Anti-gastric cancer activity in three-dimensional tumor spheroids of bufadienolides

Multicellular spheroids of cancer cells have been increasingly used to screen anti-tumor compounds, owing to their in vivo like microenvironment and structure as well as compatibility to high-throughput/high-content screening. Here we report the potency and efficacy of a family of bufadienolides to inhibit the growth of gastric cancer cell line HGC-27 in three-dimensional (3D) spheroidal models. Examining the morphological and growth patterns of several cell lines in round-bottomed ultra-low attachment microplate suggested that HGC-27 cells formed reproducibly multicellular spheroidal structures. Profiling of 15 natural bufadienolides isolated from toad skin indicated that 8 14-hydroxy bufadienolides displayed inhibitory activity of the growth of HGC-27 spheroids in a dose-dependent manner. Notably, compared to clinical drugs taxol and epirubicin, active bufadienolides were found to penetrate more effectively into the HGC-27 spheroids, but with a narrower effective concentration range and a shorter lasting inhibitory effect. Furthermore, compared to two-dimensional (2D) cell monolayer assays, active bufadienolides exhibited weaker efficacy and different potency in 3D spheroid model, demonstrating the great potential of 3D multicellular cell spheroid models in anti-cancer drug discovery and development.

bufadienolides discovered via isolation from animals and plants or biotransformation 36 . These compounds have attracted great attention, since they show a wide range of bioactivities, such as renal sodium excretion, blood pressure stimulating 37,38 , immunoregulatory 39,40 , and anti-tumor activities towards diverse cancer cell lines, including hepatoma, lung carcinoma, pancreatic, gastro-intestinal and breast cancers 36,[41][42][43] . These studies focus on their anti-tumor activities on 2D monolayer cells. Little is known about the activity of bufadienolides to inhibit the growth of cancer cells in 3D spheroid models.
Here, we aim to investigate anti-gastric cancer activity of natural bufadienolides using 3D spheroid-based functional assays.

Results
Establishment of 3D spheroid-based assays. To establish a 3D spheroid assay, we first examined the ability of multiple human cancer cell lines to form spheroid in the 96-well ULA round-bottomed plate. These cell lines included gastric cancer cell line HGC-27, colon carcinoma cell line HT-29, breast cancer cell lines MDA-MB-231 and SUM-159, lung cancer cell line A549 and hepatoma cell lines Hep G2, PLC/PRF/5 and SK-HEP-1. Results showed that different cell lines formed distinct types of spheroidal structures (Fig. 1). According to their spheroid morphology and the classification method in reference 18,44 , these spheroids were classified: tight spheroids (HGC-27, HT-29 and SUM-159), compact aggregates (MDA-MB-231 and A549), and loose aggregates (Hep G2, PLC/PRF/5 and SK-HEP-1). Interestingly, although HT-29 and SUM-159 cells formed tight spheroids, cells were dissociated from the spheroidal structure after treatment with anti-cancer drugs (data not shown), resulting in difficulty in measuring their diameters. In contrast, the gastric cancer cell line HGC-27 displayed a well-defined spheroid morphology in shape and was well packed, thus was chosen to develop 3D spheroid assays for compound profiling.
The physiological state of cells in spheroid depends on the spheroid size. A standard size of 370-400 μm after 4-day incubation was frequently selected for drug testing 29 . To obtain the optimal size of HGC-27 multicellular spheroids for screening, we further characterized its growth patterns as a function of initial seeding density. Results showed that cells with different seeding densities all formed tight and regular spheroids (Fig. 2a), but with different kinetics in increasing diameters over time (Fig. 2b). The size of spheroids was proportional to the initial cell numbers. With 300 cells per well, the spheroid diameter was approximately approaching 400 μm at day 4 with a low coefficient of variation (CV) (352.10 ± 14.62 μm, n = 240) in spheroid diameters, indicating great reproducibility in spheroid formation (Fig. 2c). Therefore, in the following studies, the cells with a seeding density of 300 cells per well cultured for 4 days in 96-well ULA round-bottomed plates were used for anti-cancer activity assays.
