A combination strategy for extraction and isolation of multi-component natural products by systematic two-phase solvent extraction-13C nuclear magnetic resonance pattern recognition and following conical counter-current chromatography separation: Podophyllotoxins and flavonoids from Dysosma versipellis (Hance) as examples

Accepted Manuscript Title: A combination strategy for extraction and isolation of multi-component natural products by systematic two-phase solvent extraction-13 C NMR pattern recognition and following conical counter-current chromatography separation: Podophyllotoxins and flavonoids from Dysosma versipellis (Hance) as examples Author: Zhi Yang Youqian Wu Shihua Wu PII: DOI: Reference:

S0021-9673(15)01872-5 http://dx.doi.org/doi:10.1016/j.chroma.2015.12.074 CHROMA 357176

To appear in:

Journal of Chromatography A

Received date: Revised date: Accepted date:

28-7-2015 23-12-2015 28-12-2015

Please cite this article as: Zhi Yang, Youqian Wu, Shihua Wu, A combination strategy for extraction and isolation of multi-component natural products by systematic two-phase solvent extraction-13C NMR pattern recognition and following conical counter-current chromatography separation: Podophyllotoxins and flavonoids from Dysosma versipellis (Hance) as examples, Journal of Chromatography A http://dx.doi.org/10.1016/j.chroma.2015.12.074 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

A combination strategy for extraction and isolation of multi-component natural products by systematic two-phase solvent extraction-13C NMR pattern recognition and following conical counter-current chromatography separation: Podophyllotoxins and flavonoids from Dysosma versipellis (Hance) as examples

Zhi Yang, Youqian Wu, Shihua Wu*

Research Center of Siyuan Natural Pharmacy and Biotoxicology, College of Life Sciences Zhejiang University, Hangzhou 310058, China

*Corresponding author. Tel: +86-571-88206287; Fax: +86-571-88206287; E-mail: [email protected]

1

Highlights

 The HEMWat systems were used systematically for extraction and separation of NPRs.

 13C-NMR pattern recognition was used to determine the optimum extraction systems.

 A novel conical CCC device was developed for the purification of Dysosma versipellis.

 Elution-extrusion mode was introduced in the conical CCC purification process.

 17 components were purified by one-step conical CCC separation from HEMWat extracts.

2

Abstract Despite of substantial developments of extraction and separation techniques, isolation of natural products from natural resources is still a challenging task. In this work, an efficient strategy for extraction and isolation of multi-component natural products has been successfully developed by combination of systematic two-phase liquid-liquid extraction-13C NMR pattern recognition and following conical counter-current chromatography separation. A small-scale crude sample was first distributed into 9 systematic hexane-ethyl acetate-methanol-water (HEMWat) two-phase solvent systems for determination of the optimum extraction solvents and partition coefficients of the prominent components. Then, the optimized solvent systems were used in succession to enrich the hydrophilic and lipophilic components from the large-scale crude sample. At last, the enriched components samples were further purified by a new conical counter-current chromatography (CCC). Due to the use of 13C NMR pattern recognition, the kinds and structures of major components in the solvent extracts could be predicted. Therefore, the method could collect simultaneously the partition coefficients and the structural information of components in the selected two-phase solvents. As an example, a cytotoxic extract of podophyllotoxins and flavonoids from Dysosma versipellis (Hance) was selected. After the systematic HEMWat system solvent extraction and

13

C NMR pattern recognition

analyses, the crude extract of D. versipellis was first degreased by the upper phase of HEMWat system (9:1:9:1,v/v), and then distributed in the two phases of the system of HEMWat (2:8:2:8, v/v) to obtain the hydrophilic lower phase extract and lipophilic upper phase extract, respectively. These extracts were further separated by conical CCC with the HEMWat systems (1:9:1:9 and 4:6:4:6, v/v). As results, total 17 cytotoxic compounds were isolated and identified. In general, whole results suggested that the strategy was very efficient for the systematic extraction and isolation of biological active components from the complex biomaterials.

Keywords:

13

C NMR-based pattern recognition; Conical counter-current chromatography; Dysosma

versipellis (Hance); Flavonoids; Natural Products isolation; Podophyllotoxins.

3

1. Introduction. Natural products are the most consistently successful source of drug leads [1, 2]. However, natural products resources such as plant and microorganism extracts are very complex and usually contain large numbers of hydrophobic and hydrophilic components. Despite of substantial developments of extraction and separation techniques, it is still a challenging task to get multiple pure targets from complex biological materials including plants, marine organisms or microorganisms [3]. Generally speaking, liquid-liquid extraction is a common and basic step after the primary solvent extraction from the biological materials in the natural product research. For example, hexane, ethyl acetate, n-butanol and water are usually used in succession to distribute the crude sample into the low polar, medium polar, polar and highly polar parts, respectively. However, there are usually overlapping components in the different extracted parts because of low extraction efficiency and similar properties of extraction solvents. In addition, the structures of the extracted components are usually not pre-determined due to the lack of standards. Therefore, it is favorable to obtain the ideal extracted parts with less overlapping components but pre-known structural types. Recently, NMR pattern recognition technique has been proved to be an efficient approach to identify metabolites [4, 5]. Jane Hubert and her collaborators demonstrated that

13

C-NMR pattern

recognition techniques together with centrifugal partition chromatographic separation could identify some prominent known metabolites [5, 6]. We also found that by combination of reversed performance liquid chromatography and

