analytes) the most “critical region” (comprised between 0.2 and strongly depend on the ratio of SC to SDS [49]. Figure 2 0.7 mole fraction of SDS) is that placed between querce- shows the variation of the electrophoretic mobility of fla- tin and galangin peaks; consequently, we calculated the vonoid aglycones in the mixed SC–SDS micellar system. Rs between the consecutive peaks for these four analytes Several selectivity changes in the system on going from (Fig. 3). As a result of this study, a 0.5 mole fraction of one-component SC or SDS micelles to a mixed SC–SDS SDS, that is, a combination of 25 mM SC and 25 mM SDS, micellar system were observed. The reversal of the migra- gave the highest resolution values for “the critical pairs” tion order of the catechin/epicatechin peaks observed at of analytes (especially for kaempferol/apigenin pair) in a the mole fraction of SDS lower than 0.2 can be justified short analysis time. Figure 4 shows the electrophero- by the preferential interaction of SDS with the cis-isomers gram of analytes obtained in these optimum conditions. (epicatechin) at the C3 carbon position in contrast with SC [40, 45, 46, 50]. Moreover, several changes in the 3.2 Analysis of food samples migration order of fisetin, morin, quercetin, and narige- nin are observed when the SDS mole fraction increases In order to test the feasibility of the method, we perform- from 0 to 0.2. ed the analysis of propolis extract, a natural complex i 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com J. Sep. Sci. 2007, 30, 2493 – 2500 Electrodriven Separation 2497

Figure 5. Electropherogram of a raw propolis extract. Peak detection was performed at 380 nm, identification and other Figure 3. Resolution, R , plotted versus the mole fraction of s experimental conditions are as in Fig. 4. SDS. Symbols: (g) Quer/Kmp, (0) Kmp/Apg, (f) Apg/Chry, and (6) Chry/Gal. Experimental conditions and peak identifi- cation as in Fig. 1. gal, antiviral, and antioxidant activities [53, 54]. These effects have been associated with the main organic con- stituents of propolis, which are flavonoid compounds and phenolic acid esters [53, 54]. Raw propolis is com- posed of 50% resin, flavonoids and related phenolic acids, 30% wax, 10% essential oils, 5% pollen, and 5% vari- ous organic and mineral compounds [55, 56]. As shown in Fig. 5, a wide matrix peak overlapped with several analyte peaks (chrysin and galangin), thus hin- dering an accurate evaluation of the content of these fla- vonoids. Sample preparation was apparently not the cause of this interference problem, since this was equally observed with different sample treatments on the same sample and with other propolis samples (data not shown). The use of another wavelength did not improve the selectivity. This wide peak could be attributed to hydrophobic wax-compounds commonly present in propolis [55, 56]. In order to improve the resolution, the addition of organic solvents (1–15%) was tested. ACN led to a clear coelution of analytes, methanol allowed a suitable sep- Figure 4. Electropherogram of a standard mixture of flavo- aration and quantitation of all analytes since the migra- noid aglycones. Running buffer: 50 mM phosphate contain- tion time of matrix peaks increased sufficiently with ing 25 mM SDS and 25 mM SC at pH 7. Capillary: 47 cm respect to the migration times of the analyte peaks. The (effective length, 40 cm)650 lm id; applied voltage, 20 kV; hydrodynamic injection, 0.5 psi65 s; detection at 214 nm. best compromise between resolution and analysis time Peak identification as in Fig. 1. for these compounds was achieved using a BGE contain- ing 25 mM SC and 25 mM SDS and 10% methanol. The presence of methanol produced a slight decrease in the sample. Propolis is a resinous substance collected by hon- electrophoretic mobilities of each solute with respect to eybees from various plant sources, mainly from the pop- those obtained in a BGE without methanol, although the lar (Populus) genus. It has been used as a natural medicine total analysis time did not increase appreciably (ca. because of its remarkable pharmacological properties, 25 min). This decrease could be attributed to the decrease which include anti-inflammatory, antibacterial, antifun- in EOF, which is due to the reduction in the electric per- i 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com 2498 J. M. Herrero-Martnez et al. J. Sep. Sci. 2007, 30, 2493 – 2500

Table 1. Repeatability and LOD for S/N = 3 of analytes in the proposed procedure

Analyte tm tm (RSD,%) Area (RSD,%) LOD (min) (lg/mL) Intra- Inter- Intra- Inter- daya) dayb) daya) dayb)

