Journal of Chromatography A, 1157 (2007) 51 55

Journal of Chromatography A, 1157 (2007) 51 55 Extraction and ultra-performance liquid chromatography of hydrophilic and lipophilic bioactive components in a Chinese herb Radix Salviae Miltiorrhizae Mei Liu a,b, Yongguo Li b,c,, Guixin Chou b,c, Xuemei Cheng c, Mian Zhang a, Zhengtao Wang a,b,c, a Department of Pharmacognosy, China Pharmaceutical University, Nanjing 210038, China b Key Laboratory of Standardization of Chinese Medicines of Ministry of Education, Shanghai University of Traditional Chinese Medicine, Shanghai 201203, China c Shanghai R&D Centre for Standardization of Chinese Medicines, Shanghai 201203, China Received 18 October 2006; received in revised form 21 April 2007; accepted 7 May 2007 Available online 10 May 2007 Abstract An ultra-performance liquid chromatography (UPLC) method has been developed and validated for the quality evaluation of a Chinese herb Radix Salviae Miltiorrhizae (Danshen root) by chemical fingerprinting analysis. Meanwhile a novel sample preparation procedure has been designed which can simultaneously extract both the hydrophilic and lipophilic components for a single run LC system. Comparing to the conventional HPLC method, UPLC showed many advantages including reduced run time, less solvent consumption and increased peak capacities. Using the established method, 20 target components were detected based on the retention times and on-line UV spectra by referencing to the available standards and reported data. Eleven crude drug samples from different sources were evaluated using the UPLC fingerprints. The developed method proved practicable and reliable for quality control of herbal products with multivalent components in a complicated matrix. 2007 Elsevier B.V. All rights reserved. Keywords: Radix Salviae Miltiorrhizae; Fingerprinting; Salvianolic acids; Tanshinone; UPLC; HPLC 1. Introduction The quality assessment of herbal medicines has always been a challenging task due to the diversity of the multi-components existing in a complicated matrix. The chromatographic fingerprinting (CFP) method has become one of the most frequently applied approaches [1,2], which can provide the whole profile of not only the marker compounds but also the unknown components. Nevertheless development and operation of CFP on HPLC is a rigorous operation as it generally needs about one hour for a single run and the resolution seems still poor due to Corresponding author at: Key Laboratory of Standardization of Chinese Medicines of Ministry of Education, Shanghai University of Traditional Chinese Medicine, No. 1200 Cai Lun Road, Zhangjiang Hi-Tech Park, Shanghai 201203, China. Tel.: +86 21 51322507; fax: +86 21 51322519. Co-corresponding author. E-mail addresses: yongguoli@yahoo.com (Y. Li), wangzht@shutcm.edu.cn (Z. Wang). the extreme complexity of the components contained in herbal products. Ultra-performance liquid chromatography (UPLC) is a newly developed Waters instrumentation with high power in separation and analysis speed over traditional HPLC [3]. There were several publications introducing the use of UPLC in pharmaceutical analysis [4 9], however, no application on quality assessment or chemical fingerprint of Chinese herbal medicines using this technique has been reported up to date. Danshen, root of Salvia miltiorrhiza, is one of the most popularly used Chinese herbs and is now experiencing a particularly strong and rapid demand for treatment and prevention of cardiovascular system disorders [10]. The major bioactive constituents of Danshen can be classified into the hydrophilic depsides derived from caffeic acid, e.g. salvianolic acids, and lipophilic components including diterpenoids tanshinones (see Supplementary data, Fig. S1). It is well known that the accumulations of the secondary metabolites in natural herb, like Danshen root are seriously affected by ecological environments and cultivation conditions, and therefore the quality 0021-9673/$ see front matter 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2007.05.018

