Sphingolipids are a class of membrane lipids conserved from yeast to mammals which determine whether a cell dies or survives. Perturbations in sphingolipid metabolism cause apoptotic cell death. Recent studies indicate that reduced sphingolipid levels trigger the cell death, but little is known about the mechanisms. In the budding yeast Saccharomyces cerevisiae, we show that reduction in complex sphingolipid levels causes loss of viability, most likely due to the induction of mitochondria-dependent apoptotic cell death pathway, accompanied by changes in mitochondrial and endoplasmic reticulum morphology and endoplasmic reticulum stress. Elevated cytosolic free calcium is required for the loss of viability. These results indicate that complex sphingolipids are essential for maintaining endoplasmic reticulum homeostasis and suggest that perturbation in complex sphingolipid levels activates an endoplasmic reticulum stress-mediated and calcium-dependent pathway to propagate apoptotic signals to the mitochondria.
Sphingolipids, especially abundant complex sphingolipids, are important structural components of eukaryotic cell membranes. Sphingolipids associate with sterols to form membrane microdomains (also referred to as lipid rafts) as a platform for sorting proteins and clustering signal transducers (van Meer et al., 2008). In addition to structural roles, sphingolipid metabolites such as ceramide, long-chain base (LCB) and LCB phosphate (LCB-P) play critical roles as signalling molecules in many cellular processes, including cell migration, stress response, survival, apoptosis, senescence, differentiation and endocytosis (Hannun and Obeid, 2008). In particular, ceramide and/or LCBs including sphingosine, dihydrosphingosine and phytosphingosine are well-known mediators of apoptotic death in multi-cellular organisms, whereas LCB-P opposes them and acts as a pro-survival signal (Cheng et al., 2003; Taha et al., 2006; Castro et al., 2008; Hannun and Obeid, 2008; Wang et al., 2008; Ponnusamy et al., 2010). The balance between the concentrations of ceramide and LCB-P could be a cellular rheostat that determines whether a cell dies or survives (Ponnusamy et al., 2010). The sphingolipid rheostat is likely evolutionarily conserved in eukaryotes, although the pro-apoptotic role of ceramide has not yet been documented in the budding yeast Saccharomyces (S.) cerevisiae as a model organism (Pereira et al., 2008; Abdelwahid et al., 2011).
While the major role of ceramide in initiating apoptosis is as a second messenger to activate downstream effectors, including ceramide-activated protein phosphatases, protein kinases and cathepsin D (Taha et al., 2006; Hannun and Obeid, 2008; Ponnusamy et al., 2010), it is becoming clear that the biophysical properties of ceramide and its effects on the biological membrane play important roles in cell death. Indeed, ceramide can form channels to permeabilize the mitochondria (Colombini, 2010), which then release apoptogenic proteins such as cytochrome c or apoptosis-inducing factor (Aif1) into the cytosol. In addition, recent studies suggest that ceramide induces an endoplasmic reticulum (ER) stress response and triggers an ER-associated apoptotic cell death pathway (Huang et al., 2011). Although it is unclear how ceramide actually induces ER stress and apoptosis, one can speculate that excess ceramide is the direct cause of ER stress and apoptosis by alteration of ER-membrane fluidity or order. There is also a possibility that ceramide accumulation in the ER impairs ER-to-Golgi protein trafficking, thereby causing ER stress via protein overload. Given the variety of target proteins and actions of ceramide, it is likely that ceramide-induced apoptosis is mediated through multiple mechanisms and pathways. Like ceramides, complex sphingolipids, and in particular gangliosides such as GD3 and GM1, also can elicit an ER stress-dependent apoptosis (Morales et al., 2007; Sano et al., 2009).
Contrary to the pro-apoptotic function of ceramide, a recent study showed that knock-down of ceramide synthase 6 (CerS6) using siRNAs induces ER stress-mediated apoptosis in human head and neck squamous cell carcinoma (HNSCCs) cells (Senkal et al., 2010). These data indicate a pro-survival role for CerS6-generated ceramide in the regulation of apoptosis in mammalian cells. Such positive correlations between downregulated ceramide synthesis and increased cell death have also been described in human skin fibroblasts (Schulz et al., 2006) and in Caenorhabditis elegans (Menuz et al., 2009). In addition, in the yeast S. cerevisiae, overexpression of YDC1 which encodes a ceramidase, leads to reduced chronological lifespan and increased apoptotic cell death (Aerts et al., 2008). Because the reduced lifespan can be suppressed by exogenous addition of ceramide, the yeast apoptosis from YDC1 overexpression is likely because of decreased ceramide levels, rather than increased sphingoid base levels (Aerts et al., 2008). Consistent with these findings, disruption of ISC1 which encodes an inositol phosphosphingolipid phospholipase C, leading to a reduced hydrolysis of yeast complex sphingolipids including inositol phosphorylceramide (IPC), mannosylinositol phosphorylceramide (MIPC) and mannosyldi(inositolphosphoryl)ceramide (M(IP)2C) and hence decreased overall ceramide content within the cell (Cowart et al., 2006) or decreased α-hydroxylated C26-phytoceramide content in mitochondria (Kitagaki et al., 2007b), exhibits defective mitochondrial phenotypes, oxidative stress hypersensitivity and stimulated caspase-dependent apoptosis during chronological ageing (Almeida et al., 2008; Barbosa et al., 2011). These studies with different organisms may suggest an evolutionarily conserved action of ceramide to protect apoptosis. However, the mechanisms by which downregulated ceramide synthesis induces apoptotic cell death remain poorly understood. One possibility is that further metabolism of ceramide to complex sphingolipids is linked to the apoptosis associated with reduced ceramide levels; this possibility needs to be elucidated.
