Wei LiZheng ChenYiping LiaoHuwei Liu

The Key Lab of Bioorganic Chemistryand Molecular Engineering,Ministry of Education,College of Chemistry and MolecularEngineering,Peking University, Beijing, P.R. China

Received March 21, 2005Revised June 9, 2005Accepted August 22, 2005

Short Communication

Study on separation of aristolochic acid I andII by micellar electrokinetic capillarychromatography and competitionmechanism between SDS and �-cyclodextrin

In this study, a rapid MEKC method using 40 mM sodium borate buffer containing50 mM SDS as surfactant was developed for the analysis of aristolochic acid (AA) inAristolochia plants. Baseline separation of AA-I and AA-II was achieved within 3 minwith high separation efficiency, satisfactory sensitivity, repeatability, and recovery.Resolution between AA-I and AA-II is above 5 and great performance with higher than200 000 theoretical plate numbers was obtained. The detection limits (based on 3 S/N)were both 1.0 mg/mL. Two kinds of AA in 35 herbal samples of Aristolochia plants weresuccessfully determined. The competition mechanism between b-CD and SDS wasalso investigated by changing the content ratio of b-CD and SDS.

Keywords: Aristolochic acid / Capillary zone electrophoresis / Chinese herbs nephro-pathy / Medicinal plants / Micellar electrokinetic chromatography

DOI 10.1002/elps.200500228

Aristolochic acids (AAs), composed of AA-I and AA-II, arestructurally related to nitrophenanthrene carboxylic acids(Fig. 1). They are the major toxic components found inAristolochia plants [1, 2]. A number of publications haveindicated that some kinds of Chinese herbs containingAAs and some related components will cause a severekidney disease, the so-called Chinese herbs nephropathy(CHN) [3], which is characterized by a progressive inter-stitial fibrosis leading to a severe atrophy of the proximaltubules and incurable damage of kidney in a short time[4–6]. Such cases occurred in many areas including Bel-gium, France, Spain, UK, Japan, and China Taiwan andhave attracted people’s attention worldwide about theclinic security [7].

In the past several years, HPLC has become the mostwidely used method for the analysis of AAs in herbal med-icines [8–13] and HPLC-MS [14, 15] has been also used forthe detection of AA-I and AA-II. Although the separationperformance is obviously improved compared with TLC

Figure 1. Structures of (a) AA and (b) AL. AA-I and AL-I:R = OMe. AA-II and AL-II: R = H.

[16, 17] or ultraviolet spectrophotometry [18] which areusually applied in the analysis of the total amount of AA-Iand AA-II or the major active component AA-I; thesemethods cannot meet the requirement of high-throughputanalysis for the longer analytical time and high cost oforganic solvent used as mobile phase. From the practicalpoint of view, CE shows more advantages in various fieldsincluding the analysis of herbal medicines [19–22]. How-ever, currently there are only a few publications on theanalysis of AA-I and AA-II by CE. Kvasnicka et al. [23]reported a CZE method to detect AA in dietary supple-ments and selected herbs with a mixture of morpholi-

Correspondence: Professor Huwei Liu, College of Chemistry andMolecular Engineering, Peking University, Beijing 100871, P.R. ChinaE-mail: hwliu@pku.edu.cnFax: 186-10-62751708

Abbreviations: AA, aristolochic acid; AL, aristololactam; CHN, Chi-nese herbs nephropathy

Electrophoresis 2006, 27, 837–841 837

2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com

838 W. Li et al. Electrophoresis 2006, 27, 837–841

nethanesulphonic acid, bistrispropane, and hydroxy-ethylcelullose in 10% methanol as a BGE. In our previouswork [24], a fast CZE method with b-CD as modifier wasfirstly developed for the analysis of AA-I and AA-II, whichwas proved to be practical in routine analysis of relatedherbal medicines. In this study, MEKC was applied to theanalysis of AA-I and AA-II in Chinese herbs using a buffercomposed of 40 mM sodium borate and 50 mM SDS assurfactant. The result showed that baseline separation oftwo kinds of AA was achieved within 3 min with highseparation efficiency. Furthermore, some other toxic AA-related neutral components such as aristololactam (AL)can be also detected by this method, which cannot beachieved by CZE method. The competition mechanismbetween b-CD and SDS was also discussed.

