TAIII suppresses morphological changes and migration of breast cancer cells induced by paracrine interactions with mammary epithelial cells
In our previous study [26], we demonstrated that conditioned medium from the mammary epithelial cell line MCF10A (MCF10A CM) regulates morphological plasticity and promotes migration of the breast cancer cell line MDA-MB-231. To identify natural compounds with the potential activity to disrupt the responses of breast cancer cells to paracrine interactions with mammary epithelial cells, we screened a natural compound library by a scratch wound assay. MDA-MB-231 cells were plated under a serum-deprived condition and treated with each natural compound at a concentration of 1 mM and MCF10A CM (Fig. 1a). As we reported previously [26], baicalein suppressed migration of MDA-MB-231 cells induced by MCF10A CM (Fig. 1b). Although saikosaponin a and d clearly suppressed migration of MDA-MB-231 cells, they also induced cell death during the assays (Fig. 1b). Among the compounds that inhibited migration of MDA-MB-231 cells, TAIII exhibited the strongest activity without cytotoxicity (Fig. 1b and c). The effect of TAIII in conditioned medium-induced migration was confirmed using a Boyden chamber (Fig. 1d). Therefore, we further analysed the effects of TAIII on breast cancer cell migration regulated by paracrine interactions.
First, we examined whether TAIII disrupted the morphological plasticity of breast adenocarcinoma cells regulated by conditioned medium from mammary epithelial cells. We treated MDA-MB-231 cells with MCF10A CM in the presence of TAIII for 1 h and then observed their morphology. As shown in Fig. 2a, MDA-MB-231 cells became elongated and changed their morphology in response to MCF10A CM. However, MCF10A CM-induced morphological changes were perturbed in the presence of TAIII (Fig. 2a). These effects of TAIII were confirmed by quantification of the cell length (Fig. 2b).
Upon MCF10A CM treatment, MDA-MB-231 cells spread and formed lamellipodia at their leading edge and started migration (Fig. 2c and d, and Supplementary Video 1). Thus, we addressed whether TAIII also influenced lamellipodia formation. Time-lapse imaging analysis revealed that TAIII-treated cells did not extend lamellipodia in response to MCF10A CM. The ratio of the lamellipodia area induced by MCF10A CM in TAIII-treated cells was significantly lower than that in control cells (Fig. 2c–e and Supplementary Video 2). These results suggested that TAIII inhibited lamellipodia formation, thereby suppressing morphological plasticity and migration of breast adenocarcinoma cells induced by paracrine interactions with mammary epithelium-derived cells.
Because TAIII suppressed lamellipodia formation upon treatment with MCF10A CM, we next examined whether TAIII also affected lamellipodia formation induced by serum. MDA-MB-231 cells in normal culture medium were treated with TAIII and their morphology was observed. In this experiment, we treated cells with TAIII at a concentration of 10 mM because we did not observe any effects of TAIII at the lower concentration, which was probably due to the presence of protein components in serum, such as albumins and lipoproteins. Time-lapse analysis by phase contrast microscopy revealed that TAIII treatment disrupted lamellipodia formation (Supplementary Fig. S1a and b, and Supplementary Video 3). The number of cells with membrane ruffling was immediately and significantly decreased up to 10 min (10.8±3.79%) and then temporary recovered at 20 min (34.7±1.52%) but decreased again at 60 min (7.88±1.21%) after treatment (Supplementary Fig. S1c). The lamellipodia area in each cell upon TAIII treatment showed similar changes (Supplementary Fig. S1d). Accordingly, the sizes of MDA-MB-231 cells became significantly smaller by approximately 50% after TAIII treatment (Supplementary Fig. S1e). Fluorescence imaging analysis revealed that actin-rich structures at the leading edge had disappeared in MDA-MB-231 cells treated with TAIII (Supplementary Fig. S1f). Together with the data from Fig. 2, TAIII negatively regulated lamellipodia formation induced by extracellular stimuli.
We also treated epidermoid carcinoma A431 cells with TAIII and observed their membrane dynamics. In the presence of serum, A431 cells tightly associated with each other and exhibited expanded lamellipodia in the cell edge and collective migration (Supplementary Fig. S2 and Video 4). However, lamellipodia formation had disappeared and migratory activity was apparently suppressed in TAIII-treated cells, while their cell-cell interactions shown by E-cadherin staining appeared intact (Supplementary Fig. S2 and Supplementary Video 5). These results corroborated the observations in breast adenocarcinoma cells that TAIII exerted inhibitory effects on cell migration by disrupting lamellipodia formation.