13

C-NMR pattern recognition, we could not only identify the known

metabolites, but also characterize some new metabolites[7]. Thus it seems possible to use the 13C-NMR pattern recognition to distinguish the kinds and structures of major components in the extracted parts after systematic solvent extraction. Compared with the traditional binary solvent extraction systems, such as ethyl acetate-water and n-butanol-water, the quaternary solvent systems composed of n-hexane-ethyl acetate-methanol-water (HEMWat) can provide various systematic formulations with wide range of polarity, thus HEMWat systems are widely applied as two-phase solvent systems for the separation of some targets by counter-current chromatography (CCC) [8-11]. However, there are few studies to focus on the extraction using the quaternary systems for the multi-component sample [12]. Therefore, the first aim of this work is to investigate the extraction efficiency by the systematic HEMWat map systems for multi-component targets. It has been known that CCC is a unique liquid-liquid partition chromatography without support matrix, which relies on the continuous partition of one sample between two immiscible solvents to 4

achieve separation [13, 14]. Due to lack of solid support matrix, CCC eliminates several complications resulting from the solid support matrix, such as irreversible adsorption and denaturation of samples. In addition, compared with the conventional chromatography, CCC is an economic instrument with the capability of high mass loading and total recovery of the injected samples [15]. Thus CCC has gained more and more popularity in natural products separation [13, 16-19]. There are numbers of CCC devices and methods developed for the isolation of natural products [13, 19-22]. Recently, we have developed a conical CCC for natural product isolation [23]. In the conical CCC, the tubes holder hub is conical instead of cylindrical, resulting in an extra centrifugal gradient, which makes the mobile phase move faster and enables CCC a much higher retention of stationary phase. In the previous conical CCC [23], the tube was winded from small to large diameter of the conical hub then from large to small diameter in turn from inner to outer layer, thus link mode of layer to layer was not the same. In this work, we introduced a new tube design for the conical CCC by changing the tube winding to form the same mode of layer to layer. Therefore, the second aim of this work was to use the new conical CCC device for the separation of multiple targets in the extracts after comprehensive liquid-liquid extraction. As an example, the crude ethanol extract of Dysosma versipellis (Hance) was selected, which is a well-known traditional Chinese medicine. It has been used as a general remedy for the treatment of snake bite, weakness, condyloma accuminata, lymphadenopathy and tumors for thousands of years. So far, there are a number of podophyllotoxins and flavonoids isolated and identified from this plant and other plant of podophyllum taxa [24-28] by multiple chromatographic techniques. However, there are only a few components isolated by CCC methods[23, 29] Therefore, the purpose of this work is to establish an efficient strategy for extraction and separation of multi-component natural products by combination of quaternary HEMWat solvent extraction,

13

C NMR pattern recognition and following new conical CCC purification. The results

demonstrated that the strategy was very efficient for the extraction and separation of bioactive podophyllotoxins and flavonoids from D. versipellis (Hance). To the best of our knowledge, this is the first report to use HEMWat system and conical CCC for the extraction and separation of multiple components from D. versipellis (Hance). 2 Experimental 2.1 Reagent and materials 5

Organic solvents for the CCC separation, including n-hexane, ethyl acetate and methanol were of analytical grades (Sinopharm Chemical Reagent Co., Shanghai, China). Methanol used for HPLC was of chromatographic grade (Merck, Darmstadt, Germany). The water was purified by means of a water purifier (18.2 MΩ, Wanjie Water Treatment Equipment, Hangzhou, China). The dry roots of D. versipellis (Hance) were bought from a drug market in Bozhou (Anhui, China). The species was identified by Institutes of Plant Sciences, College of Life Sciences, Zhejiang University, China.

2.2 Preparation of crude sample The dry roots of D. versipellis (Hance) were ground to a fine powder. The powder (2 Kg) was then extracted three times, each time with 4 L 95% ethanol at boiling point for 2 h. The extracts were combined and evaporated to dryness at 45°C under reduced pressure, and 420 g residue was obtained.

2.3 HEMWat solvent extraction 500 mg of the crude was dissolved by 9 linear HEMWat solvent systems (1:9:1:9-9:1:9:1, v/v) [11], and every system contained 50 mL upper phase and 50 mL lower phase. After dissolved completely, 100 μL of the upper phase and 100 μL of the lower phase were subjected to HPLC analysis. The remaining two immiscible phases were separated and lyophilized.

2.4 HPLC analysis of the crude and extracts of HEMWat systems HPLC analysis was performed on an Agilent 1100 system, equipped with a G1379A degasser, a G1311A Quat Pump, a G1367A Wpals, a G1316A column oven, a G1315B diode assay detector and an Agilent ChemStation. The column used was a RP SB-C18 (4.6 mm id. × 250 mm length, 5μm). The flow rate was 0.8 mL/min and the column temperature was 30°C. The injection volume was 10 μL. Methanol-0.1% trifluoroacetic acid (TFA) aqueous solution system was used as the mobile phases in a gradient mode as follows: 0-5 min, methanol from 10% to 30%; 5-35 min, methanol from 30% to 70%; 35-45 min, methanol from 70% to 100%; 45-50 min, methanol from 100% to 10%; 50-55 min, methanol was maintained at 10%. The effluent was monitored by a diode assay detector at 254 nm.

