Qe

Ia

b

a

ARRAA

KBSC

1

tshroht

mel[hs

ct(crls

0d

Analytica Chimica Acta 663 (2010) 105–108

Contents lists available at ScienceDirect

Analytica Chimica Acta

journa l homepage: www.e lsev ier .com/ locate /aca

uantitative determination of glucoraphanin in Brassica vegetables by micellarlectrokinetic capillary chromatography

ris Leea, Mary C. Boycea,∗, Michael C. Breadmoreb

School of Natural Sciences, Edith Cowan University, Perth, Western Australia 6027, AustraliaSchool of Chemistry, University of Tasmania, Hobart, Tasmania 7001, Australia

r t i c l e i n f o

rticle history:eceived 2 November 2009eceived in revised form 22 January 2010ccepted 22 January 2010

a b s t r a c t

Glucoraphanin, a glucosinolate, is found naturally in plants and is present in relatively high concentrationsin broccoli. Glucosinolates have received much attention as studies have indicated that a diet rich in themmay provide some protection from certain cancers. A micellar electrokinetic chromatography (MEKC)

vailable online 1 February 2010

eywords:roccoliodium cholate

method using sodium cholate as the micellar phase has been developed to quantify for glucoraphaninin broccoli (seeds and florets) and Brussels sprouts. The glucoraphanin peak elutes just under 5 minwith a theoretical plate number of 380,000 per metre of capillary. The method is suitable for crudeextracts of broccoli and Brussels sprouts. Glucoraphanin in broccoli seeds (1330 mg/100 g) broccoli florets(89 mg/100 g) and Brussels sprouts (3 mg/100 g) was determined and agreed with the data obtained by

hrom00 g),

apillary electrophoresis high performance liquid cin broccoli seeds (4 mg/1

. Introduction

Glucosinolates occur naturally in plants. They are found in rela-ively high concentrations in plants belonging to the Brassica familyuch as broccoli, cabbage, cauliflower and Brussels sprouts. Studiesave shown that a diet rich in Brassica vegetables may reduce theisk of some cancers and this is attributed in part to the presencef glucosinolates [1]. This has lead to an increased interest in theealth benefits of these vegetables and hence analytical methodshat can rapidly measure glucosinolate content.

High performance liquid chromatography (HPLC) has been aethod of choice for the determination of glucosinolates. Sev-

ral reversed phase (RP) methods have been outlined in theiterature for both intact glucosinolates and desulfoglucosinolates2–4]. However, due to the polarity of the analytes ion pair andydrophilic interaction chromatography methods are often neces-ary to enhance retention [5,6].

The ability of capillary electrophoresis to separate the glu-osinolates has been demonstrated. The intact glucosinolates areypically separated using micellar electrokinetic chromatographyMEKC) [7–9] employing a borate/phosphate buffer with 50 mM

etyltrimethylammonium bromide (CTAB) at pH 7 which was firsteported by Michaelson et al. [7]. For the desulfonated glucosino-ates an MEKC approach has also been adopted. Both CTAB [7,8] andodium cholate (SC) [10] have been employed as the micellar phase.

∗ Corresponding author. Tel.: +61 8 63045273.E-mail address: m.boyce@ecu.edu.au (M.C. Boyce).

003-2670/$ – see front matter © 2010 Published by Elsevier B.V.oi:10.1016/j.aca.2010.01.043

atography. The LODs were 10–100 times below the levels typically foundbroccoli florets (0.9 mg/100 g) and Brussels sprouts (0.1 mg/100 g).

© 2010 Published by Elsevier B.V.

Karcher and El Rassi used capillary zone electrophoresis (CZE) todetermine total glucosinolates by enzymatically releasing glucosefrom the glucosinolates and then converting it to gluconic acid[11]. This same group profiled individual glucosinolates by mon-itoring either the acid or enzymatic degradation products [12,13].The degradation products were reacted with a fluorescent tag andthen resolved using MEKC and the neutral micellar phase octyl-�-d-glucopyranoside [13]. Bringmann et al. used CZE with massspectrometry (MS) detection to determine the glucosinolate pro-file of Arabidopsis thaliana seeds [14]. More recently, Bellostas et al.used MEKC and a SC pseudostationary phase to monitor myrosinasecatalysed hydrolysis of 2-OH substituted glucosinolates and theircorresponding degradation products [15]. Glucoraphanin was notincluded in this study. In 2008, Fouad et al. used microchip CE todetermine the total glucosinolate levels in crude plant extracts [16].

