|Home | About | Journals | Submit | Contact Us | Français|
Stress conditions affecting the functions of the endoplasmic reticulum (ER) cause the accumulation of unfolded proteins. ER stress is counteracted by the unfolded-protein response (UPR). However, under prolonged stress the UPR initiates a proapoptotic response. Mounting evidence indicate that the ER chaperone calnexin is involved in apoptosis caused by ER stress. Here, we report that overexpression of calnexin in Schizosaccharomyces pombe induces cell death with apoptosis markers. Cell death was partially dependent on the Ire1p ER-stress transducer. Apoptotic death caused by calnexin overexpression required its transmembrane domain (TM), and involved sequences on either side of the ER membrane. Apoptotic death caused by tunicamycin was dramatically reduced in a strain expressing endogenous levels of calnexin lacking its TM and cytosolic tail. This demonstrates the involvement of calnexin in apoptosis triggered by ER stress. A genetic screen identified the S. pombe homologue of the human antiapoptotic protein HMGB1 as a suppressor of apoptotic death due to calnexin overexpression. Remarkably, overexpression of human calnexin in S. pombe also provoked apoptotic death. Our results argue for the conservation of the role of calnexin in apoptosis triggered by ER stress, and validate S. pombe as a model to elucidate the mechanisms of calnexin-mediated cell death.
The endoplasmic reticulum (ER) is a specialized organelle playing essential and central roles in the biology of the cell. The ER is the site of synthesis and folding of secreted, membrane-bound and some organelle-targeted proteins (Bukau et al., 2000 ; Fewell et al., 2001 ). Protein folding in the ER is assisted by a battery of molecular chaperones and foldases (Bukau et al., 2000 ; Fewell et al., 2001 ; Trombetta and Parodi, 2003 ; Helenius and Aebi, 2004 ). In addition, the ER contains several factors required for optimum protein folding, including, ATP, Ca2+, and an oxidizing environment to allow disulphide-bond formation. Proper ER function is critical for numerous aspects of cell physiology, including vesicle trafficking and lipid and membrane biogenesis, as well as protein targeting and secretion (Lai et al., 2007 ).
The ER is highly sensitive to stresses perturbing the cellular energy levels and ER lipid or glycolipid imbalances or changes in the redox state or Ca2+ concentration (Breckenridge et al., 2003 ; Boyce and Yuan, 2006 ; Szegezdi et al., 2006 ). Such stresses reduce the protein folding capacity of the ER, which results in the accumulation and aggregation of unfolded proteins, a condition referred to as ER stress. When the capacity of the ER to fold proteins properly is compromised or overwhelmed, a highly conserved unfolded-protein response (UPR) signal-transduction pathway is activated. The ER response to stress is basically conserved from yeast to mammalian cells (Patil and Walter, 2001 ; Ron and Walter, 2007 ). To counter ER stress, the UPR halts general protein synthesis and up-regulates the transcription of genes encoding ER resident chaperones and other regulatory components of the secretory pathway, giving the cell a chance to correct the environment within the ER (Breckenridge et al., 2003 ; Schroder and Kaufman, 2005 ; Xu et al., 2005 ; Boyce and Yuan, 2006 ; Szegezdi et al., 2006 ; Wu and Kaufman, 2006 ). Moreover, in coordination with the UPR, the cell increases the capacity of the proteasome-dependent ER-associated degradation (ERAD) to remove misfolded proteins from the ER (Ahner and Brodsky, 2004 ; Meusser et al., 2005 ; Marciniak and Ron, 2006 ).
In Saccharomyces cerevisiae, ER stress is monitored by Ire1p, an ER transmembrane protein. According to current models (Patil and Walter, 2001 ; McCracken and Brodsky, 2005 ; Boyce and Yuan, 2006 ; Ron and Walter, 2007 ), the ER lumenal domain of a single Ire1p molecule interacts with BiP/Kar2p, an ER lumenal chaperone that assists the folding of nascent polypeptides entering the lumen. Under ER stress conditions, BiP is titrated away from Ire1p due to its binding to unfolded or misfolded proteins. The removal of BiP permits the activation of Ire1p by oligomerization leading to the efficient induction of UPR, which up-regulates genes coding for ER chaperones and factors involved in all stages of the secretory pathway. The mammalian UPR retains several key features of the yeast program like the role of BiP, but is significantly more complex and is mediated through three ER transmembrane receptors: PERK, ATF6, and IRE1. In normal conditions, the suite of responses activated by the UPR pathway succeeds in restoring ER homeostasis. However, if protein aggregation is persistent and the stress cannot be resolved, the UPR signaling switches from prosurvival to proapoptotic. On prolonged ER stress the PERK, ATF6, and IRE1 proapoptotic signals ultimately execute cell death through the activation of kinases, proteases, and transcription factors, thereby affecting the expression and activities of the Bcl-2 family of proteins and their modulators (Breckenridge et al., 2003 ; Xu et al., 2005 ; Szegezdi et al., 2006 ).
