3.1. The obtaining of EKC and its analogues
Epimedium koreanum is a good resource of prenylated flavonoids. Our phytochemical study on the leaves of this plant led to the isolation of EKC and other 26 prenylated flavonoids. The isolation process and structures of the isolated prenylated compounds were shown in Additional file 1 and Additional file 2. The structure of the compound was determined by analysis of the spectroscopic data (Fig. 1A). 1H NMR and 13C NMR data of EKC were listed as below. 1H-NMR (600 MHz, DMSO-d6) δppm: 12.88 (1H, br s, 5-OH), 7.50 (1H, d, J = 1.2 Hz, H-2′), 7.39 (1H, d, J = 1.2 Hz, H-6′), 6.61 (1H, s, H-3), 6.26 (1H, s, H-6), 5.24 (1H, d, J = 4.2 Hz, H-16), 5.21 (1H, t, J = 6.6 Hz, H-12), 4.22 (1H, d, J = 4.2 Hz, H-17), 3.44 (2H, d, J = 6.6 Hz, H-11), 1.77 (3H, s, CH3-14), 1.63 (6H, s, CH3-15), 1.20 (3H, s, CH3-20), 1.13 (3H, s, CH3-19); 13C-NMR (150 MHz, DMSO-d6) δppm: 182.1 (C-4), 163.9 (C-2), 162.8 (C-7), 159.4 (C-5), 154.7 (C-9), 151.7 (C-4′), 142.2 (C-3′), 132.9 (C-5′), 131.3 (C-13), 123.9 (C-1′), 122.8 (C-12), 115.2 (C-6′), 114.6 (C-2′), 106.5 (C-8), 103.6 (C-10), 103.1 (C-3), 98.8 (C-6), 98.0 (C-17), 72.0 (C-16), 70.0 (C-18), 26.2 (CH3-14), 25.8 (CH3-19), 25.4 (CH3-20), 21.7 (C-11), 18.2 (CH3-15).
3.2. EKC induced cell death and cytoplasmic vacuolization in NCI-H292 cells.
During our screening of anticancer candidates from prenylated flavonoids, the effects on cell viability of these 27 compounds were determined by MTT assay first (Additional file 2). EKC inhibited cell growth in a dose- and time- dependent manner in non-small cell lung cancer NCI-H292 cells, the IC50 value for 48 h was 17.04 µM. Meanwhile, in an immortalized normal human bronchial epithelial cell line 16HBE, the cell viability after 48 h of EKC treatment was almost unaffected (Fig. 1B). Thus, EKC was more toxic to the tested lung cancer cells compared to normal bronchial epithelial cells. In terms of cytotoxity, EKC was not really a promising candidate, but it captured our attention because of its cytoplasmic vacuoles inducing effect. We found that EKC caused a quick and striking accumulation of numerous phase-lucent cytoplasmic vacuoles within 12 h in NCI-H292 cells, these morphological effects of EKC persisted for 48 h and beyond (Fig. 1C). In contrast, 16HBE cells and the other normal human bronchial epithelial cell line, Beas2B were relatively resistant to EKC-induced vacuolization (Fig. 1D). The vacuolization effects in NCI-H292 cells induced by EKC at a dose of 15 µM were demonstrated to be reversible, if cells were treated for 12 h, then placed in EKC-free medium for culturing additional 12 h, the vacuolization was almost completely resolved (Fig. 1E). The effects of EKC on other four cancer cell lines, including A549, Calu-1, HepG2 and PANC-1, were also evaluated. Similar to the results with NCI-H292 cells, EKC induced dramatic cytoplasmic vacuolization in all of the four cell lines (Additional file 3).
Four closely related compounds with > 75% structure similarity to EKC were isolated from E. koreanum together, they showed almost no vacuole-inducing activity or induced a small amount of vacuoles (Additional file 4). This suggested that the effect of EKC was probably due to its interaction with specific intracellular targets rather than non-specific effects.
