ABSTRACT
Cells synthesize ceramides in the endoplasmic reticulum (ER) as precursors for sphingolipids to form an impermeable plasma membrane. As ceramides are engaged in apoptotic pathways, cells would need to monitor their levels closely to avoid killing themselves during sphingolipid biosynthesis. How this is accomplished remains to be established. Here we identify SMSr (SAMD8), an ER-resident ceramide phosphoethanolamine (CPE) synthase, as a suppressor of ceramide-mediated cell death. Disruption of SMSr catalytic activity causes a rise in ER ceramides and their mislocalization to mitochondria, triggering a mitochondrial pathway of apoptosis. Blocking de novo ceramide synthesis, stimulating ceramide export from the ER or targeting a bacterial ceramidase to mitochondria rescues SMSr-deficient cells from apoptosis. We also show that SMSr-catalyzed CPE production, although essential, is not sufficient to suppress ceramide-induced cell death and that SMSr-mediated ceramide homeostasis requires the N-terminal sterile α-motif, or SAM domain, of the enzyme. These results define ER ceramides as bona fide transducers of mitochondrial apoptosis and indicate a primary role of SMSr in monitoring ER ceramide levels to prevent inappropriate cell death during sphingolipid biosynthesis.
INTRODUCTION
Sphingolipids are vital components of cellular membranes that, besides providing mechanical stability, contribute to molecular sorting, signaling and cell-cell recognition (Holthuis et al., 2001; Lopez and Schnaar, 2009; Lingwood and Simons, 2010). In addition, several intermediates of sphingolipid metabolism act as signaling molecules in key physiological and pathological processes (Alvarez et al., 2007; Hannun and Obeid, 2008). Notably, ceramides are obligate precursors for sphingolipid biosynthesis but are also implicated as mediators of stress responses, cell cycle arrest and apoptosis (Merrill, 2002; Taha et al., 2006; Hannun and Obeid, 2008). This implies that cells must control sphingolipid biosynthesis precisely to avoid jeopardizing their viability. Although recent studies provided important new insights into the mechanisms of sphingolipid homeostasis (Aronova et al., 2008; Tomishige et al., 2009; Breslow et al., 2010; Berchtold et al., 2012), how cells monitor toxic intermediates such as ceramides is poorly understood.
The involvement of ceramides in apoptosis appears clinically significant. Several anti-cancer regimens including irradiation (Deng et al., 2008), oxidative stress (Dolgachev et al., 2004) and chemotherapeutic agents (Bose et al., 1995; Chalfant et al., 2002; Perry et al., 2000) cause an increase in endogenous ceramide levels through de novo synthesis, leading to G0/G1 arrest and apoptosis, and genetic or pharmacological inhibition of ceramide production can result in apoptotic resistance (Liu et al., 2004; Mesicek et al., 2010). Consequently, some of the enzymes involved in ceramide metabolism have been investigated as potential therapeutic targets (Reynolds et al., 2004). The molecular mechanisms whereby ceramides initiate and execute apoptosis are not well understood. In vitro studies suggest that ceramides (or ceramide-derived metabolites) can act directly on isolated mitochondria to promote outer membrane permeabilization and cell death (Siskind et al., 2002; Stiban et al., 2008; Lee et al., 2011; Chipuk et al., 2012). However, the origin of mitochondrial ceramides and their actual contribution to the apoptotic response in living cells are topics of controversy (Tepper et al., 1999; Deng et al., 2008; Wang et al., 2009).
Ceramides are synthesized de novo by N-acylation of sphingoid long-chain bases on the cytosolic surface of the ER (Merrill, 2002). The bulk of newly synthesized ceramides is converted to sphingomyelin (SM) in the lumen of the Golgi (Tafesse et al., 2007). Efficient delivery of ER ceramides to the site of SM production requires the cytosolic ceramide transfer protein CERT (Hanada et al., 2003). SM production is mediated by SM synthase (SMS). Mammalian cells contain two SMS isoforms, namely SMS1 in the Golgi and SMS2 at the plasma membrane (Huitema et al., 2004; Yamaoka et al., 2004), and a closely related enzyme, sphingomyelin synthase-related protein (SMSr; also known as sterile α-motif domain-containing protein; SAMD8), resides in the ER (Vacaru et al., 2009). SMSr/SAMD8 is not a conventional SM synthase but instead synthesizes trace amounts of the SM analogue ceramide phosphoethanolamine (CPE) (Vacaru et al., 2009; Ternes et al., 2009). SMSr homologues are also found in organisms such as Drosophila, which lacks SM and synthesizes CPE as the principal sphingolipid. Although SMSr from Drosophila possesses CPE synthase activity (Vacaru et al., 2009), bulk production of CPE in this organism is mediated by a different enzyme that belongs to an insect-specific branch of the CDP-alcohol phosphotransferase superfamily (Vacaru et al., 2013). Unexpectedly, blocking SMSr catalytic activity in mammalian and insect cells causes a substantial rise in ER ceramides, leading to a structural collapse of the early secretory pathway, including Golgi fragmentation (Vacaru et al., 2009). The molecular basis of these events is unclear.
The ER is a highly efficient distribution system that is intimately associated with other organelles, including mitochondria (Holthuis and Levine, 2005). Here we show that loss of SMSr triggers mitochondrial apoptosis by causing a deregulation of ER ceramides and their mistargeting to mitochondria. Moreover, we find that SMSr catalytic activity, although required, is not sufficient to suppress ceramide-induced cell death and that the regulation of ER ceramides by SMSr is dependent on the N-terminal sterile α-motif (SAM) domain of the enzyme. Together, these results suggest that SMSr has a primary role in monitoring ER ceramide levels to protect cells against the potential risk of killing themselves during sphingolipid biosynthesis.