Apoptosis is a form of programmed cell death that is characterized by several specific morphological and biochemical changes including cell rounding and shrinkage, chromatin breakage, nuclear fragmentation, and activation of caspases (Kerr et al., 1972 ; Kerr, 2002 ). Apoptosis was first described and extensively studied in multicellular organisms, where it plays critical roles in development and the control of many diseases in adult organisms. Evidence accumulated in the last 10 y has demonstrated that apoptosis-like processes do also occur in S. cerevisiae and in Schizosaccharomyces pombe yeasts. Yeasts undergo apoptosis-like death under a variety of conditions and cell processes including DNA damage, mitotic catastrophe, replication defects, deficiency in triacylglycerols, ER stress, aging, and mating (Burhans et al., 2003 ; Zhang et al., 2003 ; Fahrenkrog et al., 2004 ; Wissing et al., 2004 ; Low et al., 2005 ; Cho et al., 2006 ; Hauptmann et al., 2006 ; Roux et al., 2006 ; Walter et al., 2006 ; Frohlich et al., 2007 ). Both S. cerevisiae and S. pombe encode several homologues of proteins characterized for their implication in apoptosis including AIF, HtrA2/Omi, and IAP. Yeasts also encode a metacaspase called Yca1/Mca1 in S. cerevisiae and Pca1 in S. pombe (Uren et al., 2000 ; Madeo et al., 2002 ; Lim et al., 2007 ). However, programmed-cell death pathways in yeast can be metacaspase-dependent or -independent (Madeo et al., 2002 ; Bettiga et al., 2004 ; Fannjiang et al., 2004 ; Herker et al., 2004 ; Wadskog et al., 2004 ; Ivanovska and Hardwick, 2005 ; Reiter et al., 2005 ; Liang et al., 2008 ; Mazzoni and Falcone, 2008 ). It has been proposed that apoptotic processes are useful in unicellular organisms for their survival under conditions of environmental stresses and are relevant in populations of cells such as colonies and biofilms (Skulachev, 2002 ; Buttner et al., 2006 ). Because of their powerful genetics, S. cerevisiae and S. pombe have become interesting models to study the core mechanisms of apoptosis (Ink et al., 1997 ; Ligr et al., 1998 ; Madeo et al., 2002 ; Priault et al., 2003 ; Hardwick and Cheng, 2004 ; Madeo et al., 2004 ; Rodriguez-Menocal and D'Urso, 2004 ; Burhans and Weinberger, 2007 ; Frohlich et al., 2007 ; Almeida et al., 2008 ; Fabrizio and Longo, 2008 ; Low and Yang, 2008 ).
The molecular chaperone calnexin plays key roles in the translocation of nascent polypeptides and in the folding and quality control of newly synthesized proteins (Bukau et al., 2000 ; Fewell et al., 2001 ; Williams, 2006 ). Structurally, calnexin is a type I ER transmembrane protein, with a large lumenal domain, a transmembrane domain (TM), and a short cytosolic tail (see Figure 1A). The lumenal domain of calnexin contains the highly conserved central domain (hcd), which is the portion of the protein that is the most conserved across species. This region of calnexin interacts with client proteins in a glycan-lectin manner or via protein–protein interactions (Arunachalam and Cresswell, 1995 ; Fernandez et al., 1996 ; Jannatipour et al., 1998 ; Beaulieu et al., 1999 ; Saito et al., 1999 ; Parodi, 2000 ; Marechal et al., 2004 ; Hebert et al., 2005 ; Thammavongsa et al., 2005 ; Williams, 2006 ).