2.5 NMR analysis of solvent extracts and data processing All samples were analyzed with the same acquisition and processing parameters. Every extract (S1-S18) of 20 mg was dissolved in 500 μL DMSO-d6 and in every sample 2.5 mg trimesic acid was 6

added. NMR analysis was performed at 298 K on a Bruker DMX-500 spectrometer equipped with cooled 1H, 13C, and two-dimensional (2D) coils and preamplifiers. The 13C NMR spectra were acquired at 125 MHz. A standard zgpg pulse sequence was used with an acquisition time of 0.865 s and a relaxation delay of 1.5 s. For each sample, 1448 scans were co-added to obtain a satisfactory signal-to-noise (S/N) ratio. The spectral width was 300 ppm, from -40 ppm to 260 ppm. The spectra data were handled with the ACD/NMR processor Academic Edition (ACD/Labs Release 12.0, Ontario, Canada). The central resonance of DMSO-d6 was calibrated at δ 39.51 ppm. Positive 13C peak signals were collected automatically with a minimum intensity threshold of 0.003 and exported to the professional software Similarity Evaluation System for Chromatographic Fingerprint of Traditional Chinese Medicine (Version 2004A) for calibration the 13C NMR shifts of all fractions. The 13C NMR spectrum of trimesic acid was selected as the reference spectrum with chemical shift window width of 0.2 ppm, multi-point calibration was operated manually and then the match was performed automatically.

Then

the

calibrated

13

C

NMR

data

were

(http://www.broadinstitute.org/cancer/software/GENE-E/index.html,

imported

Broad

into

Institute)

for

GENE-E pattern

recognition. Hierarchical clustering analysis was directly applied on raw 13C peak intensity values. The classification was performed on the rows and columns. The Eucidian distance was used to measure the proximity between samples, and the One minus pearson correlation was performed to agglomerate the 13

C NMR data. The resulting 13C NMR chemical shifts clusters were visualized as dendrograms on a

2D heat map. The deeper the red color in the map, the higher the relatively intensity of 13C signals.

2.6 New conical counter-current chromatography isolation of extracts The CCC separation was performed on a conical CCC device with new tube winding. Its major machine fabrication except for the tube winding mode was the same with previous conical CCC designed by our groups [23]. In contrast to traditional CCC, the tubes holder hub of conical CCC was conical instead of cylindrical. As shown in Fig. 1, the apparatus hold three tapered holders with three upright conical coil columns connected in series, and the CCC column was constructed by winding parallel polytetrafluoroethylene (PTFE, 4.5 mm id. and 1 mm wall thickness) tubes onto the holder hub in head-to-tail. In the previous conical CCC [23], the tube was winded from small to large diameter of the tapered holder hub then from large to small diameter in turn from inner to outer layer, thus link mode of layer to layer was not the same. In this work, all multi-layer coils tubes were winded in the same way, when one layer was fulfilled the coil was drawn back to the initial position and the next 7

winding process was continued. In this way, the arrangement of tube coils of every layer became completely the same, resulting in the more uniform of the action of the biphase in the tubes. The capacity of columns was 570 mL and the volume of inject loop was 40 mL. The revolution radius was 10 cm. The β values of the multi-layer coils varied from 0.5 at the internal terminal of vertex of the cone to 0.7 at the external terminal of cone bottom (β= r/R where r was the distance from the coil to the holder shaft and R was the revolution radius or the distance between the holder axis and central axis of the centrifuge). The revolution speed was regulated by a speed controller ranging from 0 to 1500 rpm. In addition, the CCC apparatus was equipped with a P270 gradient metering pump , a UV 230 + variable-wavelength spectrometer, an EC2000 ChemStations (Elite Analytical Instrument Co., Ltd., Dalian, China), a BSZ-100 fraction collector and a six-port medium pressure injection valve (V-541, IDEX Health &Science LLC, Oak Harbor, WA, USA). The selected two-phase solvent systems were thoroughly equilibrated in a separation funnel by repeated vigorous shaking at room temperature. The two phases were separated and then used for the CCC separation. The upper phase was used as the stationary phase while the lower phase as the mobile phase. Due to that the addition of crude sample made two-phase systems get more emulsified and had longer settling time, a relative low rotation speed was selected [30]. During the separations, the CCC column was first filled with the upper phase as the stationary phase at a flow rate of 40 mL/min for 15 min. Then the apparatus was started at 360 rpm, and then the lower phase was pumped into the column at a flow rate of 5 mL/min from the head end to the tail end. After the mobile phase front emerged and the hydrodynamic equilibrium was established, sample solution was injected into the column through the injection valve. The effluent was monitored with a UV detector at 254 nm and collected with a fraction collector.

2.7 Identification and characterization of isolated compounds by conical CCC The compounds separated by conical CCC were identified by MS and NMR spectrum. Positive and negative ESI-MSn analyses were performed using a Thermo Finnigan LCQ Deca XP ESI-MS. NMR experiments were carried out using a Bruker Advanced DMX 500 NMR spectrometer with DMSO-d6 as solvent and tetramethylsilane as the internal standard. 8

3 Results and discussion 3.1 HPLC analyses of crude sample It’s well known that the main components of D. versipellis were podophyllotoxins and flavonoids [26, 31]. HPLC analyses (Fig. 2A) indicated that the crude ethanol extract of the roots of D. versipellis (Hance) contained at least 17 major components among more than 50 peak components. Although several chromatographic methods, such as liquid chromatography [32, 33], thin-layer chromatography [34], electrochromatography [35, 36] and counter-current chromatography (CCC) [31, 37] have been explored to analyze and isolate these components, it still lacks an efficient method for preparative separation of the full components.

