2-Deprenyl-Rheediaxanthone B Isolated from Metaxya rostrata Induces Active Cell Death in Colorectal Tumor Cells

Metaxya rostrata C. Presl (Metaxyaceae) is a common tree fern in Central and South America that is used for the treatment of intestinal ulcers and tumours in ethnic medicine. Using a bioactivity-guided strategy 2-deprenyl-rheediaxanthone B (XB) has been isolated as one of the active principles in this plant. XB induced loss of cell viability in colorectal cancer cell lines at IC50 concentrations of 11–23 µM. This was caused by both accumulation of cells in the G2- and S-phase as well as by induction of active cell death in a time and concentration-dependent manner. Cells exposed to XB were incapable of undergoing regular mitosis due to down-regulation of FoxM1 and absence of chromosome condensation. The apoptosis-related proteins Bcl2 and Bclxl were up-regulated so that Caspase 3 was not activated and classical apoptosis was not observed. However, XB triggered damage pathways down-stream of ATR and activated Caspase 2 causing cell death by a mechanism similar to mitotic catastrophe. Our observations are the first to show the cytotoxic activity of 2-deprenyl-rheediaxanthone B and indicate that XB is an interesting new lead compound for cancer therapy that merits further development.


Introduction
There has been considerable progress in cancer therapies and prognosis has been improved for several malignancies -e.g. leukemias, breast cancer -while for other tumours -e.g. melanomas, late stage colon cancer -innovative therapeutic options are still urgently needed [1]. Consequently, cancer therapy is one of the most addressed issues in drug discovery. Over the past decade efforts went in two directions -targeted therapeutics and optimized combination chemotherapy. Natural products are an important source of anticancer agents, providing over 50% of the drugs in present clinical use [2] and many of these come from tropical plants used in ethnic medicine [3].
Colorectal cancer is among the most common malignant diseases in the western world and therapy is still difficult at advanced stages. Today the standard therapy consists of 5fluorouracil in combination with levamisol, oxaliplatin or irinotecan [4,5]. In addition antibodies targeting the epidermal growth factor receptor and the vascular endothelial growth factor have been introduced and improved prognosis for colon cancer patients [6]. In spite of this progress therapy of advanced colorectal cancers is still not satisfactory due to inherent mechanisms of therapy resistance in colorectal tumour cells -such as mutation of p53 [7] and over-expression of Bcl 2 [8,9], as well as multidrug-resistance genes [10,11].
Therefore we report the bioactivity-guided isolation and biological characterization of a rare bioactive natural compound from the tree fern Metaxya rostrata C. Presl (Metaxyacea) that is used in traditional Costa Rican medicine for the treatment of intestinal diseases such as ulcers and tumours [12].

Extraction and Isolation
Rootlets of Metaxya rostrata C. Presl were collected at the research center of the University of Vienna in La Gamba in southwestern Costa Rica, and authenticated in the Herbarium of the Museo National in San Jose by Dr. W. Huber (University of Vienna). Voucher specimens (Voucher number: MR0203) are deposited at the Herbarium of the Department of Pharmacognosy, University of Vienna, Austria. 920 g dried rootlets of Metaxya rostrata were pulverized and extracted by sonification with CH 2 Cl 2 yielding 5.3 g CH 2 Cl 2extract. The extract was subjected to column chromatography on Sephadex LH-20 under elution with EtOAc (column diameter and height: 4 cm and 75 cm) to obtain 16 subfractions CC-1 to CC-16. In this bioactivity guided approach, after every step of fractionation the obtained fractions were evaluated for their effect on cancer cell viability on SW480. The most active fractions were chosen for further separation. Fraction CC-14 showed highest activity reducing the cell viability in a significant manner at concentrations of 6 mg/ml. Fraction CC-14 was further fractionated using solid phase extraction. It was applied on the SPEcartridges (C-18, 20 ml, Bond Elut, Varian) and elution was performed sequentially with 4 reservoir volumes each of 60%, 70%, 80%, 90% and 100% MeOH. The fractions of identical polarity were pooled and evaluated for their effect on cancer cell viability. The 70% MeOH-fraction (CC-14/70) was the most active one decreasing cell viability significantly from a concentration of 0.6 mg/ml upwards. The major compound in HPLC analysis at Rt 24.5 min was isolated and identified as 2-deprenylrheediaxanthone B using MS and NMR analysis (Figures S1-S8, Table S1 in Text S1). As 2-deprenyl-rheediaxanthone B was the major compound with a purity of more than 98% in this fraction, it is certain that the high activity in the bioassay is due to that rare natural compound.

