Cytotoxic Sesquiterpene Lactones from Kauna lasiophthalma Griseb

Two new eudesmane derivatives (3 and 8) were isolated from the ethanol extract of the aerial parts of Kaunia lasiophthalma Griseb, together with 14 known eudesmane, germacrane, and guaiane sesquiterpenes, and four flavones. The structures and relative configurations of all the compounds were established by NMR spectroscopy and high-resolution mass spectrometry. The anticancer activity of sesquiterpenes 1, 3, 6–9, 11, 12, 14, and 16 was evaluated in vitro with the breast cancer cell lines HCC1937, JIMT-1, L56Br-C1, MCF-7, and SK-BR-3, and compared with the cytotoxicity in the non-cancerous breast epithelial cell line MCF-10A. All compounds were found to possess anticancer activity, and compound 1 was the most potent in all of the investigated cancer cell lines with IC50 values ranging between 2.0 and 6.2 μM. In order to demonstrate the importance of the α-methylene-γ-lactone/ester moiety present in all compounds for the effects on the cells, the methyl cysteine adduct 21 was prepared from 9 and found to be inactive or considerably less potent.


Introduction
The relatively small genus Kaunia (Asteraceae: Eupatorieae) comprises only 14 species of which most grow in Bolivia, but that also are found in Argentina, Brazil, Peru, and Ecuador [1,2]. Previous chemical investigations of plants classified in this genus have revealed the presence of sesquiterpene lactones, predominately guaianes, and thymol derivatives as the main constituents [3][4][5]. Kaunia lasiophthalma Griseb (syn. Eupatorium lasiophthalmum G.) is a shrub bearing white-purple flowers, locally known in Cochabamba, Bolivia by the common name of "Tuwi" and used to treat inflammation and headaches (Lic. Modesto Zárate, personal communication). It has previously been subjected to a phytochemical investigation by Gutierrez and co-workers, who isolated 19 guaiane sesquiterpene lactones [6]. In our search for biologically active compounds in the Bolivian flora, we have isolated and characterized the chemical constituents of the aerial parts of K. lasiophthalma G. Two new eudesmanolide derivatives, compounds 3 and 8, together with 18 known compounds (1, 2, 4-7, [9][10][11][12][13][14][15][16][17][18][19][20] were obtained, and this constitutes the first report of eudesmanolide and germacranolide sesquiterpenes as well as flavones from this species. As most isolated compounds possess an α-methylene-γ-lactone moiety that previously has been associated with various biological activities, we decided to investigate the effects of the sesquiterpenes available in sufficient quantities on cancer cells. Unsaturated lactones are likely to exert biological effects because they react with cell constituents as Michael acceptors, and have consequently been considered to be generally toxic. However, recent studies have shown that Michael acceptors may be selective and several are currently in clinical trials as drug candidates [7]. Human breast cancer can be classified into different molecular subtypes using gene expression profiles [8][9][10][11][12][13]. The anticancer activity of compounds 1, 3, 6-9, 11, 12, 14, and 16 in the five breast cancer cell lines MCF-7 (luminal A subtype), SK-BR-3 (luminal B), HCC1937, L56Br-C1 (basal subtype), and JIMT-1 (HER2 subtype) was compared with the cytotoxicity in the normal-like breast epithelial cell line MCF-10A [8][9][10][11][12][13]. In breast cancer, each subtype has a different prognosis and is subjected to different treatment. The luminal A and B subgroups express estrogen receptors and are amenable to hormone therapy, while the HER2 group, expressing the human epidermal growth factor receptor 2, may be subjected to trastuzumab therapy. The basal tumors lack expression of both estrogen receptors and HER2; they are biologically more aggressive and the prognosis is often poorer.

