With various concentrations of B5G9 for 12 h, cell viability was measured by MTT assay. *P 0.01, ***P 0.the presence of pyruvate and uridine. Moreover, MitoSOX Red fluorescence was markedly elevated in wt HepG2 cells treated with B5G9, whereas the fluorescence in 0 HepG2 cells only slightly increased (Fig. 7d), which further indicated that B5G9-induced ROS generation was derived from mitochondria. The sensitivity of wt and 0 HepG2 cells to B5G9 treatment was measured with MTT assays. As expected, 0 HepG2 cells were resistant to B5G9 (Fig. 7e), indicating that mitochondria play an important role in B5G9-induced ROS-dependent cell death.Discussion Traditional Chinese medicine (TCM) is an abundant source of anti-HCC agents. Recently, the modification of natural products from TCM has become a promising approach for novel drug development. To enhance the antitumour efficacy of 23-HBA and reduce its side effects, we performed a series of structural modifications of 23-HBA and conducted a bioactivity screening [24]. We found that a piperazidine derivative of 23-HBA,B5G9, showed excellent anti-HCC activity both in vitro and in vivo. In the current study, we evaluated the anti-HCC activity of B5G9 on four HCC cell lines, including HepG2, Bel-7402 and Hep3B cells (p53-null), and multidrugresistant HepG2/ADM cells. B5G9 showed comparable SB 202190 biological activity cytotoxicity in the four hepatoma cell lines, which indicates the potential of B5G9 in the treatment of various types of HCC cells, regardless of p53 status. B5G9 also exhibited outstanding anti-proliferation activity on HepG2 tumour xenografts without significant toxicity, whereas 23-HBA had a minimal effect. ROS generation is considered the key mechanism for the anti-cancer activity of BA and 23-HBA [27, 28, 53]. However, the source of ROS was unclear. In the Human Oxidative Stress Plus PCR Array, we surprisingly found that B5G9-induced mitochondria-specific SOD2 upregulation was completely inhibited by antioxidants. The results indicated that the mitochondria might undergo oxidative stress. This hypothesis was confirmed by the co-localization of H2DCFDA and Mito-Tracker as well as MitoSOX Red staining. We found that 23-HBA had aYao et al. Journal of Experimental Clinical Cancer Research (2016) 35:Page 13 ofsimilar but weaker effect, which could explain why B5G9 had greater anti-HCC activity than that of 23-HBA; however, B5G9 may have a different underlying mechanism of mitochondrial dysfunction. Moreover, we noted that B5G9 just induced a slight increase of mitochondrial ROS and apoptosis in normal liver LO2 cells compare to HepG2 cells (Additional file 2: Figure S1A B) which indicated its selective cytotoxicity against HCC cells. These results also implied the important role of mitochondrial ROS in B5G9-induced apoptosis. In addition to mitochondria, NADPH oxidase, XO, LOX, COX and cytochrome p450 contribute to ROS overproduction, but the results showed that inhibitors of these enzymes failed to abolish B5G9-induced ROS production. All these data indicate that mitochondria are the major site of B5G9-induced ROS production. Inhibition of mitochondrial complex activity (especially complex I and III) can induce ROS production [54, 55]. We further measured mitochondrial complex activities upon B5G9 treatment by cell-based assay. As shown in Additional file 2: PubMed ID:https://www.ncbi.nlm.nih.gov/pubmed/28878015 Figure S1C, B5G9 treatment (0.5, 1 and 3 h) had no effect on activities of mitochondrial complex I, III and V. In our study, B5G9induced mitochondrial-R.
