Chemical profiling of anti-hepatocellular carcinoma constituents from Caragana tangutica Maxim. by off-line semi-preparative HPLC-NMR

Abstract An EtOAc fraction from the roots of Caragana tangutica Maxim. (CTEA) displayed promising anti-hepatocellular carcinoma (HCC) activity during screening of a traditional Chinese ethnic herb library against HepG2 and Hep3B cell lines. HPLC-based activity profiling of CTEA by combination of MS-guided large-scale semi-preparative HPLC and NMR methods led to the identification of a new pterocarpan glycoside, (-)-maackiain 3-O-6′-O-methyl malonyl-β-d-glucopyranoside (1), together with three known pterocarpan glycosides, (-)-maackiain 3-O-β-d-glucopyranoside (2), 3-O-6′-O-acrylyl-β-d-galactopyranoside (3), and (-)-maackiain 3-O-6′-O-acetyl-β-d-glucopyranoside (4). Compound 1 was isolated during a drug discovery programme aimed at identifying new anti-HCC leads from a natural product library. Anti-HCC study showed that all four compounds exhibited cytotoxic activity with IC50 values range of 29.1–53.5 μg/mL against HepG2 and Hep3B cell lines.


Introduction
Hepatocellular carcinoma (HCC), also called malignant hepatoma, is the most common type of liver cancer (Kumar et al. 2016). HCC is the sixth most prevalent cancer and the third leading cause of cancer-related deaths, with over 662,000 deaths worldwide per year (Song et al. 2015). usually, males are more severely affected by this disease than females, and it is most common between the age of 30 and 50 years with about half of them occurring in China (Srivatanakul et al. 2004). HCC is resistant to commonly used chemotherapy and rarely amenable to radiotherapy, which makes anti-HCC drug discovery all the more urgent and the development of novel anticancer drugs remain challenging (Marin et al. 2008). Therefore, development of more effective therapeutic agents to fight this disease is much needed. The genus Caragana belonging to family Fabaceae, contains about 80-100 species in world. The main chemical principles of Caragana plants have been reported to be flavonoids and their glycosides (Song et al. 2010;Niu et al. 2013;Luo et al. 2015;Perveen et al. 2015). Caragana tangutica Maxim. is a perennial leguminous bush endemic to the north-west of China. It is one of the medicinal plants used as the traditional Tibetan medicine 'Zuo Mao Xing' , widely distributed in Gansu, Qinghai, Tibet and Sichuan in China (Editorial Committee for Flora of Qinghai, Northwest Institute of Plateau Biology (Chinese Academy of Sciences) 1999). The roots and heartwoods of C. tangutica have been long used to treat some cardiovascular diseases, such as atherosclerosis, hyperlipidaemia, hypertension, blood circulation disorder, blood stasis and so on (Northwest Plateau Institute of Biology (the Chinese Academy of Sciences) 1996; National Bureau of Traditional Chinese Medicine Editorial Board 2002). It is also used to treat arthritis and abnormal menstruation, remove fever, ameliorate pains, reduce trauma, diminish inflammation and dissipate swelling (Niu et al. 2014). The main chemical constituents have been reported from Caragana, including flavonoids and aromatic acids, such as melilotocarpan A, medicarpin, brachysides C and D, maackiain, 2-(2′,4′-dihydroxyphenyl)-3-methyl-6-methoxy benzofuran, cajanin, formononetin, 73′-dihydroxy-5-methoxy isoflavone, texasin, 2′,44′-trihydroxy chalone, bolusanthin III, and p-ethoxy benzoic acid (Niu et al. 2013).
As part of our continuing research into the discovery of new bioactive leads from Chinese folk herbal plants (Yang et al. 2011Wang et al. 2014), we undertook screening of a Chinese ethnic medicine library against HepG2 and Hep3B cell lines. From the screening data, we identified one ethyl acetate extract (CTEA) derived from the traditional Tibetan medicine 'Zuo Mao Xing' (Caragana tangutica) that displayed promising anti-HCC activity and no significant cytotoxicity towards a human embryonic kidney cell line (HEK293). An off-line semi-preparative HPLC-NMR approach was performed for targeted identification of four anti-HCC active principles from CTEA coupled with MS-directed purification method.