In this study, we addressed whether ceramide or complex sphingolipids can play a critical role in the regulation of apoptotic cell death in S. cerevisiae. We show that a reduction in complex sphingolipid levels but neither increase nor decrease in ceramide levels leads to loss of cell viability, which is mediated by metacaspase and apoptogenic mitochondrial protein cytochrome c. We further demonstrate that perturbation of complex sphingolipid metabolism causes ER stress and leads to a rise in cytosolic Ca2+ levels. Furthermore, the suppression of cytosolic Ca2+ levels can prevent the loss of viability. These results imply a link between the perturbation of the complex sphingolipid metabolism and ER stress-mediated and Ca2+-dependent mitochondrial apoptosis in yeast.
Inhibition of IPC synthesis by aureobasidin A (AbA) induces metacaspase-dependent mitochondrial apoptosis in S. cerevisiae
To explore a role for sphingolipids in the regulation of yeast apoptosis, we first tested the effects of AbA. AbA inhibits IPC synthase activity, leading to both increased levels of ceramides and decreased levels of complex sphingolipids (Cerantola et al., 2009). Because AbA treatment led to loss of cell viability as measured by colony-forming assay (Fig. 1A), we asked if AbA-treated cells exhibit apoptotic markers such as nuclear fragmentation and chromatin condensation. As shown in Fig. 1B, electron microscopy investigation of wild-type cells treated with AbA for 4 h revealed nuclear fragmentation and chromatin condensation. The latter morphological feature resembled the chromatin condensation seen in cells treated with H2O2 (Fig. 1B) (Madeo et al., 1999). As assessed by labelling with 2′, 7′-dichlorofluorescein diacetate (H2DCFDA), a reactive oxygen species (ROS) indicator (Madeo et al., 1999), AbA induced an increase in ROS levels (Fig. 1C and D) as was also seen after H2O2 treatment. The other hallmark of apoptosis (Madeo et al., 1999), nuclear DNA fragmentation, was also observed in AbA-treated cells in the TdT-mediated dUTP nick end-labelling (TUNEL) assay (Fig. 2A and B).
Because yeast programmed cell death induced by external stimuli or internal signals is often dependent on metacaspase activity accompanying cytochrome c release from mitochondria (Silva et al., 2005; Braun et al., 2006), we next tested the involvement of these proteins in the loss of viability caused by AbA. Deletion of the gene YCA1, which encodes a yeast metacaspase, led to partial suppression of both AbA-induced increased DNA fragmentation (Fig. 2A and B) and loss of viability (Fig. 2C). Additionally, the yca1Δ mutation conferred resistance to AbA (Fig. 2D). Similar results were also observed in a deletion mutant of RSM23 (Fig. 2A–D). RSM23 encodes the putative yeast orthologue of the mammalian mitochondrial mediator of apoptosis, DAP-3, which functions downstream of Yca1 (Madeo et al., 2002). Furthermore, deletion of the CYC1 and CYC7 genes, which encode two isoforms of cytochrome c, resulted in a higher survival (Fig. 2C) and a lower sensitivity to AbA (Fig. 2D), compared with the wild-type control cells. These results suggest that AbA-induced loss of viability depends on Yca1 activity and involves the mitochondrial release of cytochrome c.
Mitochondria undergo dynamic morphological changes during apoptotic cell death (Pereira et al., 2008; Abdelwahid et al., 2011). These include mitochondrial membrane permeabilization (MMP) and mitochondrial fragmentation (fission). MMP is associated with the release of mitochondrial apoptogenic proteins. Absence of ADP/ATP carrier (AAC) proteins, yeast orthologues of the adenine nucleotide translocator (ANT), impair MMP and cytochrome c release in acetic acid-induced cell death (Pereira et al., 2007). Consistent with the involvement of cytochrome c in AbA-induced loss of viability, we found that strains lacking the AAC gene (AAC1 or AAC3) exhibited resistance to AbA treatment (Fig. S1A). Mitochondrial fragmentation was also detectable in the AbA-treated wild-type cells (Fig. S1B). Taken together, these findings suggest that inhibition of IPC synthesis by AbA induces Yca1-dependent mitochondrial apoptosis in yeast.