3.2 Systematic HEMWat solvent extractions and fingerprinting analyses As shown in Fig. 2A, there are numbers of components in the crude ethanol extracts of D. versipellis. In spite of well resolution in analytical-scale analyses, it is still some difficult in the preparative-scale chromatographic isolation of these compounds because of different molecular proprieties including polarity. Thus, it is essential to extract the targets before further purification. Because the extracts would be purified by CCC using the HEMWat systems, here we used 9 HEMWat systems (Table 1) to extract the targets, which had been widely used for natural product isolation [8-10]. As shown in Table 1, the product mass extracted by upper phase decreased with the decrease of the volume ratio of ethyl acetate while the mass extracted by lower phase increased with the increase of the methanol volume ratio in the systems. In addition, we found that the crude extract was easily dissolved in the phases with higher ratio of ethyl acetate or methanol. Thus, in order to fully dissolve the sample, we used a large volume phase (50 mL upper phase and 50 mL lower phase) to extract the small-scale sample (500 mg). After extraction by HEMWat systems, the extracted products were further analyzed by HPLC. Besides the differences of mass (Table 1), the distributions of components in these systems were very different. As shown in Fig. 3, there were minor components in the low-polar hexane-based upper phases such as the upper phases of 9:1:9:1, 8:2:8:2 and 7:3:7:3, while large components distributed in the other lower phases or ethyl acetate-based upper phases such as 1:9:1:9 and 2:8:2:8. It should be pointed out that the components in some ethyl acetate-based two-phase systems such as 1:9:1:9, 2:8:2:8, and 3:7:3:7 could be well divided the extracts into two relative separated samples. 9

3.3 Pre-determination of major components by 13C NMR pattern recognition As described above, the extracts of D. versipellis contained a large number of compounds including podophyllotoxins and flavonoids. Some of them have similar structures, resulting in close or even overlapped

13

C NMR shifts. Thus trimesic acid (2.5 mg) was added in each sample (20 mg

dissolved in 500 μL DMSO-d6) for calibration of the shifts of components and better dereplication of the 13C NMR peaks. It might also help to determine quantitatively the contents of components in every sample. As shown in Fig. S1, besides the seven prominent signals of DMSO-d6 there were another three prominent signals at 166.11, 133.57 and 132.21 belonged to trimesic acid. Due to no overlapping signals, the trimesic acid was suitable as internal standard for shifts calibration and quantitative determination of components. When

13

C NMR shifts of all fractions were calibrated by Similarity

Evaluation System for Chromatographic Fingerprint of Traditional Chinese Medicine (Version 2004A), a two-dimensional table (Table S1) containing 18 columns (samples) and 257 rows (13C signals) was yielded. The table was then subjected to the hierarchical clustering analysis using the software of GENE-E. As shown in Fig. 4, there were several obvious clusters in the 2D heat map belonged to three main blocks (i-iii). In block (i), most of the chemical shifts were among 10-30 ppm in high magnetic field, and there were several chemical shifts between 126-130 and at 174 ppm, suggesting that the dominating components contained in those extracts were unsaturated grease. In block (ii), the primary components were recognized as mixtures of podophyllotoxins based on the characterized

13

C NMR

13

shifts of podophyllotoxins and C NMR data in the reported literatures. In block (iii), most of the 13C shifts were between 60-83 ppm and some were at about 100 ppm, which implied that the main compounds in the corresponding products were saccharides and/or glycosides. Although the major type of components in the extracts could be proposed according to the clustered 13C NMR shifts, the single component could not be identified from the dereplication of 13C NMR data because the components in the HEMWat extracts were still complex. If the selection of the used solvent system is high enough, or using some pre-purification methods such as centrifugal partition chromatography (CPC) [5, 6] and reversed-phase liquid chromatography (RPLC) [7], the 13C shifts may be possibly clustered into each sub-clusters for each components.

3.4 Determination of the optimum solvent systems for extraction and separation of the targets Although HPLC fingerprinting analyses (Fig. 3) significantly illustrated many different 10

components in the samples extracted by HEMWat two-phases, the

13

C NMR map (Fig. 4) visually

demonstrated that there were very significant differences in the extracted components. For example, in the samples S14, S16 and S18, there were almost no any podophyllotoxins signals but only the signals of low polar lipids (Fig. 4, block i), which suggested that the upper phase of 7:3:7:3, 8:2:8:2, and 9:1:9:1 could be used to degrease from the crude extract. In addition, 13C signals of samples S1, S3 and S5 were very different from those of the samples S2, S4 and S6 (Fig. 4, block ii and iii), implying the systems of 1:9:1:9, 2:8:2:8 and 3:7:3:7 were suitable for extraction with less overlapping components. The further K values analyses also showed similar results. As shown in Table 2, the K values of compounds 1,2,3,7 were less than other compounds in the several HEMWat systems, implying that these compounds were hydrophilic while other compounds were lipophilic. As is well known, the extraction is different from CCC separation. The ideal extract requires a larger or less distribution ratio (K). For example, K>10, the target in upper phase is far more than that in lower phase; K<0.1, the target in lower phase is far more than that of upper phase. Hence, the HEMWat systems of 2:8:2:8 and 3:7:3:7 were suitable to extract the prominent components. Due to that there were more lipophilic components (Fig. 2), the system of 2:8:2:8 was selected as extraction solvent system, which could also distribute the crude ethanol extract of D. versipellis into two relative separated parts (Fig. 2B, C and Fig. 4). Therefore, the crude ethanol extract of D. versipellis was first degreased by the system of 9:1:9:1, and then distributed into two parts of hydrophilic and lipophilic extracts by the two-phase HEMWat system of 2:8:2:8, which were further isolated by the conical CCC. Generally speaking, successful CCC separation largely depends on the correct selection of twophase solvent system. Usually, the ideal sweet spots are 0.45) in the selected solvent system of 4:6:4:6, an elution-extrusion mode was adopted.

