CUR5g selectively induces autophagosome accumulation in cancer cells
This work used U87 cells with stable expression of GFP-LC3B fusion protein to screen various curcumin analogs to find new small-molecule inhibitors of autophagy. Among these analogs, CUR5g (Fig. 1A) induced extensive cytoplasmic vacuolization, and GFP-LC3B signal shifted from diffuse cytosolic staining to a punctate pattern outlining autophagosomes (Fig. 1B and S1), suggesting that CUR5g might regulate autophagy. As revealed by WB assay, CUR5g up-regulated LC3B-II and sequestosome 1 (SQSTM1) levels time- and dose-dependently (Fig. 1C). This increase was not the result of enhanced transcription, as mRNA expression of SQSTM1 and LC3B were not increased within CUR5g-exposed cells (Fig. 1D), suggesting that CUR5g might block autophagic flux rather than increase autophagosome formation.
Excitingly, CUR5g also induced SQSTM1 and LC3B-II accumulation in other human cancer cells (Fig. S2), but not in normal cells (Fig. S3). The punctuate distribution of GFP-LC3B was significantly increased in CUR5g-treated A549 cells (Fig. 1E). Following 24-h exposure to 10 µM CUR5g, numerous autophagosomes were observed in A549 cells, but much fewer in untreated controls (Fig. 1F). Based on these observations, CUR5g was selected for subsequent experiments.
Cur5g Causes A Late-stage Block Of Autophagy By Suppressing The Fusion Of Autophagosome With Lysosome
Exposure to 3-MA, the autophagosome formation inhibitor, did not eliminate CUR5g-induced SQSTM1 and LC3B-II accumulation (Fig. 2A). Treatment of A549 cells with CUR5g and the late-stage autophagy inhibitor CQ showed no effect on further increasing LC3B-II relative to simply CQ (Fig. 2B). CUR5g further induced LC3B-II accumulation within A549 cells under the nutrient deficient state compared with those cells maintained in nutrient adequate medium (Fig. 2C), indicating that CUR5g causes a late-stage block of autophagy.
Congruently, CUR5g increased the GFP+RFP+ punctum number (autophagosomes, yellow), but not GFP−RFP+ punctum number (autolysosomes, red), in HEK293T cells stably expressing RFP-GFP-LC3B reporter (Fig. 2D).
Additionally, CUR5g’s role in major proteins levels involved in autophagosome formation was analyzed, namely, mechanistic target of rapamycin kinase (mTOR) and its substrate RPSK6B1, ATG14-like protein (Atg14L), BECN1, phosphatidylinositol 3-kinase catalytic subunit type 3 (PIK3C3), UV radiation resistance-associated gene (UVRAG), phosphoinositide 3-kinase regulatory subunit 4 (PIK3R4), as well as ATG5. Except for the significant reduction in UVRAG levels, the levels of other proteins in CUR5g-treated cells were not statistically different from control cells (Fig. 2E), further ruling out the possibility that CUR5g increases autophagy initiation.
The inhibition of autophagy can be induced by impaired autophagosome degradation or/and blocking of autophagosome-lysosome fusion. To clarify how CUR5g works, we detected the LC3B- LysoTracker Red colocalization within U87 cells. The overlap of LC3B-GFP signals with LysoTracker Red signals was observed in EBSS-incubated cells, rather than CUR5g-treated counterparts (Fig. 2F), indicating that CUR5g inhibits the fusion of autophagosomes and lysosomes. Notably, CUR5g did not reduce acid-dependent LysoTracker Red signals, unlike CQ, which accumulates within and alkalinizing the lysosome, suggesting that CUR5g does not block lysosome acidification. In parallel, LC3B puncta and lysosomal-associated membrane protein 1 (LAMP1, the lysosomal marker) were dramatically separated within CUR5G-treated cells, a phenomenon close to CQ-treated cells in a less pronounced manner (Fig. 2G), confirming that CUR5g blocks autophagosome-lysosome fusion.