Inhibition of glycosylphosphatidylinositol (GPI) assembly stimulates AbA-induced loss of viability
GPI assembly is required for the efficient transport of ceramides from the ER (Kajiwara et al., 2008). Because mutants defective in GPI assembly have reduced levels of complex sphingolipids (Kajiwara et al., 2008), we tested whether inhibition of GPI assembly stimulates the apoptotic cell death induced by AbA. To address this question, wild-type, gpi single mutants (arv1Δ, gpi2-7 and gpi3-10), double mutants (arv1Δ yca1Δ) and a triple mutant (arv1Δ cyc1Δ cyc7Δ) were analysed for survival and sensitivity to AbA. As shown in Fig. 3A, the loss of viability caused by AbA was significantly enhanced in mutants defective in GPI synthesis (arv1Δ, gpi2-7 and gpi3-10). The increased inviability in arv1Δ mutant strains was suppressed by deletion of the YCA1 (Fig. 3B left) or RSM23 gene (Fig. 3B right). Also, deletion of the YCA1, RSM23 (Fig. 3C) or AAC3 (Fig. S1A) gene, as well as the double deletion of the CYC1 and CYC7 (Fig. 3C) genes, rescued the hypersensitivity of the arv1Δ mutant to AbA. Interestingly, gpi mutants produce more ROS than wild-type cells under basal conditions without AbA treatment (Fig. 3D and E). Also, in the absence of AbA, loss of Arv1 function causes mitochondrial fragmentation (Fig. S1B). These results suggest that inhibition of GPI assembly exacerbates AbA-induced cell death.
Reduction in complex sphingolipid levels is associated with AbA-induced loss of viability
It is unclear whether increased ceramides or decreased complex sphingolipids induce yeast apoptosis in response to treatment with AbA. To determine if reduction in complex sphingolipids causes AbA-induced loss of viability, we next used a temperature-sensitive lcb1-100 mutant defective in serine palmitoyltransferase (SPT), the enzyme catalysing in the first step in sphingolipid synthesis (Zanolari et al., 2000). The lcb1-100 mutant only produces low levels of all sphingolipids even at permissive temperatures (Zanolari et al., 2000). As shown in Fig. 4A, AbA-induced loss of viability was aggravated in the lcb1-100 mutant strain in comparison with the wild-type strain, implying the contribution of reduced sphingolipid levels to the loss of viability. Consistent with this interpretation, treatment with myriocin (Myr), a specific SPT inhibitor (Horvath et al., 1994) led to loss of viability and it was aggravated by inhibition of GPI assembly (Fig. 4B). Like AbA treatment, the lcb1 mutation and Myr treatment resulted in ROS accumulation (Fig. 4C and D) and the latter induced mitochondrial fragmentation (Fig. S1B).
Furthermore, total amounts of newly synthesized ceramides were either increased or unchanged but not decreased (Fig. 4E and F) during the AbA treatment even at a high concentration (2 μg ml−1), which causes almost complete inhibition of incorporation of [3H]dihydrosphingosine (DHS) into complex sphingolipids (Fig. 4E and G). Thus, all together, these results suggest that reduction of complex sphingolipid levels but neither increase nor decrease of ceramide levels is critical for death of yeast cells treated with AbA.
Accumulation of immature ER forms of GPI proteins causes ER stress
Genes required for GPI assembly genetically interact with Ire1 (Ng et al., 2000), an ER stress sensor, which activates the unfolded protein response (UPR) signalling pathway. The UPR is constitutively activated in mutants defective in GPI-anchor synthesis or remodelling (Copic et al., 2009; Jonikas et al., 2009; Castillon et al., 2011; Shechtman et al., 2011). HAC1 encodes the downstream transcription factor of Ire1. Consistent with these previous studies, we observed that double deletions of ARV1 and IRE1 or HAC1 caused a synthetic sick interaction at 25°C or a lethal interaction at 37°C (Fig. 5A). In addition, the level of expression of the well-known UPR target gene, KAR2 under control of its own promoter was higher in the arv1Δ mutant than in wild-type yeast (Fig. 5B). The glycosylation inhibitor, tunicamycin (TM) is a potent inducer of the UPR. Because mutant strains that have constitutive activation of the UPR are sensitive to TM (Takeuchi et al., 2006; Han et al., 2010), we assessed the sensitivity of arv1Δ mutant to TM (Fig. 5C). As reported previously (Liu and Chang, 2008), the ire1Δ mutant showed a hypersensitivity to TM. The arv1Δ mutant grew more slowly than wild-type cells in the presence of TM, suggesting that cells defective in GPI assembly display an increased sensitivity to the ER stress induced by TM. When the yeast ER was visualized by Kar2-GFP-HDEL (CEN, URA3) which bears the TDH3 promoter and the actin terminator, as an ER marker (Okamoto et al., 2006), we also noticed that gpi mutants had altered ER structures with proliferation or expansion of the ER membrane (Fig. 5D). Similar morphological changes of the ER were observed previously in the yeast after ER stress (Bernales et al., 2006), and ER membrane expansion may alleviate ER stress by providing space to accommodate newly synthesized ER proteins or folding machinery (Schuck et al., 2009). Thus, these findings indicate that defects in GPI assembly cause ER stress. Because defects in GPI assembly result in accumulation of immature ER forms of GPI proteins (Kajiwara et al., 2008), the findings also raise the possibility that the accumulation of immature GPI proteins may cause a very strong ER membrane perturbation and UPR response leading to apoptosis. Interestingly, ARV1 deletion and gpi defects have been shown to lead to an UPR response even if Ire1, the stress sensor, is unable to sense unfolded proteins (Promlek et al., 2011).