RESULTS
Loss of SMSr triggers apoptosis
To investigate the potential engagement of ER ceramides in apoptotic pathways and gain further insight into SMSr function, we analyzed SMSr-depleted human cervical carcinoma HeLa cells for signs of apoptosis. After 3 days of treatment with either one of two independent small interfering RNAs (siRNAs) targeting SMSr (siSMSr), HeLa cells displayed membrane blebs and condensed nuclei, morphological hallmarks of apoptosis (supplementary material Fig. S1A,B). This was accompanied by proteolytic activation of caspase-3 (Casp3) and caspase-mediated cleavage of poly ADP-ribose polymerase (PARP; supplementary material Fig. S1C; Fig. 1A–D). Cleavage of Casp3 and PARP also occurred in various other siSMSr-treated carcinoma cell lines (see Fig. 1E), but not in mock or nonsense siRNA (siNS)-treated cells. Transfection of an RNAi-resistant SMSr construct (SMSr-V5-res) fully suppressed PARP cleavage in siSMSr-treated cells (Fig. 2A,B), indicating that induction of apoptosis is attributable to silencing of endogenous SMSr and not to an off-target effect.
Loss of SMSr triggers apoptosis. (A) HeLa cells, treated for 3 days with siNS or siSMSr, were imaged by Normarski or fluorescence microscopy after immunostaining for cleaved Casp3 or cleaved PARP (red) and counterstaining for DNA (DAPI; blue). Scale bars: 25 µm. (B) Quantification of HeLa cells immuno-positive for cleaved Casp3 or cleaved PARP after treatment with siRNA as in A or for 6 hours with the apoptosis-inducing agent staurosporine (Stauro; 1 µg/ml). Values are means ± s.d. (n = 3). (C) Immunoblots of HeLa cells treated as in B and stained for cleaved Casp3 and β-actin. (D) Immunoblots of HeLa cells treated for 3 days with siNS, siSMSr, siRNA targeting SMS1 (siSMS1) or siRNA targeting SMS2 (siSMS2), or for 6 hours with staurosporine, and stained for PARP, Golgin-160 and β-actin. (E) Immunoblots of ovarian carcinoma (OVCAR, IGROV and SKOV) cells treated as in B and stained for PARP. FL, full length; CL, cleaved.
Apoptosis in SMSr-depleted cells is due to ceramide accumulation in the ER. (A) HeLa cells treated with siSMSr were incubated with fumonisin B1 (FB1; 100 µM) or transfected with empty vector (EV), siRNA-resistant SMSr (SMSr-V5-res) or enzyme-dead SMSr (SMSrD348E-V5-res) and immunostained for cleaved PARP (red) and V5 (green). Counterstaining was with DAPI (blue). The white box in the bottom right image shows cells from a different field. Scale bar: 15 µm. (B) Quantification of HeLa cells immuno-positive for cleaved PARP after treatment with siNS or siSMSr and transfection with EV, SMSr-V5-res, SMSrD348E-V5-res, ER-resident SMS1 (SMS1ER-V5) or CERT (CERT–FLAG). Values are means ± s.d. (n = 3). *P<0.001 by two-tailed unpaired Student's t-test. (C) Quantification of HeLa cells immuno-positive for cleaved PARP after treatment with siNS or siSMSr and FB1 (25 and 100 µM) or myriocin (30 µM). Values are means ± s.d. (n = 3). *P<0.001 by two-tailed unpaired Student's t-test.
Loss of SMSr also triggered cleavage of Golgin-160 (Fig. 1D), a caspase substrate whose proteolytic conversion promotes Golgi fragmentation during apoptosis (Mancini et al., 2000). This suggests that the structural collapse of the Golgi complex observed previously in SMSr-depleted cells (Vacaru et al., 2009) is secondary to activation of an apoptotic pathway. Depletion of SMS1 also induced cleavage of PARP and Golgin-160 (golgin subfamily A member 3; GOLGA3), but to a lesser extent than depletion of SMSr (Fig. 1E). In contrast, depletion of SMS2 did not result in any cleavage of these caspase substrates. This is remarkable, as HeLa cells depleted for SMS1, SMS2 or SMSr accumulate ceramides to a similar extent (Tafesse et al., 2007; Vacaru et al., 2009). Together, these results suggest that ceramide accumulation per se is not sufficient to trigger apoptosis, and that the site of ceramide accumulation might be important.
Apoptosis in SMSr-depleted cells is due to ceramide accumulation in the ER
We previously showed that blocking SMSr-mediated CPE production by mutation of active site residue Asp348 to Glu is sufficient to trigger ceramide accumulation in the ER (Vacaru et al., 2009). As shown in Fig. 2A,B, the ability of siRNA-resistant SMSr to suppress PARP cleavage in siSMSr-treated cells was completely abolished by mutation of Asp348 to Glu, supporting the notion that apoptosis is due to a rise in ER ceramides. Alternatively, CPE produced by SMSr might be essential for suppressing apoptosis. To distinguish between these possibilities, siSMSr-treated cells were transfected with an ER-resident and enzymatically active form of SMS1 (SMS1ER) (Vacaru et al., 2009). Expression of SMS1ER blocked cleavage of PARP (Fig. 2B), indicating that SMSr-mediated prevention of cell death is not dependent on production of CPE. Because SMS1ER activity would consume ceramides in the ER, we next asked whether inhibiting ceramide biosynthesis or stimulating ceramide export from the ER would also suppress apoptosis. Indeed, treatment with long chain base synthase inhibitor myriocin or ceramide synthase inhibitor fumonisin B1 (FB1) or overexpression of CERT blocked PARP cleavage in siSMSr-treated cells (Fig. 2A–C). These results indicate that apoptosis in SMSr-depleted cells is due to ceramide accumulation in the ER. In line with this notion, we found that knockdown of CERT also triggers PARP cleavage and that this process is partially suppressed by blocking de novo ceramide synthesis with myriocin (supplementary material Fig. S2).