Chemical Analysis
HPLC was performed on a Shimadzu instrument using LC solution software.
Analytical-grade solvents for extraction and fractionation were obtained from VWR (Vienna, Austria). NMR and MS details are given in supplementary material.

Cell Lines
SW480, SW620, DLD1, HCT116, and Caco2 colon carcinoma cells were obtained from ATCC and were kept under standard tissue culture conditions using minimal essential medium (MEM) containing 10% fetal calf serum (FCS). F331 cells are fibroblasts isolated from fetal colon [13] and were kept under standard culture conditions using Dulbeccos minimal essential medium (DMEM) supplemented with 10% FCS. LT97 cells were established in our group from early adenoma tissue and grown in a semi-defined medium [14].

Determination of Viability and Cytotoxic Effect
Cells were plated at 5610 4 /well in 24-well plates, left to attach for 24 h (SW480, SW620, CaCo2, DLD1, HCT116) or 72 h (F331, LT97). When cultures were semi-confluent they were exposed to XB for 48 hours (stock solutions were prepared in DMSO and stored at -20uC). Control media contained the appropriate volume of DMSO. Cell viability was determined by MTT or neutral red uptake with both methods yielding similar results [15].

Hoechst Staining
After removal of the neutral red dye, fixed cells were stained with 800 ng/ml Hoechst 33285. Nuclei with chromatin condensed at the nuclear margin or in the centre of the nucleus were classified as apoptotic [15].

Cell Cycle Analysis
Cells were trypsinized and homogenized with 0.5 M citric acid and 0.5% Tween-20 and the nuclei isolated by centrifugation. After staining with 500 mg/ml PI (Sigma) containing 100 mg/ml RNAse, nuclear DNA content was analyzed by flow cytometry on a FACSCalibur (Becton Dickinson, Mountain View, CA) equipped with a 15 mW argon laser exciting at 488 nm.
CellQUEST software was used for data acquisition and MOD-FIT (Becton Dickinson) for data evaluation [15].

Determination of Mitochondrial Membrane Potential (MMP)
MMP was measured by the lipophilic cationic probe JC-1 (Alexis, Lausanne, Switzerland) that is selectively concentrated in intact mitochondria to form multimer J-aggregates emitting fluorescent light at 590 nm. The monomeric form emits light at 527 nm after excitation at 490 nm. Cells were incubated in MEM with 10% FCS containing 10 mg/ml JC-1 for 10 min at 37uC in the dark. Stained cells were washed, resuspended in PBS and immediately analyzed by flow cytometry [15].

Statistical Analysis
IC 50 concentrations were determined by non-linear regression and differential sensitivity by 2-way ANOVA using GraphPad Prism software. All other results were analysed by either ANOVA or Kruskal-Wallis Test as appropriate.

Isolation and Chemical Structure
Rootlets of Metaxya rostrata were extracted with CH 2 Cl 2 and fractionated in a bioactivity-guided approach, which focused on the cytotoxicity on SW480 colon carcinoma cells. An active principle was isolated by different chromatographic methods and identified by extensive MS and NMR analysis (find a detailed description of the methods in Text S1) as the rare natural compound 2-deprenyl-rheediaxanthone B (XB) (Figure 1). Details of the isolation and identification process are given in supplementary materials (Figures S1-S8, Table S1 in Text S1).

