The natural anticancer agent cantharidin alters GPI-anchored protein sorting by targeting Cdc1-mediated remodeling in endoplasmic reticulum

Cantharidin (CTD) is a potent anticancer small molecule produced by several species of blister beetle. It has been a traditional medicine for the management of warts and tumors for many decades. CTD suppresses tumor growth by inducing apoptosis, cell cycle arrest, and DNA damage and inhibits protein phosphatase 2 phosphatase activator (PP2A) and protein phosphatase 1 (PP1). CTD also alters lipid homeostasis, cell wall integrity, endocytosis, adhesion, and invasion in yeast cells. In this study, we identified additional molecular targets of CTD using a Saccharomyces cerevisiae strain that expresses a cantharidin resistance gene (CRG1), encoding a SAM-dependent methyltransferase that methylates and inactivates CTD. We found that CTD specifically affects phosphatidylethanolamine (PE)-associated functions that can be rescued by supplementing the growth media with ethanolamine (ETA). CTD also perturbed endoplasmic reticulum (ER) homeostasis and cell wall integrity by altering the sorting of glycosylphosphatidylinositol (GPI)-anchored proteins. A CTD-dependent genetic interaction profile of CRG1 revealed that the activity of the lipid phosphatase cell division control protein 1 (Cdc1) in GPI-anchor remodeling is the key target of CTD, independently of PP2A and PP1 activities. Moreover, experiments with human cells further suggested that CTD functions through a conserved mechanism in higher eukaryotes. Altogether, we conclude that CTD induces cytotoxicity by targeting Cdc1 activity in GPI-anchor remodeling in the ER.

growth of hepatocellular carcinoma (16), leukemia (17), pancreatic (18), colorectal (19), gallbladder (20), oral (21), and breast cancer (22). The serine-threonine protein phosphatases, PP1 and PP2A, are the only reported molecular targets of CTD (23,24). The inhibition of PP2A causes cell cycle arrest (25,26) and apoptosis (27,28). CTD also impairs different cellular processes such as heat shock response (29), autophagy (22), DNA damage response, and mitogen-activated protein kinase signaling (18,21). One of these studies also demonstrated a PP2A-or PP1-independent alteration in heat shock response (29), suggesting the existence of additional molecular targets of CTD (29,30). Most of the studies performed with CTD were based on mammalian cell lines, making it difficult to decipher a conserved mechanism of action of the drug due to their tissuespecific origin and differential gene regulation. Hence, yeast (Saccharomyces cerevisiae) serves as an appropriate model system to identify the conserved molecular targets of the drug (31)(32)(33). Previous studies showed that yeast YHR209W gene was essentially required for CTD resistance (34), which was subsequently named as cantharidin-resistant gene (CRG1) (35). Later, Crg1 was characterized as a SAM-dependent methyltransferase that detoxifies CTD by methylation (30). Deletion of CRG1 enables the identification of the molecular targets of CTD more easily, so we utilized budding yeast as a model organism to dissect the molecular mechanism of CTD toxicity.
Our study was focused on the identification of the conserved cellular pathways targeted by CTD. Interestingly, we found that CTD impaired the GPI-anchored protein sorting by targeting the remodeling process in ER. More specifically, it affected the Cdc1 activity, leading to multiple cellular changes, such as missorting and aggregation of GPI-anchored proteins, temperature sensitivity, cell wall damage, and decreased UPR. Most of the CTD-induced phenotypes observed in yeast cells were also reproducible in human cells. Our comprehensive genetic and cell biology-based experiments revealed that the Cdc1 activity is a molecular target of CTD in eukaryotic cells. Overall, we identified the GPI-anchor remodeling as a direct target of CTD.

Supplementation of ethanolamine (ETA) suppresses the cytotoxic effect of CTD
Previous studies have shown that CTD treatment affects the lipid homeostasis in budding yeast by inhibition of the elongation of short-chain phospholipids to long-chain phospholipids (30). The phospholipid imbalance can be restored with exogenous supplementation of the precursor molecules. For example, supplementation of ETA and choline (CHO) activates the synthesis of phosphatidylethanolamine (PE) and phosphatidylcholine (PC), respectively, via an alternative pathway, i.e. the Kennedy Pathway (Fig. 1F) (36). Inositol (INO) and Ser enter into the canonical pathways of phosphatidylinositol (PI) and phosphatidylserine (PS) biosynthesis, respectively (Fig. 1F) (37,38). Based on these phenomena, we sought to identify the specific phospholipid affected by CTD. We supplemented the medium with specific precursor molecules, ETA, CHO, and INO, with or without CTD and measured the growth of WT and crg1⌬ strains ( Fig. 1A and Fig. S10). CTD exposure pro-duced a lethal effect on crg1⌬ mutant compared with WT (30). However, ETA supplementation completely rescued the growth of the crg1⌬ strain from CTD cytotoxicity ( Fig. 1A and Fig. S1, A-D). On the other hand, CHO and INO supplementation failed to rescue the growth of crg1⌬ strain in CTD-containing medium (Fig. 1A). This observation suggests that CTD specifically targets PE. The exclusive rescue in the growth of the CTD-treated cells by ETA supplementation was a surprising phenomenon, because PE and PC are both synthesized in the same pathway (37). Thus, we believe that CTD may not affect the PE biosynthesis pathway, but it might be altering the PEassociated structures or functions. PE plays an essential role in maintaining membrane and cell wall integrity under heat stress (38,39), so we examined the fitness profile of WT and crg1⌬ strains in heat stress (37°C) with a permissible dose of CTD (2 M). Interestingly, we found complete inhibition of growth of crg1⌬ mutant at 37°C in the presence of CTD, whereas the growth was unaffected at optimum (30°C) or below the optimum (25°C) temperature (Fig. 1B). CTD cytotoxicity was suppressed again at 37°C by supplementation of ETA (Fig. 1C). PE biosynthesis takes place in mitochondria and Golgi/vacuole with the help of Psd1 and Psd2, respectively (40). A major fraction (Ͼ90%) of the net PE in a cell is synthesized by Psd1 in mitochondria (40), so we created a double-deletion mutant, crg1⌬psd1⌬, to check synthetic lethality between PSD1 and CRG1 in the presence of CTD. For this purpose, WT, crg1⌬, psd1⌬, and crg1⌬psd1⌬ strains were grown in CTD-containing medium. We found that the crg1⌬psd1⌬ mutant was hypersensitive to CTD than crg1⌬, suggesting that PE is essentially required to tolerate CTD toxicity ( Fig. 1D and Fig. S1, E-H). crg1⌬psd1⌬ mutant followed the same trend at higher temperature as well (37°C) (Fig. 1E). The synthetic lethality between CRG1 and PSD1 in the presence of CTD suggests an essential role of PE to tolerate CTD toxicity. These observations suggest that CTD affects the PE-associated functions (Fig. 1F); therefore, enhanced synthesis of PE helps to overcome the CTD toxicity.