Cat 5.41 0.32 1.12 1.35 2.41 1.38 Ec 5.80 0.36 1.16 1.46 2.54 1.24 Mor 9.68 0.42 1.22 2.35 4.56 3.96 Nar 11.46 0.44 1.28 2.64 4.23 2.25 Fis 13.80 0.56 1.33 2.83 4.72 2.54 Quer 19.11 0.86 1.74 3.18 5.04 2.54 Kmp 21.45 1.23 2.11 3.81 5.22 2.78 Apg 22.34 1.25 2.22 3.91 5.27 3.30 Chry 24.63 1.42 2.6 4.43 5.86 2.47 Gal 28.23 1.63 3.1 4.62 5.95 3.30

a) As RSD (n = 8). b) As RSD (n = 24).

mittivity–viscosity ratio, and also a decrease in the zeta potential of the silica surface. As shown in Fig. 6A, the peaks of analytes and interfer- ence in propolis sample were satisfactorily separated. Under these conditions, several food samples (G. biloba, orange, and wine) were analyzed. In Fig. 6, traces B and C, two representative electropherograms of the food extracts are shown. Analytical performance characteristics of analytes in the selected MEKC system are shown in Table 1. Linear calibration curves of peak areas with r A0.999 were obtained by injecting six standard solutions in the range of 5–200 lg/mL. Migration time and peak area repeat- abilities were obtained by injecting a 50 lg/mL solution of each analyte. Inter- and intraday repeatabilities were calculated from three series of eight experiments in each series, which were injected at the rate of one series per day for three consecutive days. The LODs of the solutes were obtained for an S/N = 3. The recommended MEKC procedure was applied to the determination of flavonoid aglycones in several food samples. Peak identification of analytes was performed by comparing the migration times and absorption spec- tra with those of the standards, and when necessary also by spiking the sample extracts with the standards. Fur- thermore, standard addition calibration curves were obtained by adding to the extracts at least four solutions with increasing concentrations up to 100 lg/mL. The curves were linear with r A0.998, and in all cases the slope of calibration curve was the same as with the exter- nal calibration method. The contents of the flavonoid aglycones present in the Figure 6. Electropherograms of flavonoid aglycones present analyzed food extracts are given in Table 2. In general, in several food samples: (A) raw propolis, (B) wine, and (C) orange juice. A BGE with the same composition as in Fig. 5, the flavonoids found in these natural products are simi- but supplemented with 10% methanol. Peak identification as lar to those reported in the literature [57]. In raw propolis in Fig. 1. extracts, we found that chrysin and galangin were the i 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com J. Sep. Sci. 2007, 30, 2493 – 2500 Electrodriven Separation 2499