CUR5g does not affect the lysosomal proteolytic function and cytoskeleton, but blocks incorporation of syntaxin 17 (STX17) on autophagosomes
We examined the effect of CUR5g on lysosomal pH with the use of Lysotracker red and acridine orange (AO) to further clarify whether CUR5g affects lysosomal function. We found that CUR5g did not attenuate red fluorescence in the above two dyes (Fig. 3A), indicating that CUR5g did not induce lysosomal alkalization. CUR5g did not suppress the activities of lysosomal proteases, including acid phosphatase (ACP), CTSB, and CTSD (Fig. 3B-D). In parallel, according to WB assay, CUR5g had no effect on lysosomal marker LAMP1 expression, nor the conversions of proCTSB and proCTSD to mature cathepsins (Fig. 3E), supporting the notion that CUR5g does not disrupt lysosomal proteolytic function. Moreover, CUR5g treatment did not change the distribution of F-actin, vinculin and β-tubulin (Fig. 3F). Vinculin and β-tubulin expression in CUR5g-treated cells were not significantly different from untreated cells (Fig. 3G), suggesting that CUR5g does not affect the cytoskeleton.
Autophagosomal membrane-localized STX17 interacts with endolysosome-localized vesicle-associated membrane protein 8 (VAMP8) and synaptosomal-associated protein of 29 kDa (SNAP29) to form a soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) complex, and they play a key role in driving autophagosome-lysosome fusion [21]. As revealed by WB assay, CUR5g made no difference to reduce STX17, SNAP29 or VAMP8 expression (Fig. 4A). Immunostaining of cells treated with CUR5g showed that STX17 and SNAP29 were not recruited to LC3B-positive autophagosomes, whereas a significant colocalization of STX17 and SNAP29 with LC3B was observed in cells maintained in EBSS medium (Fig. 4B and C), suggesting that CUR5g prevented the incorporation of STX17 onto autophagosomes.
In addition to the SNARE complex, the role of Rab GTPase Rab7 in the fusion of autophagosomes with lysosomes has been extensively studied [22, 23]. We found that CUR5g did not influence Rab7 level (Fig. S4A). Transfection of A549 cells with wild-type pEGFP-Rab7 or the mutants Q67L (a constitutively active form) and T22N (a dominant negative form) did not reverse CUR5g-induced accumulation of LC3B-II and SQSTM1 (Fig. S4B), ruling out the possibility that CUR5g suppressed autophagosome-lysosome fusion in a Rab7-dependent manner.
Overexpression Of Uvrag Reverses Cur5g-induced Autophagosome Accumulation
In addition to being a key component in autophagosome formation, UVRAG is also thought to be necessary for autophagosome-lysosome fusion [30]. The fact that UVRAG levels remarkably decreased within CUR5g-exposed cells (Fig. 2E and S5), making us to investigate whether the failure of autophagosome-lysosome fusion is due to UVRAG downregulation. We found that in UVRAG overexpressing cells, CUR5g failed to induce the accumulation LC3B-II and SQSTM1, increase the number of GFP+RFP+ puncta, or prevent the recruitment of STX17 to autophagosomes (Fig. 5A-C). Overexpression of UVRAG did not prevent CUR5g-induced SQSTM1 and LC3B-II accumulation within STX17 knockdown cells (Fig. 5D). These results collectively demonstrated that CUR5g inhibits STX17 targeting to autophagosomes by down-regulating UVRAG.
The molecular modeling and docking studies predicted that CUR5g appears to bind preferably to the region 270–442 of UVRAG (Fig. S6) which was sufficient for interaction with VPS16 [24]. Efficient binding to VPS16 is necessary for UVRAG to fully facilitate autophagosome-endosome/lysosome fusion [24], raising the possibility that CUR5g might weaken the interaction between UVRAG and VPS16 and therefore impairs the autophagosomes maturation. Supporting this notion, we found that CUR5g weakened UVRAG-VPS16 interaction by immunofluorescence observation and coimmunoprecipitation (co-IP) analysis (Fig. S7).