CTD alters ER homeostasis by inhibition of UPR
ER is the organelle for the synthesis of the major phospholipids. Imbalance in the phospholipid composition of lipid bilayer membrane is reported to induce ER stress (41)(42)(43)(44). Existing evidence suggests that CTD also perturbs ER-synthesized phospholipids (30); thus, we proposed that CTD might be altering the ER homeostasis. We examined ER stress in crg1⌬ cells in the presence of CTD. First, WT and crg1⌬ cells were co-treated with CTD and ER stress (or UPR) inducers, dithiothreitol (DTT) or tunicamycin (TM), to check whether there was any synergistic effect between the two molecules. For this, we chose a permissible dose of CTD (4 M) for the crg1⌬ mutant, at which it survived, but survival was lower than that of the WT. Both the strains were spotted on CTD-containing medium, with and without TM or DTT. Interestingly, the co-treatments (CTD ϩ TM and CTD ϩ DTT) inhibited the growth of crg1⌬ cells more severely compared with only CTD treatment ( Fig. 2A and Fig. S1, I-L and Fig. S11, A). The synergistic lethal effect on the growth of crg1⌬ cells upon co-treatments suggests that CTD perturbs ER homeostasis. Next, we measured UPR by Cantharidin targets GPI-anchor remodeling ␤-gal assay with the help of UPRE-LacZ reporter plasmid (45,46). We found that the basal level of UPR was lower in crg1⌬ cells compared with WT, and CTD treatment further inhibited UPR in both the strains (WT and crg1⌬) (Fig. 2B). Because ETA supplementation rescues the yeast cells from CTD toxicity, we decided to measure UPR upon ETA supplementation. Surprisingly, ETA supplementation could not rescue UPR inhibited by CTD (Fig. 2B), suggesting that CTD inhibits UPR via a distinct mechanism independent of PE in ER. To gain more insight into this mechanism, UPR was measured upon co-treatments of cells with CTD ϩ DTT and CTD ϩ TM. We found decreased UPR levels in crg1⌬ cells upon DTT and TM treatment (Fig.  2C). Moreover, CTD treatment strongly inhibited UPR induced by DTT or TM in WT as well as crg1⌬ mutant (Fig. 2C). Consistent with these observations, we also found inhibition of HAC1 mRNA splicing in crg1⌬ cells compared with WT. The splicing of HAC1 mRNA was further inhibited in both of the strains, WT and crg1⌬, upon CTD treatment (Fig. 2D). DTT and TM treatment strongly induced HAC1 mRNA splicing; however, the presence of CTD with DTT or TM suppressed HAC1 mRNA splicing (Fig. 2D). These results suggest that CTD inhibits UPR by making an obstruction in HAC1 mRNA splicing, although the mechanism remains unclear.
Our data suggest that CTD exposure leads to ER stress that cannot be rescued by ETA supplementation. The ER-lumen maintains higher oxidation potential with the help of a low Figure 1. CTD specifically targets PE in crg1⌬ cells. A, B, C, and E, growth sensitivity assays. Equal numbers of cells were serially diluted and spotted on SC agar medium. Images were captured after 72 h of incubation. A, supplementation of ETA rescues crg1⌬ mutant from CTD toxicity. The phospholipid precursors ETA, INO, and CHO were added into SC agar medium with or without CTD. WT and crg1⌬ cells were spotted and incubated at 30°C. B, CTD toxicity increases with rising temperature. WT and crg1⌬ cells were spotted on SC agar medium containing CTD and incubated at different temperatures (25, 30, and 37°C). C, ETA supplementation rescues the crg1⌬ mutant from CTD toxicity at higher temperature. WT and crg1⌬ cells were spotted on SC agar medium containing CTD with and without ETA supplementation and incubated at different temperatures (25, 30, and 37°C). D and E, CRG1 shows synthetic lethality with PSD1 under CTD stress. D, growth curve assay. Equal numbers of cells of WT, crg1⌬, psd1⌬, and crg1⌬psd1⌬ were grown at 30°C with or without CTD in liquid medium. A 600 was measured at the time interval of 30 min using an automated plate reader for 23 h. E, WT, crg1⌬, psd1⌬, and crg1⌬psd1⌬ cells were spotted on SC agar medium containing CTD with or without ETA and incubated at two different temperatures (30 and 37°C). F, phospholipid biosynthesis pathways in yeast (37,66,77,78). INO and Ser in medium are directly utilized to synthesize PI and PS with the help of Pis1 and Cho1, respectively. PE and PC biosynthesis has two pathways. The first pathway is canonical biosynthesis of PE/PC, which takes place in mitochondria and the ER. The first reaction starts in the ER, where Cho2 synthesizes PS from Ser. PS is transported to mitochondria, where Psd1 catalyzes its decarboxylation to synthesize PE. (A similar mechanism also takes place in Golgi and vacuole by Psd2, which contributes a very minor fraction of the net PE content). Next, PE is transported again to the ER, where Cho2 and Opi3 convert it into PC via a sequence of methylation reactions. The second pathway is noncanonical PE or PC synthesis, also known as the Kennedy pathway. In this pathway, externally supplemented precursors (ETA/CHO) are utilized and converted into PE or PC, respectively, via a series of enzymatic reactions.