Table 2. Mean values (n = 3) of flavonoid aglycone content the previous HPLC procedures [8, 15, 16, 18, 29, 56] that (mg/100 g) found in several food samples involved analysis times of 50–60 min (column equilibra- tion time not included). Other non-negligible advantages Food sample Analyte Average content (mg/100 g)a) that are present in our method are the low solvent con- sumption and minimal sample requirement (ca. 6 nL/ Raw propolis Nar 25.10 l 1.12 sample) in comparison with HPLC procedures. l Quer 8.60 0.22 Furthermore, the feasibility of our method for flavo- Kmp 7.00 l 0.34 Chry 64.10 l 2.54 noid aglycones has been demonstrated with different Gal 37.70 l 1.83 and complex real samples, in contrast to other CE proto- G. biloba Quer 0.27 l 0.03 cols focused on the separation of standard solution prep- Kmp 0.11 l 0.01 arations or just one type of sample. Red wine Cat 23.67 l 1.34 Ec 9.71 l 0.32 Quer 9.18 l 0.23 Orange 4 Concluding remarks Peel Nar 3.67 l 0.14 The simultaneous separation of several compounds Pulp Nar 2.52 l 0.13 belonging to different classes of flavonoid aglycones has a) Concentrations in red wine are expressed in mg/L. been effectively accomplished by MEKC with the use of SC–SDS mixed micelles. With the proposed method, a complete baseline separation of analytes close to 25 min flavonoids at the highest concentration, which is in good in complex matrix samples could be achieved, without agreement with the results reported in literature [37, 38, any cleanup pretreatment, which constitutes an attrac- 54, 55]. However, in comparison to other CE methods tive alternative to the HPLC methods to analyze these reported for the analysis of propolis [37, 38], our method analytes in such matrices. The analytical procedure provided a significant improvement in the resolution of described here could be especially useful to analyze the analytes peaks from matrix bands in real samples, which main flavonoid aglycones present in natural and food allows an accurate quantitation of the main flavonoid products common in the human diet, and also to exam- aglycones present in these types of samples. ine the physiological activities of these compounds in The G. biloba extract showed the flavonol quercetin as different natural samples. the main flavonoid aglycone, followed by kaempferol, which agrees with the studies of Sticher et al. [47]. The We are thankful for joint support from MCYT of the Spanish Gov- electropherogram of wine extract (Fig. 6B) showed the ernment and FEDER of EU (project CTQ2004-00965/BQU). presence of catechin, epicatechin, and quercetin, typical components found in most of the red wines. The flavo- noid content found was similar to the findings reported 5 References by Rodrguez-Delgado et al. [32, 58]. [1] Robards, K., Antolovich, M., Analyst 1997, 122, 11R – 34R. Flavanone was detectable in orange fruit (Fig. 6C). The [2] Harborne, J. B., The Flavonoids: Advances in Research Since 1986, Chap- naringenin content found in orange pulp was in agree- man and Hall, London 1994. ment with other amounts reported in the literature [14, [3] Rice-Evans, C., Current Med. Chem. 2001, 8, 797 – 807. 16]. To the best of our knowledge, no CE analysis of flavo- [4] Magnani, L., Gaydou, E. M., Hubaud, J. C., Anal. Chim. Acta 2000, noid aglycones in orange fruit has been described. 411, 209 – 216. The application of mixed MEKC to the analysis of com- [5] Rice-Evans, C. A., Miller, N. J., Paganga, G., Free Radic. Biol. Med. 1996, 20, 933 – 956. plex real samples like the ones described here offers sev- [6] Cao, G., Sofic, E., Prior, R. L., Free Radic. Biol. Med. 1997, 22, 749 – eral important advantages over the HPLC methods com- 760. monly employed in the literature to analyze these types [7] Netzel, M., Strass, G., Kaul, C., Bitsch, I., Dietrich, H., Bitsch, R., of samples. Thus, HPLC analysis of complicated or dirty Food Res. Int. 2002, 35, 213 – 216. samples like propolis [11, 29, 55], G. biloba [8, 11, 18, 47], [8] Hasler, A., Sticher, O., Meier, B., J. Chromatogr. 1990, 508, 236 – 240. usually implies a cleanup pretreatment in order to [9] Hertog, M. G. L., Hollman, P. C. H., Venema, D. P., J. Agric. Food remove contaminants from the chromatographic col- Chem. 1992, 40, 1591 – 1598. umn. In our protocol, the samples could be analyzed [10] Hertog, M. G. L., Hollman, P. C. H., Katan, M. B., J. Agric. Food Chem. directly without any cleanup step, since any contami- 1992, 40, 2379 – 2383. nant in CE capillary was removed by rinsing with suit- [11] Merken, H. M., Beecher, G. R., J. Agric. Food Chem. 2000, 48, 577 – able solvents (NaOH, water, and BGE) after each analysis. 599. [12] Molnar-Perl, I., Fuzfai, Z., J. Chromatogr. A 2005, 1073, 201 – 227. Concerning analysis time, mixed MEKC protocol pro- [13] Wulf, L. W., Nagel, C. W., J. Chromatogr. 1976, 116, 271 – 279. vided a total analysis time between 15 and 25 min [14] Justensen, U., Knuthsen, P., Leth, T., J. Chromatogr. A 1998, 799, (depending on the analyte and the sample), in contrast to 101 – 110. i 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com 2500 J. M. Herrero-Martnez et al. J. Sep. Sci. 2007, 30, 2493 – 2500