Results and discussion
The MTT assay in 96-well microplates was used to evaluate anti-HCC activities of extracts against HepG2 and Hep3B cell lines, while preliminary toxicity towards human cells was investigated using a human embryonic kidney cell line, HEK293. When tested at 200 μg/mL against HepG2 and Hep3B cell lines, CTEA displayed momentous antiproliferative activity with 77.1 and 69.6% inhibitions, respectively.
HPLC-based activity profiling is a miniaturised and highly effective approach for localisation, dereplication and characterisation of bioactive compounds from natural medicinal plants (Yang et al. 2011Wang et al. 2014). This protocol linked with various cell-based assays has been successfully built in our group. In the present research, CTEA was submitted to HPLC-based activity profiling using a validated protocol (Yang et al. 2011). The chromatogram of a semi-preparative separation of 5 mg extract and the corresponding anti-HCC potentiation against HepG2 and Hep3B by the timed-based fractionations (11 microfractions of 120 s each) are shown in Figure 1. Meanwhile, according to the protocol described in Supplementary Material 2.4, microfractions from the semi-preparative HPLC separations were dissolved in 5 μL of DMSO, 2 μL of this solution was applied to ESIMS analysis with direct injection, and 1 μL of this solution was subject to cytotoxicity bioassay in 96-well plates. In Figure 1(B) and (C), fractions 3 exhibited the weak anti-HCC potentiation with inhibition rates of between 31.2 ± 1.9% and 34.5 ± 2.1% against HepG2 and Hep3B, and 4 displayed the highest anti-HCC potentiation with inhibition rates between 67.8 ± 4.1% and 63.2 ± 2.8%. Fractions 5 potentiated anti-HCC activity to a lesser extent with inhibition rates of 62.6 ± 3.4% and 57.7 ± 3.2% against HepG2 and Hep3B, respectively. While, other fractions were inactive against HepG2 and Hep3B cell lines.
In order to elucidate the bioactive components from CTEA effectively, we established a hyphenated approach based on mass-directed fractionation of the large-scale organic extract and the off-line semi-preparative HPLC-NMR. 1.10 g of CTEA was submitted to the semi-preparative HPLC for large-scale isolation by a reverse phase C 18 column (Waters, Sunfire, 250 mm × 20 mm i.d., 5 μm) with the gradient mobile-phase comprising of acetonitrile (0.1% formic acid) and water (0.1% formic acid). HPLC gradients were shown in Supplementary Material 2.7. The flow rate used in this large-scale separation was 9.0 mL/min. The large-scale preparative chromatogram by separation of 1.10 g of CTEA was shown in Supplementary Material Figure S1.
Sixty fractions (60 tubes, 60 s each) were collected from time = 0 min. About 120 μL of each collected fraction is aspirated and combined by each 5 fractions to yield 12 combined fractions for direct ESIMS analysis. In the chromatogram of the large-scale preparative separation of CTEA, fractions 21-25 contained an interesting ion cluster at m/z 447/469/893, fractions 26-30 contained an ion cluster of interest at m/z 501/523/1023 and fractions 31-35 contained two ion clusters of interest at m/z 547/1093/1115 and 489/511/977. Fractions containing target ions of interest were combined, and then submitted to Sephadex micro-column and semi-preparative HPLC for further purification to yield compounds 1-4 from CTEA. 1D and 2D NMR spectra of four isolated pure compounds were tested by 1.4 mm heavy wall Micro-NMR tubes (date was shown in Supplementary Material Figure S2).
The active compound 1 was isolated as light yellow powder from the active fractions 31-35 in the large-scale separation of CTEA with IC 50 values of 40.3 and 36.4 μg/mL against HepG2 and Hep3B cell lines, respectively. The molecular formula of C 26 H 26 O 13 was determined by a quasi-molecular ion [M + H] + at m/z 547.1448 in the HRESIMS, which agreed well with its 13 C-NMR data. The uV spectrum showed absorption maxima at 279, 286, and 311 nm, classic of pterocarpanes (Song et al. 2010). The off-line 1 H NMR spectrum showed signals of two isolated aromatic protons at δ H 6.98 (1H, s) and 6.52 (1H, s), two isolated oxygenated