3.3. EKC induced cell death was distinct from apoptosis
Apoptosis is the most common form of programmed cell death. To investigate if the cell death induced by EKC was related with apoptosis, annexin V-FITC/PI double staining and DAPI staining were performed. As shown in Fig. 2A and 2B, neither significant increase of the number of apoptotic cell nor obvious morphological changes in the nuclei were observed with EKC treatment, indicating that apoptosis may not be critically involved in EKC-induced cell death. For further determination, the activation of apoptosis markers including caspase-9, caspase-3 and PARP was also examined. As depicted in Fig. 2C, no obvious cleavage of caspase-9, caspase-3 and PARP was observed in 24 h after EKC exposure. Consistent with this assumption, pretreatment with pan-caspase inhibitor Z-VAD(OMe)-FMK (Z-VAD) for 1 h prior to EKC treatment did not block both the loss of cell viability and vacuolization induced by EKC (Fig. 2D and 2E). These findings suggested that EKC-induced cell death was non-apoptotic and caspase-independent.
3.4. EKC induced methuosis in NCI-H292 cells
Cytoplasmic vacuolation has been observed in several types of non-apoptotic cell deaths. To determine the origin of EKC-induced vacuoles, different subcellular compartments staining dyes including acridine orange (AO), MitoTracker red, LysoTracker red and ERTracker red were used. Live cell imaging showed that the vacuoles could sequester AO to change to orange fluorescence (Fig. 3A), which indicated the acidic nature of the vacuoles. None of MitoTracker red, LysoTracker red and ERTracker red was observed to be incorporated into the vacuoles (Fig. 3B-3D), indicating that the vacuoles were not derived from mitochondria, lysosome or endoplasmic reticulum.
The formation of large vacuoles is one of the most important characteristics of macropinocytosis. We next turned our attention to determine whether the vacuoles induced by EKC were correlated with macropinocytosis. One of the typical features of macropinocytosis is the incorporation of extracellular-phase fluid tracers. Lucifer Yellow (LY) is a well-established fluorescent tracer used to define micropinocytosis [18]. As shown in Fig. 3E, LY incorporated into most of the phase-lucent vacuoles induced by EKC, suggested that EKC-induced vacuoles might be derived from macropinosomes.
Normally, once internalized into cells, macropinosomes pesist for only about 5 to 20 minutes, during which they either recycle back to the cell surface or merge with lysosomes [18, 19]. While in methuosis, macropinosomes bypass the normal endosomal trafficking pathway, immediately recruit Rab7 and form progressively larger LAMP1-positive structures [20]. For further investigation of whether EKC induced methuosis, immunofluorescence assay against two late endosome markers, Rab7 and LAMP1, was performed. As seen in Fig. 3F, Rab7 and LAMP1 were detected on the membrane of EKC-induced vacuoles. Rab5, a marker of early endosome, was also detected, but it was not labeled on the membrane of the vacuoles. The protein levels of Rab7 and LAMP1 were also detected, as shown in Fig. 3G, EKC treatment increased the levels of Rab7 and LAMP1 in dose- and time- dependent manners. Combined with aforementioned LysoTracker staining results, we could conclude that the vacuoles induced by EKC possessed the characteristics of late endosomes, but did not merge with lysosomal compartments. These characteristics were consistant with methuosis, suggesting that EKC was a methuosis inducer.
3.5. EKC induced PIKfyve phosphoinositide kinase inhibition and Akt suppression
PIKfyve, a class Ш phosphoinositide (PI) kinase, plays crucial role in the regulation of trafficking events associated with the endocytic pathway [21, 22]. It has been reported that PIKfyve was one candidate protein target of a methuosis inducer, MOMIPP [23]. To explore whether EKC inhibited PIKfyve expression, the protein level of PIKfyve was detected by immunoblot assay. As depicted in Fig. 4A, EKC treatment decreased the protein levels of PIKfyve dose- and time-dependently. Akt suppression was required for the cytotoxicity of PIKfyve inhibitors, the phosphorylation of Akt was shown to be down-regulated concurrently (Fig. 4A).
