Introduction

Glucosinolates are β-d-thioglucoside-N-hydroxysulfates found in the plant family Cruciferae [1, 2]. More than 120 glucosinolates with different R groups, e.g. alkyl, alkenyl, hydroxyalkenyl, aryl, or indole have been identified [3]. The variety of the R group results in a wide range of biological activity and polarity of the glucosinolates [4].

Glucosinolates are found in the seeds, roots, stems, and leaves of plants in 11 families of dicotyledonous angiosperms [5]. Hydrolysis of glucosinolates is catalyzed by myrosinase present in seeds and tissues of the cruciferous plants [6] and produces goitrogenic and potentially hepatoxic compounds, for example isothiocyanates, thiocyanates, nitriles, and thiones, depending on pH, temperature, metal ions, protein cofactors, and the R group [6, 7]. Some glucosinolates result in a wide range of flavors and toxic effects on consumption [8]. Glucosinolate hydrolysis products, isothiocyanates, have adverse and beneficial effects on human health [9]. It has been suggested glucosinolates and derived products prevent carcinogens from reaching the target site, activate important hepatic enzymes against several carcinogens, or interact with carcinogens [10, 11].

A variety of analytical methods have been developed either for direct determination of the intact glucosinolates or for indirect measurement of the enzymatic degradation products [12]. Direct analysis of the intact glucosinolates is important, because it can reflect the specificity of each glucosinolate [12]. Gas chromatography (GC) and high-performance liquid chromatography (HPLC) have been used for direct determination of glucosinolates [13]. GC analysis is not amendable to some glucosinolates containing polar side chains [13].

Natural herbs have been used in traditional Chinese medicines over two-thousand years. During long applications, some herbs were replaced by herbs unavailable locally, if they had the same or similar pharmaceutical effectiveness. As a result, herbs belonging to totally different classes share similar or the same Chinese names in different regions. For example, banlangen (root of Isatis indigotica Fort.) is a cruciferous herb that was used in Northern China whereas nan-banlangen (root of Baphicacanthus cusia (Nees) Bremek.) was used as its replacement in Southern China. A non-cruciferous herb Guangdong-daqingye (leaf of Baphicacanthus cusia (Nees) Bremek.) has been used in Guangdong province to replace a cruciferous herb daqingye (leaf of Isatis indigotica Fort.). Thus banlangen and daqingye have been commonly confused with nan-banlangen and Guangdong-daqingye, respectively. In addition, a non-cruciferous herb baijiangcao (Patrinia scabiosaefolia Fisch. ex Trev.) has often been confused with the cruciferous herb ximing (Thlaspi arvense L.) because ximing has also been called baijiangcao in mid-southern China. In specific circumstances these three pairs of herbs may have similar or the same effectiveness and thus can be replaced with each other in some traditional medicine prescriptions. They also have different activity, however, so when they are replaced by each other and used in other medicines they may cause side-effects or even toxic effects. To standardize Chinese herbs it is necessary to develop a reliable and specific method for differentiating the original herbs from those with which they may be confused.

Chemical analysis has been used for differentiation of herbs and their active components [14]. Banlangen, daqingye and ximing are in the Cruciferae family and contain glucosinolates. Nan-banlangen and Guangdong-daqingye belong to the Acanthaceae and baijiangcao belongs to the Vaerianceae. Acanthaceae and Vaerianceae plants contain no glucosinolates, so glucosinolate determination may provide scientific evidence for differentiation of cruciferous traditional Chinese plants from the corresponding non-cruciferous plant. This paper describes the use of HPLC and ESI–MS for direct determination of twelve intact glucosinolates in ten Chinese herbs. The results were used to distinguish between herbs with the same core name with or without various regional sub-names.

Experimental

Chemicals and reagents

Epiprogoitrin, glucocheirolin, glucoerucin, glucoiberin, gluconapin, gluconasturtiin, glucoraphenin, glucosibarin, glucotropaeolin, progoitrin, and sinalbin were obtained from KVL (Frederiksberg C, Denmark). Sinigrin monohydrate was from Sigma (St Louis, MO, USA). Daqingye was a gift from Professor Zhong-Zhen Zhao. Hancai (Rorippa indica (Linn.) Hiern) was a gift from Professor Albert Wai-Ming Lee. Other herbs were purchased from the Chinese Medicine Clinical Practice Centre, Hong Kong Baptist University. HPLC-grade hexane and methanol were purchased from Riedel-de Haën (Hanover, Germany). Dichloromethane and ethyl acetate were from Tedia (Fairfield, USA). Ammonium acetate was obtained from Panreac (Barcelona, Spain) and formic acid was from Acros organics (New Jersey, USA). Milli-Q water was produced by a Milli-Q ultrapure water-purification academic system from Millipore (Billerica, USA).