Because sphingolipid synthesis is required for the efficient ER exit of GPI proteins (Horvath et al., 1994; Watanabe et al., 2002), it is possible that inhibition of complex sphingolipid synthesis causes accumulation of immature GPI proteins in the ER. To address this possibility, we determined the steady-state levels of the immature and mature forms of Gas1, a GPI protein, and carboxypeptidase Y (CPY), a non-GPI protein, by Western blot analysis. By measuring the levels of these proteins, we tested the effect of perturbation of sphingolipid metabolism by AbA. As shown in Fig. 5E, wild-type cells treated with AbA revealed a significant increase in the level of immature Gas1 but not immature CPY (p1 and p2). The increased level of immature Gas1 is not a consequence of ceramide accumulation because blocking all sphingolipid synthesis with Myr also increased the immature Gas1 level. Next, we analysed the localization of mRFP-Gas1 (Fig. 5F). As reported previously, in wild-type cells, mRFP-Gas1 was localized to the plasma membrane (Fujita et al., 2006) and the sites of the mother-daughter neck region (Rolli et al., 2009). However, when the wild-type cells were treated with AbA or Myr, mRFP-Gas1 became colocalized with Kar2-GFP-HDEL, suggesting that sphingolipid metabolism is required for proper localization of Gas1 and its perturbation leads to accumulation of Gas1 in the ER. Mislocalization of Gas1 to the ER was also observed when ER stress was induced by TM treatment leading to a severe underglycosylation of Gas1 (Fig. 5B and E) (Pittet et al., 2006) or when a misfolded mutant mRFP-gas1* was expressed in wild-type cells as reported previously (Fujita et al., 2006). We also found that, like TM-treated wild-type cells, AbA-treated, Myr-treated, mRFP-gas1*-expressing wild-type cells (Fig. 5F), and lcb1-100 mutant cells (Fig. 5D) all displayed aberrant ER morphology similar to the ERs of gpi mutants. Additionally, data show that KAR2 expression was increased by treatment with AbA or Myr in a time- and dose-dependent manner (Fig. 5E and G). The increased expression levels of KAR2 were significantly higher than that seen in the cells treated with TM, indicating an increased activation of the UPR in AbA- or Myr-treated cells. Therefore, these results reinforce the idea that accumulation of immature GPI proteins in the ER causes ER stress.
Cytosolic-free Ca2+ is involved in AbA-induced loss of viability
Prolonged ER stress can activate apoptotic cell death (Perrone et al., 2008; Schroder, 2008; Giorgi et al., 2009). Because Ca2+ is a potential signalling molecule linking ER stress and mitochondrial apoptosis (Perrone et al., 2008; Giorgi et al., 2009), we assessed if cytosolic-free Ca2+ ([Ca2+]c) levels increase in response to ER stress induced by perturbations of sphingolipid metabolism and GPI assembly. For this experiment we measured the expression of PMC1–lacZ, a Ca2+/calmodulin-induced calcineurin-dependent reporter gene (Cunningham and Fink, 1996). As shown in Fig. 6A, expression of PMC1–lacZ was increased by treatment with AbA. Moreover, treatment with Myr and TM showed increased PMC1–lacZ expression similar to that seen with AbA treatment. The increases of PMC1–lacZ expression were almost completely abolished by treatment with FK506, an inhibitor of calcineurin, as expected (Stathopoulos and Cyert, 1997; Bonilla and Cunningham, 2003). Consistent with ER stress response to Myr, the expression level of PMC1–lacZ in the lcb1-100 mutant was much higher than that in the wild-type cells (Fig. 6B left). These results suggest that when the sphingolipid metabolism is perturbed, the [Ca2+]c becomes elevated. Furthermore, similar to pmr1Δ cells that lack the gene encoding a Golgi Ca2+ pump and can elevate [Ca2+]c (Bonilla et al., 2002), arv1Δ and gpi2-7 mutants exhibited higher expression levels of PMC1–lacZ than wild-type cells (Fig. 6B middle and right), suggesting that [Ca2+]c levels are elevated in mutants defective in GPI assembly. ER stress caused by treatment with TM or DTT and membrane stress by treatment with azole-class antifungal drugs stimulate Ca2+ influx through a plasma membrane Ca2+ channel composed of Mid1 and Cch1 and activate calcineurin signalling (Bonilla et al., 2002; Bonilla and Cunningham, 2003). This process known as capacitive Ca2+ entry serves to increase the magnitude and duration of Ca2+ signals and to replenish the Ca2+-depleted organelles (Parekh and Putney, 2005). We tested the effect of MID1 gene deletion on the expression of PMC1–lacZ in the arv1Δ mutant. We found that the increased expression of PMC1–lacZ in the arv1Δ mutant is suppressed by deletion of MID1 (Fig. 6B right). Therefore, these results imply that accumulation of immature GPI proteins in the ER stimulates Ca2+ influx via the Mid1-Cch1 channel.