PIKfyve also plays a crucial role in ensuring phagosomal maturation [24]. Under the same conditions in which EKC induced the accumulation of cytoplasmic vacuoles, immunoblotting analysis revealed a dose- as well as time-dependent increase of two biomarkers of autophagosome, LC3-П and SQSTM1, indicating the accumulation of autophagosomes (Fig. 4B). To determine whether the accumulation of autophagosomes resulted from induction of autophagy or from disruption of autophagic flux, cells were transfected with mRFP-GFP-LC3 adenovirus, as shown in Fig. 4C, distinct yellow puncta were observed in EKC treated cells. Therefore, EKC disrupted autophagic flux by preventing autolysosome formation.
Bafilomycin A1 (Baf A1), an H+-vacuolar ATPase inhibitor, has been shown to prevent formation of cytoplasmic vacuoles that are induced by inhibition of PIKfyve PI kinase [25]. To determine whether or not Baf A1 also prevented induction of the cytoplasmic vacuoles that were induced by EKC, NCI-H292 cells were treated by Baf A1, EKC, or both Baf A1 and EKC, the results showed that Baf A1 prevented EKC-induced cytoplasmic vacuolation almost completely (Fig. 4D), suggesting that EKC inhibited PIKfyve activity.
3.6. EKC affected the activities of Rac1 and Arf6 GTPases
It has been reported that methuosis triggered by over-expression of activated H-Ras required activation of the Rac1 GTPase with a concomitant deactivation of another GTPase, Arf6 [26]. In EKC treated cells, we first detected the total protein level of Rac1 and Arf6, the results revealed that EKC treatment had no significant effect on the expression of Rac1, while decreased the total level Arf6 (Fig. 5A). The mRNA levels were examined by qPCR, and the mRNA level of Rac1 was not changed, while the mRNA level of Arf6 decreased (Fig. 5B). Futhermore, the active GTP-bound forms of Rac1 and Arf6 were measured in pull-down assays using fusion proteins that bind specifically to the activated forms of Rac1 and Arf6. The results showed that the amounts of active Rac1 (Rac1-GTP) increased although the total level of Rac1 did not change in EKC treated cells. In contrast, EKC had the opposite effect on Arf6, both total level and active Arf6 decreased, however Arf6-GTP decreased more significant than total Arf6 after EKC treatment, that is the ratio of active/total Arf6 decreased (Fig. 5C). These results indicated that EKC treatment induced activation of the Rac1 GTPase, with a concomitant reduction in the activation state of Arf6 GTPase. For further exploration, NCI-H292 cells were pretreated with a highly specific Rac inhibitor EHT 1864 for 24 h, then incubated with EKC in the presence of EHT 1864 for additional 24 h, as illustrated in Fig. 5D, EKC induced cell viability decrease was antagonized by EHT 1864; in Fig. 5E, most of the vacuoles were precluded by EHT 1864. These results suggested that the mechanisms of vacuolization induced by EKC is probably similar with that induced by Ras, that is, might be depended on reverse regulation of Rac1 and Arf6.
3.7. EKC inhibited NCI-H292 cell migration
It has been demonstrated that Arf6 regulates cancer cell invasion through the activation of MEK/ERK signaling pathway, and downregulation of Arf6 correlate with impaired cell migration [27, 28]. We evaluated the effect of EKC on the migration of NCI-H292 cells by performing a scratch wound-healing assay. As shown in Fig. 6A, EKC treatment at the dose of 6.5, 12.5 µM could significantly reduce cell migration in NCI-H292 cells. We also examined the phosphorylation as well as total ERK levels by immunoblot analysis. EKC treatment decreased the levels of phosphorylation ERK in dose- and time- dependent manners, while no obvious change of total ERK was observed (Fig. 6B).
3.8. EKC sensitized NCI-H292 cells to doxorubicin and etoposide
To determine the effects of EKC on the sensitivity to conventional chemotherapy agents, NCI-H292 cells were exposed to doxorubicin/etoposide in the presence of 6.5 µΜ of EKC, the viability of cells was monitored using a real time cell analyzer (RTCA). As shown in Fig. 7, both doxorubicin and etoposide displayed a reduction in cell growth, whereas treatment with 6.5 µΜ of EKC alone resulted in slight growth inhibition. However, co-treatment of EKC with doxorubicin/etoposide significantly reduced cell growth as judged by decreased cell index (CI) compared with doxorubicin and etoposide treatment alone.