Results and Discussion
The EtOH extract of the leaves and flowers of K. lasiophthalma G. were subjected, separately, to sequential liquid-liquid partition with hexane, CH 2 Cl 2 , and EtOAc. The major constituents were found to be fats, which were not further investigated, and the compounds reported here are minor metabolites of this plant. Vacuum liquid chromatography (VLC) of the CH 2 Cl 2 fractions followed by silica gel and Sephadex LH-20 chromatography as well as HPLC fractionation afforded the two new natural products 3 and 8, along with 18 known compounds, 14 sesquiterpenes and four flavonoids (see Figure 1 for chemical structures).
Compound 3 was obtained as a colourless gum. The HR-ESI-MS indicated that its elemental composition is C 17 H 25 O 5 , which suggests six degrees of unsaturation. The IR spectrum showed absorption bands corresponding to a hydroxyl group (3473 cm −1 ), an α,β-unsaturated-γ-lactone function (1764 cm −1 ), and an ester group (1730 cm −1 ). The NMR spectra displayed the characteristic signals of an eudesmane lactone, see Table 1 for 1D NMR data and Figure 2 for 2D data. The presence of an exomethylene-γ-lactone ring was established by the 13-H 2 proton signals at δ H 5.94 and 4.79 and their HMBC correlations to C-7, C-11, and C-12, and the 6-H lactone proton signal at δ H 3.70 and its strong 1 H-1 H coupling with 7-H and HMBC correlation to C-12. In addition, the NMR data indicated the presence of the acetoxylated tertiary carbon (C-1) at δ H 4.67 and δ C 77.  [17] indicated that they are similar. The difference is that the C-1 hydroxyl group in 5 is acetylated in 3 and that the configuration at C-1 is inversed, which consequently identifies 3 as 4-epi-1α-acetoxy arbusculin A. The anticancer activities of sesquiterpenes 1, 3, 6-9, 11, 12, 14, and 16 were assessed in five breast-cancer cell lines, HCC1937, JIMT-1, L56Br-C1, MCF-7, and SK-BR-3, and compared with the cytotoxicity in the breast-derived non-cancerous cell line MCF-10A using the MTT colorimetric assay. The inhibitory concentration 50 values (IC 50 ) were deduced from the obtained dose-response curves and are presented in Table 2.
Interestingly, the cancer cell lines were more sensitive to all of the compounds than the normal-like MCF-10A cells. No obvious patterns related to the breast cancer cell line subgroup (vide supra) was found, and all compounds possessed activity. Compound 1 was found to be the most active in all of the cell lines with IC 50 values ranging from 2.0 to 6.2 µM in the cancer cell lines, while compounds 3 and 16 exhibited the lowest activity. 16 is an unsaturated ester and differs in that respect from the other compounds, but the lower activity of 3 compared to the similar compounds was unexpected. Costunolide (1) differs from the eudesmane sesquiterpenes by having the unsaturated lactone fused with a macrocyclic system instead of a cyclohexane ring, and the tension of the lactone ring is likely to be lower in 1. This would render 1 less reactive and possibly more selective. Indeed, 1 together with the two guaianes 12 and 14, shows a slightly higher selectivity for the cancer cells compared to the eudesmanes. The difference in the activities of 7 and 8 may depend on the higher lipophilicity of 8, facilitating its absorption into the cells. In 11 and 14, the presence of a second Michael acceptor function may influence the activity. It is difficult to speculate from these data, but a trend is that the MCF-7 cells seem slightly less affected than the other cancer cell lines. MCF-7 has a normal wild type p53 gene, which the MCF-10A cells also have, while the others have a mutated p53. Thus, MCF-7 and MCF-10A cells may share a property of being blocked in the G 1 phase of the cell cycle, which has a protective function, while the other cancer cell lines do not.
Previous investigations of sesquiterpenoid α-methylene-γ-lactones have indicated that the cytotoxic and antitumor activities are related to their ability to react as Michael acceptors [7]. We therefore added methyl cysteine to the exocyclic methylene group of 9 to give 21, and compared its cytotoxicity with the natural products'. As can be seen from the results in Table 2, 21 is significantly less potent, however, it is not devoid of activity and its cytotoxicity towards the normal-like MCF-10A cells is similar to that of the α-methylene-γ-lactones. This may depend on the reversibility of Michael additions, by which 21 slowly can eliminate methyl cysteine and regenerate 9 during the assay conditions [7]. In conclusion, the molecular causes for the lower cytotoxicity of the compounds (especially compound 1) in the normal-like breast epithelial MCF-10A compared to the breast cancer cells lines need to be further exploited and may find clinical use by showing less off-target cytotoxicity.