All experiments were performed with an Agilent 3D CEsystem with air-cooling and a diode array detector (Agi-lent Technologies, Palo Alto, CA, USA). MEKC was per-formed in an uncoated fused-silica capillary of 48.5 cmlength (effective length 40.0 cm)650 mm ID (Ruifeng,Hebei, China). Other conditions are as following: capillarytemperature 257C, applied voltage 30 kV, UV detection at254 nm, and samples injection at 50 mbar for 5 s. Thecapillary was washed with 1.0 M sodium hydroxide(10 min), water (10 min), and the running buffer (10 min) inturn. Between consecutive analyses, the capillary wasflushed with the running buffer for 5 min to guaranteegood repeatability. The standard mixture of AA-I (29%)and AA-II (66%) was obtained from Sigma, AL was kindlyprovided by School of Pharmaceutical Sciences ofPeking University, and b-CD was provided by TokyoChemical Industries (Japan). SDS was purchased fromFluka (Buchs, Switzerland), and 35 kinds of medicinalplants samples of Aristolochia plants were identified byProfessor Shilin Hu at China Academy of Traditional Chi-nese Medicine. All chemicals were of analytical-reagentgrade: methanol and sodium borate were from BeijingChemical Factory (Beijing, China); pure water was pre-pared by Aquapro Reverse Osmosis system (Aquapro,Chongqing, China).

For sample preparation, 0.5 g of each dried plant wasmarinated in 5 mL anhydrous methanol for 16 h. After that,it was extracted by ultrasonication and marinated at roomtemperature for 30 min and then centrifuged at 3000 rpmfor 8 min. The extraction was repeated twice (1.5 mL62),and the extracts were combined and diluted to 5 mL withmethanol, which was then passed through a 0.45 mmmembrane filter before analysis.

A standard sample of AA-I and AA-II was briefly studiedapplying a series of sodium borate buffers with variouspH ranging from 8.0 to 10.5 and concentration in therange of 20–120 mM, showing that 40 mM sodium borate

buffer (pH 8.8) was the best condition. Surfactant and itsconcentration have significant effects on separation effi-ciency. Our experiments showed that SDS was the mostsuitable surfactant for this separation. Furthermore, theconcentration of SDS was systematically optimized, il-lustrating very important effect on the separation. It wasdemonstrated that baseline separation of AA-I and AA-IIcould not be achieved when SDS concentration wasbelow 20 mM, and resolution and theoretic plate numberof AA were both improved with the increase of SDSconcentration in the range of 20–50 mM, but decreasedin the range of 50–100 mM. Consequently, the buffercomposed of 40 mM sodium borate and 50 mM SDS(pH 8.8) with the capillary temperature of 257C andapplied voltage of 30 kV was confirmed to be the optimalconditions, under which, baseline separation of AA-I andAA-II can be successfully achieved within 3 min, which isdramatically shorter than that in the most of HPLCmethods. Great separation performance with resolutionbetween two components higher than five, and theoreticplate numbers greater than 200 000 per meter was alsoobtained. A typical electropherogram of standard mix-ture of AA-I and AA-II is illustrated in Fig. 2a. Comparedwith HPLC [9] and CZE [24] method, this rapid and eco-nomic MEKC method is believed to be more valuable forapplication. Furthermore, the experimental results indi-cate that the sensitivity, repeatability, and recovery are allsatisfactory.

The calibration curves of the peak area (A) versus amountof AA (c) are as following: AA-I: A = 754.38C 1 1.05(r = 0.9982); AA-II: A732.13C 1 3.09 (r = 0.9997). It canbe seen that the linearity was satisfactory with a correla-tion coefficient (r) greater than 0.998. The detection limitsof two AAs were found to be both 1.0 mg/mL (corre-sponding to 10 mg/g dried plants), which were the sameas those obtained by CZE method [24]. The repeatabilityof this MEKC method was tested by the consecutiveanalysis of AA-I and AA-II for five runs, and the RSD(n = 5) of migration time for AA-I and AA-II were bothbelow 0.6%, and the RSD (n = 5) of peak area was 1.8and 1.0%, respectively, showing that the repeatability iswell acceptable. The recoveries of the AA-I and AA-IIwere 97.6 and 76.1% (n = 4) with the RSD of 3.0% forAA-I and 3.8% for AA-II, demonstrating that the simplepreparation method could provide acceptable extractionefficiency in a short period.

By applying the proposed MEKC method, 35 samples ofAristolochia plants collected from various regions in Chinaincluding eight kinds of Aristolochia fangchi Wu, 21 kindsof Caulis Aristolochiae Manshuriensis, and six kinds ofother Aristolochia medicinal plant samples have beenanalyzed. Figure 2b shows a typical electropherogram of

2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com

Electrophoresis 2006, 27, 837–841 CE and CEC 839

Figure 2. Typical electro-pherograms of AA and AL byMEKC. (a) Standard sample ofAA-I and AA-II; (b) a real sampleof C. Aristolochiae Manshur-iensis from Mudanjiang, Hei-longjiang region by MEKCmethod; (c) standard sample ofAA-I and AA-II and AL. Separa-tion conditions: (a) and (b):40 mM sodium borate buffer(pH 8.8) containing 50 mM SDS;(c) 100 mM sodium borate buffer(pH 8.8) containing 50 mM SDS;other conditions: 50 MmID648.5 cm (40.0 cm effective)fused-silica capillary, appliedvoltage of 30 kV, capillary tem-perature of 257C, UV detectionat 254 nm, injection at 50 mbarfor 5 s.

