Cytotoxicity and Pharmacogenomics of Medicinal Plants from Traditional Korean Medicine

Aim. The present study was designed to investigate the cytotoxicity of a panel of 280 Korean medicinal plants belonging to 73 families and 198 species against human CCRF-CEM leukemia cells. Selected phytochemicals were investigated in more detail for their mode of action. Methods. The resazurin assay was used to determine cytotoxicity of the plant extracts. Microarray-based mRNA expression profiling, COMPARE, and hierarchical cluster analyses were applied to identify which genes correlate with sensitivity or resistance to selected phytochemicals of the Korean plants. Results. The results of the resazurin assay showed that cytotoxicity extracts tested at 10 μg/mL from 13 samples inhibited proliferation more than 50% (IC50 < 10 μg/mL) and the most active plants are Sedum middendorffianum (15.33%) and Lycoris radiata (17.61%). Out of 13 selected phytochemicals from these plants, hopeaphenol and deoxynarciclasine were the most cytotoxic ones. Genes from various functional groups (transcriptional or translational regulation, signal transduction, cellular proliferation, intracellular trafficking, RNA metabolism, endoplasmic/sarcoplasmic reticulum function, etc.) were significantly correlated with response of tumor cell lines to these two compounds. Conclusion. The results provide evidence on the possible use of selected Korean medicinal plants and chemical constituents derived from them for the treatment of tumors.


Introduction
Since ancient times, humans have derived many benefits from medicinal plants. A variety of different medicinal plants have traditionally been used in Asian cultures as medicinal plants to treat cancers [1]. The pharmacological screening of plants is an important mean for the discovery of new, safe, and effective drugs. Over 50,000 plants, possess therapeutic virtues in the world, and about 80% of human use herbal medicines at least once in their life [2,3]. Medicinal plants through the multiplicities of their chemical constituents are important for the discovery of new substances active against tumors and other cancers. Screenings of medicinal plants used as anticancer drugs have provided modern medicine with effective cytotoxic pharmaceuticals. More than 60% of the approved anticancer drugs in the United State of America (from 1983 to 1994) were from natural origin [4,5].
Traditional Korean medicine is widely used in Korea and is the primary health care system for more than 20% of the population [6,7]. As previously emphasized in a regional demographic survey, about 30-40% of the Korean population had used complementary and alternative medicine (including traditional Korean medicine) within a 5-year period [6,8].
In the pharmacopoeia of many countries worldwide including Korea, there is still a serious lack of information on the use of medicinal plants in the treatment of cancer. However, it has been recommended that ethnopharmacological usages such as for immune and skin disorders, inflammatory, infectious, parasitic, and viral diseases should be taken into account when selecting plants used to treat cancer, since these reflect disease states bearing relevance to cancer or cancerlike symptoms [9,10]. In general, leukemia cells are often more sensitive to cytotoxic drugs than adherent cancer cells, making them a suitable tool for primary bioactivity screenings. Hence, the present work was designed to investigate the cytotoxicity of a panel of 280 Korean medicinal plants against human CCRF-CEM leukemia cells. Based on the bioactivity of the extracts, selected phytochemicals of active plants were analyzed in more detail. Genes determining sensitivity or resistance to selected compounds were identified by microarray-based mRNA expression profiles and hierarchical cluster analyses in a panel of tumor cell lines of the National Cancer Institute, USA.  betulin; betulinic aldehyde; 3-acetoxybetulinic aldehyde, -sitosterol [45] Camellia japonica L.