Cantharidin targets GPI-anchor remodeling
GSH/GSSG ratio (1:1 to 3:1) compared with the high GSH/ GSSG ratio (30:1 to 100:1) of the cytosol (47). GSH provides a redox buffer for the catalytic activity of the protein-folding enzymes in the ER (48,49). The imbalance in GSH/GSSG ratio in ER impairs oxidative protein folding that causes ER stress (50,51). Based on these previous findings, we predicted that CTD-induced ER stress might be due to imbalance in the GSH/ GSSG ratio in ER. To test this hypothesis, we checked the effect of GSH on CTD toxicity. We used the permissible dose of CTD (4 M) for crg1⌬ mutant and supplemented the medium with a high dose of GSH and NAC. We found that the growth of crg1⌬ mutant was suppressed in the presence of either of the two reducing molecules, GSH or NAC, along with CTD. However, GSH or NAC alone did not show any effect on the growth of Figure 2. CTD treatment inhibits UPR by alteration of the ER-redox homeostasis. A, UPR inducers (DTT/TM) synergistically enhance CTD toxicity. Equal numbers of serially diluted WT and crg1⌬ cells were spotted on CTD-containing SC agar medium with or without DTT/TM in the presence or absence of ETA and incubated at 30°C for 72 h. B, CTD inhibits UPR. WT and crg1⌬ strains transformed with pPW344 (UPRE-LacZ) plasmid were grown in SC-URA medium at 30°C. Cells were treated with CTD (6 M) with or without ETA (2.5 mM) at the mid-exponential phase (A 600 ϭ 0.8) and incubated for 2 h. A ␤-gal assay was performed to measure the UPR. The graph shows a scatter plot of each data point of three independent experiments with mean (horizontal green line) Ϯ S.D. (error bars). Statistical analysis was done with GraphPad Prism version 5, applying two-way ANOVA and Bonferroni post hoc test, where p Յ 0.05 (*), p Յ 0.01 (**), and p Յ 0.001 (***). C, CTD inhibits UPR in presence of DTT and TM. WT and crg1⌬ strains carrying pPW344 vector were grown until the mid-exponential phase and treated with CTD (3 M) in combination with DTT (0.5 mM) or TM (0.25 g/ml) for 2 h. A ␤-gal assay was performed to measure the UPR. The graph shows a scatter plot of each data point of three independent experiments with mean Ϯ S.D. (error bars). Statistical analysis was done using GraphPad Prism version 5, applying two-way ANOVA and Bonferroni post hoc test, where p Յ 0.05 (*), p Յ 0.01 (**), and p Յ 0.001 (***). D, CTD inhibits HAC1 mRNA splicing. WT and crg1⌬ strains were grown in the conditions mentioned above (C), and HAC1 mRNA splicing was measured by RT-PCR. HAC1(u), unspliced HAC1; HAC1(i), spliced HAC1. The figure represents one of the three independently performed experiments. E, GSH or NAC supplementation enhances the CTD cytotoxicity. Equal numbers of WT and crg1⌬ cells were serially diluted and spotted on CTD-containing SC agar medium with or without reducing agents (GSH and NAC) in the presence or absence of ETA, incubated at 30°C for 72 h. F, GSH and NAC supplementation reduces UPR. WT and crg1⌬ strains transformed with pPW344 (UPRE-LacZ) were grown in SC-URA medium at 30°C until mid-exponential phase. The cells were treated with CTD (3 M) in the presence or absence of GSH (20 mM) or NAC (20 mM) for 2 h and processed for the ␤-gal assay. The graph shows a scatter plot of each data point of three independent experiments with mean Ϯ S.D. (error bars). Statistical analysis was done using GraphPad Prism version 5, applying two-way ANOVA and Bonferroni post hoc test, where p Յ 0.05 (*), p Յ 0.01 (**), and p Յ 0.001 (***). G, GSH and NAC supplementation enhances the CTD-mediated inhibition of HAC1 splicing. WT and crg1⌬ cells were grown in SC medium at 30°C until mid-exponential phase under the same conditions mentioned above (F), and the HAC1 mRNA splicing was measured by RT-PCR. The figure represents one of the three independently performed experiments.

Cantharidin targets GPI-anchor remodeling
crg1⌬ mutant ( Fig. 2E and Fig. S11, B). Furthermore, the supplementation of ETA did not rescue the growth of crg1⌬ mutant upon CTD ϩ GSH or CTD ϩ NAC treatments. This result supports the previous observation where, upon ETA supplementation, UPR suppressed by CTD treatment could not be rescued (Fig. 2B). Similar observations were also made in liquid growth culture (Fig. S2, A-F). Next, we measured UPR using a ␤-gal assay. Interestingly, we observed that GSH or NAC supplementation results in the reduction in UPR in WT and crg1⌬ cells (Fig. 2F). As CTD treatment also inhibits UPR, we observed severe reduction in UPR upon co-treatments with CTD ϩ GSH or CTD ϩ NAC (Fig. 2F). We also found an inhibition in HAC1 mRNA splicing upon the addition of GSH and NAC (Fig. 2G), which was more when the cells were co-treated with CTD ϩ GSH or CTD ϩ NAC (Fig.  2G). These observations suggest that CTD-mediated inhibition of UPR is probably due to imbalance in ER-redox homeostasis, which is enhanced with the addition of GSH. It also explains the reason why ETA supplementation failed to rescue the UPR.

CTD exposure perturbs the cell wall integrity via ER stress
Yeast cell wall biosynthesis and maintenance largely depend on functional ER (7,52,53). Dysfunctional ER affects the synthesis, modifications, folding, and transport of the proteins destined to the plasma membrane or cell wall. Based on our results, we proposed that CTD-induced ER stress could also perturb cell wall integrity. To examine the effects of CTD on cell wall integrity, we measured chitin content in the cell wall of WT and crg1⌬ cells by calcofluor white (CFW) staining (54). We found substantial increase in chitin content in crg1⌬ cells upon CTD treatment, suggesting that CTD treatment induced cell wall damage (Fig. S3A). To gain more insight on the effect of CTD on cell wall integrity, we co-treated the cells with CTD and cell wall-perturbing agents, Congo red (CR) and CFW. We used a permissible dose of CTD (4 M) in combination with cell wallperturbing agents to measure the growth of WT and crg1⌬ cells. We found that crg1⌬ mutant did not grow in either of the co-treatments (CTD ϩ CR or CTD ϩ CFW), whereas the growth was unaffected in individual treatments ( Fig. 3A and Fig. S11, C and D). We also supplemented sorbitol (SRB) into the medium to maintain the osmotic balance across the cell membrane. SRB rescued the growth of crg1⌬ mutant upon CTD ϩ CFW treatment, but not upon CTD ϩ CR treatment. That suggests the CTD ϩ CR-induced cell wall damage is irreversible, although the mechanism remains to be identified (Fig.  3A). We obtained similar results in liquid growth culture under similar conditions (Fig. S3, B-E). Yeast cell wall damage is sensed by many sensor proteins residing in the cell wall, which in turn activate downstream signaling via Slt2 (55). Activation of Slt2 triggers the transcription of cell wall maintenance genes via Rlm1 and Swi4 -Swi6 transcription factors (52,56). Hence, we did Western blot analysis of Slt2 phosphorylation in WT and crg1⌬ cells upon CTD treatment. We observed increased phosphorylation of Slt2 in crg1⌬ cells upon CTD treatment at 25°C. Slt2 phosphorylation increased further when the cells were grown at 37°C, and CTD treatment induced Slt2 phosphorylation strongly in crg1⌬ cells at this temperature (Fig. 3, B and C). As we knew that CTD cytotoxicity could be neutralized by ETA supplementation, we decided to measure Slt2 phosphorylation in CTD-treated cells supplemented with ETA. We did not find significant decrease in Slt2 phosphorylation upon ETA supplementation in CTD-treated cells (Fig. 3, B and C). Next, we challenged the WT and crg1⌬ cells with the combined treatment of CTD and UPR inducers (DTT and TM) to measure the synergistic effect on Slt2 phosphorylation. We found strong induction in Slt2 phosphorylation upon co-treatments with CTD ϩ DTT or CTD ϩ TM compared with individual treatments (CTD/DTT/TM) (Fig. 3, D and E). This observation suggests that CTD-induced cell wall damage might be due to ER stress. Furthermore, we checked Slt2 phosphorylation upon co-treatments with CTD ϩ GSH and CTD ϩ NAC. We found that both the co-treatments did not cause any significant change in Slt2 phosphorylation compared with CTD alone. Moreover, only GSH or NAC did not affect Slt2 phosphorylation (Fig. S12, A and B), suggesting a distinct mechanism of GSH-induced ER stress unlike DTT, TM, and CTD. We conclude that CTD perturbs cell wall integrity via ER stress. Our data indicate the linked phenotypes of ER stress and cell wall damage, which is illustrated briefly in the schematic diagram (Fig. 3F).