3.5 Preparative conical CCC separations As shown in Fig. 5, the hydrophilic extract was isolated by the conical CCC using the system of 1:9:1:9, and all the targeted components were separated with an elution time about 230 min. Peaks 1, 3 and 5 were all well purified with the purity more than 90%. Due to the close K values of peaks 2 and 7, 11

the two compounds were overlapped in the separation. Then the lipophilic extract was isolated by the conical CCC with an elution-extrusion mode using the system of 4:6:4:6. As shown in Fig. 6A, the upper phase was used as stationary phase and the lower phase as mobile phase. When the compounds 6, 4, 11, 8, 10 were eluted with lower phase for 280 min, the conical CCC device kept the same rotation speed (360rpm) but it altered to pump the upper phase as mobile phase. The components 9 and 12 were eluted with sweep elution of lower phase retained in the column while other compounds including 13, 14, 15, 16 and 17 were extruded by upper phase [41]. As shown in Fig.6B, the purity of most of compounds was satisfactory except for the components 8, 13, 14 and 15. The compounds purified by CCC were all first analyzed by HPLC, and the results were listed in Table 3. Compounds with unsatisfied purity were further purified. As shown in Fig. S2, 1H NMR spectrum showed that the purities of all obtained components were satisfactory. It's worth noting that although the components 4 and 5, 12 and 13 had same retention time in the HPLC analyses (Fig. 2), they could be well resolved by conical CCC (Fig. 5, and 6) because of the different separation mechanisms. Thus, conical CCC is an efficient separation technique and may be a useful alternative choice for users.

3.6 Identification of the separated components The structures of the separated compounds were determined by MS, MS n, and 1D and 2D NMR. As shown in Fig. 7, the structures of 17 components purified by conical CCC were identified as α-peltatin glucoside (1), quercetin-3-O-glucopyranoside (2), kaemferol-3-O-glucopyranoside (3), α-peltatin (4), β-peltatin glucoside (5), 4′-demethylpodophyllotoxin (6), podophyllotoxin glucoside (7), quercetin (8), isopicropodophyllxone (9), β-peltatin (10), podophyllotoxin (11), kaemferol (12), podophyllotoxone (13), podoverine A (14), podoverine F (15), podoverine D (16), podoverine E (17). Their ESI-MS, ESI-MS/MS data were summarized in Table 4, and 1H and 13C NMR data were listed in supporting material and Table S2-S4, respectively.

4. Conclusion In this work, the two-phase HEMWat solvent systems were systematically used as the solvent systems both for selective extraction and CCC separations of D. versipellis (Hance). The results indicated that

13

C NMR pattern recognition was an efficient visual technique to characterize the

metabolites and distinguish the solvent systems extracts. Although the simple HEMWat solvent extraction could not well resolve the components into each single compound,

13

C NMR pattern 12

recognition together the known structural data could still pre-determine the types of the prominent components in the samples (Fig. 4). Especially,

13

C NMR map (Fig. 4) could clearly illustrate the

quantitative and qualitative difference of compounds in the samples, which was very important for the selection of solvent system for extraction and separation. Once the optimum extraction and separation solvent systems were selected, the crude sample were extracted and further separated by the new conical CCC. As a result, 17 components were well separated and identified. In summary, this work established a new combination strategy for systematic extraction and separation of complex natural products. As shown in Fig. 8, the whole process included three main parts: (1) Preparation of crude samples; (2) Systematic HEMWat solvent systems extraction and

13

C

NMR pattern recognition; (3) Targeted conical CCC separations. It combines liquid-liquid extraction, 13

C NMR-based pattern recognition and following targeted conical counter-current chromatography

separation and will be useful for the selective extraction and targeted separation of natural products.

Acknowledgement We would like to thanks Prof. Shuqun Zhang for valuable suggestions to improve the manuscript. This work was supported in part by National Natural Science Foundation of China (grant no.: 21272209, 20972136).

Appendix A. Supplementary data. Supplementary data associated with this article can be found in the online version.

13

References [1] A.L. Harvey, R. Edrada-Ebel, R.J. Quinn, The re-emergence of natural products for drug discovery in the genomics era. Nat. Rev. Drug Discov. 14 (2015) 111-129. [2] J.W.H. Li, J.C. Vederas, Drug discovery and natural products: end of an era or an endless frontier? Science 325 (2009) 161-165. [3] F. Bucar, A. Wube, M. Schmid, Natural product isolation - how to get from biological material to pure compounds. Nat. Prod. Rep. 30 (2013) 525-545. [4] A. Dubey, A. Rangarajan, D. Pal, H.S. Atreya, Pattern recognition-based approach for identifying metabolites in nuclear magnetic resonance-based metabolomics. Anal. Chem. 87 (2015) 7148-7155. [5] J. Hubert, J.M. Nuzillard, S. Purson, M. Hamzaoui, N. Borie, R. Reynaud, J.H. Renault, Identification of natural metabolites in mixture: a pattern recognition strategy based on C-13 NMR. Anal. Chem. 86 (2014) 2955-2962. [6] S.K. Oettl, J. Hubert, J.M. Nuzillard, H. Stuppner, J.H. Renault, J.M. Rollinger, Dereplication of depsides from the lichen Pseudevernia furfuracea by centrifugal partition chromatography combined to C-13 nuclear magnetic resonance pattern recognition. Anal. Chim. Acta 846 (2014) 60-67. [7] Zhi Yang, Youqian Wu, Hui Zhou, Xiaoji Cao, Xinhang Jiang, Kuiwu Wang, S. Wu, A novel strategy for screening new natural products by combination of reversed-phase liquid chromatography fractionation and 13C NMR pattern recognition: the discovery of new anti-cancer flavone dimers from Dysosma versipellis (Hance). RSC Adv. 5 (2015) 77553-77564. [8] J.B. Friesen, G.F. Pauli, GUESS - A generally useful estimate of solvent systems in CCC. J. Liq. Chromatogr. R. T. 28 (2005) 2777-2806. [9] S. Dubant, B. Mathews, P. Higginson, R. Crook, M. Snowden, J. Mitchell, Practical solvent system selection for counter-current separation of pharmaceutical compounds. J. Chromatogr. A 1207 (2008) 190-192. [10] A. Frey, E. Hopmann, M. Minceva, Selection of Biphasic Liquid Systems in Liquid-Liquid Chromatography Using Predictive Thermodynamic Models. Chem. Eng. Technol. 37 (2014) 1663-1674. [11] J. Liang, J. Meng, D. Wu, M. Guo, S. Wu, A novel 9*9 map-based solvent selection strategy for targeted counter-current chromatography isolation of natural products. J. Chromatogr. A 1400 (2015) 27-39. [12] R.J. Case, Y.H. Wang, S.G. Franzblau, D.D. Soejarto, L. Matainaho, P. Piskaut, G.F. Pauli, Advanced applications of counter-current chromatography in the isolation of anti-tuberculosis constituents from Dracaena angustifolia. J.Chromatogr. A 1151 (2007) 169-174. 14