Apoptosis in SMSr-depleted cells is independent of ER stress
A rise in ER ceramide levels has been proposed to sensitize cells to ER stress (Swanton et al., 2007; Wang et al., 2009). As chronic ER stress can initiate apoptosis (Szegezdi et al., 2006), we next examined whether apoptosis in SMSr-depleted cells is triggered by ER stress. The onset of ER stress is detected by ER transmembrane receptors IRE1, ATF6 and PERK that initiate the unfolded protein response (UPR) to restore normal ER function (Ron and Walter, 2007). Activated IRE1 catalyzes splicing of X-box-binding protein 1 (XBP1) mRNA, which leads to transcriptional activation of the ER stress markers BiP and calnexin, whereas activation of ATF6 or PERK induces expression of the DNA-damage-inducible gene 153 GADD153/CHOP (Ron and Walter, 2007). Using RT-PCR, we found no evidence of XBP1 mRNA splicing in siSMSr-treated cells (Fig. 3A). Additionally, immunoblot analysis showed that these cells lacked detectable amounts of GADD153/CHOP and contained similar levels of BiP and calnexin as mock or siNS-treated cells (Fig. 3B). Conversely, XBP1 splicing and GADD153/CHOP expression were readily detectable in cells treated with tunicamycin (Tm; Fig. 3A,B), a potent inducer of ER stress. Hence, loss of SMSr does not activate any of the three known branches of the UPR. This indicates that apoptosis in SMSr-depleted cells occurs independently of ER stress.
SMSr-depleted cells are devoid of ER stress. (A) Detection of XBP1 mRNA splicing by RT-PCR in HeLa cells treated with siNS or siSMSr for 3 days, or with tunicamycin (Tm; 2 µg/ml) for the indicated times. RT-PCR of GAPDH mRNA served as controls. XBP1u, unspliced; XBP1s, spliced. (B) Immunoblots of HeLa cells treated as in A were stained for calnexin, Grp94, BiP, GADD153 (CHOP) and β-actin.
Loss of SMSr triggers a mitochondrial pathway of apoptosis
Ceramides have been proposed to induce apoptosis by increasing the permeability of the outer mitochondrial membrane to apoptogenic intermembrane proteins such as cytochrome c (Siskind et al., 2002; Stiban et al., 2008; Lee et al., 2011). The origin of mitochondrial ceramides is unclear, but might involve transfer of newly synthesized ceramides from the ER to mitochondria (Deng et al., 2008; Stiban et al., 2008). To investigate how deregulation of ER ceramides can trigger cell death, we first asked whether loss of SMSr activates a mitochondrial pathway of apoptosis. Dissipation of the mitochondrial membrane potential (ΔΨm) is an event associated with mitochondrial apoptosis that can be monitored by labeling cells with the potentiometric dye JC-1 (Simeonova et al., 2004). Cells treated with staurosporine, a well-known inducer of mitochondrial apoptosis, showed a more diffuse JC-1 staining with a shift from red to green fluorescence when compared with mock or siNS-treated cells (Fig. 4A,B). A similar shift in JC-1 fluorescence was observed in cells treated with siSMSr, indicating that loss of SMSr disrupts ΔΨm. This was accompanied by translocation of cytochrome c from mitochondria to the cytosol, as evidenced by immunofluorescence microscopy (Fig. 4C) and immunoblot analysis of cytosolic fractions (Fig. 4D). Release of cytochrome c into the cytosol promotes activation of caspase-9 (Casp9), an initiator caspase associated with mitochondrial apoptosis (Li et al., 1997). Indeed, siSMSr-treated cells displayed proteolytic activation of Casp9 and a nearly fourfold induction of Casp9 activity relative to mock- or siNS-treated cells (Fig. 4E,F). Finally, overexpression of Bcl-2, a negative regulator of mitochondrial apoptosis (Zhai et al., 2008), fully suppressed PARP cleavage in siSMSr-treated cells (Fig. 4G). Together, these data show that apoptosis in SMSr-depleted cells is mediated by a mitochondrial pathway.
Loss of SMSr triggers a mitochondrial pathway of apoptosis. (A) Staurosporine (Stauro) and siSMSr-treated HeLa cells incubated with the potentiometric dye JC-1 show a more diffuse and green fluorescence relative to siNS-treated control cells, indicative of a disrupted mitochondrial membrane potential. Scale bar: 15 µm. (B) Flow cytometric quantification of staurosporine (Stauro) and siRNA-treated HeLa cells showing green-shifted JC-1 fluorescence. Values are means ± s.d. (n = 3). FL1-H, green fluorescence; FL2-H, red fluorescence. (C) HeLa cells treated with staurosporine or siRNA were immunostained for cytochrome c (red) and counterstained with DAPI (blue). Scale bar: 15 µm. (D) Immunoblots of cytosol and mitochondrial pellets (mito) of HeLa cells treated with staurosporine (Stauro) or siRNA were stained for cytochrome c, mitochondrial marker GRP75 and β-actin. (E) HeLa cells treated as in (D) were lysed and analyzed for Casp9 activity by colorimetric assay. Values are means ± s.d. (n = 3). (F) Immunoblots of HeLa cells treated as in D were stained for Casp9. FL, full length; CL, cleaved. (G) Quantification of HeLa cells immuno-positive for cleaved PARP after treatment with siNS or siSMSr and transfection with EV or Bcl-2. Values are means ± s.d. (n = 3).