To ascertain the role of Ca2+ in ER stress-mediated cell death, we asked whether loss of cell viability caused by AbA treatment and stimulated by ARV1 deletion depends on Pmr1 or Mid1. We monitored the survival of arv1Δ pmr1Δ and arv1Δ mid1Δ double mutants after AbA treatment. Although deletion of two genes ARV1 and PMR1 does not seem to have an additive effect on [Ca2+]c elevation (Fig. 6B right), the survival of arv1Δ cells was decreased by the additional deletion of PMR1 (Fig. 6C). Consistent with this, the combination of arv1Δ and pmr1Δ mutations resulted in synthetic growth defects at 25°C (Fig. 6D and E) and 37°C (Fig. 6E). In contrast, absence of Mid1 suppressed the AbA-induced loss of viability in arv1Δ cells, as well as in wild-type cells. A similar suppressive effect of the MID1 deletion was observed in the hypersensitivity of arv1Δ cells to AbA and the mid1Δ cells showed more resistance to AbA compared with the wild-type cells (Fig. 6D). Deletion of CCH1 also conferred resistance to AbA. Taken together, these findings provide evidence that AbA-induced loss of viability is dependent on Pmr1 and Mid1, which may suggest that [Ca2+]c elevation has a critical role in cell death induced by perturbations of sphingolipid metabolism and GPI assembly.
Here, we show that perturbation of sphingolipid metabolism by AbA treatment leads to loss of cell viability in S. cerevisiae, and the scenario we propose is presented in Fig. 7. AbA-induced loss of viability appears due to apoptotic cell death because of nuclear fragmentation, chromatin condensation and DNA strand breaks. This conclusion is further supported by the findings that, in addition to ROS accumulation and mitochondrial fragmentation, the metacaspase Yca1 and cytochrome c are required for AbA-induced loss of viability. An elevated [Ca2+]c that can be activated by ER stress, is also required. These findings provide evidence that the loss of viability is caused by an ER stress-mediated and Ca2+- and mitochondria-dependent apoptotic pathway.
Our results also suggest that reduced levels of complex sphingolipids but neither increased nor decreased levels of ceramides trigger AbA-induced loss of viability. It was previously reported that an ipt1 mutant lacking M(IP)2C, which has concomitant increases in MIPC and IPC (Dickson et al., 1997), displays both increased resistance to oxidative stress and increased chronological lifespan (Aerts et al., 2006). The pro-survival roles of MIPC and IPC are consistent with our results and further observation that mutations in gpi genes, which have a defect in complex sphingolipid synthesis (Kajiwara et al., 2008), reduce chronological lifespan (Fig. S2). However, it is unclear whether the ipt1 mutant suppresses AbA-induced cell death or if the ipt1 mutant's effect on the complex sphingolipids, MIPC and IPC, contributes to the pro-survival role. We found that two deletion mutants (csg1Δ, csh1Δ) lacking MIPC do not lead to a significant increase in sensitivity to AbA (data not shown), suggesting that lack of IPC but not lack of MIPC may induce apoptotic cell death in response to AbA.
Lipid homeostasis in ER and its role in ER stress are of hot interest. Although contradictory results have been reported concerning the effect of gene deletion on ceramide levels (Breslow et al., 2010; Han et al., 2010), an orm1Δ orm2Δ mutant with impaired sphingolipid homeostasis has a constitutively active UPR. The fact that the orm1Δ orm2Δ mutant has a delay in the ER-to-Golgi GPI protein transport (Han et al., 2010), together with our results, suggest that accumulation of immature GPI proteins in the ER caused by impaired sphingolipid homeostasis can be principally responsible for ER stress and consequent cell death (Fig. 7). However, it is also possible that alterations in other lipid metabolisms may be the direct cause of ER stress. For example, excess cholesterol in the ER can activate the UPR and subsequently trigger apoptosis in macrophages (Feng et al., 2003). Furthermore, a recent study has shown that cellular accumulation of saturated fatty acids and ergosterol, the major sterol in yeast, induces the UPR in yeast (Pineau et al., 2009). Because sphingolipids and sterols interact functionally (Guan et al., 2009) and because there is a link between GPI assembly deficiency accompanied by a decrease in complex sphingolipid levels and increased sterol levels (Kajiwara et al., 2008), aberrant sterol metabolism may contribute to AbA-induced ER stress and apoptosis.