CTD alters GPI-anchored protein sorting
To identify the major pathway affected by CTD treatment, we did functional clustering (57) of the genetic interactors of CRG1 that show synthetic lethality in the presence of CTD (30). We found that the majorly affected pathways were associated with the ER-Golgi traffic system (Table S4). Yeast cell wall biosynthesis and maintenance mainly depend on the GPI-anchored proteins, sorted by the ER-Golgi traffic system (1,7,58). PE also plays a crucial role in the regulation of this traffic system (4,7,44,58). Thus, we hypothesized that the CTD-induced cell wall damage might be due to the defect in GPI-anchored protein sorting. We decided to study the GPI-anchored protein sorting upon CTD treatment. We used Gas1-GFP as a model GPI-anchored protein and tracked its localization in WT and crg1⌬ cells upon CTD treatment (3,4,59). We observed that CTD induced missorting and aggregation of Gas1-GFP in crg1⌬ cells (Fig. 4A). Additionally, Gas1-GFP protein, mature (M) and immature (IM), decreased considerably after CTD treatment in crg1⌬ mutant, probably due to the degradation of the aggregated proteins (Fig. 4, B-D) (60 -62). We also observed slower migration and shift of the Gas1-GFP(M) band in SDS-PAGE (Fig. 4B) upon CTD treatment in crg1⌬ mutant, indicating the direct effect on the maturity of GPI-anchored proteins (Figs. 4B and 6E). Supplementation of ETA completely rescued the sorting of Gas1-GFP in CTD-treated crg1⌬ cells (Fig. 4A). This might be the reason for ETA-mediated rescue against CTD treatment (Fig. 1, A and C). Furthermore, we measured DTT-extractable cell surface proteins (CSPs) integrated into the cell wall and plasma membrane through GPI-anchors. These proteins were extracted as described previously (63). We found a high yield of CSPs from CTD-treated crg1⌬ cells compared with WT cells. However, the supplementation of ETA restored the cell surface proteins to normal level, equal to that of WT (Fig. S4, A and B). High yield of CSPs from CTD-treated crg1⌬ cells are maybe due to inappropriate anchorage to cell Cantharidin targets GPI-anchor remodeling wall or cell membrane; hence, they become easily extractable from the surface. In contrast, ETA supplementation restabilizes the binding of the GPI-anchors, reversing the phenotype to normal and equivalent to WT. To further ascertain the role of CTD on GPI-anchor biosynthesis, we checked the genetic interaction of CRG1 with a few GPI-anchor biosynthesis genes (GPI2, GPI13, and MCD4) (1). Because these genes are essential for the cell survival, we used their heterozygous deletion mutants (gpi2⌬/GPI2, gpi13⌬/GPI13, and mcd4⌬/MCD4). We deleted CRG1 to create double-deletion mutants, crg1⌬/ ⌬gpi2⌬/GPI2, crg1⌬/⌬gpi13⌬/GPI13, and crg1⌬/⌬mcd4⌬/ MCD4, and performed a growth assay upon CTD treatment. Surprisingly, the double-deletion mutants (crg1⌬/⌬gpi2⌬/ GPI2, crg1⌬/⌬gpi13⌬/GPI13, and crg1⌬/⌬mcd4⌬/MCD4) showed better growth compared with single-deletion mutant, crg1⌬/⌬, in CTD-treated medium (Fig. 4 (E and F) and Fig. S3, F-I). That suggests that the molecular target of CTD may be downstream of the GPI biosynthesis cascade (4). We conclude that CTD alters the GPI-anchored protein sorting, which can be rescued by ETA supplementation.

Cdc1-mediated GPI-anchor remodeling is the major target of CTD
GPI-anchor remodeling is the step successive to biosynthesis. Based on the results discussed above, we hypothesized that GPI-anchor remodeling may be the target of CTD (30). We A, CTD and cell wall-perturbing agents (CR or CFW) are synergistically lethal to crg1⌬ mutant. Equal numbers of WT and crg1⌬ cells were serially diluted and spotted on SC agar medium containing CTD with and without CR or CFW. The cells were incubated at 30°C for 72 h. B-E, Western blot analysis of the Slt2 phosphorylation. Whole-cell lysates were prepared from WT and crg1⌬ cells grown in different conditions. Tbp1 was taken as a loading control. B, CTD-induced cell wall damage increases with heat stress. WT and crg1⌬ strains were grown at two different temperatures, 24 and 37°C, until mid-exponential phase (0.8 A 600 ) and then treated with CTD in the presence or absence of ETA for 2 h. The data represent one of the three independently performed experiments. C, densitometry quantification of the three biological repeats of the Western blots shown in B, with the help of ImageJ software. The graph shows a scatter plot of each data point of three independent experiments with mean Ϯ S.D. (error bars). Statistical analysis was done, applying Student's t test, where p Յ 0.05 (*), p Յ 0.01 (**), and p Յ 0.001 (***). D, CTD-induced cell wall damage increases with UPR induction. WT and crg1⌬ strains were grown at 24°C until mid-exponential phase and treated with CTD with or without DTT or TM for 2 h. The figure represents one of the three independently performed experiments. E, densitometry quantification of the three biological repeats of the western blots shown in D, with the help of ImageJ software. The graph shows a scatter plot of each data point of three independent experiments with mean Ϯ S.D. (error bars). Statistical analysis was done, applying Student's t test, where p Յ 0.05 (*), p Յ 0.01 (**), and p Յ 0.001 (***). F, a hypothetical model connecting two majorly affected pathways, UPR and CWI, by CTD. The CTD inhibits HAC1 mRNA splicing and subsequent UPRE activation, which promotes ER stress. Yeast cell wall biosynthesis is an ER-dependent process; the CTD-induced ER stress may alter the cell wall integrity, evident in this study by Slt2 activation. The probable link between ER stress and cell wall damage could be the GPI-anchored protein sorting, and it might be the direct target of CTD.