Mitochondria of SMSr-depleted cells accumulate ceramides
To investigate whether ceramides that accumulate in the ER of SMSr-depleted cells can reach mitochondria, we analyzed the ceramide content of ER-deficient mitochondrial pellets (Pmito) from control and siSMSr-treated cells using mass spectrometry. Quantification of ceramide levels in ER-enriched microsomal pellets (Pmicro) served as a reference. As shown in Fig. 5B, mitochondrial pellets of SMSr-depleted cells contained threefold higher ceramide levels than controls (2.0±0.3 versus 0.7±0.1 mol%). In contrast, SMSr depletion had no effect on the levels of cardiolipin (CL), a phospholipid found almost exclusively in the inner membrane of mitochondria (8.0±0.4 versus 8.2±1.1 mol%). Consistent with our previous findings (Vacaru et al., 2009), mitochondrion-deficient microsomal pellets of SMSr-depleted cells also contained higher ceramide levels than controls (1.1±0.2 versus 0.4±0.1 mol%). Immunoblot analysis showed that mitochondrial pellets of control and SMSr-depleted cells in each case contained at best only a very minor fraction of ER (below 3%; Fig. 5A). As the ER and mitochondria represent ∼50–60% and 20–40% of total membranes in mammalian cells, respectively (Alberts et al., 2008), it appears unlikely that the accumulation of ceramides in mitochondrial pellets of SMSr-deficient cells is solely due to contamination with ER. Interestingly, the increase in mitochondrion-associated ceramides concerned all major species (Table 1), mirroring those that accumulate in the ER of SMSr-depleted cells (Vacaru et al., 2009). Collectively, our results suggest that ceramides that accumulate in mitochondria of SMSr-depleted cells originate from the ER.
Apoptosis in SMSr-depleted cells is due to mislocalization of ER ceramides to mitochondria. (A) Postnuclear supernatants (PNS), mitochondrial pellets (Pmito) and microsomal pellets (Pmicro) from siNS and siSMSr-treated HeLa cells were immunoblotted and stained for GRP75 (mitochondria) and calnexin (ER). (B) Mass spectrometric analysis of ceramide (Cer) and cardiolipin (CL) levels in mitochondrial and microsomal pellets from siNS and siSMSr-treated HeLa cells. Values are means ± range (n = 2). (C) HeLa cells transfected with cytosolic or mitochondrially targeted bacterial ceramidase (bCDase-myc or Mito-bCDase-myc) and immunostained for myc (green) and GRP75 (red). Scale bar: 15 µm. (D) Mass spectrometric analysis of ceramide (Cer) levels in mitochondrial pellets from sins- or siSMSr-treated HeLa cells transfected with empty vector (EV) or mitochondrially targeted bacterial ceramidase (Mito-bCDase-myc). Values are means ± range (n = 2). (E) Quantification of HeLa cells immuno-positive for cleaved PARP after treatment with siNS or siSMSr and transfection with EV, bCDase-myc, Mito-bCDase-myc or enzyme-dead Mito-bCDaseH96/98A-myc. Values are means ± s.d. (n = 3). *P<0.001 by two-tailed unpaired Student's t-test.
Targeting ceramidase to mitochondria blocks apoptosis in SMSr-depleted cells
To address whether the arrival of ER ceramides in mitochondria is necessary for induction of apoptosis, we next investigated whether targeting a ceramide-consuming enzyme to mitochondria would prevent PARP cleavage in SMSr-depleted cells. To this end, the mitochondrial signal peptide of human cytochrome c oxidase subunit VIII was fused to the N-terminus of a soluble Mycobacterium-derived ceramidase, bCDase (Okino et al., 1999), and a myc epitope was added to its C-terminus to facilitate detection. bCDase-myc expressed in HeLa cells had a cytosolic distribution. In contrast, the enzyme carrying the mitochondrial signal peptide, Mito-bCDase-myc, localized exclusively to mitochondria (Fig. 5C). Expression of Mito-bCDase-myc normalized mitochondria-associated ceramide levels (Fig. 5D) and effectively blocked PARP cleavage in SMSr-depleted cells (Fig. 5E). This rescuing effect was abolished by removal of the mitochondrial signal peptide or mutation of two invariant histidine residues in the active site (His96/98) of the enzyme (Inoue et al., 2009), indicating that a catalytically active and mitochondrion-associated form of CDase is required to prevent apoptosis. Thus, transfer of ER ceramides to mitochondria appears essential for committing cells to death when SMSr is lost.