How reduction in complex sphingolipid levels causes ER stress is currently unknown. But, as Aur1, the target for AbA is localized to the Golgi compartment (Levine et al., 2000), it is tempting to speculate that Golgi apparatus and/or plasma membranes, which sense low levels of complex sphingolipids may send the signal preventing exit of GPI proteins to the ER (Fig. 7). Interestingly, it was recently shown that plasma membrane stress induced by inhibition of sphingolipid synthesis activates the target of rapamycin kinase complex 2 (Berchtold et al., 2012), which is functionally linked to the Pkc1 and Mpk1 proteins (De Virgilio and Loewith, 2006). Thus, ER stress and plasma membrane stress induced by reduction of complex sphingolipids may trigger signalling through Pkc1/Mpk1 pathway, resulting in activation of the Mid1-Cch1 Ca2+ channel (Bonilla and Cunningham, 2003). The identification of the lipid that accounts directly for ER stress, and the mechanisms by which the perturbation of lipid metabolism induces ER stress are critical issues in further studies.
Ca2+ is a signal that can link perturbations in the ER homeostasis by stresses, such as lipid imbalance and misfolded protein accumulation, to mitochondria-dependent apoptosis (Perrone et al., 2008; Giorgi et al., 2009). In mammalian cells, accumulation of the ganglioside, GM1, and ceramides at the ER membrane alter Ca2+ dynamics and trigger mitochondria-mediated apoptosis (Morales et al., 2007; Giorgi et al., 2009; Sano et al., 2009; Novgorodov et al., 2011). Similarly, inducers of ER stress such as thapsigargin and TM also change Ca2+ levels and can trigger apoptosis (Deniaud et al., 2008; Giorgi et al., 2009). In yeast, mitochondria-dependent apoptosis associated with the rise of cytosolic Ca2+ has also been reported in pheromone- and amiodarone-induced cell deaths (Pozniakovsky et al., 2005).
We show that perturbations to sphingolipid metabolism and GPI assembly increase the cytosolic Ca2+ concentration. Also, the MID1 deletion alleviates AbA-induced loss of viability, while the PMR1 deletion augments, suggesting a causative role of Ca2+ influx via capacitive Ca2+ entry in mitochondrial apoptosis in response to AbA. Interestingly, amiodarone-induced Ca2+ influx is required for its toxicity (Pozniakovsky et al., 2005; Muend and Rao, 2008), while the Ca2+ influx in response to other stresses plays a pro-survival role, rather than a pro-death role (Bonilla et al., 2002; Courchesne et al., 2011) Consistent with the functional similarity of Ca2+ influx induced by AbA and amiodarone, we found that deletion of ARV1 conferred a sensitivity to amiodarone-induced loss of viability (Fig. S3), suggesting that a GPI assembly defect may stimulate apoptosis via the same pathway as amiodarone. Because the decreased survival in arv1Δ cells in response to amiodarone was alleviated by loss of Aac3, MMP might be a common process downstream of Ca2+ influx (Fig. 7).
These results are consistent with the plausible model that the influx of extracellular Ca2+ across the plasma membrane in response to AbA induces apoptotic cell death. However, AbA-induced loss of viability was not rescued by addition of the membrane-impermeable Ca2+ chelator 1,2-bis(2-aminophenoxy) ethane-N, N, N′, N′-tetra acetic acid (BAPTA, 1 mM) (data not shown). Interestingly, Mid1 has been reported to localize not only in the plasma membrane, but also in the ER membrane (Yoshimura et al., 2004). Thus, it is conceivable that Mid1 located in the ER membrane releases Ca2+ into the cytosol in response to AbA. MID1 deletion could reduce calcium stores of the cells, specially of the ER, and in this scenario, the damage caused by AbA to the ER membrane integrity or Ca2+ pumps might have less of an effect on cytosolic calcium levels, since there is not much calcium to be released from the ER.
Calcineurin acts downstream of the elevated [Ca2+]c to induce or suppress cell death (Lotem et al., 1999; Bonilla et al., 2002; Dudgeon et al., 2008). In yeast, for example, calcineurin is known to suppress the non-apoptotic death of cells treated with TM (Lotem et al., 1999; Bonilla et al., 2002; Dudgeon et al., 2008) and this is consistent with the fact that deletion of CNB1, a gene for the regulatory subunit of calcineurin is hypersensitive to TM (Fig. S4) as reported previously (Bonilla et al., 2002; Chen et al., 2005). On the contrary, deletion of CNB1 confers resistance to AbA. These data suggest that AbA-induced apoptosis is mediated by calcineurin activation. However, since calcineurin negatively regulates sphingolipid synthesis (Aronova et al., 2008), another potential explanation is that the resistance to AbA is due to a direct effect of the CNB1 deletion on sphingolipid synthesis. The fact that the cell death induced by a 4 h treatment with AbA was not prevented by concomitant addition of FK506 (data not shown) may support the latter possibility. Although further studies will be necessary to determine the site of action of calcineurin, the results presented here, together with other studies, imply that in yeast, stress-mediated cell deaths can be controlled by bimodal functions of the Ca2+-dependent pathway.