Relatively few of these papers deal in any detail with real sam-ples and only one paper quantitatively determines glucoraphanin(Fig. 1) the main glucosinolate of interest in broccoli and other Bras-sica vegetables [9]. Trenerry et al. used MEKC with CTAB, however,the analysis time was relatively long with glucoraphanin elutingafter 17 min and the internal standard, sorbic acid, eluting after23 min. The internal standard was necessary to improve migra-tion time repeatability. Furthermore, a solid phase extraction (SPE)step was necessary to remove interfering peaks and to improve

repeatability [9].

In this paper we report a method for the determination ofintact glucoraphanin from broccoli (florets and seeds) and Brusselssprouts. It is an improvement on the previously reported methodas (1) glucoraphanin elutes in under 5 min (compared to 17 min),

106 I. Lee et al. / Analytica Chimica A

(tanb

2

2

d(bt(eFtR

2

abUpob7pGi

2

df

2

eflbtfipa

Fig. 1. Structure of glucoraphanin.

2) an SPE cleanup step is not necessary allowing crude extractso be analysed directly and (3) this method has been shown to bepplicable to other Brassica vegetables. The quantitative determi-ation of glucoraphanin in vegetables by this method was verifiedy HPLC.

. Materials and methods

.1. Chemicals

All chemicals were of AR, HPLC grade or purity stated. Sodiumihydrogen orthophosphate (phosphate), di-sodium tetraborateborate) and CTAB >98% were obtained from Fisher Scientific (Mel-ourne, Australia) SC >98% sodium dodecyl sulphate >99% (SDS) andetramethylammonium bromide >98% (TMAB) from Sigma–AldrichSydney, Australia), acetonitrile and methanol from Lomb Sci-ntific (Sydney, Australia) and formic acid 90% from APS Ajaxinechem (Sydney, Australia). Milli-Q 18.2 M� cm water was usedhroughout the experiment. Glucoraphanin was obtained from theesearch Centre for Industrial Crops (Bologna, Italy).

.2. Buffers and solutions

Stock solutions (100 mM) of sodium tetraborate, SC, SDS, CTABnd 150 mM sodium dihydrogen orthophosphate were preparedy dissolving appropriate amounts of these chemicals in water.sing these stock solutions, the following running buffers were pre-ared: 20 mM borate; 20 mM borate with varying concentrationsf SC (25, 50, 80 mM); 20 mM borate and 50 mM SDS; and 18 mMorate, 30 mM phosphate and 50 mM CTAB adjusted to a final pH of.00 with 0.1 M NaOH. A 1 M formic acid (pH 2.10) buffer was alsorepared as a running buffer by dilution of the concentrated acid.lucoraphanin standards (25, 50, 100 and 200 ppm) were prepared

n water from a 500 ppm stock aqueous solution.

.3. Food samples

Broccoli seeds were purchased from Green Harvest Organic Gar-en Supplies. Broccoli florets and Brussels sprouts were purchasedrom a local supermarket.

.4. Extraction of glucoraphanin

A previously described method with some modifications wasmployed [9]. Three grams of broccoli seeds or 10 g of broccoliorets were added to approximately 30 mL of boiling water and

oiled for ∼5 min. The seeds or florets and residual water werehen transferred to a mortar and ground to paste. The paste wasltered, under vacuum, through a Whatman no. 4 filter paper. Theaste was washed with a sufficient amount of hot water so thatfinal volume of 50.0 mL was attained when made up in a vol-

cta 663 (2010) 105–108

umetric flask. All extracts were stored in a freezer at −20 ◦C andfiltered through a 0.45 �m nylon filter prior to analytical HPLC andCE analyses. Glucoraphanin in Brussels sprouts was extracted asfor broccoli floret with two amendments: the extraction was per-formed with 70% methanol:30% water and the extract was madeup to a final volume of 10.0 mL.

2.5. Instrumentation and conditions

2.5.1. Capillary electrophoresisThe extracts were analysed on a Hewlett Packard 3D CE (Wald-

bron, Germany) and controlled by Agilent Chemstation software.Separations were performed using uncoated fused silica capillaries(Polymicro Technology, Phoenix AZ) of length 60 cm × 50 �m inter-nal diameter (i.d.) with an effective length of 52 cm for SC, SDS, CZEseparations and 77 cm × 75 �m i.d. with an effective length of 69 cmfor CTAB separations. The separations were performed at 30 ◦C and+30 kV (SC, SDS and CZE-borate analyses) or −15 kV (CZE-formicacid and CTAB analyses). The samples were loaded onto the col-umn by pressure injection at 50 mbar for 5 s. The UV–vis diodearray detector was set at 230 nm. Between runs, the capillaries werewashed with water (2 min) followed by buffer (2 min) for SC, SDSand CZE separations, and with 1 M NaOH (2 min), water (2 min) fol-lowed by buffer (3 min) for CTAB separations. Running buffers werechanged after every 5 analyses. Data were collected and processedby Agilent Chemstation software.