52 M. Liu et al. / J. Chromatogr. A 1157 (2007) 51 55 Table 1 Data of normalized peak areas of UPLC fingerprints of Danshen roots from different sources Peak D01 D02 D03 D04 D05 D06 D07 D08 D09 D10 D11 RT RRT Average RSD (%) 1 0.06 0.05 0.09 0.57 0.21 0.03 0.03 1.55 0.49 0.22 0.06 0.63 0.04 0.31 149 2 0.41 0.11 0.00 0.29 0.15 0.00 0.00 0.67 0.22 0.53 0.00 0.91 0.06 0.22 108 3 2.07 1.16 0.60 2.99 1.17 0.64 0.65 2.30 3.32 4.43 1.05 1.91 0.14 1.85 69 4 0.19 0.00 0.00 0.34 0.00 0.00 0.00 0.64 0.25 0.72 0.00 2.13 0.15 0.19 138 5 0.11 0.10 0.39 0.16 0.25 0.05 0.00 0.32 0.10 0.13 0.09 3.28 0.23 0.15 76 6 0.19 0.00 0.60 0.12 0.14 0.06 0.09 0.17 0.00 0.17 0.19 3.56 0.25 0.16 104 7 0.30 0.26 0.00 0.20 0.14 0.82 0.27 0.92 0.14 0.31 1.33 4.16 0.29 0.43 96 8 1.03 0.60 0.49 0.68 0.92 1.73 0.08 2.39 0.32 0.93 0.23 4.73 0.34 0.85 80 9 1.58 1.97 2.64 1.14 1.55 1.97 0.94 0.37 1.65 0.84 0.15 4.87 0.35 1.35 55 10 3.41 4.96 2.43 3.26 2.11 1.52 1.23 3.50 4.05 5.37 1.90 5.09 0.36 3.07 44 11 51.41 49.48 49.94 39.55 50.01 42.94 33.91 29.27 46.22 42.26 60.81 5.52 0.39 45.07 20 12 1.08 2.75 2.01 0.64 0.95 0.38 0.84 0.76 0.98 1.43 1.12 6.01 0.43 1.18 57 13 0.56 0.24 0.18 0.15 0.34 0.99 0.39 0.33 0.10 0.00 0.21 7.19 0.51 0.32 85 14 0.54 0.72 0.71 0.49 0.18 3.15 1.83 0.82 0.23 0.27 0.85 10.30 0.73 0.89 98 15 1.71 1.01 1.91 1.78 1.31 1.08 5.78 1.33 0.83 0.62 1.59 11.08 0.79 1.72 82 16 0.75 1.23 0.59 0.76 0.85 1.84 1.67 0.49 0.34 0.38 0.68 11.63 0.82 0.87 58 17 1.42 0.60 1.50 2.39 1.52 1.06 3.99 0.98 0.75 0.84 1.39 12.49 0.89 1.49 64 18 3.81 5.05 5.18 4.68 3.91 5.52 6.25 5.91 2.52 2.89 4.03 12.64 0.90 4.52 27 19 1.88 2.20 3.07 2.00 1.73 6.36 6.99 2.71 1.06 1.07 2.96 13.21 0.94 2.91 68 20 8.98 9.93 7.95 4.78 10.19 11.21 9.19 10.96 3.71 4.44 6.60 14.11 1.00 7.99 34 Note: Peak normalization was calculated based on the integrative peaks (the peaks area larger than 3000 V, or the peaks height larger than 30 V). RRT (relative retention time) was calculated by referring to peak 20 (tanshinone IIA). D01 D11: Danshen root collected from different sources. D01, from Shanghai, D02 from Bozhou, Anhui, D03 from Hefei, Anhui, D04 from Nanning, Guangxi, D05 from Zhengzhou,Henan, D06 from Anguo,Hebei, D07 from Jinan, Shandong, D08 from Shanghai, D09 from Guangzhou, Guangdong, D10 from Zhangshu, Jiangxi and D11 from Germany. evaluation of Danshen root has drawn extraordinary attention [11 15]. CFP analyses of Danshen using HPLC or high-speed countercurrent chromatography (HSCCC) have been reported [16,17], nevertheless, in those methods, the hydrophilic or lipophilic components were extracted and analysed separately, which were undoubted time-consuming and inconvenient. A specific designed pattern for fingerprinting analysis of Danshen to extract simultaneously both the hydrophilic and lipophilic components proved a challenging task due to their diverse chemical and physical properties of the two groups of compounds. The present research aimed to develop a new CFP method using UPLC to demonstrate both the hydrophilic and lipophilic components coexisting in Danshen in a single run. As a comparison, a routine HPLC chromatographic method was also developed and optimized. The Danshen roots from different sources were evaluated using the developed CFP method as well (See Supplementary data, Fig. S1). 2. Experimental 2.1. Chemicals, reagents and samples HPLC grade acetonitrile and phosphoric acid (Merck, Darmstadt, Germany) were used for HPLC and UPLC analyses. Deionised water was purified by a Milli-Q system (Millipore, Bedford, MA, USA). Other chemicals and solvents were of analytical grade. The reference standards of protocatechuicalhydryde, danshensu, caffeic acid, rosmarinic acid, lithospermic acid, salvianolic acid B, salvianolic acid A and tanshinone IIA were provided by Shanghai R&D Centre for Standardization of Chinese Medicines and their purities are of more than 98% by HPLC-Diode-Array-Detector analysis based on a peak area normalization method. The samples were collected from Germany and several sources within China (Table 1) and were authenticated by Professor ZhengtaoWang, and voucher samples were deposited in the laboratory of institute of Chinese Materia Medica, Shanghai University of Traditional Chinese Medicine. 