Cytochrome c, Yca1 and the downstream effector Rsm23 participate in AbA-induced loss of viability. This process is accompanied by mitochondrial fragmentation and ROS generation, and is dependent on mitochondrial AAC proteins, suggesting a mitochondrial-dependent apoptosis. Consistent with mitochondrial fragmentation, absence of Fis1, which inhibits mitochondrial fission and accelerates cell death (Kitagaki et al., 2007a; Pereira et al., 2008; Abdelwahid et al., 2011), conferred sensitivity to AbA (Fig. S5). Previous studies have reported that a mutation in cytochrome c rescues amiodarone- or pheromone-induced yeast programmed death but does not suppress ROS accumulation (Pozniakovsky et al., 2005). In the same manner, the arv1 null mutation- or AbA-induced ROS accumulation was not suppressed by deletion of CYC1 and CYC7 (Fig. S6A), suggesting that cytochrome c acts downstream of ROS generation (Fig. 7). Similarly, the deletion of YCA1 did not rescue the ROS accumulation. We also showed that the arv1Δ yca1Δ double mutant was still defective in complex sphingolipid synthesis (Fig. S6B). These results suggest that Yca1 acts downstream of calcium elevation and mitochondrial dysfunction to execute the cell death programme rather than on upstream events such as ER stress or sphingolipid biosynthesis (Lee et al., 2008). A functional link between cytochrome c release and metacaspase Yca1 activation in AbA-induced cell death and the order these events occur remain to be determined.
Finally, since recent studies have shown positive correlations between downregulated ceramide synthesis and increased cell death (Schulz et al., 2006; Menuz et al., 2009; Ponnusamy et al., 2010; Senkal et al., 2010), ceramide and/or complex sphingolipids have a pro-survival role, in addition to a pro-death role, in animal cells. It remains unknown, however, how the reduced levels of sphingolipids trigger apoptosis in animal cells. Our data in the more primitive S. cerevisiae may indicate that some of apoptotic cell death induced by imbalanced sphingolipid metabolism might occur via an evolutionarily conserved mechanism in which the ER stress-mediated and Ca2+-dependent pathway propagates apoptotic signals to the mitochondria. However, we cannot formally exclude the possibility that the real death signal comes from the Golgi or the plasma membrane as a consequence of absence of GPI anchors and of mature sphingolipids.
Strains and plasmids
The yeast strains and plasmids used in this study are listed in the Tables S1 and S2. Yeast cultures, genetic manipulations and strain construction were carried out as described previously (Kajiwara et al., 2008). S. cerevisiae wild-type strain BY4741 and BY4742, and BY4742-derived single-deletion KanMX strains were purchased from Open Biosystems (Huntsville, AL). RH401-7C, RH3802, RH6082, W303-1B, DL2828 and DL2829 yeast strains were the kind gifts from Dr H. Riezman, Dr Y. Jigami and Dr P. Orlean. Kar2-GFP-HDEL (pMO13), mRFP-Gas1 (pMF608) and mRFP-gas1* (pMF617) plasmids were provided by Dr Y. Jigami. The pmc1::lacZ (pKC190) plasmid provided by Dr K. Cunningham has been described (Cunningham and Fink, 1996).
Yeast cells were fixed with 2% glutaraldehyde, washed with water and post-fixed with 4% KMnO4 as described (Mazzoni et al., 2003). After washes with water, cells were dehydrated in increasing (50–100%) concentrations of ethanol, followed by ethanol-propylene oxide (1:1) and propylene oxide. Then the samples were infiltrated with mixtures of Epon 812 and propylene oxide 1:3 for 5 h, 1:1 for 9 h, 2:1 for 3 h and finally in pure Epon 812, and polymerized for 24 h at 60°C. Thin sections were cut, stained with 2% uranyl acetate for 20 min and 2% lead citrate for 4 min, and processed for electron microscopy.
TUNEL staining were performed as described previously (Kitagaki et al., 2007a). Cells were also stained with DAPI to quantify total number of cells, and observed with a fluorescence microscope (Olympus BX51, Tokyo, Japan). TUNEL-positive nuclei were counted blindly to avoid operator bias, and the percentages of TUNEL-positive per over 300 cell nuclei stained with DAPI in each sample were determined. For ROS detection, cells were incubated with H2DCFDA (Molecular Probes, Eugene, OR) for 1 h at 25°C as described (Madeo et al., 1999). The cells were then washed with water and viewed under fluorescence microscopy. Images were acquired with same exposure time and the fluorescence intensities of H2DCFDA stained cells were measured from individual cells by the Image J software. More than 30 cells were analysed in each experiment and intensities were averaged. Three independent experiments were performed and the average values and standard deviations of relative signal intensity were calculated. To study the morphology of the ER and the localization of Gas1, Kar2-GFP-HDEL and mRFP-Gas1 or mRFP-gas1* plasmids were transformed into yeast cells. The transformants were grown overnight at 25°C in synthetic minimal medium (SD), lacking uracil and leucine (Kajiwara et al., 2008), and incubated with or without drugs before being observed by fluorescence microscopy.