Screening of anti-cancer activity of bufadienolides. We first profiled the anti-cancer activity of 15 bufadienolides (Fig. 3), each at 400 nM, using the HGC-27 3D spheroid growth assays. These bufadienolides belong to 14-hydroxy compounds (Group A) and 14, 15-epoxy compounds (Group B). Results showed that after treated with Group A compounds (except No. 4 bufotalin) the 4-day-old spheroids stopped growing as its size remained the same as the 4 th day for the next 6 days (Fig. 4a,c). On the other hand, compounds in Group B displayed overall weak inhibitory effects in spheroid growth (Fig. 4b,d). These results suggested that 14-hydroxy compounds had stronger anti-gastric cancer activities than 14, 15-epoxy compounds.
Pharmacology of active bufadienolides. We next examined the potency of active bufadienolides (compounds in Group A except No. 4) on HGC-27 3D multicellular spheroids. Results showed that all eight compounds tested completely stopped the spheroid growth when assayed at high concentration (Fig. 5). However, different compounds had different potency (Table 1), as well as different effective concentration range. Among them, bufalin (No. 1) had the widest effective concentration range, and telocinobufagin-3-suberoylarginine ester (No. 8) had the narrowest effective concentration range (Fig. 5a). For comparison, taxol and epirubicinboth had wide concentration range (Fig. 6a). These results suggest that active bufadienolides possessed the narrow effective concentration range to inhibit the growth of HGC-27 cells in spheroid.
Based on their inhibitory potency (Figs 3 and 5b, Table 1), bufalin (No. 1) was the most potent inhibitor. Compared to bufalin, introduction of 3-ketone (No. 5), 5-hydroxyl (No. 2) and 16-hydroxyl (No. 3) substituents reduced its potency. Introduction of ) and 14,15-epoxide (No. 10) rendered compounds inactive. Introduction of 11-ketone and 12-hydroxyl (No. 6) made no significant difference in the degree of inhibitory activity. However, transformation of their positions (No. 7) would obviously lower the activity. Compounds No. 8 and No. 9 displayed weaker inhibitory activities than compounds No. 2 and No. 7, respectively, meaning that the large substituent at the C-3 position was unfavorable to the inhibitory activity.
Interestingly, different compounds also gave rise to distinct treatment duration dependent potency (Fig. 7). All bufadienolides exhibited decreasing potency as the treatment duration increases, opposite to both taxol and epirubicin, suggesting that these two clinical drugs exhibited longer lasting inhibitory effect on HGC-27 spheroids than bufadienolides. On the other hand, examining the morphology of HGC-27 spheroids after treatment for 6 days with these compounds (Fig. 5c) or drugs (Fig. 6c), revealed that similar to taxol and epirubicin, bufadienolides effectively penetrated the multiple layers of HGC-27 spheroids, as the outer-layer cells detached from spheroids due to exposure to compounds, possibly due to reduced cell-cell adhesion. Comparison anti-gastric cancer activity of bufadienolides in 3D multicellular spheroids with 2D monolayer cells. Finally, we compared the pharmacology of bufadienolides in 3D tumor spheroids with 2D monolayer cells. Results showed that these compounds generally exhibited poorer efficacy, but different potency in 3D spheroid assay, compared to 2D cell monolayer assay (Fig. 8). Furthermore, the inhibitory potency of compounds in 3D spheroid-based assays decreased in the order: 8. However, the order of inhibitory potency in 2D cell monolayer assays was 8. Both 3D tumor spheroids and 2D cell monolayer assays showed three compounds No. 1, No. 6 and No. 2 all had better inhibitory potency than market drugs epirubicin and taxol. They may offer new therapeutic candidates to develop new treatment of gastric cancer. These results suggest that drug molecules may display different pharmacology in 3D multicellular spheroid models.