2.2. Apparatus UPLC was performed using a commercial Acquity System from Waters and 100 mm columns with an I.D. of 2.1 mm packed with 1.7 m Acquity C 18 BEH particles. The UPLC system was equipped with a 500 nl flow cell and a Rheodyne injector, an ultraviolet detector and an auto sampler. HPLC analysis was performed on an Agilent 1100 series HPLC system with an auto sampler and a diode array detector. The analysis was conducted on an Agilent Zorbax SB-C 18 column 250 mm 4.6 mm and 5 m packing material. 2.3. Preparation of sample solutions One gram of the Danshen root powder, accurately weighed, was transferred to a 50 ml tampon centrifuge tube. The preparations in triplicate were extracted with 10 ml of 10% methanol for 30 min by ultra-sonication and subsequently centrifuged for 10 min at 2000 g; the supernatant was taken as solution 1. The residue was treated with 25 ml of 90% methanol for another 30 min by ultra-sonication, after centrifugation, the supernatant was taken as solution 2. Finally, 5 ml of solution 1 and 20 ml of

M. Liu et al. / J. Chromatogr. A 1157 (2007) 51 55 53 solution 2 were mixed in a 25 ml volumetric flask, 75% methanol was added to make up to the mark if necessary. The final solution was allowed to stand for 30 min, and filtered through a 0.45 m membrane before injection. 2.4. Preparation of the reference solutions Accurately weighed quantities (5 mg for each) of the reference compounds were combined and made to a 10 ml volume using 75% methanol in a volumetric flask, This solution was then filtered through a 0.45 m membrane and used as the mixed reference solution for UPLC and HPLC analyses. 2.5. Chromatographic conditions The HPLC fingerprinting analysis was carried out on an Agilent Zorbax SB-C 18 column (250 mm 4.6 mm I.D., 5 m). Chromatographic separation was carried out using a gradient elution of (A) acetonitrile, (B) 0.1% phosphoric acid in water and (C) methanol as follows: 0 30 min, A 5 25%, B 95 75%, 30 40 min, A 25 50%, B 75 50%, 40 70 min, A 70%, B 10%, C 20%. The flow rate was kept at 1.0 ml/min, column temperature at 30 C, injection 10 l, the wavelength of PDA detector ranged from 200 to 400 nm and detected at 280 nm. The condition of UPLC fingerprinting was conducted on an Acquity C 18 BEH column (100 2.1 mm I.D., 1.7 m), using a binary mobile phase of (A) acetonitrile and (B) 0.1% phosphoric acid: 0 5 min, A 10 25%; 5 10 min, A 25 50%; 10 16 min, A 80%. The flow rate was kept at 0.5 ml/min whilst the column temperature kept at 30 C, injection volumes were 2 l, and the chromatogram was recorded at 280 nm. 3. Results and discussion 3.1. Sample procedure development Due to the diverse physicochemical properties, the hydrophilic and lipophilic components co-existing in Danshen root had been extracted individually and determined separately with time-consuming procedures [18]. A recent report [19] described a method to extract both the two types of components, but which seemed unsatisfactory as the lipophilic components gave lower recovery, and furthermore, the hydrophilic component salvianolic acid B displayed so prominent a peak that the profiling figure seemed extremely disproportionate. In our present research, a new sample preparation approach was developed by integrating two extraction steps into one procedure for simultaneous analysis of both the hydrophilic and the lipophilic components. Briefly, the first step was to extract the hydrophilic components with dilute aqueous methanol. Four concentrations of methanol (0, 5, 10 and 15%) were tested and the results showed that 0, 5 and 10% methanol gave the same high extraction efficiencies as indicated by the recovery of salvianolic acid B (97.4%), whilst there was no lipophilic component taken out at this strength of methanol. Consequently 10% methanol was used for extraction of the hydrophilic components (solution 1). For the extraction of the lipophilic components, three concentrations (90, 95 and 100%) of methanol were tried which gave the similar extraction efficiencies, as estimated by the recovery of tanshinone IIA (>97%) and at last, 90% methanol was applied to obtain lipophilic components (solution 2). Finally, 5 ml of solution 1 and 20 ml of solution 2 were mixed well in a 25 ml volumetric flask and make to the mark with 75% methanol if necessary. Using this integrated method both the two types of chemically diverse analytes were sufficiently extracted and more importantly a perfect CFP image was constructed. 