SMSr-mediated CPE production is required but not sufficient to suppress ceramide-induced apoptosis
Since SMSr produces only trace amounts of CPE (<0.05 mol% of total cellular lipids), the amount of ceramides accumulating in SMSr-depleted cells (∼1 mol%) cannot be ascribed to a block in ceramide consumption for CPE production (Vacaru et al., 2009). This enigma could be explained if SMSr-derived CPE directly influences the activity of enzymes responsible for ceramide biosynthesis or turnover in the ER. Our finding that mutation of active site residue Asp348 in SMSr is sufficient to deregulate ER ceramides (Vacaru et al., 2009) and trigger apoptosis (Fig. 2A,B) is consistent with this notion. Besides an active site comprising a catalytic triad of histidine and aspartate residues, SMSr carries a sterile α-motif, or SAM, domain at its N-terminus (residues 12–78; Fig. 6A). SAM domains mediate protein–protein interactions through formation of homo- or heterotypic oligomers (Meruelo and Bowie, 2009). A global study of human SAM domains produced in Escherichia coli revealed that SMSr-SAM forms polymers (Knight et al., 2011). To determine whether SMSr requires its SAM domain to prevent ceramide-induced cell death, we created a SAM-deficient SMSr mutant and analyzed its ability to suppress ceramide accumulation and caspase-mediated cleavage of PARP. Removal of the SAM causes a redistribution of SMSr from the ER to the Golgi (Fig. 6B) (Vacaru et al., 2009). To retain SMSrΔSAM in the ER, a KKxx motif (KKAS) was added to its C-terminus (Fig. 6B). Contrary to mutation of active site residue Asp348, addition of the KKxx motif or removal of the SAM did not affect the ability of the enzyme to synthesize CPE (Fig. 6C). Transfection of siRNA-resistant SMSr-KKAS in siSMSr-treated cells effectively blocked ceramide accumulation (Fig. 6D) and PARP cleavage (Fig. 6E). As expected, mutation of active site residue Asp348 in SMSr was sufficient to deregulate ER ceramides and trigger apoptosis. Importantly, the ability of SMSr to prevent ceramide accumulation and cell death was also completely abolished by removal of its SAM domain (Fig. 6D,E). Together, these results demonstrate that SMSr-catalyzed CPE production, although essential, is not sufficient to suppress ceramide-induced apoptosis and that regulation of ER ceramides by SMSr requires the N-terminal SAM domain of the enzyme.
SMSr catalytic activity is essential but not sufficient to suppress apoptosis. (A) Model of SMSr with SAM domain (white), membrane spans (gray) and active-site residues (red) are indicated. (B) HeLa cells transfected with SMSr-V5, SMSrΔSAM-V5 or SMSrD348E-V5, or with these constructs containing a C-terminal ER-retention signal (KKAS) were immunostained for the V5 epitope (green) and the Golgi marker GM130 (red). Scale bar: 5 µm. (C) TLC analysis of the reaction products formed when NBD-ceramide (NBD-Cer) was incubated with lysates of HeLa cells transfected with EV, SMSr-V5-KKAS, SMSrD348E-V5-KKAS or SMSrΔSAM-V5-KKAS. (D) HeLa cells treated with siNS or siSMSr and transfected with EV or SMSr constructs as in C were labeled for 5 hours with [14C]serine. Levels of radiolabeled ceramides were quantified by TLC and autoradiography and expressed as the percentage of the control (siNS-treated). Values are means ± range (n = 2). (E) Quantification of HeLa cells immuno-positive for cleaved PARP after treatment with siNS or siSMSr and transfection with EV or SMSr constructs as in C. Values are means ± s.d. (n = 3); *P<0.001 by two-tailed unpaired Student's t-test.
DISCUSSION
The essential but potentially lethal nature of ceramides would suggest that cells monitor ceramide levels closely to avoid jeopardizing their viability. The present study identifies SMSr, an ER-resident ceramide phosphoethanolamine synthase, as a suppressor of ceramide-induced cell death. Disruption of SMSr function causes a rise in ER ceramides and their flow into mitochondria, triggering a mitochondrial pathway of apoptosis. Any approach used to normalize the ceramide concentration in the ER proved effective in preventing apoptosis when SMSr is lost, including: (1) a block in de novo ceramide production through inhibition of long chain base synthase or ceramide synthase; (2) consumption of excess ceramides by a SM synthase targeted to the ER; and (3) stimulation of ceramide export from the ER by overexpression of CERT. Conversely, knockdown of CERT triggers an apoptotic response that is dependent on de novo ceramide production. These findings firmly establish ER ceramides as authentic transducers of mitochondrial apoptosis and suggest that SMSr mediates ceramide homeostasis in the ER to avoid inappropriate cell death during sphingolipid biosynthesis.
We previously reported that loss of SMSr is accompanied by a structural collapse of the Golgi complex (Vacaru et al., 2009). Here we show that SMSr depletion results in cleavage of Golgin-160, a caspase substrate whose proteolytic conversion promotes Golgi fragmentation during apoptosis (Mancini et al., 2000). Moreover, we found that the Golgi fragmentation phenotype can be suppressed by the same treatments that rescue the apoptotic phenotype, i.e. the presence of enzymatically active SMSr, overexpression of CERT, targeting SM synthase to the ER, or blocking ceramide biosynthesis by drugs (Vacaru et al., 2009). This suggests that Golgi fragmentation in SMSr-depleted cells is secondary to the initiation of apoptosis.