[15] Menghinello, P., Cucchiarini, L., Palma, F., Agostini, D., Dacha, [37] Cao, Y. H., Wang, Y., Yuan, Q., Chromatographia 2004, 59, 135 – M., Stocchi, V., J. Liq. Chromatogr. Relat. Technol. 1999, 22, 3007 – 140. 3018. [38] Volpi, N., Electrophoresis 2004, 25, 1872 – 1878. [16] Merken, H. M., Beecher, G. R., J. Chromatogr. A 2000, 897, 177 – [39] Jac, P., Polasek, M., Pospisilova, M., J. Pharm. Biomed. Anal. 2006, 184. 40, 805 – 814. [17] Crozier, A., Jensen, E., Lean, M. E. J., McDonald, M. S., J. Chroma- [40] Tonin, F. G., Jager, A. V., Micke, G. A., Farah, J. P. S., Tavares, M. F. togr. A 1997, 761, 315 – 321. M., Electrophoresis 2005, 26, 3387 – 3396. [18] Pietta, P. G., Mauri, P. L., Rava, A., Sabbatini, G., J. Chromatogr. [41] Bumgarner, J. G., Khaledi, M. G., Electrophoresis 1994, 15, 1260 – 1991, 549, 367 – 373. 1266. [19] Seitz, U., Oefner, P. J., Nathakarnkitkool, S., Popp, M., Bonn, G. [42] Clothier, J., Tomellini, S., J. Chromatogr. A 1996, 723, 179 – 187. K., Electrophoresis 1992, 13, 35 – 38. [43] Wiedmer, S. K., Jumppanen, J. H., Haario, H., Riekkola, M. L., Elec- [20] Ng, C. L., Ong, C. P., Lee, H. K., Li, S. F. Y., Chromatographia 1992, trophoresis 1996, 17, 1931 – 1937. 34, 166 – 172. [44] Fuguet, E., Rfols, C., Bosch, E., Abraham, M. H., Ross, M., J. Chro- [21] Morin, P., Villard, F., Dreux, M., J. Chromatogr. 1993, 628, 153 – matogr. A 2001, 942, 237 – 248. 160. [22] Morin, P., Villard, F., Dreux, M., Andre, P., J. Chromatogr. 1993, [45] Herrero-Martnez, J. M., Rfols, C., Ross, M., Torres, J. L., Bosch, 628, 161 – 169. E., Electrophoresis 2003, 24, 707 – 713. [23] McGhie, T. K., J. Chromatogr. 1993, 634, 107 – 112. [46] Herrero-Martnez, J. M., Rfols, C., Ross, M., Bosch, E., Lozano, C., Torres, J. L., Electrophoresis 2003, 24, 1404 – 1410. [24] Pietta, P. G., Mauri, P. L., Bruno, A., Zini, L., J. Chromatogr. 1993, 638, 357 – 361. [47] Sticher, O., Meier, B., Hasler, A., in: van Beek, T. A. (Ed.), The Anal- ysis of Ginkgo Flavonoids, Harwood Academic Publishers, Amster- [25] Bjergegaard, C., Michaelsen, S., Mortensen, K., Sorensen, H., J. dam 2000, p. 179. Chromatogr. 1993, 652, 477 – 485. [48] Careri, M., Elviri, L., Mangia, A., Musci, M., J. Chromatogr. A 2000, [26] McGhie, T. K., Markham, K. R., Phytochem. Anal. 1994, 5, 121 – 126. 881, 449 – 460. [27] Pietta, P. G., Mauri, P. L., Zini, L., Gardana, C., J. Chromatogr. 1994, [49] Wiedmer, S. K., Riekkola, M. L., Nyden, M., Soderman, O., Anal. 640, 175 – 179. Chem. 1997, 69, 1577 – 1584. [28] Delgado, C., Toms-Barbern, F. A., Talou, T., Gaset, A., Chroma- tographia 1994, 38, 71 – 78. [50] Kreimeyer, J., Petereit, F., Nahrstedt, A., Planta Med. 1998, 64,63– 67. [29] Ferreres, F., Blzquez, M. A., Gil, M. I., Toms-Barbern, F. A., J. Chromatogr. A 1994, 669, 268 – 274. [51] Herrero-Martnez, J. M., Sanmartin, M., Ross, M., Bosch, E., Rfols, C., Electrophoresis 2005, 26, 1886 – 1895. [30] Gil, M. I., Ferreres, F., Toms-Barbern, F. A., J. Liquid Chromatogr. 1995, 18, 3007 – 3019. [52] Jovanovic, S. V., Steenken, S., Tosic, M., Marjanovic, B., Simic, M. G., J. Am. Chem. Soc. 1994, 116, 4846 – 4851. [31] Liang, H. R., Siren, H., Riekkola, M. L., Vuorela, P., Vuorela, H., Hiltunen, R., J. Chromatogr. A 1996, 746, 123 – 129. [53] Burdock, G. A., Food Chem. Toxicol. 1998, 36, 347 – 363. [32] Rodrguez-Delgado, M. A., Prez, M. L., Corbella, R., Gonzlez, G., [54] Bankova, V., Popova, M., Bogdanov, S., Sabatini, A. G., Z. Natur- Garca Montelongo, F. J., J. Chromatogr. A 2000, 871, 427 – 438. sforch. 2002, 57, 530 – 533. [33] Dadakova, E., Prochazkova, E., Krizek, M., Electrophoresis 2001, 22, [55] Bankova, V., Christoy, R., Stoev, G., Popov, S., J. Chromatogr. 1992, 1573 – 1578. 607, 150 – 153. [34] Fonseca, F. N., Kato, M. J., Oliveira Jr., L., Neto, N. P., Maggi [56] Pietta, P. G., Gardana, C., Pietta, A. M., Fitoterapia 2002, 73,S7– Tavares, M. F., J. Microcol. Sep. 2001, 13, 227 – 235. S20. [35] Cao, Y., Chu, Q., Fang, Y., Ye, J., Anal. Bioanal. Chem. 2002, 374, [57] Hollman, P. C. H., Arts, I. C. W., J. Sci. Food Agric. 2000, 80, 1081 – 294 – 299. 1093. [36] Suntornsuk, L., Kasemsook, S., Wongyai, S., Electrophoresis 2003, [58] Rodrguez-Delgado, M. A., Malovana, S., Prez, J. P., Borges, T., 24, 1236 – 1241. Garca-Montelongo, F. J., J. Chromatogr. A 2001, 912, 249 – 257.

i 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com