3.2. UPLC fingerprint analysis of Danshen root To obtain good separation and ideal peak distribution of CFP, the system conditions including mobile phase, column temperature, detection wavelength and flow rate were investigated respectively. By referring to the established HPLC conditions, it was determined that acetonitrile was a better organic mobile phase than methanol for the separation of salvianolic acids. In order to improve the resolution and symmetry of the peaks, different strengths of buffers in the mobile phase was tested. It was found that 0.1% of phosphoric acid achieved better separation and suppressed any tailing peaks. It was also suggested that the optimised separation could be achieved when the column temperature was kept at 30 C rather than 25 or 40 C and the flow rate set at 0.5 ml min 1. For detection, both the tanshinones and salvianolic acids displayed the maximal UV absorption at 280 nm. CFP of Danshen root obtained by UPLC as shown in Fig. 1A using the given chromatographic parameters, in which the target peaks were well separated from each other as well as from other unidentified components with a minimum resolution (R s, min) of 1.8 being obtained using a binary solvent system. By referring to the retention times and UV spectra of the standards, eight peaks (2, 4, 5, 9,10,11,13 and 20) in the UPLC profile were identified. These were danshensu, protocatechuicalhydryde, caffeic acid, rosmarinic acid, lithospermic acid, salvianolic acid B, salvianolic acid A and tanshinone IIA, respectively. 3.3. Comparison of the fingerprints between UPLC and HPLC The CFP analysis of Danshen root by HPLC-PDA was also constructed using the given chromatographic parameters (Fig. 1B). In this CFP profile, the target peaks were also well separated from each other with a minimum resolution (R s, min ) of 1.1, but a ternary solvent system had to be used for the separation of peaks 18 and 19. Eight peaks (2, 4, 5, 9, 10, 11, 13 and 20) in the HPLC fingerprint were characterized as the same as in UPLC profile. By comparing the obtained data and chromatograms generated from the UPLC and HPLC, the advances in UPLC over HPLC could be summarized as follows. (1) the single running time of UPLC (17 min) was about five times shorter than that of HPLC (70 min) which may be most important for huge batch analysis; (2) a combination of the shortened running time with a smaller flow rate of 0.5 ml/min reduced solvent (mostly harmful to human body) consumption to only 7.5 ml, while solvent

54 M. Liu et al. / J. Chromatogr. A 1157 (2007) 51 55 be caused by the different geographic or cultivating conditions. (See Supplementary data, Fig. S2). 4. Conclusion Fig. 1. Chromatographic fingerprint of Danshen root obtained by (A) UPLC and (B) HPLC. 1 20 were the coexisting peaks: 2, Danshensu; 5, Protocatechuic alhydryde; 6, caffeic acid; 9, rosmarinci acid; 10, lithospermic acid; 11, salvianolic acid B; 13, salvianolic acid A; 20, tanshinoneiia. The chromatograms were detected at 280 nm. usage for a single run in HPLC was up to 70 ml, (3) the peak capacities obtained by UPLC [20] were 87, more efficient than that obtained by HPLC (peak capacities of 65), (4) in the condition described above, the two critical pairs of peaks 9 and 10, 18 and 19 were well separated by UPLC. In summary, the UPLC method had its advantages over HPLC in terms of time-saving, solvent-saving, high performance and high efficiency, and has been successively applied to CFP analysis and quality evaluation of Danshen