TAIII attenuates membrane dynamics regulated by Rac
To further investigate the effect of TAIII on membrane dynamics, we next accessed the effect of TAIII on cell spreading onto ECM. HeLa cells were transfected with the expression plasmid for LifeAct-mCherry, a widely used actin-binding peptide fused to mCherry protein [30], and replated onto a fibronectin-coated glass bottom dish. When the cells had attached to the glass-bottom dish, we treated them with TAIII and examined their behaviours by time-lapse microscopy (Fig. 3a). After attachment to fibronectin, DMSO-treated cells showed membrane blebbing and then formed membrane ruffling and protrusions, followed by rapid spreading to the largest extension (Fig. 3b and Supplementary Video 6). We also observed dynamic remodelling of the actin cytoskeleton in response to the interaction with fibronectin (Fig. 3b and Supplementary Video 6). Moreover, vinculin-positive structures were scattered in the cell edge (Fig. 3c). Conversely, TAIII-treated cells did not form membrane blebs or ruffling. They adhered to the culture dish after replating and formed filopodia-like protrusions but could not spread on the dish (Fig. 3b and Supplementary Video 7). Vinculin-positive structures in TAIII-treated cells were obviously tiny compared with those in DMSO-treated cells (Fig. 3c). Accordingly, the cell area of TAIII-treated cells became significantly smaller, which was dependent on the dose of TAIII (Fig. 3d). These results demonstrated the specific activity of TAIII in the regulation of membrane dynamics, which is important for cell spreading.
Because it is well-known that lamellipodia formation and membrane ruffling is regulated by activated Rac, a Rho family GTPase [31], we addressed whether TAIII influenced activation of Rac. To this end, we first transfected HeLa cells with expression plasmids for green fluorescent protein (GFP) and LifeAct-mCherry and then subjected them to time-lapse imaging analyses (Fig. 4a). As shown by fluorescence signals, HeLa cells continuously formed membrane protrusions and showed ruffling under normal culture conditions (Fig. 4b and Supplementary Videos 8 and 9). However, similar to the other cultured cells examined in this study, these membrane dynamics were suppressed by 60 min of TAIII treatment, followed by expansion of intercellular gaps (Fig. 4b and Supplementary Videos 8 and 9). We also confirmed attenuation of membrane dynamics in response to TAIII treatment by kymograph analysis (Fig. 4b). We next examined membrane and actin cytoskeleton dynamics upon TAIII treatment in HeLa cells that overexpressed GFP-tagged constitutively activated Rac1 (GFP-Val12 Rac1) together with LifeAct-mCherry. Overexpression of GFP-Val12 Rac1 induced cell spreading and frequent membrane ruffling, followed by drastic remodelling of the actin cytoskeleton at the cell edge (Fig. 4c and Supplementary Videos 10 and 11). Interestingly, upon TAIII treatment, membrane ruffling was immediately attenuated and cell contraction was observed (Fig. 4c and Supplementary Videos 10 and 11). The rapid changes of membrane dynamics in Val12 Rac1-expressing cells induced by TAIII treatment were also shown by kymograph analysis (Fig. 4c). These data demonstrated that TAIII controlled membrane ruffling independently of the regulation of Rac activity. We also examined the effects of TAIII on filopodia formation and actin stress fibres regulated by constitutively active CDC42 (Val12 CDC42) and RhoA (Val14 RhoA), respectively. Unlike Val12 Rac1-expressing cells, Val12 CDC42- and Val14 RhoA-expressing cells retained membrane and actin cytoskeleton dynamics even after TAIII treatment (Supplementary Fig. S3 and Supplementary Videos 12–15), which implied that TAIII specifically inhibited membrane dynamics regulated by Rac signalling.
Immunofluorescence analysis revealed that HeLa cells treated with TAIII also showed morphological changes with accumulation of vinculin-positive structures at the cell edge and expanded intercellular gaps (Supplementary Fig. S4). Therefore, we also assessed the structure-activity relationship among saponins isolated from the rhizome of Anemarrhena asphodeloides in the regulation of membrane dynamics (Supplementary Fig. S4a). TAI is a deglycosylated derivative of TAIII, which lacks glucose in its saccharide moiety. TBII and AMS share the steroid core and disaccharide moiety that are identical to those of TAIII. TBII contains an extra sugar moiety at the end of the steroidal side chain. However, AMS does not harbour additional sugar moieties, whereas a hydroxyl group is linked to the C-15 position in its steroid core. We found that HeLa cells treated with these TAIII derivatives did not show any obvious morphological changes similar to those observed after treatment with TAIII (Supplementary Fig. S4b). We also examined sarsasapogenin, an aglycone of TAIII, and found that it had no significant effects on cell morphology or vinculin localisation (Supplementary Fig. S4b). We further tested the natural compound library to determine whether it contained inhibitory activities in the regulation of membrane dynamics. Among the compounds in the library, only shikonin and alkannin affected the morphology of HeLa cells. The cells treated with these compounds became rounded and detached from the culture dish upon treatment, probably because of their cytotoxicity (Supplementary Table S1). Triterpenoid saponins, such as astragaloside IV, ginsenosides (Rb1, Rc, Rd, Re, and Rg1), glycyrrhizic acid, and saikosaponins (a, b2, c, and d), and cardiotonic steroid bufadienolides, such as bufalin and bufotalin (Supplementary Table S1), are listed in the library. However, unlike TAIII, these compounds did not exert any obvious effects on cell morphology. Collectively, these results supported the unique activity of TAIII in the regulation of membrane ruffling.