Discussion
Bufadienolides display inhibitory activities towards various cancer cells 43,45,46 , which have great potential for becoming anti-cancer drug candidates. Current studies have focused on evaluation of the inhibitory activity of bufadienolides via 2D cell monolayer assays. However, owing to this highly artificial environment, 2D monolayer cells cannot mimic the pathophysiology of in vivo tumor, resulting in limitation for prediction of in vivo anti-cancer activities. 3D multicellular spheroids closely reflect the in vivo tumor characteristics, leading to better prediction power as a useful and effective technique for investigating anti-cancer activity of drugs [4][5][6] . Given that natural bufadienolides possess better inhibitory activity than derived bufadienolides 41,43 , 15 natural bufadienolides isolated from the skin of Bufo bufo gargarizans Cantor (toad skin) were screened and evaluated using 3D spheroid gastric cancer model. Here, HGC-27 cells were found to spontaneously form the tightest spheroids in 96-well ULA round-bottomed microplate with high reproducibility. So 3D spheroid-based assays of HGC-27 cell line were used for compound screening and evaluation. These assays consisted of three steps: (1) spheroid formation via culturing cells at the optimal density for 4 days in 96-well ULA plate; (2) microscope imaging from the 4 th day to the 10 th day after addition of drugs or compounds on the 4 th day; (3) data process and analysis. According to procedures of 3D spheroid-based assays, 15 natural bufadienolides were screened, including 9 14-hydroxy compounds and 6 14, 15-epoxy compounds. Among them, 14-hydroxy compounds (except No. 4 bufotalin) had stronger anti-cancer activities than 14, 15-epoxy compounds, suggesting the importance of a hydroxy group in the C-14 position for the anti-cancer activities, which was consistent with result obtained on 2D cell monolayer assays and reported by Kamano et al. 41 . Interestingly, bufotalin (No. 4) belonged to 14-hydroxy compounds but displayed weak inhibitory activities. Overall, bufadienolide-induced growth inhibitory activity was validated in 3D spheroid-based assays.
Given that 3D spheroid-based assays can provide real-time quantitative kinetic analyses and detailed morphological investigations 18,29,34 , the growth kinetic curves, concentration-dependent curves and morphology of HGC-27 spheroids after treatment by active bufadienolides can be obtained. This discovery is appealing for investigation of pharmacology characteristics of drugs in vivo. According to their growth kinetic curves, active bufadienolides had relative narrow effective concentration range compared to taxol and epirubicin, which was unfavorable for their use in clinic due to their severe cardiotoxicity associated with specific inhibition of Na + / K + -ATPase [47][48][49] . As reported, bufalin was a potent inhibitor of Na + /K + -ATPase, with a cardiotoxicity range of 1 ~100 nM 50 . And its concentration range for inhibiting the growth of 3D multicellular spheroids was 0.64 ~16 nM. Thus, the narrow effective concentration range of bufadienolides for inhibiting the growth of 3D multicellular would be within the concentration range of cardiotoxicity, limiting their usage in clinic. Based on their inhibitory potency calculated by concentration-dependent curves, reasonable structure-activity relationships (SARs) were summarized and the change trend of IC 50 values with time was acquired. It was found that large hydrophilic substituent at the C-3 position reduced the inhibitory potency, which could be attributed to the possibility that the hydrophilicity of the compounds may have an adverse effect on delivering compounds into the inner layer of 3D spheroids 51 . Compared to the long-lasting inhibitory effect on HGC-27 spheroids of the two clinical drugs, the inhibitory effect of active bufadienolides lasted for about three days and higher concentration of compounds was needed for longer time treatment, possibly due to drug-resistance 52,53 . According to spheroid morphology, active bufadienolides could effect cell-cell adhesion and effectively penetrate the multiple layers of HGC-27 spheroids, resulting in inhibiting their growth. These results provided important parameters for the development of their clinical usage.