Cantharidin targets GPI-anchor remodeling
decreased GPI biosynthesis can rescue the growth defect of the mutants lacking GPI-anchor remodeling, perhaps by decreasing the GPI traffic on the remodeling factors. Because of the dynamic behavior of cdc1 alleles (cdc1-314 and cdc1-310) against CTD, we hypothesized that the Cdc1 activity could be a specific target of CTD in the remodeling process ( Fig. 5A and Figs. S5 and S6 (C-F)). As the activity of Cdc1 is Mn 2ϩ -dependent (3, 5, 6), we decided to examine the effect of CTD by controlling Mn 2ϩ concentration in the medium. We added the diionic chelator EGTA into the medium along with CTD and checked the fitness profile of the mutants. We observed that the growth of single and double mutants (crg1⌬, crg1⌬cdc1-310, and crg1⌬cdc1-314) was suppressed gradually with increasing concentration of EGTA ( Fig. 5C and Fig. S6A). Moreover, exogenous supplementation of MnCl 2 recovered the growth of crg1⌬cdc1-310 mutant in CTD-treated medium (Fig. 5D and Fig. S6 (B and G-J)). In this study, the two cdc1 alleles exhibit reproducible phenotype in a condition different from that reported previously (3,6). In summary, we conclude that CRG1 and CDC1 work in two different axes; CRG1 works as a guard to resist CTD, whereas CDC1 participates in the remodeling process. Loss of CRG1 results in the increased availability of active CTD that impairs the remodeling process by targeting the Cdc1 activity (Fig. 5E). Although both of the genes work in two different axes, they are required in parallel to tolerate CTD toxicity. We conclude that Cdc1 activity is essential to tolerate CTD cytotoxicity, and it may serve as a mechanistic target of the drug.

CTD-induced phenotypes strongly correlate with the loss of Cdc1 activity
Analysis of the phenotypes observed in this study and the investigations conducted previously suggest that CTD treatment induces phenotypes similar to cdc1 mutants (cdc1-314, Figure 5. CTD targets Cdc1 activity involved in GPI-anchor remodeling. A-D, growth sensitivity assay. Equal numbers of serially diluted cells of WT and the indicated mutants were spotted on SC agar medium with various treatments. Images were captured after 72 h of incubation. A, CRG1 shows synthetic lethality with GPI-anchor-remodeling genes under CTD stress. A spot assay on medium with increasing doses of CTD (1-8 M) was followed by incubation at 25°C. B, CRG1 shows synthetic lethality with GPI-anchor-remodeling genes under CTD and heat stress. A spot assay was done on medium containing CTD and incubated at 25 and 37°C. C, Mn 2ϩ chelation increases CTD toxicity. The yeast strains indicated above were spotted on medium containing CTD with and without EGTA and incubated at 25°C. D, Mn 2ϩ supplementation decreases CTD toxicity. Yeast strains indicated above were spotted on medium containing CTD and MnCl 2 and incubated at 25°C. E, schematic representation of CTD-dependent genetic interaction of CRG1 with GPI-anchor-remodeling genes; PER1, GUP1, and CDC1.CRG1 show synthetic lethality with PER1, GUP1, and CDC1. cdc1-310 shows dose-dependent interaction with crg1⌬: synthetic rescue at lower dose (2-4 M) and synthetic lethality at higher dose (6 -8 M).

Cantharidin targets GPI-anchor remodeling
Next, we measured the growth of GPI-anchor-remodeling mutants in presence of CTD or antioxidants with increasing temperature. We found that the mutants of GPI-remodeling genes (per1⌬, gup1⌬, cdc1-310, cdc1-314, per1⌬cdc1-314, and gup1⌬cdc1-314) were sensitive to the higher doses of CTD, and the sensitivity increased with increasing temperature (Fig. S7). We also found these mutants to be hypersensitive to a reducing environment developed by supplementation of GSH or NAC into the medium (Fig. S7). The sensitivity to GSH as well as NAC increased again with elevated temperature. The result showed the essential role of redox balance in the remodeling process of the GPI-anchors and described the synergistic lethal phenotype generated by the co-treatments with CTD ϩ GSH or CTD ϩ NAC (Fig. 2E). Additionally, ETA supplementation did not rescue the growth defect of cdc1-314 and cdc1-310 at higher temperature (Fig. S7), suggesting that ETA-mediated rescue in Gas1-GFP sorting in CTD-treated cells did not occur via the GPI-anchor-remodeling mechanism. The hypersensitivity of the single mutant cdc1-314 to the higher doses of CTD indicates the involvement of a CRG1-independent pathway targeted by the drug (Fig. S7). We also observed that a higher dose of CTD (300 M) altered Gas1-GFP sorting even in WT strain (Fig. S8A). Furthermore, to investigate whether CTD-induced alteration in GPI-anchored protein sorting was PP2A/PP1-dependent or not (24), we analyzed Gas1-GFP localization in sit4⌬ (PP2A) and GLC7/glc7⌬ (PP1) strains (Fig. S8B). We did not find any defect in Gas1-GFP localization in both of the mutants, implying the CTD-induced alteration in GPI-anchored protein sorting was independent of PP2A and PP1. Overall, the CTD-induced phenotypes strongly correlate to that of the cdc1-314 allele, so perhaps CTD targets Cdc1 activity. The entire molecular mechanism of CTD cytotoxicity can be illustrated in a graphical model that summarizes the complete sequence of events induced by CTD (see Fig. 8, A and B).