Cytotoxic Activity of XB
For biological testing XB was dissolved in DMSO and then diluted directly into the medium of semi-confluent cell cultures. Cytotoxic activity of XB was determined using MTT assays in the colon cancer cell lines SW480, SW620, DLD1, HCT116, and CaCo2 as well as the non-malignant LT97 adenoma cell line and the normal colonic fibroblasts F331. XB significantly reduced cell viability in a concentration-dependent manner in all carcinoma cell lines ( Figure 2). For 3 of the cell lines -namely SW480, HCT116 and LT97 cells -the dose-response curve was steep and the 95% confidence interval for the IC 50 concentrations were narrow ( Table 1). For Caco2, DLD1 and SW620 cell loss was much more gradual with sensitivity becoming less in the sequence Caco2. SW620. DLD1 indicating two different cellular mechanisms of action.
SW480 and Caco2 were used as typical examples for the investigation of the cellular mechanism underlying the cytotoxic effect of XB. Cell cycle distribution was assessed by FACS analysis and demonstrated a concentration-dependent decrease of the proportion of cells in the G0-G1 phase in both cell lines. In SW480 cultures the percentage of cells in both S-phase and G2-M increased dose-dependently and significantly after 48 h ( Figure 3A) suggesting an XB-induced G2-M phase cell cycle arrest. In Caco2 cultures cells accumulated in S-phase ( Figure 3B). In order to confirm the cell cycle blockade, cell cycle regulatory proteins were analysed by western blot from cell lysates obtained after exposure to XB. In SW480 cells the results showed increased levels of Cyclin A and Cyclin B1, while Cyclin E was decreased compared to the control ( Figure 3C) confirming the G2-M phase cell cycle arrest seen with FACS analysis. The increase of Cyclin A could be confirmed in Caco2 cells ( Figure 3D).
Protein levels of the forkhead transcription factor FoxM1 -an important regulator of gene expression during the G2 phase and mitosis -was highly expressed in SW480. The protein was no longer detectable in XB exposed SW480 cells, however. In addition, phosphorylation at Ser10 of Histone H3 which is tightly correlated with chromosome condensation was blocked by XB ( Figure 3C). In Caco2 control cells the FoxM1 level was lower than in SW480 and the impact of XB on both FoxM1 and histone phosphorylation was less pronounced ( Figure 3D).
Staining of cells with Hoechst 33258 revealed the appearance of few pycnotic and fragmented nuclei in XB-exposed SW480 and Caco2 cells. However, the most prominent change was an increase in the size of the nuclei compared to the control ( Figure 4A-D).
Involvement of the mitochondrial pathway of apoptosis induction in XB-induced cell loss was investigated by JC-1 staining and FACS analysis.    proteins BAX and BAK as well as a slight decrease of Survivin as compared to the control (data not shown). The anti-apoptotic proteins Bcl xl and Bcl 2 were up-regulated and activated Caspase 3 was slightly reduced compared to the control. In agreement with this observation XB did not induce but slightly diminished cleavage of Poly(ADP-ribose) polymerase (PARP) ( Figure 5A). By contrast, an increase was observed in the levels of both Pro-Caspase 2 and activated Caspase 2 ( Figure 5B). Induction of XB-induced damage response was investigated by determination of the DNA damage markers by phospho-specific western blot analysis. The results showed a slight increase of ATR but not ATM protein level ( Figure 6A). Down-stream of ATR an increase in both total and phosphorylated Chk1 and a slight increase in total and phosphorylated Cdc2 was observed. Cdc25C was clearly decreased compared to the control ( Figure 6B).