[13] G.F. Pauli, S.M. Pro, J.B. Friesen, Countercurrent separation of natural products. J. Nat. Prod. 71 (2008) 1489-1508. [14] A. Berthod, S. Ignatova, I.A. Sutherland, Advantages of a small-volume counter-current chromatography column. J. Chromatogr. A 1216 (2009) 4169-4175. [15] M. Gu, O.Y. Fan, Z.G. Su, Comparison of high-speed counter-current chromatography and high-performance liquid chromatography on fingerprinting of Chinese traditional medicine. J. Chromatogr. A 1022 (2004) 139-144. [16] A.P. Foucault, L. Chevolot, Counter-current chromatography: instrumentation, solvent selection and some recent applications to natural product purification. J. Chromatogr. A 808 (1998) 3-22. [17] A. Marston, K. Hostettmann, Developments in the application of counter-current chromatography to plant analysis. J. Chromatogr. A 1112 (2006) 181-194. [18] I.A. Sutherland, D. Fisher, Role of counter-current chromatography in the modernisation of Chinese herbal medicines. J. Chromatogr.A 1216 (2009) 740-753. [19] J.B. Friesen, J.B. McAlpine, S.-N. Chen, G.F. Pauli, Countercurrent Separation of Natural Products: An Update. J. Nat. Prod. (2015). [20] R.L. Hu, Y.J. Pan, Recent trends in counter-current chromatography. Trac-Trends in Anal. Chem. 40 (2012) 15-27. [21] S. Wu, J. Liang, Counter-current chromatography for high throughput analysis of natural products. Comb. Chem. High T. Scr. 13 (2010) 932-942. [22] L.H. Yin, Y.N. Li, B.N. Lu, Y.J. Jia, J.Y. Peng, Trends in counter-current chromatography: applications to natural products purification. Sep. Purif. Rev. 39 (2010) 33-62. [23] J. Liang, J. Meng, M. Guo, Z. Yang, S. Wu, Conical coils counter-current chromatography for preparative isolation and purification of tanshinones from Salvia miltiorrhiza Bunge. J. Chromatogr. A 1288 (2013) 35-39. [24] J.V. Marques, K.-W. Kim, C. Lee, M.A. Costa, G.D. May, J.A. Crow, L.B. Davin, N.G. Lewis, Next generation sequencing in predicting gene function in podophyllotoxin biosynthesis. J. Bio. Chem. 288 (2013) 466-479. [25] Y.-Q. Liu, J. Tian, K. Qian, X.-B. Zhao, S.L. Morris-Natschke, L. Yang, X. Nan, X. Tian, K.-H. Lee, Recent progress on C-4-modified podophyllotoxin analogs as potent antitumor agents. Med. Res. Rev. 35 (2015) 1-62. [26] X. Xu, X. Gao, L. Jin, Antiproliferation and cell apoptosis inducing bioactivities of constituents from Dysosma versipellis in PC3 and Bcap-37 cell lines. Cell Div. 6 (2011) 14. [27] Y. Sun, Z. Li, H. Chen, W. Zhou, H. Hua, Advances in studies on chemical constituents of podophyllum taxa and their bioactivities. Chinese Trad. Herb Drugs 43 (2012) 1626-1634. [28] Y.Q. Liu, L. Yang, X. Tian, Podophyllotoxin: current perspectives. Curr. Bioactive Compounds 3 (2007) 37-66. 15