2.5.2. High performance liquid chromatographyThe analyses were performed on a Varian ProStar 240 Solvent

Delivery system equipped with a Varian ProStar 400 Auto Sampler(20 �L sample loop) and a Varian ProStar 330 Photodiode ArrayDetector set at 230 nm. Data were collected and processed by Var-ian Star Chromatography Workstation software. The separationswere completed on a Luna 5 �m C18, 250 mm × 4.6 mm column(Phenomenex). An isocratic mobile phase of 5 mM TMAB in 2% (v/v)methanol/water and a flow rate of 1 mL min−1 was employed [9].

3. Results and discussion

Glucoraphanin has previously been quantitatively determinedin broccoli by MEKC [9]. Using CTAB as the micellar phase, glu-coraphanin eluted at approximately 17 min, however, due to poormigration time reproducibility an internal standard (IS), sorbic acid,was employed which eluted after 23 min. The inclusion of the ISincreased the run time further. Options such as using a shorter cap-illary or a higher voltage to reduce the run time were hampered bythe presence of a closely migrating peak.

Our attempts to use CTAB met with the same issues describedby Trenerry et al. [9]. In particular, the non-reproducibility of thesystem was not easily resolved. Extensive washing of the capillarybetween runs made the relatively long method even longer and ledus to investigate an alternative method with better performance.Initially CZE was investigated by trying to separate a crude broc-coli floret extract using both a high pH borate buffer and a lowpH formic acid buffer. In the high pH buffer system glucoraphanineluted in approximately 4 min, however, it was unresolved fromother analytes in the crude extract (Fig. 2a). The low pH bufferhad the advantage of being MS compatible and had been reportedfor determination of glucosinolates in A. thaliana seeds [14] andwhile glucoraphanin eluted much later in this buffer it was alsounresolved (Fig. 2b).

In an effort to improve the resolution of glucoraphanin inthe high pH buffer system, a borate buffer with 50 mM SDS wasalso used, however, it did not succeed in resolving glucoraphanin(Fig. 2c). The potential of SC was then examined as it has beenused previously for the separation of desulfonated glucosinolates

I. Lee et al. / Analytica Chimica Acta 663 (2010) 105–108 107

Fig. 2. Electropherograms of crude broccoli extracts separated by (a) CZE, 20 mMbpG

[tflrni

Fo

orate, pH 9.00; (b) CZE, 1 M formic acid, pH 2.10; (c) 50 mM SDS in 20 mM borate,H 9.00. The electropherograms were obtained using different broccoli extracts.= glucoraphanin.

10]. Varying concentrations of SC (25, 50 and 80 mM) were addedo the 20 mM borate buffer and used to separate a crude broccoli

oret extract. Adding SC to the buffer reduced the mobility of gluco-aphanin and improved resolution between glucoraphanin and theearest peak (Fig. 3). At 25 mM SC the resolution was 1.3 which

ncreased to 1.7 for the 50 mM SC buffer and further increased

ig. 3. The effect of adding SC to the buffer on the relative (to EOF) migration timef glucoraphanin.

Fig. 4. Electropherogram for (a) crude broccoli extract and (b) SPE broccoli extractseparated by MEKC using a 50 mM SC 20 mM borate buffer. G = glucoraphanin.

to 2.3 for the 80 mM buffer. At all concentrations tested the peakfor glucoraphanin was sharp and symmetrical: the tailing factorwas consistently 0.5 for glucoraphanin. The peak efficiency for theglucoraphanin peak was superior for the 50 mM SC buffer sys-tem (380,000 theoretical plates per metre of capillary) particularlywhen compared to the 80 mM SC buffer systems (140,660 theo-retical plates per metre of capillary). As 50 mM provided sufficientresolution in the shortest time and peak efficiency was optimal, thismethod was adopted (Fig. 4a).

As SPE has been recommended for cleanup of broccoli extracts,the extract was then passed through a C18 SPE cartridge. The result-ing electropherogram for this extract is recorded in Fig. 4b. Thedetector response was maintained for the glucoraphanin peak andthere were only minor changes to the extract which did not impacton the analysis of glucoraphanin.