a real sample, indicating that the analysis of AA-I and AA-II did not interfere with other components, which was dif-ficult to be avoided by HPLC. The identifications of AA-Iand AA-II peaks in real sample solutions were confirmedby the comparison of their migration times and online UVspectra with those of the standard samples in order todetermine the contents of AAs exactly. Some of theresults, which can present the distribution of AA-I and AA-II among different samples are listed in Table 1, revealing

AA-I was the major component and there was high varia-bility in the contents of AAs due to the difference of spe-cies and regions. In generally, the contents of AAs in C.Aristolochiae Manshuriensis samples were higher thanthose in Aristolochia fangchi Wu samples. Furthermore,AA-I and AA-II were not detected in the medicinal plantscollected in Zhongshan, Qujiang, Gaoyao, and Zhaoqingin Guangdong province, which was in accordance withthe results from HPLC [9] and CZE [24] methods.

2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com

840 W. Li et al. Electrophoresis 2006, 27, 837–841

Table 1. Contents of AA-I and AA-II in real samples of Chinese medicinal plants determined byMEKC

Species Cultivation region AA-I (%, w/w) AA-II (%, w/w)

Aristolochia fangchi Wu Zhaoqing, Guangdong –a) –Zhongshan, Guangdong – –Qujiang, Guangdong – –Gaoyao, Guangdong – –Mengshan, Guangxi 0.12 –Cangwu, Guangxi 0.045 0.011

C. Aristolochiae Manshuriensis Tumen, Jilin 0.28 0.085Antu, Jilin 0.062 –Dunhua, Jilin 0.12 0.037Huanren, Liaoning – –Dandong, Liaoning 0.087 0.042Mudanjiang, Heilongjiang 0.11 0.031

Stephania tetrandra S. Moore Abei, Sichuan 0.028 –

a) “–” means that the content of AA is lower than the detection limit.See the experimental section for separation conditions; the detection limits of AA-I and AA-II in realsamples of Aristolochia plants were both 10 mg/g (0.001%).

By the MEKC approach, not only charged componentsbut also some AA-related neutral compounds can bedetected. It was reported that AL, a metabolite of AA, is akind of toxic component to people’s kidney, so it is nec-essary to study efficient detection method for these neu-tral compounds. By CZE method, neutral substancescannot be separated because they will comigrate withelectroosmotic flow marker. By this MEKC method, ALwas successfully separated from AAs (see Fig. 2c), andthe improvement of separation efficiency of AL-I and AL-IIis under investigation.

In this study, we also investigated the competition mech-anism between b-CD and SDS with the conversion ofseparation mechanism in MEKC mode. The results indi-cated that when b-CD and SDS were both used in thebuffer as pseudostationary phase, inclusion complexa-tion of b-CD with the analytes was effectively competitivetoward partition of the analytes into SDS micelles [25].The separation of AA-I and AA-II was typical to validatethis competitive mechanism. We systematically investi-gated a number of buffer compositions in which the con-centration of b-CD was kept constant (10 mM) while theconcentrations of SDS were changed, showing differentseparation profiles of AAs. It was the ratio of SDS to b-CDthat exactly determines separation mechanism of AA-Iand AA-II. The separation efficiency of AA-I and AA-II wassatisfactory with high resolution when the ratio of SDS tob-CD was below 2:1. It was the inclusive ability of b-CDthat mainly affected the separation and it followed theCZE mechanism. With the increase of SDS concentration,

the separation efficiency was dramatically deterioratedbecause of the competition between SDS and b-CD. Theinclusion complexation of the AA into the CD cavity wasdeeply affected by the partition of the solute towards themicelle when SDS concentration reached the criticalmicellar concentration. When the ratio of SDS to b-CDreached 2:1, the peaks of AA-I and AA-II were merged toone because SDS and b-CD disturbed each other mostseriously at this ratio. After that point, separation wasimproved again with the continuous increase of SDSconcentration, but the migration order of AA-I and AA-IIwas reversed, showing that the separation mechanismwas converted to MEKC mechanism by which AA-I andAA-II were separated due to the difference in their parti-tion between the pseudostationary phase and aqueousphase. It can be verified by the fact that the migrationorder of the peaks of AA-I and AA-II was as same as theirelution order in HPLC [9].