Material and Methods
Cosmetic protectant to keep the skin and hair healthy and as a soothing agent [47] Fruits Antibacterial activity [48], inhibitor of human immunodeficiency virus type 1 protease [49], Epstein-Barr virus inhibitor [50], antimetastasis activity [51], antioxidant activity [52,53], inhibitor of human type I procollagen production [54], and antiallergic responses [55], anti-inflammatory [47] 3 ,18 -dihydroxy-28-norolean-12-en-16-one; 18 -hydroxy-28-norolean-12-ene-3,16-dione; camelliagenin A, B, and C [56],  28-hydroxy-d-friedoolean-14-en-3-one (); myricanol Rubus corchorifolius L. f.  the NCI cell line panel has been described [12,13] and the date has been deposited at the NCI website (http://dtp.nci.nih. gov/). For hierarchical cluster analysis, objects were classified by calculation of distances according to the closeness of between-individual distances by means of hierarchical cluster analysis. All objects were assembled into a cluster tree (dendrogram). The merging of objects with similar features leads to the formation of a cluster, where the length of the branch indicates the degree of relation. The distance of a subordinate cluster to a superior cluster represents a criterion for the closeness of clusters as well as for the affiliation of single objects to clusters. Thus, objects with tightly related features appear together, while the separation in the cluster tree increases with progressive dissimilarity. Previously, cluster models have been validated for gene expression profiling and for approaching molecular pharmacology of cancer [12][13][14]. Hierarchical cluster analyses applying the WARD method were done with the WinSTAT program (Kalmia, Cambridge, MA, USA). Missing values were automatically omitted by the program, and the closeness of two joined objects was calculated by the number of data points they contained. In order to calculate distances between all variables included in the analysis, the program automatically standardizes the variables by transforming the data with a mean = 0 and a variance = 1. For COMPARE analysis, the mRNA expression values of genes of interest and IC 50 values for selected phytochemicals of the NCI cell lines were selected from the NCI database (http://dtp.nci.nih.gov/). The mRNA expression has been determined by microarray analyses as reported [12]. COMPARE analyses were performed by a web-based algorithm (http://dtp.nci.nih.gov/) to produce rank-ordered lists of genes expressed in the NCI cell lines. The methodology has been described previously in detail [15]. Briefly, every gene of the NCI microarray database was ranked for similarity of its mRNA expression to the IC 50 values for the corresponding compound. Scale indices of correlations coefficients ( -values) were created. Pearson's correlation coefficients with positive algebraic signs indicate that greater mRNA expression in cell lines correlate with enhanced drug resistance, whereas coefficients with negative algebraic signs indicate that greater mRNA expression in cell lines was associated with drug sensitivity. Pearson's correlation test and 2 test were implemented into the WinSTAT Program (Kalmia). The one-way ANOVA at 95% confidence level was used for statistical analysis.

Cytotoxic Activity.
In the present work, we screened 280 plant extracts derived from traditional Korean medicine belonging to 73 plant families and 198 species (see Supplementary Table) for their cytotoxicity against human CCRF-CEM leukemia cells.
The results of the cytotoxicity assay ( Figure 1; Supplementary Table) indicated that is among the 280 studied plant extracts, 58 promoted the growth (% growth > 100%) of CCRF-CEM cells, while the remaining samples showed various extents of growth inhibition. Out of these extracts, 207 extracts showed no or a weak inhibition of cancer cell proliferation (<50%) at the tested concentration of 10 g/mL. However, 13 samples induced the proliferation of less 50% cells (IC 50 < 10 g/mL). These samples included  Table 1.

Cytotoxicity of Phytochemicals Derived from Koran Medicinal Plants.
We searched the literature on the chemical constituents of the cytotoxic Korean plants (Table 1). Subsequently, we mined the NCI database for these compounds. If the average IC 50 values over the entire range of cell lines were diversified regarding their tumor types, colon cancer and melanoma cell lines were most sensitive towards hopeaphenol, whereas leukemia and kidney cancer cell lines were most resistant (Figure 3(a)). The cell lines reacted in a different manner towards deoxynarciclasine. Leukemia and colon cancer were most sensitive towards this compound, whereas breast cancer and prostate cancer cell lines were most resistant (Figure 3(b)).

Cross-Resistance of the NCI Cell Line Panel towards
Hopeaphenol and Deoxynarciclasine. The log 10 IC 50 values of hopeaphenol and deoxynarciclasine were correlated with clinically well-established anticancer drugs. The cell line panel showed statistically significant correlations between deoxynarciclasine and doxorubicin, daunorubicin, vincristine, paclitaxel, cisplatin, melphalan and carmustin. By contrast, cross-resistance was not found between hopeaphenol and these standard drugs, indicating that hopeaphenol might be useful for the treatment of otherwise drug-resistant tumors (Table 2). Interestingly, an inverse correlation was found between the log 10 IC 50 values for hopeaphenol and the epidermal growth factor receptor tyrosine kinase inhibitor, erlotinib. Such a relationship was not found between deoxynarciclasine and erlotinib (Table 2).