Discussion
2-deprenyl-rheediaxanthone B (XB) is a rare natural compound and has been isolated before only from two species of the family of the Guttiferae, from the roots of Rheedia benthamiana Pl. Triana [16] and the roots of Hypericum roeperanum Schimp. ex A. Rich. and also from Garcinia viellardii Pierre (Clusiaceae) [17]. Hostettmann and his group showed an anti-fungal activity against Candida albicans for this compound [18]. In addition, XB has been tested for antimalarial activity [17] and effects on angiogenesis [19]. Further biological activities of 2-deprenyl-rheediaxanthone B have not yet been studied.
In our study the compound was isolated from the Costa Rican tree fern Metaxya rostrata. The plant is common in the rain forests of Central and South America and used in ethnic medicine for the treatment of intestinal ulcers and tumours [20], which made it a promising source of cytotoxic lead compounds. XB is not the only bioactive compound in Metaxya rostrata. The polyphenols cinnamtannin B-1 and aesculitannin B have been isolated in a previous study. With an IC 50 of about 11 mM XB is much more effective in SW480 cells than the polyphenols, that induced cell death with an IC 50 of about 50 mM [20]. Viability of other colon carcinoma cell lines was similarly reduced by XB with IC 50 concentration ranging from 10-23 mM. Non-malignant colon adenoma cells also had an XB induced a cell cycle block in S and G2/M as characterised by increased levels of Cyclins A and B, while Cyclin E decreased. Protein levels of Cdc2, the kinase partner of Cyclin B, were increased, but Cdc2 phosphorylation was little affected. In addition, Cdc25C that activates CyclinB/Cdc2 in G2 [21] was decreased. Together these observations place the exact point of the cell cycle block at S and G2. This may be due to down-regulation the FoxM1 that was observed in both SW480 and Caco2 cells. The transcription factor is essential for the progress of the cells into mitosis [22] and in both cell lines phosphorylation of Histone 3-a prerequisite of chromosome condensation in mitosis [23] -was much reduced. SW480 and Caco2 cells differ in their control levels of FoxM1 protein, which are much higher in SW480. The XBinduced decrease of FoxM1 level was less in Caco2, however, as was the reduction of Histone phosphorylation. This correlates with both a more gradual dose-response curve with regard to loss of cell viability and a cell cycle block rather in S than in G2. Whether this is a mechanistic consequence of the lesser FoxM1 level warrants further investigation which could not be performed in this study due to the limited amount of XB available.
In addition XB-induced cell death was an active process involving loss of MMP. The IC 50 concentration for this process was in the same range (10-13 mM) as the IC 50 for loss of viability indicating that the active cell death component is the main mechanism for cell loss. The underlying cellular mechanism of cell death was not indicative of apoptosis, however. In spite of the time and concentration-dependent loss of MMP, no activation of Caspase 3 occurred and PARP degradation was even reduced. In addition, Bcl 2 and Bcl xl were not decreased but slightly upregulated. By contrast the caspase activated by XB was Caspase 2. In addition the compound induced the damage response markers ATR, Chk1 and Cdc2, and down-regulated Cdc25C. Together these alterations are characteristic for the induction of mitotic catastrophe [24,25]. The accumulation of giant nuclei indicates that the cells go through several cycles of DNA-replication in spite of their inability to initiate chromosome condensation and mitosis.
While no information has previously been available on the effect of XB in other tumour types, different prenylated xanthones have been reported to possess activity against MCF-7, NCI-H460 [26], HL60 leukemia cells [27], HeLa and MDA-MB-231 cancer cell lines [28]. a-Mangostin, the best studied prenylated xanthone from Garcinia mangostana L., induced cell-cycle arrest in G1 and apoptosis through the activation of the intrinsic pathway in DLD-1 human colon cancer cells [29][30][31]. In general various xanthone derivatives exert growth inhibitory and cell death inducing effects in different cell lines and the cellular mechanisms involved can vary widely. In addition to classical apoptotic pathways e.g. [32][33][34] caspase-independent forms of cell death [35] and autophagy [36] have been observed.
Pyranoxanthones are suspected to induce apoptosis by directly binding to Bcl-family proteins [37]. Other xanthone derivatives interfered with down-stream survival signalling [29,34], induced oxidative stress [32,38] or directly interacted with the DNA [33,39]. XB triggered the damage response pathway in SW480 cells by activating Chk1 indicating the involvement of oxidative or replicative stress in the cell response. In a p53-defective tumour cell like SW480 this should cause a G2 cell cycle arrest by inactivating cyclin B/cdc2 kinase activity [40], which is in agreement with our results. It should also protect cells from DNA-damage induced cell death [41] and actually we did not observe classical damage-induced apoptosis in our experiments. By contrast the cells form polyploid giant nuclei without any indication of chromosome condensation indicating an additional target at the G2/M border. This target could be FoxM1 that has gained much interest in this context [42] and is down-regulated in our experiments.  Mitotic catastrophe (MC) has gained interest as response to cancer drugs as well as radiation especially in solid tumours where the induction of classical apoptosis turned out to be rare (for review see [43]). In this context it is a considerable advantage that MC can also be induced independent of p53 function. It can be regarded as a delayed response of p53-mutant tumours that are resistant to replicative damage. Compounds inducing MC like XB could improve the therapy options for solid tumours bearing an inactive p53 protein [24,44] including colon cancer.
A difficulty arises from the rarity of the substance that is mainly contained in the roots of M. rostrata and also only at low concentrations. This is the reason why the analysis of the biological mechanisms underlying XB's cellular mechanism of action had to be limited to proof of principle in this study. It also hampers further development. Subsequent projects have to investigate alternative sources and/or semi-synthetic approaches to produce the compound. Chances are promising, because some several other xanthones have shown growth inhibitory and cell deathinducing effects in different cell lines and trigger various cellular mechanisms of cell death. Text S1 Supplementary information on the isolation and chemical analysis of 2-Deprenyl-rheediaxanthone B (XB). The text contains the supplemental materials and methods as well as table S1 listing published NMR-data of XB.