CTD alters GPI-anchored protein sorting in human cancer cells
The pathway for the biosynthesis and sorting of GPI-anchored proteins is conserved from yeast to higher eukaryotes (1,64). Therefore, we reasoned that the fundamental mechanism of action of CTD would be similar in yeast and human cells. To study the GPI-anchored protein sorting in human cells, we used GFP-CD59 as a model GPI-anchored protein (5). We observed that CTD induced aggregation of GFP-CD59 in HeLa cells, whereas the untreated cells showed normal distribution of the protein (Fig. 7A and Fig. S8C). This observation suggests that the molecular mechanism of action of CTD is conserved from yeast to human cells. Furthermore, we also checked the total expression of GFP-CD59 in HeLa cells upon treatment with CTD. Unlike yeast, we did not find any change in GFP-CD59 expression (Fig. S13C), suggesting a distinct mechanism for the clearance of aggregated proteins in human cells. Next, we measured the phosphorylation of p44/42 (a human homologue of yeast Slt2). We found a significant induction in p44/42 phosphorylation in HeLa and HepG2 cells upon CTD treatment (Fig. 7D). CTD treatment also decreased the XBP1 mRNA expression (Fig. 7, B and C), suggesting a down-regulation of UPR similar to the yeast cells (Fig. 2, B and C) (65). To rescue the cells from CTD toxicity, we supplemented ETA, CHO, and INO into the Dulbecco's modified Eagle's medium. Interestingly, ETA supplementation rescued the HepG2 cells from CTD-induced cell death, but the rescue of HeLa cells was not considerable (Fig. 7, E and F), suggesting a cell type-specific utilization of ETA perhaps due to diverse origin. On the other hand, CHO and INO supplementation could not rescue the human cells (Fig. S9, A-D) from CTD toxicity as observed in yeast cells (Fig. 1A). Because the phenotypes induced by CTD treatment in human and yeast cells are quite similar, we propose a conserved mechanism of action of CTD in eukaryotes (Fig. 8).

Discussion
GPI-anchored proteins control essential biological functions in animal cells by regulating the cell-to-cell communication, adhesion, and signal transduction (2,9). GPI-anchored proteins are also shown to be involved in tumorigenesis and metastasis (10 -12). Targeting an essential cellular pathway is one of the key aspects of anticancer chemotherapeutics. In this study, employing extensive genetic and cell biological approaches, we identified Cdc1 (yeast homologue of human PGAP5)-mediated GPI-anchor remodeling as a mechanistic target of CTD in addition to PP2A and PP1. However, biochemical validations will further support our observations. CTD has been shown to disturb phospholipid homeostasis in crg1⌬ mutant (30). To understand the underlying mechanism of its action, we screened the crg1⌬ mutant for the auxotrophy of different phospholipids upon CTD treatment. This approach helped us to conclude that CTD specifically affects PE, which can be rescued by exogenous supplementation of ETA. CTD treatment induced phenotypes similar to psd1⌬ (39, 44), and we found that PSD1 was synthetically lethal in combination with CRG1 under CTD stress. The reason for PE auxotrophy upon CTD treatment may be either inhibition of PE biosynthesis or alteration in PE-associated structures (e.g. GPI-anchors). The biosynthesis of PC mainly depends on the availability of PE in ER, which suggests that PE deficiency can lead to the deficiency of PC as well (66). However, the supplementation with PC did not rescue the growth defect induced upon CTD treatment. Thus, we conclude that CTD probably alters the PE-associated structures or functions rather than PE biosynthesis. PE deficiency is also known to induce ER stress and UPR in yeast (44). On the contrary, we found decreased UPR upon CTD treatment in the absence as well as the presence of the UPR inducers (TM and DTT). Our further investigations revealed that the drop in UPR upon CTD treatment was due to alteration in ER-redox homeostasis and Cdc1 activity (4), where we found that increased GSH level or lack of Cdc1 activity diminished UPR. The oxidative environment in ER is maintained by low GSH/GSSG (1:1 to 3:1) ratio for correct folding and modifications of the proteins (47,49,50). Our study demonstrates that the oxidative environment is also essential for the process of GPI-anchor remodeling (Fig. S7). ER is the site of synthesis and fate determination of the secretory proteins in the cell. Biosynthesis and maintenance of the yeast cell wall majorly depends on these secretory proteins (1). We believe that the CTD-induced cell wall damage (30, 34) is due to alteration in ER homeostasis. The synergistic lethal effect of CTD with ER stress Cantharidin targets GPI-anchor remodeling inducers (heat, DTT, and TM) and cell wall-perturbing agents (CR and CFW) supports this hypothesis. CTD-induced Slt2 phosphorylation also increases synergistically with ER stress inducers. Thus, we conclude that CTD-induced ER stress triggers cell wall damage. We also found rescue from CTD-induced cytotoxicity upon ETA supplementation, the reason for which may be the restoration of the GPI-anchored protein sorting (38,44). However, we do not know the exact mechanism by which the increased PE level restores the GPI-anchored protein sorting against CTD.
Next, we investigated the molecular mechanism for the ER stress and cell wall damage upon CTD treatment. The genetic interaction profile of CRG1 suggests that the ER-Golgi traffic system is a major pathway affected by CTD (30). In yeast cells, the proteins that travel from the ER to the cell wall are mostly the GPI-anchored proteins. GPI-anchored proteins constitute a major part of the total cell wall proteins and are required for the biosynthesis and maintenance of the yeast cell wall (1, 2, 7, 58, 64). Alteration in biosynthesis or sorting of GPI-anchored proteins induces ER stress and cell wall damage (4). Interestingly, we observed missorting and aggregation of the GPI-anchored protein (Gas1-GFP) upon CTD treatment. CRG1 showed synthetic rescue with GPI-anchor biosynthesis genes (GPI2, GPI13, and MCD4) and synthetic lethality with GPIanchor-remodeling genes (GUP1, PER1, and CDC1) upon CTD stress, indicating that the CTD alters GPI-anchored protein sorting by targeting the remodeling process (4). These results also support the genetic interaction profile of CRG1 reported previously (30). In addition, we identified CDC1 as an additional new gene that showed synthetic lethality with CRG1 in the presence of CTD. CDC1 encodes for Mn 2ϩ -dependent mannose-EtNP phosphodiesterase required for the removal of EtNP from the first mannose of the GPI-anchor (4). The crg1⌬cdc1-314 double mutant shows strong sensitivity to CTD compared with crg1⌬ and cdc1-314 single mutants. The triple mutant strains crg1⌬ per1⌬cdc1-314 and crg1⌬gup1⌬cdc1-314