COMPARE and Hierarchical Cluster Analyses of mRNA Microarray Data.
We further investigated the microarraybased transcriptomic mRNA expression by COMPARE analyses to test whether the responses of the tumor cells lines to hopeaphenol and deoxynarciclasine were associated with specific gene expression profiles. For this reason, we mined the transcriptome-wide mRNA expression database of the NCI and correlated the expression data with the IC 50 values for both compounds. This represents a hypothesis-generating bioinformatical approach, which allows the identification of novel putative molecular determinants of cellular response for hopeaphenol and deoxynarciclasine. The scale rankings of genes obtained by COMPARE computation were subjected to Pearson's rank correlation tests. The thresholds for correlation coefficients were > 0.50 for direct correlations and < −0.50 for inverse correlations. The genes fulfilling these criteria are shown in Table 3 (for hopeaphenol) and Table 4 (for deoxynarciclasine). These genes were from diverse functional groups for hopeaphenol (Table 3). As a next step, the genes identified by COMPARE analyses and Pearson's rank correlation tests were subjected to hierarchical cluster analyses. Only the mRNA expression levels but not the IC 50 values of the compounds for the cell line panel were used for cluster analyses. Four cluster branches were found in the hopeaphenol-related dendrogram ( Figure 4) and three clusters were obtained in the deoxynarcilasine-related cluster analysis ( Figure 5). Remarkably, the distribution of cell lines sensitive or resistance to both drugs considerably varied between the different clusters of the dendrograms. Since the IC 50 values of the compounds were not included into the cluster analysis, we could address the question whether the gene expression profiles alone are sufficient to predict sensitivity or resistance of cell lines to the compounds.
As shown in Table 5, the distributions of cell lines sensitive or resistant to hopeaphenol and deoxynarciclasine, respectively, were significantly different between the clusters of the dendrograms, indicating that the expression of these genes was not only responsible for the branching of the dendrograms, but also predicted sensitivity or resistance of these cell lines towards these compounds.

Cytotoxic Activity.
The cytotoxicity observed in Isodon. japonicus in this work is in concordance with previous reports. In effect, the cytotoxic effect of I. japonicus extract was reported against five human cancer cell lines (IC 50 values below 10 g/mL on stomach MKN-45, breast MCF-7, leukemia K562, colon HT29, and lung A549 cell lines) with leukemia K562 (IC 50 : 2.70 g/mL) being the most sensitive [16]. Also, the ethanol extract of Lycors radiata exhibited significant antiproliferative effect against B16F10 melanoma cells and induced apoptosis through the activation of p38 and AP-1 pathway [17]. The present report, therefore, provides evidence on the activity of L. radiata not only against cell lines derived from solid tumors but also derived from the   hematopoietic system, that is, CCRF-CEM leukemia cells. Sorbaria sorbifolia was cytotoxic towards HepG-2 cells via induction of apoptosis and cell cycle arrest [18], and evidence of this activity towards CCRF-CEM leukemia cells is provided herein. In the US NCI plant screening program, a crude extract is generally considered to have in vitro cytotoxic activity, if the IC 50 value following incubation between 48 and 72 h is less than 20 g/mL [19]. In this study, we reduced the cutoff point to 10 g/mL. All extracts were tested at this concentration and only samples with an inhibitory effect >50% were considered to have highly promising activities against leukemia cells. Under this condition, 13 samples from 12 medicinal plants ( Figure 1) were identified as promising anticancer products and should be screened for more cancer cell lines. It is noteworthy that only Lycoris radiata exhibited a significant (<50% growth proliferation) activity with both aerial and subterraneal parts, suggesting that different parts of the plant should be considered when screening the cytotoxicity of medicinal plants.
To the best of our knowledge, the cytotoxic effect of several active plants (Table 1) 55.47 M [20]. Interestingly, the extract of the stem bark of this plant exhibited significant antiproliferative activity on leukemia CCRF-CEM cells (Figure 1), inducing less than 25% growth at 10 g/mL. In addition, Kim et al. [21] reported the presence of two wellknown cytotoxic compounds, betulin and lupeol [22] in the extract of this plant. Apoptosis induction via Fas-mediated pathway was also reported in human MCF-7 breast adenocarcinoma cells by prodelphinidin B-2 3,3 -di-O-gallate, a constituent of M. rubra [23]. The presence of such cytotoxic compounds could explain the good activity of the crude extract of A. japonica and M. rubra. Similarly, the presence of cytotoxic compounds such as quercetin or ferulic acid [22]  and Sedum takesimense [26] could also provide some explanation on their antiproliferative potentials. Nonetheless, it is interesting to know that the activity does not only depend on the presence of a cytotoxic substance in a plant, but also their quantities and possible interaction with other plant constituents. For example, cytotoxic compounds such as incanone {known to be active on HL60 leukemia cells (IC 50 value of 6 M) [25])} or verbascoside [27] {active on human HEP-2 larynx epidermoid carcinoma, human RD rhabdomyosarcoma and human MCF-7 breast adenocarcinoma cell lines [28]} were reported in Caryopteris incana. But extract from Caryopteris incana did not show significant activities against CCRF-CEM leukemia cells in the present study. We screened the NCI database of the Developmental Therapeutics Program of the NCI for the constituents of our panel of Korean medicinal plants. The two most cytotoxic compounds were hopeaphenol and deoxynarciclasine. Both compounds have been previously reported to be cytotoxic [29][30][31][32]. While hopeaphenol's activity has been demonstrated in mouse tumors in vitro and in vivo [31,32], the present investigation shows that this compounds also active against human tumor cells. To the best of our knowledge, the mechanisms of action of these two compounds have not been investigated in the past and are addressed in our analysis for the first time.