Each of the twelve intact glucosinolate standards (10 mg) was dissolved in 1 mL Milli-Q water to prepare a stock standard mixture. The mixture was used as a spiking and recovery standard. The stock mixture was further diluted to prepare calibration standard solutions at 0.5, 0.4, 0.3, 0.2, 0.1, 0.05, and 0.005 mg mL−1, respectively. All standard solutions were stored at 4 °C until use.

Traditional Chinese herb sample preparation

Extraction

Dried plant powder (5 g) was weighed and blended twice with 50 mL methanol at 70 °C for 15 min immediately after grinding. The mixture was then filtered through a Whatman (Maidstone, UK) No. 40 filter paper by suction filtration and washed twice with 50 mL methanol. The collected filtrate was evaporated to dryness under vacuum by use of a Rotovap (Caframo, Germany) at 55 °C and 60 rpm, and reconstituted with 10 mL methanol. The extracts were then centrifuged by use of a cyclone centrifuge at 13,000 rpm for 15 min. The supernatant was collected for cleaning.

Florisil adsorbent (Fisher Certified ACS, 60–100 mesh; Sigma, St Louis, USA) was activated overnight at 200 °C before solid-phase extraction. A polypropylene syringe barrel (5 mL) was filled with 0.4 g activated Florisil between two 20-μm polypropylene frits. The flow rate was kept at 1–2 mL min−1 throughout the clean-up process. The activated Florisil column was pre-cleaned with 5 mL 30% (v/v) dichloromethane in hexane. The supernatant (300 μL) was mixed with 5 mL 30% (v/v) dichloromethane in hexane and applied to the column. Interferences were washed from the column with 5 mL 30% (v/v) dichloromethane in hexane and the column was then aspirated for 5 min. The glucosinolates were then eluted from the column with 5 mL 30% (v/v) ethyl acetate in methanol. The extract was evaporated to dryness under a slow stream of dry nitrogen at 35 °C and reconstituted with 300 μL Milli-Q water. The extract was centrifuged at 13,000 rpm for 15 min. The supernatant were stored at 4 °C until HPLC analysis.

Method recovery test

A recovery test was conducted by using a dried herb matrix, shegan (root of Belamcanda chinensis (L.) DC.) that had been repeatedly analyzed and showed to contain no target glucosinolates. The matrix (5 g) was spiked with the glucosinolate standards (500 μg) and the spiked sample was extracted as described above. The peak areas for each of the twelve glucosinolates determined were compared with those from a 100 μg g−1 standard mixture. The experiment was repeated three times and recoveries and precision (relative standard derivations, RSD, n = 3) were calculated for each glucosinolate.

Reversed-phase HPLC analysis

HPLC analysis was performed with a Hewlett–Packard (San Francisco, USA) HP1100 series instrument equipped with a diode-array detector. A reversed-phase Hypersil BDS C18 column (250 mm × 4.6 mm i.d., 5 μm; Alltech, Deerfield, USA) was used to separate the glucosinolates in the extracts. Calibration standards and plant extracts were injected to the column via a 20-μL sample loop. Individual glucosinolates were detected at 233 nm. The mobile phase was a gradient prepared from 30 mmol L−1 ammonium acetate containing formic acid at pH 5.0 (component A) and methanol (component B). The gradient program was: 100% A–0% B for 5 min; increased to 70% A–30% B from 5 to 17 min; and kept at 70% A–30% B for 3 min; the flow rate was 1 mL min−1. Under these chromatographic conditions glucoiberin, glucocheirolin, progoitrin, sinigrin, epiprogoitrin, glucoraphenin, sinalbin, gluconapin, glucosibarin, glucotropaeolin, glucoerucin, and gluconasturtiin standards were eluted at retention times (R T) of 4.77 ± 0.02, 5.14 ± 0.02, 5.83 ± 0.03, 6.25 ± 0.03, 7.04 ± 0.03, 8.48 ± 0.04, 11.23 ± 0.03, 11.90 ± 0.04, 15.08 ± 0.03, 15.82 ± 0.03, 16.16 ± 0.03, and 19.11 ± 0.03 min, respectively.