Cur5g Exerts Anti-cancer Effects On A549 Cells
A549 cell number slightly decreased after treatment with 10 µM CUR5g, while CUR5g at 20 µM decreased the number of A549 cells significantly (Fig. 6A), suggesting that CUR5g exhibited great toxicity to A549 cells at 20 µM. Similar results were obtained by MTT assays (Fig. 6B). However, CUR5g at 40 µM showed no discernable activity in healthy human umbilical vein endothelial cell (HUVEC) viability (Fig. S8). We found that UVRAG was significantly higher in both mRNA and protein levels in HUVECs than in A549 cells (Fig. S9A and B). More excitingly, at the concentration below or equal to 50 µM, CUR5g did not induce a statistically significant reduction in UVRAG levels (Fig. S9C), and the interaction between UVRAG and VPS16 is not significantly affected by CUR5g (Fig. S9D), suggesting that A549 cells might be more sensitive to CUR5g because they exhibit lower basal levels of UVRAG. These results at least partially explained why CUR5g showed more selective autophagy inhibitory effects and cytotoxic activity against NSCLC cells than normal HUVECs. In addition, 10 µM CUR5g almost completely suppressed the colony formation in A549 cells (Fig. 6C). Upon CUR5g treatment, G0/G1-phase cell proportion declined, whereas S-phase proportion elevated (Fig. 6D), suggesting that CUR5g promoted cell cycle arrest of A549 cells at S phase.
CUR5g at 10 µM almost completely inhibited colony formation, promoting us to examine whether CUR5g affects the mobility of A549 cells. As revealed by a scratch assay, CUR5g significantly inhibited A549 cell migration (Fig. 6E). After treatment with CUR5g, Annexin V-positive cell proportion (Fig. 6F) and levels of cleaved PARP1 or cleaved Caspase-3 did not increase (Fig. 6G), nor did the activity of lactic acid dehydrogenase (LDH) (Fig. 6H), suggesting that CUR5g did not induce apoptosis and necrosis of A549 cells. Similarly, CUR5g made no difference to intracellular and mitochondrial ROS, nor the MMP levels (Fig. S10 and S11).
CUR5g exhibits potent synergistic anti-cancer effects with cisplatin on A549 cells and inhibits autophagic flux in vivo
We further tested whether CUR5g could synergize with cisplatin to exhibit stronger anti-cancer activity against NSCLC. Different from the cells exposed to CUR5g or cisplatin alone, the number of A549 cells co-treated with CUR5g and cisplatin displayed a rapid decrease, and this trend maintained until the end of the experiment (Fig. 7A). Likewise, data from colony formation assay and wound-healing assay showed that CUR5g enhanced cisplatin’s anticancer activity in A549 cells (Fig. 7B and C). Moreover, we compared the effectiveness of CUR5g with other three autophagy regulators including curcumin, CQ and CA-5f in combination with cisplatin. At the same concentration of 10 µM, CUR5g/cisplatin combination induced more effective synergistic cytotoxicity to A549 cells than the combination of cisplatin with curcumin or CA-5f. The effectiveness of 10 µM CUR5g combined with cisplatin is similar to 30 µM CQ combined with cisplatin (Fig. S12).
To explore whether the combination of CUR5g and cisplatin serves a similar function in vivo, a xenograft nude mouse model of NSCLC was established through subcutaneously injecting A549 cells. CUR5g alone or combined with cisplatin showed well tolerance in nude mice, and abnormal physical signs were not reported in the whole procedure. Differences in average body weight (BW) among all treatments were not significant (Fig. S13). No signs of mortality or macroscopic abnormalities in the organs were discovered, and no obvious pathological alterations were observed in the major organs (Fig. S14). The size and weight of the dissected tumors in the CUR5g and cisplatin combination group remarkably decreased compared with the remaining groups (Fig. 7D and E). As expected, CUR5g or cisplatin alone retarded the growth of xenografted tumors, whereas the combination treatment almost completely inhibited tumor growth (Fig. 7F), demonstrating that CUR5g potently enhanced cisplatin's anti-cancer activity in A549 cells in vitro as well as in vivo. Consistent with in vitro data, as revealed by WB assay, CUR5g increased LC3B-II and SQSTM1 levels within tumor tissues (Fig. 7G). The level of LC3B-II in CUR5g plus cisplatin group significantly increased relative to cisplatin group (Fig. 7G). Immunofluorescence staining showed a significant increase of punctate LC3B and SQSTM1 signals in tumor tissues obtained from CUR5g-treated group as relative to control (Fig. 7H and I). CUR5g synergistically promoted the effect of cisplatin on LC3B accumulation in tumors (Fig. 7H and I). In conclusion, the above findings suggested that CUR5g promoted the cisplatin sensitivity of A549 cells by inhibiting autophagic flux.