Cantharidin targets GPI-anchor remodeling
were found to be even more sensitive to CTD compared with single (crg1⌬, per1⌬, gup1⌬, cdc1-310, and cdc1-314) and double mutants (crg1⌬per1⌬ and crg1⌬gup1⌬), suggesting that GPI-anchor remodeling is the major target of CTD. On the contrary, another allele of CDC1, cdc1-310, shows a dynamic phenotype upon CTD treatment. It shows synthetic rescue at lower dose and synthetic lethality at higher dose of CTD. Such dynamic and contrasting phenotypes of the two different alleles of CDC1 suggest a possibility of direct interaction of the enzyme with the small molecule CTD. To obtain more evidence in support of this hypothesis, we manipulated the Mn 2ϩ concentrations in the medium. We found that CTD toxicity enhanced with decreasing concentrations of Mn 2ϩ in the medium and vice versa, indicating an essential requirement of the Cdc1 activity to tolerate the CTD toxicity. Based on these results, we believe that CTD inhibits Cdc1 activity. CTD shows stronger affinity to Cdc1-314 than Cdc1-310, probably due to the specific protein confirmation. Previous studies suggest that CTD acts as a potent inhibitor of protein phosphatases PP2A and PP1 (23,24,67). However, our observations suggest that it can also inhibit lipid phosphatases such as Cdc1. CTD-dependent synthetic lethality of SAC1 (phosphatidylinositol phosphate phosphatase) with CRG1 supports this hypothesis (30,68). Furthermore, the sit4⌬ (PP2A) and GLC7/glc7⌬ (PP1) mutants do not show any defect in GPI-anchored protein sorting, suggesting that CTD-induced alteration in GPI-anchored protein sorting is independent of its known protein targets PP2A and PP1 (23,24,30). Moreover, we also found that the higher dose of CTD induces the same phenotypes in WT and cdc1-314 as it does in crg1⌬ mutant at a sublethal dose, suggesting that CTD-targeted pathways are independent of CRG1.
The enzymes involved in GPI-anchor biosynthesis and remodeling in yeast are mostly conserved in higher eukaryotes, suggesting that CTD can act through a similar mechanism in Figure 8. Schematic model illustrating the molecular targets and mechanism of CTD toxicity in yeast and higher eukaryotes. A, the model describes yeast Crg1 as a key defense molecule, localized in the cytoplasm, which protects the cell from CTD-induced cytotoxicity by methyltransferase activity. Loss of Crg1 enhances the binding of CTD to its molecular targets and perturbs the related biological functions. In the absence of Crg1, CTD enters into the ER and disturbs the ER homeostasis by altering the GSH/GSSG ratio and GPI-anchor remodeling, leading to missorting and aggregation of the proteins in the cytoplasm. B, illustration of the GPI-anchor-remodeling process in budding yeast. The C-terminal end of the protein is transferred to the ethanolamine phosphate of the third mannose of the GPI-anchor, catalyzed by a complex of enzymes, GPI-transamidase. In the subsequent process, Bst1 removes the acyl group from the inositol of GPI, Cdc1 removes ethanolamine phosphate from the first mannose, Per1 removes the unsaturated fatty acid (C18:1) from the sn-2 position of the GPI-lipid, Gup1 adds C26:0 saturated fatty acid at the sn-2 position of the GPI-lipid, and at last Cwh43 replaces the diacyglycerol type lipid with ceramide in GPI. Finally, the GPI-anchor is transferred to the plasma membrane or cell wall by Dfg5 or Dcw1. In this sequence of events, CTD targets Cdc1 activity, resulting in mislocalization and aggregation of GPI-anchored proteins.

Cantharidin targets GPI-anchor remodeling
higher eukaryotes. To validate the existence of a conserved mechanism of CTD toxicity, we extended our studies to human cell lines HeLa and HepG2. We observed similar phenotypes induced by CTD in human cells. We found missorting and aggregation of GPI-anchored GFP-CD59 in the cytoplasm of HeLa cells upon CTD treatment, which was very similar to that of Gas1-GFP in yeast. Similarly, CTD also induced phosphorylation of p44/42 (yeast Slt2), supporting the previous observations of CTD-mediated activation of different mitogen-activated protein kinases (18,21). We also found decreased expression of XBP1 (yeast HAC1) upon CTD treatment, which might be via ATF6 signaling that regulates the target gene XBP1 (65). The similar phenotypes produced by CTD in yeast and human cell lines suggest that the drug functions through a conserved mechanism.
Our study provides explanations for various observations reported upon CTD treatment in different organisms. CTDinduced alteration in GPI-anchored protein sorting can be a reason for the acantholysis (69 -71) and inhibition of cancer metastasis (12). CTD-induced perturbation in adhesion, morphogenesis, and membrane trafficking in Candida albicans may be due to alteration in GPI-anchored protein sorting (72). The G 2 /M cell cycle arrest by CTD (18,19,67) is probably due to inhibition of Cdc1 activity, as the loss of Cdc1 functions also induces G 2 /M cell cycle arrest (73). CTD has been a traditional medicine to cure warts and molluscum contagiosum caused by viral infections. Our study suggests that CTD can be further explored as an antifungal, antiviral, or antiprotozoan drug, utilizing its property of altering the ER-Golgi traffic system (8,15,74).
Because CTD targets a conserved and essential pathway, its exposure can also lead to lethal side effects. Therefore, the drug delivery is required to be very specific. A cancer-or tumorspecific delivery of CTD is the only way to make it a successful chemotherapeutic anticancer drug. Similarly, the poisoning of cattle foods by the contamination with the blister beetle is another challenge, as there is no antidote available against the beetle toxin. Our study suggests that ETA can serve as a potent antidote against CTD poisoning. In summary, we identified a novel target of CTD in addition to PP2A/PP1 and a potent antidote that neutralizes its lethal cytotoxicity.

Yeast strains, plasmids, and growth conditions
Unless otherwise stated, S. cerevisiae strains used in this study were isogenic with S288c (BY4741 or BY4743). All of the strains, plasmids, and primers used in this work are listed in Tables S1-S3, respectively. Yeast strains were grown in synthetic complete (SC) or yeast peptone dextrose (YPD) medium at 30°C, maintaining the optimum growth excluding some temporal stress conditions. Various reagents used in different experiments were purchased from Sigma, Merck, Himedia, Invitrogen, Bio-Rad, and Applied Biosystems.

Growth sensitivity assays
Serial dilution assay-Equal number (A 600 ϭ 1.0) of overnight grown yeast cells were serially diluted, 10-fold, five times and then spotted on SC agar medium. The spotted cells were incubated at different temperatures according to the various experimental conditions.
Growth curve assay-An equal number (A 600 ϭ 0.2) of exponentially growing yeast cells were inoculated in 96-well plates with and without different treatments and grown for 23-28 h in the automated plate reader (Biotek) acquiring a reading at A 600 in an interval of every 30 min.