A deregulation of ER ceramides has previously been implicated in sensitizing cells to ER stress (Wang et al., 2009) and the induction of GADD153/CHOP (Swanton et al., 2007), an apoptotic regulator of the unfolded protein response (UPR). However, removal of SMSr did not activate any of the three known branches of the UPR, at least not within the time frame required to initiate apoptosis. This shows that ceramide-induced apoptosis in SMSr-depleted cells occurs independently of ER stress. Instead, we found that SMSr-depleted cells accumulate ceramides in mitochondria, that the molecular species of ceramides accumulating in mitochondria correspond to those that accumulate in the ER, and that targeting a bacterial ceramidase to mitochondria blocked apoptosis. Collectively, these results suggest that translocation of ER ceramides to mitochondria is essential for committing cells to death when SMSr is lost. Although the mechanism by which ER ceramides induce mitochondrial apoptosis remain to be elucidated, our findings complement previous in vitro studies indicating that ceramides can act directly on isolated mitochondria to promote outer membrane permeabilization and cell death (Siskind et al., 2002; Stiban et al., 2008; Lee et al., 2011, Chipuk et al., 2012). Owing to their propensity to modulate membrane structure, ceramides or ceramide-derived metabolites might serve as crucial effectors of the mitochondrial protein release process that initiates the execution phase of apoptosis. Although ceramides are virtually insoluble in water, our data provide evidence for the existence of a mechanism that facilitates their transfer from ER to mitochondria. This transfer could require intimate membrane contact between these organelles (Kornmann et al., 2009), a cytosolic transfer protein acting at the ER–mitochondrial interface, or a combination of both.
Ceramides are routinely synthesized on the cytosolic surface of the ER as precursors for sphingolipids to form an impermeable plasma membrane. So what keeps these ceramides from flowing into mitochondria to initiate inappropriate apoptosis? By capturing newly synthesized ceramides from the cytosolic surface of the ER for delivery to the site of SM production in the Golgi, CERT might constitute a first line of defense against ceramide-induced cell death. This idea finds support in the observation that overexpression of CERT rescues SMSr-depleted cells from apoptosis, whereas CERT knockdown triggers apoptosis (this study). Moreover, a genome-wide search for genes influencing the sensitivity of cancer cells to chemotherapeutic agents identified CERT as a suppressor of taxane-induced cell death (Swanton et al., 2007). We postulate that SMSr complements CERT function and provides an additional line of defense against accidental cell death by exerting tight control over ceramide levels in the ER. Thus, SMSr-mediated prevention of undesired fluctuations in ER ceramides would allow cells to synthesize the sphingolipids they need without compromising their viability.
How does SMSr control ceramide levels in the ER? The substantial rise in ER ceramides observed upon downregulation of SMSr is difficult to account for by loss of the rather modest CPE-producing activity of the enzyme (Vacaru et al., 2009). We now show that SMSr-catalyzed CPE production, although required, is not sufficient to suppress ceramide accumulation and cell death. Importantly, this finding rules out metabolic conversion of ceramides as the primary mechanism by which SMSr controls ER ceramide levels and also argues against CPE being a direct modulator of ceramide-metabolizing enzymes in the ER. Instead, we find that SMSr-mediated ceramide homeostasis requires the N-terminal SAM domain of the enzyme. Although removal of the SAM does not affect SMSr-catalyzed CPE production, it abolishes the ability of the enzyme to suppress ceramide accumulation and cell death. Preliminary work by us and others suggests that SMSr self-associates through its SAM domain to form higher-order complexes (Knight et al., 2011) (F.G.T. and J.C.M.H., unpublished data). Dynamic polymerization of SAM-containing proteins has been recognized as an important mechanism in the regulation of signal transduction (Harada et al., 2008). We propose that SMSr is a ceramide sensor that requires its SAM domain to modulate ceramide-metabolizing enzymes or transport machinery in response to fluctuations in ER ceramides. Because its active site faces the ER lumen (Huitema et al., 2004; Vacaru et al., 2009), SMSr would only ‘sense’ ceramides that slipped through the first line of defense, i.e. molecules that escaped CERT-mediated extraction from the cytosolic surface of the ER and that flipped to the luminal side. We envision that during CPE production, SMSr reaches a conformation that transduces a signal, conceivably mediated through its SAM domain, to inhibit ceramide biosynthesis, stimulate ceramide turnover and/or accelerate ceramide export from the ER.
Our recent finding that loss of SMSr has no effect on the rate of ceramide biosynthesis in isolated microsomes suggests that SMSr is not a direct regulator of ceramide synthases (P. Krumpochova and J.C.M.H., unpublished data). SMSr also does not seem to participate in the feedback regulation of ceramide biosynthesis by ORMDL proteins (Siow and Wattenberg, 2012), which are negative homeostatic regulators of long chain base synthase in mammalian cells (Breslow et al., 2010). This suggests that the mechanism by which SMSr controls ceramide levels in the ER occurs downstream of ceramide biosynthesis. Whether this mechanism involves ER-resident ceramidases (Mao et al., 2003), conventional sphingomyelin synthases or proteins that control ceramide export from the ER (e.g. CERT) is subject to ongoing investigations.
In conclusion, our study defines SMSr as a critical component of the mechanism by which cells monitor ceramide levels in the ER to protect themselves against the inherent danger of sphingolipid biosynthesis. SMSr inactivation causes a deregulation of ER ceramides, which, in turn, mislocalize to mitochondria to induce mitochondrial apoptosis. In view of the emerging involvement of ER ceramides in sensitizing cancer cells to stress-induced apoptosis, interfering with SMSr activity might provide a novel therapeutic modality to promote cell death in tumors.
MATERIALS AND METHODS
Chemicals
Propidium iodide (PI), staurosporine, myriocin and tunicamycin were obtained from Sigma-Aldrich; z-VAD-fmk was from Calbiochem; fumonisin B1 was from Cayman Chemicals; and L-[3-14C]serine was from GE Healthcare. C6-NBD-labeled ceramide (NBD-Cer) and 2-oleoyl-1-palmitoyl-sn-glycerol-3-phosphoethanolamine (POPE) were obtained from Avanti Polar Lipids, Inc.