Compared to 2D cell monolayer assays, compounds displayed different inhibitory activities in 3D spheroid-based assays. Firstly, all active bufadienolides showed weaker inhibitory efficacy in 3D spheroid-based assays, indicating that reduced compounds assessed to the center of multicellular spheroids. For instance, bufotalin (No. 4) had a higher inhibitory activity in 2D cell monolayer assay 41 , but almost lost its activities in 3D spheroid screening assays. Secondly, less bufadienolides were more active than clinical drug epirubicin in 3D spheroid-based assays. 7 bufadienolides had higher activities than epirubicin in 2D cell monolayer assays, whereas only 3 bufadienolides had higher activities than epirubicin in 3D spheroid-based assays. Moreover, their potency orders were very different. Thirdly, a more referable potency was obtained in 3D spheroid-based assays.  , IC 50 5 and IC 50 6 are the IC 50 values of bufadienolides on HGC-27 3D spheroids after treatment for 1, 2, 3, 4, 5 and 6 days, respectively. (2) "-" means that the inhibitory active is weak.
Scientific RepoRts | 6:24772 | DOI: 10.1038/srep24772 The IC 50 values of taxol and epirubicin on HGC-27 3D spheroids after treatment for 6 days were 0.062 ± 0.031 μM and 0.036 ± 0.015 μM, respectively. These values were within the range used in treatment 54,55 , confirming that 3D spheroid-based assays developed in this work could provide highly relevant screen for compounds. Therefore, the inhibitory activity of bufadienolides studied in this work would offer useful information on their related drug optimization and development.

Materials.
Taxol and epirubicin hydrochloride were obtained from Melone biotechnology Co., Ltd. (Dalian, China). All bufadienolides were separated and prepared from toad skin in house using high performance liquid chromatography 56,57   SK-HEP-1 cell lines were cultured in MEM medium. Except RPMI 1640 medium with 20% fetal bovine serum (FBS), all culture mediums contained 10% FBS. In addition, all culture mediums contained 50 μg/mL penicillin and 100 μg/mL streptomycin. All cell lines were cultured in a humidified 37 °C/5% CO 2 incubator. Spheroid formation and screening of bufadienolides. Spheroids of different cell lines were formed in 96-well ULA round-bottomed plates by dispensing 200 μL cell suspensions with seeding density of 5000 cells per well. The morphology of multicellular spheroids were recorded using an inverted light microscope (Olympus CKX41, Olympus Co. Ltd., Tokyo, Japan) to choose proper cell lines.
Spheroids of HGC-27 cells were formed in 96-well ULA round-bottomed plates by dispensing 200 μL cell suspensions at different seeding densities 20000, 10000, 5000, 2500, 625, 313 and 156 cells per well. Optimal density for spheroid formation and screening drugs was assessed based on the size and shape of each seeding density from day 1 to day 4.
Cell suspension at the optimal density was seeded in 96-well ULA round-bottomed plates and 3D multicellular spheroids were spontaneously generated after 4 days culture. On day 4, 110 μL of the old media was exchanged by 100 μL of the fresh media in each well, and 10 μL of compound solution was then added. The size and shape of multicellular spheroids were recorded using an inverted light microscope from day 4 to day 10. During media changes, the pipet tip was held at about a 45° angle away from the center of the well to protect spheroids from disruption.  Cell viability assay. The effects of all bufadienolides, taxol and epirubicin on HGC-27 cell proliferation in 2D monolayer culture were evaluated using CCK8 assay, each in four replicates. Specifically, HGC-27 cells were seeded in 96-well flat-bottomed plates at 1 × 10 4 cells per well. After culture for 24 h, 100 μL of the old media was exchanged by 90 μL of the fresh media in each well, and 10 μL of compound solution was then added. The control was the wells that cells were treated by DMSO vehicle at a concentration equal to that in compound-treated cells. After treatment for 24 h, the media was removed from the wells and 100 μL of the media without FBS but containing 10% CCK8 was added in each well. After incubation at 37 °C for 2 h, the absorbance of each well was measured at 450 nm using a microplate reader (Bio-Rad, iMark, USA). Data analysis. The diameters of 3D multicellular spheroids were measured using NIH ImageJ (http://imagej. nih.gov/ij/). The suppression of the multicellular spheroid growth was normalized by the control. Data process and analysis were performed on Microsoft excel 2010 and GraphPad Prism 6.02 (GraphPad Sofware Inc., San Diego, CA, USA). All IC 50 values described in this work were calculated based on two independent measurements, each in duplicate (n = 4).