COMPARE and Hierarchical Cluster Analyses of mRNA
Microarray Data. To gain insight into possible modes of action of both phytochemicals, we investigated gene expression profiles of the NCI cell line panel. By microarray-based gene expression and COMPARE analyses, we correlated the IC 50 values for both compounds of 60 tumor cell lines with transcriptomic mRNA expression levels of this cell line panel [12]. This approach has been successfully used to unravel the mode of action of novel compounds [33]. Cluster and COMPARE analyses are also useful for comparing gene expression profiles with IC 50 values for investigational drugs to identify candidate genes for drug resistance [34,35] and to identify prognostic expression profiles in clinical oncology [36,37].
Eleven genes passing the correlation thresholds of > 0.5 and < −0.5 were identified to significantly correlate with sensitivity or resistance to hopeaphenol. Except for fibroblast growth factor 9 [38], none of them have been associated with response of tumor cells to cytostatic drugs. This point of view is supported by the fact that the NCI cell lines did not exert cross-resistance between hopeaphenol and anticancer drugs, such as doxorubicin, vincristine, cisplatin, and others.
The genes correlating with the response of tumor cells to deoxynarciclasine were also not known to confer drug resistance as of yet, but many of them belong to functional  groups, which are associated with response to chemotherapy. For example, transcriptional regulation, signal transduction, and cell proliferation are well-known processes influencing the success of cancer chemotherapy [39][40][41][42]. This might also explain why the NCI cell line panel exhibited cross-resistance between deoxynarciclasine and standard chemotherapy. On the other hand, genes involved in cellular trafficking or endoplasmic/sarcoplasmic reticulum functions have not been recognized as possible mechanisms of drug resistance of tumors. This finding merits more detailed investigations in the future.

Conclusion.
The present work provides evidence that some plants derived from Korean medicine could be useful in the treatment of leukemia and supports the advanced investigation of the most active plants extracts. It also provides first pharmacological data on the cytotoxicity of some plants, such as Adenophora racemosa, Cinnamomum japonicum, Eurya japonica, Sedum middendorffianum, and Vitis flexuosa. The identification of cytotoxic phytochemicals from these plants, for example, hopeaphenol, and deoxynarciclasine, and the investigation of their possible molecular modes of action may foster the development of novel treatment strategies against otherwise drug-resistant and refractory tumors.

Conflict of Interests
No potential conflict of interests was disclosed.

Authors' Contribution
Victor Kuete and Ean-Jeong Seo contributed equally to this paper.