Antibodies
The following antibodies were used: mouse monoclonal anti-β-actin and mouse monoclonal anti-FLAG (Sigma-Aldrich); mouse monoclonal anti-V5 (Invitrogen); rabbit polyclonal anti-calnexin, rabbit polyclonal anti-GRP75, mouse monoclonal anti-KDEL, mouse monoclonal anti-GADD153, mouse monoclonal anti-myc and mouse monoclonal anti-PARP (Santa Cruz Biotechnology, Inc.); rabbit polyclonal anti-caspase-9 and rabbit monoclonal anti-cleaved caspase-3 (Cell Signaling); rabbit polyclonal anti-cytochrome c (Clontech); mouse monoclonal anti-GM130 (BD); rabbit polyclonal anti-PARPp85 Fragment (Promega); rabbit polyclonal anti-Golgin-160 (kindly provided by Carolyn Machamer, John Hopkins University School of Medicine). Horseradish-peroxidase-conjugated secondary antibodies were obtained from PerBio and secondary antibodies conjugated to Alexa Fluor dyes were obtained from Jackson Immunoresearch Laboratories.
DNA constructs
Human SMSr and SMS1 cDNAs were cloned into pcDNA3.1/V5-His-TOPO (Invitrogen) as described previously (Huitema et al., 2004). A siRNA-resistant form of SMSr was created by introducing five silent point mutations in the siRNA target sequence. All point mutations, including those creating catalytically inactive SMSr (D348E), were generated by the megaprimer PCR method (Angeliccio and Bonaccorsi di Patti, 2002). The SAM-deficient SMSr truncation mutant, SMSr-ΔSAM-V5-KKAS, was obtained by deleting the first 68 N-terminal residues of SMSr and adding the ER retention sequence KKSA at its C-terminus. The construct encoding an ER-resident form of human SMS1-V5, SMS1ER-V5, was described by Vacaru et al. (Vacaru et al., 2009). The cDNA of bacterial ceramidase (bCDase) from Mycobacterium tuberculosis (kindly provided by Makoto Ito, Kyushu University, Japan) was PCR amplified and cloned into pCMV/myc/cyto (Invitrogen). To target bCDase to mitochondria, the cDNA was cloned down stream of the mitochondrial signal sequence of cytochrome c oxidase subunit VIII (amino acid residues 1–30) into pCMV/myc/mito (Invitrogen). An enzyme-dead mutant of bCDase, bCDaseH96A/H98A, was created by the megaprimer PCR method. Myc-tagged human Bcl-2 and FLAG-tagged human CERT were kindly provided by John Reed (Burnham Institute for Medical Research, La Jolla, CA) and M. Olayioye (University of Stuttgart, Germany), respectively.
Cell culture and RNAi
HeLa and SKOV cells were grown in high-glucose DMEM supplemented with glutamine and 10% FCS. IGROV and OVCAR cells were grown in RPMI medium with 10% FCS. Treatment with siRNA (Qiagen) was performed using Oligofectamine reagent (Invitrogen) as described previously (Tafesse et al., 2007). siRNA target sequences were: NS (nonsense), 5′-AAUUCUCCGAACGUGUCACGU-3′; LA (lamine A), 5′-CUGGACUUCCAGAAGAACA-3′′; SMSr no. 1, 5′-CAAGAAGCUGGAAUUUCUUGC-3′; SMSr no. 2, 5′-AAUCUUCUUCAUCUUGGCUGC-3′; SMS1, 5′-AACUACACUCCCAGUACCUGG-3′; SMS2, 5′-CCCAAGAGCUUAUCCAGUG-3′; CERT, 5′-GAACAGAGGAAGCAUAUAATT-3′. Unless indicated otherwise, SMSr was depleted using siSMSr no. 1. For rescue experiments, medium was replaced after 9 hours of siRNA treatment and cells were transfected with DNA constructs using Effectene (Qiagen) according to the manufacturer's protocol. Medium was replaced 16 hours after plasmid transfection. SMSr constructs were made siRNA resistant by introducing five silent point mutations in the siRNA target sequence using the megaprimer PCR method. Fumonisin B1 (25 or 100 µM) and myriocin (30 µM) were added immediately after the start of siRNA treatment and kept throughout the 3 days of siRNA treatment. Staurosporine (1 µg/ml) was added 6–8 hours before the end of siRNA treatment.
RT-PCR analysis
RNA was extracted from HeLa cells using the RNeasey kit (Qiagen). Semi-quantitative RT-PCR was performed using the SuperScript™III kit (Invitrogen) and the following primer sets: XBP1 forward 5′-CTGGAACAGCAAGTGGTAGA-3′, reverse 5′-ACTGGGTCCTTCTGGGTAGA-3′; SMSr forward 5′-TGGATTTACATCTTTCATTATGGTT-3′, reverse 5′-TCCAATTAGTCTTTTCATTATTGCTG-3′; GAPDH forward 5′-AGAAGGCTGGGGCTCATTTG-3′, reverse 5′-AGGGGCCATCCACAGTCTTC-3′.