protons at δ H 5.94 (1H, s) and 5.90 (1H, s), and a 124-trisubstituted aromatic ring at δ H 7.37 (1H, d, J = 8.5 Hz), 6.67 (1H, dd, J = 8.5, 2.1 Hz) and 6.53 (1H, d, J = 2.1 Hz), and signals for a -OCHCHCH 2 O-moiety at δ H 5.55 (1H, d, J = 6.8 Hz), 4.26 (1H, dd, J = 10.2, 3.7 Hz), 3.61 (1H, t, J = 10.2 Hz), and 3.60 (1H, m), good agreement with those of (-)-maackiain (Song et al. 2010). The off-line 1 H NMR spectrum also displayed the residue of an anomeric proton at δ H 4.87 (d, J = 7.6 Hz), and a methylene at δ H 3.46 (d, J = 16.1 Hz) and 3.51 (d, J = 16.1 Hz) as well as a methoxy group at δ H 3.59 (s), revealing that 1 was a derivative of maackiain glycoside. Acid hydrolysis with HCl yielded d-glucose, determined by GC according to the protocols described in the literature (Song et al. 2010). The whole molecular framework was constructed by the HMBC spectrum (date was shown in Supplementary Material Figure S3). The key HMBC correlations of three aromatic protons at δ H 7.37 (1H, H-1), 6.67 (1H, H-2) and 6.53 (1H, H-4) to the quaternary aromatic carbon at δ 158.1 were observed, suggesting the carbon was attached at C-3. The key HMBC correlation of the anomeric proton at δ H 4.87 to C-3 was also observed, confirming the linkage between the sugar unit and the aglycone. The above partial structures made up a maackiain 3-O-β-d-glucopyranoside moiety, taking up 445 mass units of the whole molecular mass 546. Furthermore, the key HMBC correlations of the methylene at δ H 3.46 and 3.51 to two ketone carbons at δ C 166.4 and 166.8, and a methoxy group at δ H 3.59 to the ketone carbon at δ C 166.8 confirmed the presence of a methyl malonate moiety, good coincidence with the remaining mass units of 101. The HMBC correlations between two protons at δ H 4.36 (1H, d, J = 11.7 Hz) and 4.10 (1H, d, J = 11.7, 7.0 Hz) assignable to Glu H 2 -6′ and the ketone carbons at δ C 166.3 confirmed that the methyl malonate moiety was attached to Glu C-6′. The CD spectrum of 1 showed a positive absorption (Δε, + 3.9) at 311 nm, a negative absorption (Δε: -2.2) at 279 nm and negative absorption (Δε: -8.9) at 231 nm, suggesting that the aglycon of 1 was identified as 6aR,11aR-3-hydroxy-89-methylenedioxypterocarpane, (-)-maackiain on the basis of NMR and CD data (Song et al. 2010). Thus, compound 1 was concluded as (-)-maackiain 3-O-6′-O-methyl malonate-β-d-glucopyranoside. It was reported that maackiain is a pharmacology potential molecule occupied with vary pharmacological activities including larvicidal activity, antiallergic, etc (Bezerra-Silva et al. 2015;Mizuguchi et al. 2015). Our findings demonstrate that the maackiain derivatives for pterocarpan type have the potential activity against liver cancer (IC 50 values of 40.3 and 36.4 μg/mL against HepG2 and Hep3B cell lines). The results will offer the evidence of exploring the anticancer activity of maackiain in further study.

Conclusion
Natural products (NPs) have been one of the important sources in anticancer drug discovery (Newman et al. 2003). A great amount of NP-derived anticancer drugs has been entered into clinical evaluations (Kinghorn et al. 2009). There is the tremendous potential for developing new drugs from natural products. However, to develop an efficient and rapid protocol with few costs to discover novel anti-HCC natural products is extremely important. In the present study, the HPLC-based activity profiling approach was utilised to characterise the main bioactive components from the active EtOAc extract of C. tangutica by combination of MS-guided large-scale semi-preparative HPLC and NMR methods, which constituted a highly efficient platform for discovery of bioactive natural products. The miniaturised NMR microtube was used to increase the mass sensitivity and allows the measurement of 1D H NMR and 2D inverse heteronuclear NMR experiments with the mass-and volume-limited samples, such as mg −1 or μg −1 scale. The ideal 2D NMR spectra could be obtained with trace amount of compounds in the reasonable time.

Supplementary material
Supplementary material relating to this article is available online, including Figures S1-S9 and Table S1.