root. 3.4. Fingerprint analysis of representative Danshen samples using UPLC Eleven Danshen samples collected from different locations were analysed using the established methods and the UPLC fingerprint profiles were obtained. Twenty peaks were found common in all the analysed samples. The normalized peak areas of these 20 co-existing peaks in CFP of Danshen from different sources were summarized in Table 1 and the comparison of the chromatograms were shown in Supplementary data, Fig. S2. The data revealed that the main components and the peaks distribution were much similar among these samples, but the variation of normalization peak area clearly appeared, for instance, the predominant hydrophilic component (peak 11, salvianolic acid B), and the lipophilic component (peak 20, tanshinone IIA) showed their RSD of 20 and 34%, respectively, which might UPLC methods have been developed in recent years for pharmaceutical analysis but not yet utilized in quality control for Chinese herbal medicines until this research. In the present study a UPLC method has been established and validated for the first time in the chemical fingerprint analysis of Danshen root. The advantages of UPLC in the analysis of multi-components in a complicated matrix has been demonstrated, with shorter analysis times, reduced solvent consumption and improved peak resolution compared to conventional HPLC. Meanwhile, an integrated sample preparation method has been established to demonstrate both the hydrophilic and lipophilic components in a single run and obtain perfect chromatographic profiles. In addition, 20 common peaks were detected in all the analysed samples and eight of them were unambiguously identified by comparing their retention times and the on-line UV spectra with the available reference compounds, which will further contribute to the quality assessment of Danshen root from different populations and geographic conditions. Most importantly, the UPLC fingerprinting method established in the present report will serve as a valuable reference for quality evaluation and standardization of Chinese herbal drugs and final products. Acknowledgement The authors gratefully acknowledge the technical support of Mr. Jie Chen in Waters Corporation, Shanghai, China. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.chroma.2007.05.018. References [1] P.S. Xie, S.B. Chen, Y.Z. Liang, X.H. Wang, R.T. Tian, R. Upton, J. Chromatogr. A 1112 (2006) 171. [2] Y.Z. Liang, P.S. Xie, K. Chan, J. Chromatogr. B 812 (2004) 53. [3] J.W. Thompson, J.S. Mellors, J.W. Eschelbach, J.W. Jorgenson, LC GC 24 (2006) 16. [4] L. Novakova, L. Matysova, P. Solich, Talanta 68 (2006) 908. [5] D.T.T. Nguyen, D. Guillarme, S. Rudaz, J.L. Veuthey, J. Chromatogr. A 1128 (2006) 105. [6] A.D. Villiers, F. Lestremau, R. Szucs, S. Gélébart, F. David, P. Sandra, J. Chromatogr. A 1127 (2006) 60. [7] E. Barceló-Barrachina, E. Moyano, M.T. Galceran, J.L. Lliberia, B. Bagó, M.A. Cortes, J. Chromatogr. A 1125 (2006) 195. [8] G. Desmet, J. Chromatogr. A 1116 (2006) 89. [9] C.C. Leandro, P. Hancock, R.J. Fussell, B.J. Keely, J. Chromatogr. A 1103 (2006) 94. [10] Z.H. Dong, J. She, H.Z. Zheng (Eds.), Modern Study of Traditional Chinese Medicine, vol. 2, Xue Yuan Press, Beijing, 2001, p. 1093. [11] X.H. Fan, Y.Y. Cheng, Z.L. Ye, R.C. Lin, Z.Z. Qian, Anal. Chim. Acta 555 (2006) 217. [12] J.L. Zhang, M. Cui, Y. He, J. Pharm. Biomed. Anal. 36 (2005) 1029.

M. Liu et al. / J. Chromatogr. A 1157 (2007) 51 55 55 [13] P. Hu, G. An Luo, Z.Z. Zhao, Z.H. Jiang, Chem. Pharm. Bull. 53 (2005) 705. [14] L.M. Zhou, M. Chow, Z. Zuo, J. Pharm. Biomed. Anal. 41 (2006) 744. [15] D. Yun, Y.N. Pan, W.W. Fu, Chem. Pharm. Bull. 53 (2005) 508. [16] P. Hu, Q.L. Liang, G. An Lou, Z.Z. Zhao, Z.H. Jiang, Chem. Pharm. Bull. 53 (2005) 677. [17] M. Gu, Z.G. Su, J. Chromatogr. A 1022 (2004) 139. [18] Z.H. Shi, J.T. He, T.T. Yao, W.H. Chang, M.P. Zhao, J. Pharm. Biomed. Anal. 37 (2005) 481. [19] P. Hu, G. An Luo, Z.Z. Zhao, Z.H. Jiang, Chem. Pharm. Bull. 53 (2005) 481. [20] S.A.C. Wren, J. Pharm. Biomed. Anal. 38 (2005) 337.