The intraday repeatability of the 50 mM SC method using acrude extract was determined. The migration time repeatabilityover six runs was very good and recorded as 0.4% CV. The peak arearepeatability was also satisfactory at 4.0%. The method was thenapplied to determine the glucoraphanin content in extracts of broc-coli floret, broccoli seeds and Brussels sprouts. The extracts werealso analysed by HPLC [9]. Standards in the range of 25–500 ppm

were prepared for both HPLC and CE. A linear relationship wasobserved for both methods (r = 0.9999 and 0.9996 respectively).The samples were analysed in triplicate for HPLC and quadru-plicate for CE. There was excellent agreement between the twomethods (Table 1). The data agrees with values previously quoted

Table 1Quantitative data for glucoraphanin extracts analysed by HPLC and MEKC.

Samples Concentration mg/100 g (% CV)

MEKC HPLC

Broccoli 1 32 (6.1) 31 (0.4)Broccoli 2 89 (1.8) 91 (0.6)Broccoli seeds 1330 (4.0) 1300 (0.2)Brussels sprouts 3 (2.9) 3 (2.5)

For HPLC, n = 3.For CE, n = 4 for broccoli floret and n = 3 for seeds and sprouts.

1 mica A

iarTiemt4B

4

utasiv

A

s

[

[[[

08 I. Lee et al. / Analytica Chi

n the literature for broccoli [9]. The linear range between 25nd 500 ppm allowed samples with varying amounts of gluco-aphanin to be analysed without any modification of the method.he broccoli seeds having large amounts of glucoraphanin typicallyn excess of 1000 mg/100 g while the Brussels sprouts have low lev-ls typically only 1–10 mg/100 g were both analysed using the sameethod. The LOD for the different sample types were well below

he expected glucoraphanin content. The LODs were 0.9 mg/100 g,mg/100 g and 0.1 mg/100 g for broccoli florets, broccoli seeds andrussels sprouts respectively.

. Conclusions

This MEKC method employing SC rather than the commonlysed CTAB is rapid, accurate and reproducible in its ability to quan-itatively determine glucoraphanin from Broccoli. The method islso suitable for other vegetables such as Brussels sprouts. Thehort analysis time (glucoraphanin elutes at 5 min) and the abil-ty to cope with crude extracts (avoiding an SPE step) make it aiable alternative to HPLC.

cknowledgements

Dr. Renato Lori for the kind donation of the glucoraphanintandard. Dr Rob Trengove for the use of the Agilent capillary elec-

[

[

[

cta 663 (2010) 105–108

trophoresis unit. MCB acknowledges receipt of an ARC QEII Fellowfrom the Australian Research Council (DP0984745).

References

[1] N. Tawfiq, R.K. Heaney, J.A. Plumb, G.R. Fenwick, S.R. Musk, G. Williamson,Carcinogenesis 16 (1995) 1191.

[2] S. Perez-Balibrea, D.A. Moreno, C. Garcia-Viguera, J. Sci. Food Agric. 88 (2008)904.

[3] F. Vallejo, F. Tomas-Barberan, C. Garcia-Viguera, J. Agric. Food Chem. 51 (2003)3029.

[4] M. Meyer, S.T. Adam, Eur. Food Res. Technol. 226 (2008) 1429.[5] N. Rangkadilok, M.E. Nicolas, R.N. Bennett, R.R. Premier, D.R. Eagling, P.W.J.

Taylor, Sci. Hortic. 96 (2002) 27.[6] J.K. Troyer, K.K. Stephenson, J.W. Fahey, J. Chromatogr. A 919 (2001) 299.[7] S. Michaelson, P. Moller, H. Sorensen, J. Chromatogr. A 608 (1992) 363.[8] P. Morin, F. Villard, A. Qunisac, M. Dreux, J. High Resolut. Chromatogr. 15 (1992)

271.[9] V.C. Trenerry, D. Caridi, A. Elkins, O. Donkor, O.R. Jones, Food Chem. 98 (2006)

179.10] C. Bjergegaard, S. Michaelsen, P. Moller, H. Sorensen, J. Chromatogr. A 717

(1995) 325.11] A. Karcher, Z. El Rassi, Anal. Biochem. 267 (1999) 92.12] A. Karcher, A.M. Hassan, Z. El Rassi, J. Agric. Food Chem. 47 (1999) 4267.13] A. Karcher, Z. El Rassi, J. Liq. Chromatogr. Relat. Technol. 21 (1998) 1411.

14] G. Bringmann, I. Kajahn, C. Neusub, M. Pelzing, S. Laug, M. Unger, U. Holzgrabe,

Electrophoresis 26 (2005) 1513.15] N. Bellostas, J.C. Sorensen, H. Sorensen, J. Chromatogr. A 1130 (2006)

245–252.16] M. Fouad, M. Jabasini, M. Kaji, K. Terasaka, M. Tokeshi, H. Mizukami, Y. Baba,

Electrophoresis 29 (2008) 2280.