[29] H. Zhou, J. Liang, D. Lv, Y. Hu, Y. Zhu, J. Si, S. Wu, Characterization of phenolics of Sarcandra glabra by non-targeted high-performance liquid chromatography fingerprinting and following targeted electrospray ionisation tandem mass spectrometry/time-of-flight mass spectrometry analyses. Food Chem. 138 (2013) 2390-2398. [30] S.H. Wu, Y.P. Tai, Y.J. Pan, C.R. Sun, Effect of gravitational force on type-J counter-current chromatography by mathematical analysis. J. Chromatogr. A 1103 (2006) 243-247. [31] Z. Yang, X.M. Liu, K.W. Wang, X.J. Cao, S.H. Wu, Novel linear and step-gradient counter-current chromatography for bio-guided isolation and purification of cytotoxic podophyllotoxins from Dysosma versipellis (Hance). J. Sep. Sci. 36 (2013) 1022-1028. [32] C.L. Liu, B.H. Jiao, LC determination of podophyllum lignans and flavonoids in Podophylium emodi Wall.var.chinesis Sprague. Chromatographia 64 (2006) 603-607. [33] L. Zhao, X. Tian, P.C. Fan, Y.J. Zhan, D.W. Shen, Y. Jin, Separation, determination and identification of the diastereoisomers of podophyllotoxin and its esters by high-performance liquid chromatography/tandem mass spectrometry. J. Chromatogr. A 1210 (2008) 168-177. [34] N. Mishra, A.P. Gupta, B. Singh, V.K. Kaul, P.S. Ahuja, A rapid determination of podophyllotoxin in podophyllum hexandrum by reverse phase high performance thin layer chromatography. J. Liq. Chromatogr. R. T. 28 (2005) 677-691. [35] M. Ganzera, R.M. Moraes, I.A. Khan, Separation of Podophyllum lignans by micellar electrokinetic capillary chromatography (MECC). Chromatographia 49 (1999) 552-556. [36] S.H. Liu, X. Tian, X.G. Chen, Z.D. Hu, Micellar electrokinetic capillary chromatographic separation of diastereoisomers of podophyllum lignans at the C4 position. Chromatographia 56 (2002) 687-691. [37] W. Ping, L. Yongling, C. Tao, X. Wenhua, Y. Jinmao, L. Yongjun, L. Yulin, One-step separation and purification of three lignans and one favonol from Sinopodophyllum emodi by mdium-pressure liquid chromatography and high-speed counter-current chromatography. Phytochem. Analysis 24 (2013) 603-607. [38] A. Berthod, M.J. Ruiz-Angel, S. Carda-Broch, Elution-extrusion countercurrent chromatography. Use of the liquid nature of the stationary phase to extend the hydrophobicity window. Anal. Chem. 75 (2003) 5886-5894. [39] A. Berthod, J.B. Friesen, T. Inui, G.F. Pauli, Elution-extrusion countercurrent chromatography: Theory and concepts in metabolic analysis. Anal. Chem. 79 (2007) 3371-3382. [40] J.B. Friesent, G.F. Pauli, Performance characteristics of countercurrent separation in analysis of natural products of agricultural significance. J. Agr. Food Chem. 56 (2008) 19-28. [41] D. Wu, X. Cao, S. Wu, Overlapping elution-extrusion counter-current chromatography: A novel method for efficient 16

purification of natural cytotoxic andrographolides from Andrographis paniculata. J. Chromatogr. A 1223 (2012) 53-63.

Fig. 1. The new conical CCC device. (A) The design principle with three upright conical coils columns connected in series. (B) The photographs of holders and a part of coils. (C) The schematic tube winding mode.

17

Fig. 2. HPLC analysis of the crude and solvent extraction products for targeted separations. (A) The crude ethanol extract of Dysosma versipellis, (B) and (C) the products extracted by HEMWat (2:8:2:8, 18

v/v), (B) hydrophilic lower phase extract, (C) lipophilic upper phase extract. The column used was a RP SB-C 18 column (4.6 mm id×250 mm, 5μm). The flow rate was 0.8 mL/min and the column temperature was 30°C. The injection volume was 10 μL. Methanol-0.1% TFA aqueous solution system was used as the mobile phases in a gradient mode as follows: 0-5 min, methanol from 10% to 30%; 5-35 min, methanol from 30% to 70%; 35-45 min, methanol from 70% to 100%; 45-50 min, methanol from 100% to 10%; 50-55 min, methanol was maintained at 10%. The effluent was monitored by a diode assay detector at 254 nm.

19

Fig. 3. HPLC Fingerprinting analyses of HEMWat solvent extracts. 9 different HEMWat solvent systems (1:9:1:9-9:1:9:1, v/v)(Table 1) were used to extract the components of Dysosma versipellis, and then the upper phase and lower phase were separated, yielding 18 extracts.

20

Fig. 4.

13

C NMR map of HEMWat extracts. The

13

C NMR chemical shifts were clustered by

hierarchical clustering analysis using software of GENE-E and three blocks were obtained, 21

representing three different kinds of components. (i) grease, (ii) podophyllotoxins and flavonoids, (iii) saccharides and/or glucosides.

Fig. 5. (A) The CCC profile and (B) HPLC analyses for the separation of the hydrophilic extracts. (A) CCC conditions: the solvent system, HEMWat (1:9:1:9, v/v); upper phase as stationary phase and lower phase as mobile phase; the retention of the stationary phase, 64.4%; flow rate, 5 mL/min; rotation speed, 360 rpm; detection wavelength, 254 nm; sample, 500 mg was dissolved in a mixed solution composed of 5 mL upper phase and 5 mL lower phase. (B) HPLC conditions were same to Fig.2. 22

Fig. 6. (A) The CCC profile and (B) HPLC analyses for the separation of the lipophilic extracts. (A) CCC conditions: the solvent system, HEMWat (4:6:4:6, v/v); upper phase as stationary phase and 23

lower phase as mobile phase; the retention of the stationary phase, 64.7%; flow rate, 5 mL/min; rotation speed, 360 rpm; detection wavelength, 254 nm; sample, 300 mg was dissolved in a mixed solution composed of 4 mL upper phase and 4 mL lower phase; An elution-extrusion mode was used, and the switch time of extrusion was set at 280 min. (B) HPLC conditions were same to Fig.2.