Drug sensitivity, temperature sensitivity and survival assay
For drug sensitivity assays, cells were grown overnight to log phase (OD600 = 0.5–1) at 25°C in SD medium. The next day cells at an initial concentration of OD600 = 1 were serially diluted in 1:5 in sterile water. Eight microlitres of each fivefold serial dilution was spotted onto SD plates containing drugs and incubated for 3 days at 25°C. Stock solutions of AbA and Myr were prepared at 10 mg ml−1 and 5 mg ml−1 in ethanol, respectively, diluted to the concentrations required for each experiments and added to media with a maximum volume of 10 μl per 9 cm dish (approximately 20 ml). Temperature sensitivity was determined by spotting diluted yeast cultures on SD plates, followed by incubating at 25°C and 37°C. Survival assays were performed as previously described (Kitagaki et al., 2007a). Yeast strains (OD600 = ∼ 0.5) that had been grown overnight in an enriched non-selective yeast growth medium (YPD) were diluted into fresh YPD medium to OD600 = 0.1 and incubated at 25°C in the presence or absence of drugs at the indicated concentrations and time periods. The diluted yeast cultures were spread onto YPD plates, and allowed to grow into colonies for 3–5 days at 25°C. The colonies were counted, and the data were expressed as percentage (mean ± standard error) of colony-forming unit of untreated cells.
Labelling of yeast sphingolipids with [3H]DHS (American Radiolabelled Chemical) was performed as described (Kajiwara et al., 2008). Cells grown overnight in SD medium at 25°C were resuspended in SD medium, and labelled with 4 μCi of [3H]DHS at 25°C for the indicated times. The reaction was stopped by placing the mixture on ice. The cells were washed with cold water and subjected to lipid extraction. The lipids labelled with [3H]DHS were subjected to mild alkaline hydrolysis to deacylate glycophospholipids as described (Kajiwara et al., 2008). The pooled organic phases were desalted by partitioning with n-butanol and lipids were dried under nitrogen. Radiolabelled lipids were analysed by TLC using solvent system I, chloroform-methanol-4.2 N ammonium hydroxide (9:7:2, v/v), and were quantified by FLA-7000 system (Fujifilm). For ceramide analysis, the lipids labelled with [3H]DHS were first separated on TLC plates as above. Subsequently, fractions containing ceramides, which are approximately 2–3 cm from the top of the TLC plates (Funato and Riezman, 2001), were collected by scraping and eluting with chloroform-methanol (1:1, v/v), and were analysed by TLC using solvent system II, chloroform-methanol-acetc acid (190:9:1, v/v), as described previously (Haak et al., 1997; Cremesti and Fischl, 2000).
Western blot analysis
Yeast cells were grown overnight in YPD, harvested and incubated with the indicated concentrations of drugs for 4 h. Proteins were extracted from cells, separated on SDS-PAGE and analysed by Western blotting as described previously (Schonbachler et al., 1995). Blots were probed with rabbit polyclonal antibodies against Gas1, CPY, Wbp1 (gifts from Dr Riezman) and a peroxidase-conjugated affinity-purified anti-rabbit IgG antibody (Sigma, St Louis, MO). For analysis of Kar2 protein, membranes were prepared as described (Schonbachler et al., 1995) and Kar2 was detected using a rabbit polyclonal antibody against Kar2 (y-115) (Santa Cruz Biotechnology, Santa Cruz, CA).
Determination of [Ca2+]c levels
Yeast cells carrying the pmc::lacZ reporter gene plasmid (pKC190) were grown to log phase overnight at 25°C in SD lacking uracil, and treated with 2.0 μg ml−1 FK506 and the indicated concentrations of drugs for 4 h at 25°C. The cells were then harvested and assayed for β-galactosidase activity (Miller units) as described previously (Cunningham and Fink, 1996).
We thank Dr Howard Riezman (University of Geneva) for the RH401-7C, RH3802 and RH6082 strains and antibodies against Gas1, CPY and Wbp1; Dr Peter Orlean (University of Illinois) for the DL2828 and DL2829 strains; Dr Yoshifumi Jigami (National Institute of Advanced Industrial Science and Technology) for the W303-1B strain and pMO13, pMF608 and pMF617 plasmids; Dr Kyle W. Cunningham (Johns Hopkins University) for the pKC190 plasmid; Dr Janet M. Shaw (University of Utah) for the p413GPD-mtGFP plasmid. This work was supported by a grant from the Graduate School of Biosphere Science (Hiroshima University) to K.K and by Grants-in-Aid for Scientific research from the Japan Society for the Promotion of Science and from the Ministry of Education, Culture, Sports, Science, and Technology of Japan to K.F.