Microscopy and image analysis
Cells were fixed in 4% paraformaldehyde/PBS, processed for immunofluorescence microscopy and mounted in Vectashield medium containing DAPI (Vector Laboratories) as described (Tafesse et al., 2007). Images were captured at room temperature using a confocal microscope (LSM 510 Meta; Carl Zeiss, Inc.) with a 63×1.40 NA Plan Apo oil objective (Carl Zeiss, Inc.). The fluorochromes used were DAPI λex = 360 nm and λem = 460 nm, FITC and Alexa Fluor 488, λex = 488 nm and λem = 515 nm, Texas Red and Alexa Fluor 568, λex = 568 nm and λem = 585 nm. Projections were obtained by collecting series of 0.4–1.0 mm sections that were acquired and combined using LSM software (Carl Zeiss, Inc.). Images were further processed using Photoshop software (version 7.0.1; Adobe). The total number of cells was counted and the percentage of cells immuno-positive for cleaved PARP or cleaved Casp-3 in randomly chosen fields was determined. For rescue experiments, the percentage of cleaved PARP- or cleaved Casp-3-positive cells in the total population of plasmid-transfected cells was determined. Averages were derived from three to five independent experiments, with each average consisting of >300 (or >150 plasmid-transfected) cells per condition. The statistical significance of the data was assessed by two-tailed unpaired Student's t-tests. P<0.001 was considered significant and is marked by an asterisk in the figures.
Apoptosis assays
To measure mitochondrial membrane permeability and depolarization, cells were stained with JC-1 and subjected to fluorescence microscopy or FACS analysis using the MitoPT-JC1 assay kit (ImmunoChemistry Technologies). The release of cytochrome c from mitochondria into the cytosol was investigated using the ApoAlert cell fractionation kit (Clontech). The caspase-9 enzyme activity assay was performed on HeLa cell lysates using a colorometric assay kit (BioVision) following the manufacturer's protocol.
Isolation of mitochondria
Isolation of mitochondrial pellets was performed according to Vance (Vance, 1990) with some modifications. HeLa cells were washed twice with ice-cold 0.25 M sucrose and then harvested by scraping in ice-cold homogenizing buffer (HB; 250 mM mannitol, 5 mM Hepes-KOH, pH 7.0, 0.5 mM EGTA, 1 mM protein inhibitor cocktail and 0.1 mM phenylmethanesulfonyl fluoride) and centrifugation (600 gmax, 10 minutes). Cells were homogenized in HB by flushing through a Balch homogenizer 20–30 times using a 5 ml syringe. The homogenate was centrifuged twice at 600 gmax for 5 minutes to pellet nuclei. The post-nuclear supernatant was centrifuged at 10,300 gmax for 10 minutes in a Sorvall SM24 rotor to pellet mitochondria. Mitochondrial pellets were washed in HB and centrifuged again twice (10,300 gmax, 10 minutes) to remove contaminating ER, which was monitored by immunoblotting using organelle-specific antibodies.
Lipid tandem mass spectrometry analysis and metabolic labeling
Lipid extracts from HeLa cells and mitochondrial pellets were prepared according to Folch et al. (Folch et al., 1957), but omitting the salt. Tandem mass spectrometry (MS/MS) analysis of lipid extracts was performed as described in Vacaru et al. (Vacaru et al., 2009). For SMSr enzyme activity assay, HeLa cells were lysed in ice-cold reaction buffer (0.3 M sucrose, 15 mM KCl, 5 mM NaCl, 1 mM EDTA, 20 mM Hepes-KOH, pH 7.0) containing freshly added protease inhibitors by passing 20 times through a 26G 3/4-needle. 200 µl of post-nuclear supernatant (PNS; 700 g, 10 minutes, 4°C) were combined with 200 µl reaction buffer containing 0.002% Triton X-100, 40 nmol POPE and 50 µM NBD-Cer, and incubated at 37°C for 2 hours. Reactions were stopped by adding 1 ml methanol and 0.5 ml CHCl3 and lipids were extracted according to Bligh and Dyer (Bligh and Dyer, 1959). The lower phase was evaporated under nitrogen and the reaction products analyzed by thin layer chromatography (TLC), which was developed first in acetone, and then in CHCl3/methanol/25% NH4OH (50/25/6, v/v/v). For metabolic labeling, HeLa cells were incubated with 1 µCi L-[3-14C]serine in F-12 medium supplemented with 10% FCS for 5 hours and then washed in PBS. Lipids were extracted using CHCl3/methanol/10 mM acetic acid (1/4.4/0.2, vol/vol/vol), processed according to the method of Bligh and Dyer (Bligh and Dyer, 1959) and then separated by TLC in CHCl3/methanol/2 M NH4OH (40/10/1, v/v/v). Radiolabeled lipids were detected by exposure to BAS-MS imaging screens (Fuji Photo Film Co.), scanned on a Bio-Rad Personal Molecular Imager, and quantified with Quantity One software.
Acknowledgments
We thank Carolyn Machamer, Makoto Ito, John Reed and Monilola Olayioye for gifts of plasmids or antibodies, and Tarja Grundström for technical assistance.
Footnotes
Competing interests
The authors declare no competing interests.
Author contributions
F.G.T. and J.C.M.H. conceived the project. F.G.T. designed and performed most biochemical and molecular biological experiments with critical input from A.M.V., E.F.B., A.J. and A.H. M.H. performed all lipid mass spectrometry analyses and interpreted the data with valuable input from P.S. J.C.M.H. provided project management and wrote the manuscript, with hypothesis development, experimental design and data interpretation contributed by all authors.
Funding
This work was supported by the European Union (7th FP Marie-Curie ITN “Sphingonet” 289278), the Dutch Organization of Sciences (NWO-CW ECHO 700.59.002); and the Deutsche Forschungsgemeinschaft Sonderforschungs-bereich/SFB944, project P14 to J.C.M.H.
Supplementary material available online at http://jcs.biologists.org/lookup/suppl/doi:10.1242/jcs.138933/-/DC1
- Received July 22, 2013.
- Accepted October 30, 2013.
- © 2014. Published by The Company of Biologists Ltd