Fig. 7. Structures of the compounds separated by the conical CCC. The compounds in the top panel were separated in Fig. 5 from hydrophilic extract, and the compounds in the down panel were isolated in Fig. 6 from lipophilic extract by the conical CCC with elution-extrusion mode.

24

Fig. 8. A general protocol for the extraction and separation of complex natural product by combination of selective HEMWat solvent extraction-13C NMR-based pattern recognition and following targeted conical counter-current chromatography separation.

25

Table 1. The two-phase solvent systerms and extracted yield ratio for the extraction of the crude sample. System compostion (v/v)

Lower phase

Upper phase

Sample No.

Extracted ratio (%w)

Sample No.

Extracted ratio (%w)

9

S1

49.3

S2

50.7

2

8

S3

55

S4

45

7

3

7

S5

57.5

S6

42.5

4

6

4

6

S7

70.4

S8

29.6

E

5

5

5

5

S9

85

S10

15

F

6

4

6

4

S11

85.4

S12

14.6

G

7

3

7

3

S13

86.2

S14

13.8

H

8

2

8

2

S15

89.5

S16

10.5

I

9

1

9

1

S17

91.3

S18

8.7

Label n-hexane

Ethyl acetate

Methanol Water

A

1

9

1

B

2

8

C

3

D

26

Table 2. K values of the prominent components in the selected HEMWat solvent system. K values in the HEMWat (v/v) systems Components 1 2 3 4∕5* 6 7 8 9 10 11 12∕ 13* 14 15 16 17

1919

2828

3737

4646

5555

0.17 0.78 1.33 1.00 18.31 0.61 / / / / / / / / /

0.07 0.19 0.46 0.72 10.1 0.27 / / 25.18 21.04 / / / / /

0.04 0.07 0.19 0.43 1.70 0.10 10.59 13.81 9.90 7.02 32.56 35.58 / 46.77 /

/ 0.01 0.05 0.14 0.35 0.05 1.50 3.26 1.65 1.26 4.89 8.56 6.59 12.51 16.09

/ / / 0.02 0.04 / 0.12 0.28 0.27 0.2 0.49 0.69 0.28 1.62 2.74

* In the HPLC analyese, these components had same retention time, thus the K values were possible inaccurate for the single compounds although the following CCC may resolve the components.

27

Table 3. Yields and purity of the compounds separated by the new conical CCC Compounds obtained in Fig. 6 Peaks No.

Compounds Obtained in Fig. 5

weight(mg)

purity(%)

Peaks No.

weight(mg)

purity(%)

6

2.9

72.2

1

15.6

90.6

4

42.3

91.8

5

45.5

87.6

11

55.1

97

2, 7

25.1

42, 50

8, 10

62.3

43, 54

3

5.9

85

10

37.4

93.7

9

6.7

68.3

12

10.2

94.9

13, 14, 15

12.3

15, 23, 32

16

3.4

76

17

4.2

72

28

Table 4. ESI-MS/MS analyses of purified compounds from Dysosma versipellis.

Compounds 1 2

Compounds

22.76

α-peltatin glucoside

23.06

ESI-MS

Tentative assignment

Rt. (mins)

quercetin-3-O-glucopyranoside

Molecular

M.W.

Fomular

(Da)

C27H30O13

562

C21H20O12

464

ESI-MS/MS

Ion formular

m/z

ESI-MS/MS m/z

[M+Na]+

585

[585]: 423(50)

[M-H]

-

463

[463]: 301(100)

-

447

[447]: 327(20), 284(100)

3

26.08

kaemferol-3-O-glucopyranoside

C21H20O11

448

[M-H]

4

27.63

α-peltatin

C21H20O8

400

[M-H]-

399

[399]: 384 (90), 371 (100), 356 (50)

-

575

5

27.63

β-peltatin glucoside

C28H32O13

576

[M-H]

6

28.63

4'-demethylpodophyllotoxin

C21H20O8

400

[M-H]-

399

[575]: 426(20), 413(100), 368(50), 336(30), 228(65) [399]: 384(100), 371(10)

-

575

[575]: 413(55), 395(100), 383(60)

7

29.44

podophyllotoxin glucoside

C28H32O13

576

[M-H]

8

30.17

quercetin

C15H10O7

302

[M+H]+

303

9

31.46

isopicropodophyllxone

C22H20O8

412

[M-H]-

411

10

32.49

β-peltatin

C22H22O8

414

[M-H]-

413

[303]: 285(70), 274(20), 257(100), 247(30), 229(80), 165(25) [411]: 396(55), 383(25), 366(20), 352(100), 337(50), 321(35) [413]: 398(100), 385(95), 370(40), 355(20)

-

413

[413]: 398(95), 385(100), 369(35)

11

32.81

podophyllotoxin

C22H22O8

414

[M-H]

12

34.2

kaemferol

C15H10O6

286

[M-H]-

285

[285]: 257(60), 240(50), 229(100), 169(75)

-

411

13

34.2

podophyllotoxone

C22H20O8

412

[M-H]

14

38.31

podoverine A

C21H20O7

384

[M-H]-

383

15

39.74

podoverine F

C30H18O13

586

[M-H]-

585

[411]: 396(40), 383(15), 367(50), 352(100), 337(45) [383]: 368(25), 351(40), 299(20), 179(100), 152(20) [585]: 299(100), 283(60)

16

42.28

podoverine D

C36H28O13

668

[M-H]-

667

[667]: 515(10), 381(15), 283(100)

682

-

681

[681]: 381(100), 299(5)

17

44.81

podoverine E

C37H30O13

[M-H]

29