TAIII inhibits integrin internalisation
We next clarified the mechanisms underlying how TAIII suppressed membrane ruffling. To this end, we labelled TAIII with a fluorescence dye, Alexa Fluor 568, in accordance with the results from the structure-activity relationship among TAIII derivatives and examined its intracellular behaviour. Time-lapse analysis revealed that, after attachment to the plasma membrane, labelled TAIII formed intracellular vesicle-like structures and accumulated in cytoplasm (Fig. 5a and b, and Supplementary Video 16). Intriguingly, these intracellular vesicles and accumulation of labelled TAIII were less observed when cells were pretreated with a dynamin inhibitor, dynasore [32] (Fig. 5a and b, and Supplementary Video 17), which suggested that TAIII was internalised through a dynamin-dependent pathway.
Considering the observation that TAIII was internalised through dynamin-dependent endocytic pathways (Fig. 5), we focused on endocytic trafficking of cell surface proteins. In particular, we examined behaviours of integrin b1 in response to TAIII treatment (Fig. 6a) because TAIII-treated cells showed impaired formation of adhesion complexes positive for vinculin during spreading on fibronectin (Fig. 3c). Cell surface integrin b1 was labelled on ice with an Alexa Fluor 488-conjugated antibody, allowed to internalise at 37°C, and then subjected to time-lapse imaging analysis. Intracellular fluorescence signals from traced integrin b1, which was transported via endosomes on the basis of their estimated size (0.5–2 mm) [29], were observed in cells treated with DMSO up to 20 min after incubation at 37°C and its fluorescence intensity was increased over time (Fig. 6b and c, and Supplementary Videos 18 and 19). Conversely, we did not detect any intracellular signals of internalised integrin β1 in TAIII-treated cells after incubation at 37°C, in which fluorescent signals of LifeAct-mCherry indicated cell contraction, although signals from labelled integrin traced cell shapes prior to TAIII treatment (Fig. 6b and c, and Supplementary Videos 20 and 21). Interestingly, internalisation of integrin β1 was clearly suppressed, whereas continuous internalisation and intracellular accumulation of labelled TAIII were observed (Supplementary Fig. S5 and Supplementary Videos 22 and 23). We next examined the intracellular dynamics of integrin b1 and TAIII. We labelled cell surface integrin b1 and incubated the cells at 37°C for 1 h to allow internalisation and then treated the cells with labelled TAIII (Supplementary Fig. S6a). Whereas any colocalisation signals were not observed on the cell surface, some internalised integrin b1 and TAIII had merged in the cytoplasm (Supplementary Fig. S6b and Supplementary Videos 24–26). We also found that TAIII suppressed internalisation of E-cadherin induced by EGTA in A431 cells (Supplementary Fig. S7). Collectively, these results suggested that TAIII inhibited machineries that regulated internalisation of specific cargo proteins such as integrin b1.
To further elucidate the effect of TAIII on integrin mediated cell functions, we performed cell adhesion assays with various ECM proteins. Trypsinised HeLa cells were treated with TAIII for 30 min and replated on culture dishes coated with each ECM protein. After incubation for another 30 min, cells associated with the ECM were measured. We examined purified fibronectin, laminin-332, and vitronectin that serve as ligands for integrins a5b1, a3b1, and avb1, respectively, which are expressed in HeLa cells [33]. Adhesion assays revealed that TAIII significantly interfered with the substantial interaction of HeLa cells with purified integrin ligands in a dose-dependent manner (Fig. 6d), which demonstrated that TAIII targeted integrin heterodimers consisting of at least integrin b1. Thus, the results suggested that the effect of TAIII on the regulation of internalisation of cell surface proteins, such as integrin b1, contributed in part to its inhibitory activity against cell migration through suppression of lamellipodia formation and membrane ruffling.