From the results discussed above, it can be concludedthat as to the separation of hydrophobic compounds, thestrong association with the highly hydrophobic SDSmicelles asks for a considerably stronger competition bythe aqueous pseudostationary phase [26], which canresult in the significant improvement of selectivity. How-ever, for comparatively hydrophilic analytes, which haveweaker hydrophobic interactions with SDS, the CD-MEKC approach is sometimes not efficient because thecompetition between SDS micellar and the aqueouspseudostationary phase is too strong and it will lead tonegative effect on separation. These results should sup-

2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com

Electrophoresis 2006, 27, 837–841 CE and CEC 841

port the necessity to perform further studies in order tochoose the best combination of CD and surfactant forthe components with quite different hydrophobic char-acters.

In our work, a MEKC method using 40 mM sodium boratebuffer (pH 8.8) containing 50 mM SDS under 257C tem-perature and 30 kV applied voltage was developed for theseparation of AA-I and AA-II in medicinal plant samples.This method was proved to be effective, fast, and eco-nomic with great separation performance. It was not onlysuitable for the analysis of AA but also for other neutralAA-related components like ALs, which was prior to CZEmethod. The investigation of competition mechanism ofSDS and b-CD in MEKC was instructive to the corre-sponding research, which needs these two compoundsas additives.

This study is supported by NSFC, Grant No. 20575003,20335010, and 90209056. The authors would like toacknowledge Agilent Technologies for providing anHP3DCE system. Professor Shilin Hu, Xuefeng Feng,Shaoqing Cai, and Mr. Baoquan Che were also appre-ciated for their assistance in the experiments.

References

[1] Jong, T. T., Lee, M. R., Hsiao, S. S., J. Pharmaceut. Biomed.2003, 33, 831–837.

[2] Schaneberg, B. T., Khan, I. A., J. Ethnopharmacol. 2004, 94,245–249.

[3] Bagnis, C. I., Deray, G., Baumelou, A., Le Quintrec, M., et al.,Am. J. Kidney Dis. 2004, 44, 1–11.

[4] Cosyns, J. P., Drug Safety 2003, 26, 33–48.[5] Arlt, V. M., Stiborova, M., Schmeiser, H. H., Mutagenesis

2002, 17, 265–277.[6] Li, X. M., Wang, H. Y., Nephrology 2004, 9, 109–111.

[7] Lee, T. Y., Wu, M. L., Deng, J. F., Hwang, D. F., J. Chroma-togr. B 2001, 766, 169–174.

[8] Hashimoto, K., Higuchi, M., Makino, B., J. Ethnopharmacol.1999, 64, 185–189.

[9] Li, W., Li, R. K., Bo, T., Liu, H. W., et al., Chromatographia2004, 59, 233–236.

[10] Blatter, A., Reich, E., J. Liq. Chromatogr. Rel. Technol. 2004,27, 2087–2100.

[11] Jon, J. H., Chen, S., Wu, T. S., J. Liq. Chromatogr. Rel.Technol. 2003, 26, 3057–3068.

[12] Chang, Y. S., Deng, J. S., Ku, Y. R., J. Liq. Chromatogr. Rel.Technol. 2002, 25, 961–975.

[13] Nortier, J. L., Munz, M-C., Schmeiser, H. H., Arlt, V. M., et al.,New Engl. J. Med. 2000, 342, 1686–1692.

[14] Chan, S. A., Chen, M. J., Liu, T. Y., Talanta 2003, 60, 679–685.

[15] Lee, M. C., Tsao, C. H., Iou, S. C., J. Sep. Sci. 2003, 26, 818–822.

[16] Ioset, J. R., Raoelison, G. E., Hostettmann, K., Planta Med.2002, 68, 856–858.

[17] Ioset, J. R., Raoelison, G. E., Hostettmann, K., Food Chem.Toxicol. 2003, 41, 29–36.

[18] Shang, M.Y., Li, J., Hu, B., Yang, S., et al., Chin. Trad. HerbalDrugs (in Chinese) 2002, 31, 899–900.

[19] Yang, J., Long, H., Liu, H., Huang, A., et al., J. Chromatogr. A1998, 811, 274–279.

[20] Ku, Y. R., Lin, Y. T., Lin, J. H., Wen, K. C., et al., J. Chroma-togr. A 1998, 805, 301–308.

[21] Cheung, H. Y., Cheung, C. S., Kong, C. K., J. Chromatogr. A2001, 930, 171–176.

[22] Song, J. Z., Xu, H. X., Tian, S. J., But, P. P.-H., J. Chromatogr.A 1999, 857, 303–311.

[23] Kvasnicka, F., Sevcik, R., Voldrich, M., Kratka, J., CentralEur. J. Chem. 2004, 2, 417–424.

[24] Li, W., Gong, S. X., Wen, D W., Che, B. Q., et al., J. Chro-matogr. A 2004, 1049, 211–217.

[25] Gotti, R., Fiori, J., Mancini, F., Cavrini, V., Electrophoresis2004, 25, 3282–3291.

[26] Gotti, R., Fiori, J., Hudaib, M., Cavrini, V., Electrophoresis2002, 23, 3084–3092.

2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com