Electrospray ionization mass spectrometric analysis

HPLC fractions of the target glucosinolates detected in herb extracts were collected and analyzed by electrospray ionization–quadrupole time-of-flight mass spectrometry (ESI–QTOF-MS) in negative-ion mode with MS–MS analysis. The QTOF mass spectrometer was equipped with a turbo ion-spray source (Sciex Q-Star Pulsar i; Applied Biosystems, Canada). The MS conditions, optimized for the glucosinolates, were: ion-spray potential −4000 V, declustering potential (1) −65 V, focusing potential −165 V, declustering potential (2) −15 V, ion source gas (1) 25, ion source gas (2) 8, curtain gas 15, collision gas 3, and temperature of ion source gas (2) 200 °C. The collected HPLC fractions were introduced directly into the ESI–MS source from an infusion pump and analyzed at a mass resolving power of 10,000. Product-ion scanning was performed to confirm the identities of the glucosinolates by investigating fragmentation of the molecular ions.

Results and discussion

Sample preparation

The herbs were extracted with methanol at 70 °C for 15 min to inactivate endogenous myrosinase that may catalyze hydrolysis of the target glucosinolates. Both activated Florisil columns and C18 cartridges were investigated for clean-up and both achieved recoveries of target glucosinolates better than 85%. The C18 clean-up was achieved by retaining non-polar interferences on the adsorbent whereas most polar interferences were co-eluted with the target glucosinolates. The activated Florisil clean-up process, on the other hand, efficiently removed the polar interferences and provided cleaner extracts. In addition, the non-polar interferences were washed from the column. Recovery was tested three times for the target glucosinolates to obtain average recoveries and precision data (RSD, n = 3). Average recoveries for the twelve glucosinolates ranged from 86.1 to 122.2% with RSD from 5.3 to 14.6% (n = 3) (Table 1).

Table 1 Recovery, precision, and detection limit data for the glucosinolates
Full size table

Reversed-phase HPLC analysis

Spiked solutions and herb extracts were analyzed for the twelve glucosinolates under same HPLC conditions. Other glucosinolates were not analyzed because of the lack of authentic standards. Typical HPLC chromatograms obtained from banlangen (Fig. 1a), nan-banlangen, daqingye, Guangdong-daqingye, baijiangcao, ximing, hancai, jiezi (Seed of Sinapis alba L.), laifuzi (Seed of Raphanus sativus L.), and tinglizi (Seed of Lepidium apetalum Willd) were compared with those obtained from the twelve glucosinolate standards (Fig. 1b). Initial qualitative detection was based on comparison of retention times. Because the retention time varied slightly because of the effect of complicated matrices, retention time determination was confirmed by spiking the sample extracts with glucosinolates. Seven-point calibration plots were applied over the range 0.005 to 0.5 mg mL−1 and good linearity was achieved with correlation coefficients, R 2, better than 0.9993. Detection limits were calculated for a signal-to-noise ratio as 3 described elsewhere [15] and ranged from 0.06 μg g−1 dry weight for sinalbin to 0.36 μg g−1 dry weight for glucoraphenin, when 5 g dried herbs were analyzed. Limits of quantification, calculated on the basis of a signal-to-noise ratio of 10, ranged from 0.20 μg g−1 dry weight for sinalbin to 1.20 μg g−1 dry weight for glucoraphenin (Table 1).

Fig. 1
figure 1

Chromatograms obtained from (a) banlangen and (b) 0.3 mg mL−1 glucosinolate standard mixture

Full size image

ESI-QTOF-MS analysis

Initial detection of the glucosinolates was confirmed by mass spectrometric analysis. The HPLC fractions were collected, diluted with methanol, then analyzed by ESI–QTOF-MS in negative-ion mode with MS–MS analysis. ESI–MS analysis of the epiprogoitrin fraction from the banlangen extract at R T 6.971 min revealed the presence of the [M − H]− ion at m/z 388.0377, which matched the theoretical value with a mass difference of 1.3 ppm. Similarly, analysis of the gluconapin fraction from lafuzi at R T 11.875 min revealed the presence of the [M − H]− ion at m/z 372.0432, which matched the theoretical value with a mass difference of 2.4 ppm. For the further confirmation, MS–MS analysis of the epiprogoitrin and gluconapin were conducted. The MS–MS fragmentation enabled structure elucidation of the glucosinolates. From the MS–MS analysis of epiprogoitrin from banlangen (Fig. 2a), the observed fragment ion at m/z 275 resulted from the loss of SO4 and OH from the [M − H]− ion. The peak at m/z 146 corresponded to loss of C6H10O5 and SO3 from the [M − H]− ion. The MS–MS spectrum of gluconapin from laifuzi (Fig. 2b) contained peaks corresponding to loss of SO3 and HSO4 from the [M − H]− ion at m/z 292 and 275, respectively. The peak at m/z 130 corresponded to the loss of C6H10O5 and SO3 from the [M − H]− ion. In both MS–MS spectra the peak at m/z 195 corresponded to the fragment ion of the d-thioglucose group C6H11O5S− and the peaks at m/z 97 and m/z 80 represented the fragment ions \( HSO^{ - }_{4} \) and \( SO^{ - }_{3} \), respectively, from the glucosinolates.