General
Optical rotations were measured with a Perkin Elmer Model 341 polarimeter. IR spectra were recorded with a Bruker Alpha-P FT-IR instrument in the ATR geometry with a diamond ATR unit. HR-ESI-MS was performed with a Waters Q-TOF Micro system spectrometer (using H 3 PO 4 for calibration and as internal standard). 1D and 2D NMR spectra were recorded at room temperature on the Bruker Avance II 400 MHz and Bruker Avance 500 MHz spectrometers, operating at 400 and 500 MHz for 1 H and 100 and 125 MHz for 13 C, respectively. The chemical shifts (δ) are reported in ppm relative to solvent signals δ H 7.16 and δ C 128.39 for C 6 D 6 , and δ H 7.26 and δ C 77.00 for CDCl 3 , while the coupling constants (J) are given in Hz. Vacuum liquid chromatography (VLC) separations were carried out on the Merck Silica gel 60G (Merck), while column chromatography (CC) was performed using the Silica gel 60 (230-400 mesh, Merck), silver nitrate-impregnated Silica gel 60 [30], and gel permeation on Sephadex LH-20 (GE Healthcare). TLC analyses were carried out using aluminium-backed silica gel 60 F 254 (0,2 mm thickness, Merck). Chromatograms were visualized under a UV lamp at 254 nm then sprayed with vanillin and KMnO 4 /K 2 CO 3 /NaOH solution followed by heating. Preparative TLC (PTLC) was run on 20x20 cm glass-coated plates (1 mm thickness, Analtech) and doped TLC plates in MeCN:AgNO 3 solution [31]. HPLC was performed on the Agilent 1260 Infinity Quaternary LC system, equipped with a Standard Autosampler (G1329B), Thermostated Column Compartment (G1316A TCC), a Diode Array Detector VL (G1315D), and a semipreparative column (XTerra RP18, 10x150 mm, 5 μm i.d, Waters).

Extraction and Isolation
The air-dried and ground leaves (788.5 g) and flowers (1064.0 g) were extracted separately by maceration with 95% EtOH for 24 hours, two times at room temperature. After filtration, the combined extracts were concentrated under reduced pressure and the following crude extracts were obtained: 90.0 g from leaves and 91.0 g from flowers.

Dose-Response Assay
Stock solutions (10 or 100 μM) of the compounds were made in 100% DMSO. These were further diluted in PBS to obtain the correct concentrations used for the MTT assay. Appropriate DMSO controls were used. In general, the highest DMSO concentration was 0.1%, however, when treating with 100 µM and starting from a 10 µM stock in 100% DMSO, a final concentration of 1% DMSO was used as the control.
The MTT assay was performed as previously described [34]. Briefly, the cells were trypsinized and counted in a hemocytometer. Aliquots of 180 µl cell suspensions containing 3000 (MCF-10A) and 6000 (MCF-7, SK-BR-3, JIMT-1, L56Br-C1, and HCC1937) cells were seeded in the wells of 96-well plates. Compounds were added 24 hours after seeding to allow the attachment of cells. A concentration range between 0.1 to 100 μM was used in the MTT assays and appropriate DMSO controls. At 72 h of drug treatment, 20 µl of MTT solution (5 mg/ml MTT in PBS) was added to each well and the 96-well plates were returned to the CO 2 incubator for 1 hour. The MTT-containing medium was removed. The blue formazan product formed by the reduction in live attached cells was dissolved by adding 100 µl of 100% DMSO per well after removal of the MTTcontaining medium. The plates were swirled gently at room temperature for 10 minutes to dissolve the precipitate. Absorbance was monitored at 540 nm in a Labsystems iEMS Reader MF (Labsystems Oy, Helsinki, Finland) using the DeltaSoft II v.4.14 software (Biometallics Inc., Princeton, NJ, USA). Dose-response curves were drawn based on the % of the control in Excel. The IC 50 was deduced from the curves.

Acknowledgement
The financial support from The Swedish International Development Agency (SIDA) in a bilateral collaboration between Lund University (Sweden) and San Simon University (Bolivia) is fully acknowledged. We also wish to thank to Dr. Karl-Erik Bergquist for his helpful comments on the manuscript.

Supporting Information
Supporting information containing the 1 H-NMR and 13