Preparation of yeast whole-cell protein extract for Western blot analysis
Protein extraction from yeast cells was done by following the TCA protein extraction protocol (75). The equal number of cells were harvested and washed twice with 20% of TCA. Cell pellets were resuspended in 20% TCA with an equal volume of glass beads and vortexed rigorously to lyse the cells. TCA-precipitated protein extract was washed with ethanol and resuspended in 0.5 M Tris-Cl (pH 7.5) with 2ϫ loading buffer. The sample was boiled at 100°C for 10 min and centrifuged at maximum for 10 min to remove the debris. The supernatant was taken ahead for SDS-PAGE and Western blot analysis. The primary antibodies used in this study for immunoblotting experiments were as follows: anti-phospho-p44/42 (Cell Signaling, catalogue no. 4370S) to detect human phospho-44/42 and yeast phospho-Slt2 (pSlt2), anti-Mpk1 (Santa Cruz Biotechnology, Inc., catalogue no. SC-6803) to detect total Slt2 (Mpk1) in the cells, anti-GFP (Sigma, catalogue no. G1544) to detect Gas1-GFP and GFP-CD59, and anti-GAPDH (Cell Signaling, catalogue no. 5174S) as loading control for human cells. Primary antibody used for Tbp1 was the yeast loading control; it is a polyclonal antisera raised in rabbit. Densitometry quantification was performed with the help of ImageJ software in which the protein of interest was normalized with Tbp1 loading control.

Cell surface protein extraction
Yeast cells were washed twice with sodium phosphate buffer (0.1 M, pH 8.0). Wash-out solution was kept at 4°C. Collected cells were resuspended again in sodium phosphate buffer with 2 mM DTT and incubated at 4°C for 2 h, maintaining gentle agitation. Now the cells were pelleted down by centrifugation, and the supernatant along with the washout fraction was precipitated using 20% TCA in the final volume. TCA-precipitated cell surface proteins were separated via 8% SDS-PAGE and stained with 0.1% Coomassie Brilliant Blue R-250 (63).

RNA extraction, cDNA synthesis, and quantitative PCR for HAC1 mRNA splicing
RNA isolation was performed by using the heat/freeze RNA isolation protocol (76). Briefly, cells were grown until mid-exponential phase, harvested by centrifugation, and washed twice with 1ϫ PBS. Harvested cells were lysed with 1% of SDS in AE buffer (50 mM sodium acetate, 10 mM EDTA, pH 5.3). An equal volume of acidic phenol of pH 4.2 was added and incubated at 65°C for 4 min, followed by freezing at Ϫ85°C for 4 min, centrifuged for 2 min at maximum speed to separate the aqueous layer. The aqueous phase was mixed with an equal volume of phenol/chloroform/isoamyl alcohol (25:24:1) and separated again from the phenol phase. The total RNA present in the Cantharidin targets GPI-anchor remodeling aqueous phase was precipitated by adding sodium acetate (0.3 M) and 2.5 volumes of absolute ethanol. The cDNA synthesis was done by following the standard protocol provided by the iScript TM cDNA synthesis kit (Bio-Rad, catalogue no. 1708891). HAC1 mRNA splicing was measured using primers specified in Table S3, following the PCR conditions as described previously (53).

␤-gal assay
Exponentially growing yeast cells were harvested and washed twice with LacZ buffer (10 mM KCl, 1 mM MgSO 4 , 50 mM ␤-mercaptoethanol, and 100 mM NaPO 4 , pH 7.0). Cells were lysed using 0.01% SDS, and 22.7% chloroform in LacZ buffer in a final volume of 250 l. Subsequently, 500 l of ortho-nitrophenyl-␤-D-galactoside (2 mg/ml) was added and incubated at 30°C until the appearance of pale yellow color. The reaction was quenched by adding 500 l of sodium bicarbonate (1 M). The reaction mixture was centrifuged at maximum speed for 15 min, and the supernatant was collected to measure the absorbance at a wavelength of 420 nm. Miller unit for the ␤-gal activity was determined by applying the following formula: Miller unit ϭ (A 420 /A 600 ϫ time (min)) ϫ 1,000 (45).

Fluorescence microscopy
Fluorescence microscopy was done to study the sorting of GPI-anchored protein (Gas1-GFP) with or without CTD. Yeast cells were grown in YPD medium for different time points and harvested by centrifugation (6,000 rpm, 2 min, 4°C). Cell pellets were resuspended in PBS and kept on ice for at least 30 min (59). Gas1-GFP localization was observed using a ZEISS-Apotome.2 fluorescence microscope under a ϫ60 oil emulsion objective lens. For the microscopic localization study of the GFP-CD59, the overnight-grown HeLa cells with 50% confluence were treated with CTD (5 M) for 12 h and visualized under a ϫ20 emulsion oil objective lens (5).

Cell culture and maintenance of human cell lines
Human cell lines (HeLa and HepG2) were maintained in Dulbecco's modified Eagle's medium (Lonza) having 10% fetal bovine serum (Gibco) and antibiotics (i.e. penicillin (100 units/ ml) and streptomycin (100 g/ml). Both of the cell lines were grown at 37°C with 5% CO 2 .

Cell survival assay
Percentage survivability of the cells against CTD exposure was measured by an MTT assay. HeLa and HepG2 cells were seeded in a 96-well plate equal in number (5,000 cells) in each well. Cells were incubated for 24 h. Medium was removed, and fresh medium was added to the cells; simultaneously, cells were also challenged with CTD with or without supplementation of ETA for 48 h. 10 l of MTT solution (5 mg/ml in 1ϫ PBS) was added and incubated at growth conditions for 4 h. 100 l of DMSO was added and mixed well. Absorbance was recorded at 570 nm using a microplate reader (Biotek) (16).

Statistical data analysis
Statistical analysis for the ␤-gal assay was performed by using GraphPad Prism version 5 software. Each graph shows the indi-vidual data points with mean value as a horizontal green line. The error bars represent S.D. of a minimum of three individual repeats. We applied two-way ANOVA and Bonferroni post hoc test, where p Յ 0.05 (*), p Յ 0.01 (**), and p Յ 0.001 (***). Statistical significance of the relative -fold change in Slt2 phosphorylation or Gas1 expression with and without CTD treatment was calculated by applying Student's t test, where p Յ 0.05 (*), p Յ 0.01 (**), and p Յ 0.001 (***).
Author contributions-The hypothesis and experiments were designed by R. S. T. and P. K. S. Experiments were performed by P. K. S., and the results were analyzed by P. K. S. and R. S. T. The manuscript was written by P. K. S. and R. S. T.