Fig. 2
figure 2

MS–MS spectra obtained from (a) m/z 388 for epiprogoitrin identified in banlangen and (b) m/z 372 for gluconapin identified in laifuzi

Full size image

Quantitative analysis of glucosinolates in the plant samples

Concentrations of the glucosinolates in ten plants were quantified on the basis of external standard calibration plots, although internal standard calibration would have been more accurate. The quantitative data obtained provided information about the patterns and concentration ratios of the twelve glucosinolates in the herb samples. There was great diversity in the content and composition of the twelve glucosinolates in the samples (Table 2). The total amounts of the twelve glucosinolates in the cruciferous herbs ranged from 81.3 μg g−1 (hancai) to 13199.9 μg g−1 (tinglizi). Glucosinolate concentrations in seeds from the plants were very high, so extracts of tinglizi, laifuzi, and jiezi were diluted twentyfold, fifteenfold, and fivefold, respectively, before HPLC analysis.

Table 2 Concentrations of the glucosinolates in Chinese plants
Full size table

None of the target glucosinolates was found in the non-cruciferous herbs, including nan-banlangen, Guangdong-daqingye, and baijiangcao, but at least two glucosinolates were detected in each of the cruciferous herbs. Concentrations of the total glucosinolates in herb seed samples, for example jiezi, laifuzi, and tinglizi were approximately 2 to 162-fold higher than in other cruciferous herbs such as banlangen, daqingye, ximing, and hancai. In banlangen the predominant glucosinolates were progoitrin (795.3 μg g−1), epiprogoitrin (5397.0 μg g−1), glucoraphenin (124.5 μg g−1), and gluconapin (636.1 μg g−1); sinigrin, sinalbin, and gluconasturtiin were detected only at relatively low levels. Daqingye, a different part of the same plant as banlangen, contained glucoraphenin (488.8 μg g−1), only, as the predominant glucosinolate; progoitrin and glucosibarin were detectable at low levels.

Detection of glucosinolates for differentiation of herbs which are sometimes confused with each other

This method was used for differentiation of cruciferous plants and the corresponding plants with which they can be easily-confused, i.e. banlangen and nan-banlangen, daqingye and Guangdong-daqingye, and baijiangcao and ximing. The analytical results indicated that the cruciferous plants banlangen, daqingye, and ximing contained glucosinolates at total concentrations of 3727.4, 545.4, and 1354.0 μg g−1, respectively, whereas the non-cruciferous plants nan-banlangen, Guangdong-daqingye, and baijiangcao contained no detectable glucosinolates. Although amounts of individual and total glucosinolates in any herb may vary substantially depending on genetic origin, nature of growth, age, and cultural and environmental factors associated with the growth of particular plants, the results obtained clearly demonstrated the feasibility of applying analytical science for differentiation of plants which are, traditionally, easily confused by determination of glucosinolates. The study revealed that some glucosinolate-containing cruciferous herbs had been misused in Chinese medicinal prescription in mistake for other non-cruciferous herbs, and vice versa.

Conclusion

The method developed, using activated Florisil column and reversed-phase HPLC analysis, resulted in sufficient retention and baseline separation for analysis of twelve intact glucosinolates—glucoiberin, glucocheirolin, progoitrin, sinigrin, epiprogoitrin, glucoraphenin, sinalbin, gluconapin, glucosibarin, glucotropaeolin, glucoerucin, and gluconasturtiin—in ten traditional Chinese plants. The glucosinolates in the cruciferous herb extracts were detected and identified by combined use of ESI–QTOF-MS and MS–MS analysis in negative-ion mode. The quantitative results obtained were used to differentiate three pairs of herbs which are easily confused with each other.