Pointing to its critical biological roles, the knockout of calnexin causes early-postnatal death in mouse and is lethal in the fission yeast S. pombe (Jannatipour and Rokeach, 1995 ; Parlati et al., 1995a ; Denzel et al., 2002 ). Intriguingly, some S. pombe calnexin mutants devoid of chaperone activity are viable, demonstrating therefore that the essential cellular role of calnexin is not its chaperone function (Elagoz et al., 1999 ; Marechal et al., 2004 ; Hajjar et al., 2007 ).
Cumulating evidence point to the implication of calnexin in apoptosis induced by ER stress. First indication resulted from studies in S. pombe that led to the identification of calnexin as a lethal partner of the proapoptotic mammalian proteins Bak and Bax (Torgler et al., 1997 ). The authors showed that the expression of Bak is lethal in S. pombe cells expressing full-length calnexin but not in cells expressing a calnexin mutant devoid of the C-terminal tail, suggesting a role as signal transducer for calnexin by recruiting lethal partners (Torgler et al., 1997 ). In the case of mammalian cells, it has been shown that cells devoid of calnexin are relatively resistant to ER-stress–induced apoptosis (Zuppini et al., 2002 ; Groenendyk et al., 2006 ). It was also proposed that mammalian calnexin acts as a scaffold for the cleavage by caspase-8 of the ER transmembrane protein Bap31 under ER-stress conditions, such as in the presence of tunicamycin (Delom et al., 2006, 2007 ). Moreover, mammalian calnexin was shown to be cleaved by caspases under stress conditions, and it was suggested that this cleavage may have a role in the mediation of the apoptotic signal (Takizawa et al., 2004 ). Nevertheless, much remains to be elucidated about the mechanistic details of the role of calnexin in ER-stress–induced apoptosis. Certainly, the study of the involvement of calnexin in apoptosis caused by ER stress could greatly benefit from the use of a model organism.
Here, we demonstrate that calnexin is involved in the induction of apoptosis triggered by ER stress in S. pombe. We further show that the apoptotic effect of calnexin is counteracted by overexpression of Hmg1/2p, the S. pombe homologue of the mammalian antiapoptotic protein HMGB1 (high-mobility group box-1 protein). Interestingly, the overexpression of mammalian calnexin also induced apoptosis in S. pombe, suggesting the functional conservation of the role of calnexin in apoptosis.
Experiments were carried out using the S. pombe strains described in Table 1. All strains were cultured in Edinburgh minimal medium (here denoted as MM) supplemented with required supplements (Moreno et al., 1991 ) at 30°C. DNA transformations into S. pombe were performed by the PEG-lithium acetate procedure, as previously described (Elbe, 1992 ). S. pombe homozygous, haploid strains deleted for the homologue of IRE1 (geneID SPAC167.01), for the homologue of YCA1 (geneID SPCC1840.04, pca1::kanMX4; strain SP8167R), and for the gene encoding the homologue of Bap31 (geneID SPAC9E9.04, dma1::kanMX4; strain SP8083R) were purchased from Bioneer (Daejeon, Korea).
pREP is a family of multicopy expression vectors for S. pombe. All vectors contain the S. pombe ars1 origin of replication and a version of thiamine-repressible nmt promoter (Maundrell, 1993 ). The nmt1 promoter in pREP1, pREP2, and pREP3X is full strength, and these vectors were used to overexpress wild-type calnexin (strains SP8007R, SP8231R, SP8136R, and SP8081R), mutants of calnexin (strains SP7965R, SP16053, SP8056R, and SP8125R), and controls (strains SP16058, SP15302, SP8227R, SP8145R, and SP8178R; Forsburg, 1993 ). The pREP3X vector was derived from the pREP family, has the promoter strength like pREP1 and pREP2, but has an XhoI site replacing the ATG at the 5′ end of its polylinker (Forsburg, 1993 ). This empty vector was used as control (strain SP7975R). The nmt promoter in the vectors pREP41 and pREP42 is medium strength and expressed calnexin at levels comparable to the endogenous genomic copy of the gene (Jannatipour et al., 1998 ). pREP41 differs from pREP42 in that it contains the S. pombe LEU2 marker instead of ura4+, and likewise for the pREP1 and pREP2 vectors, respectively. To test the ability to confer viability the calnexin mutants lumenalTM_cnx1 and C-termTM_cnx1_cmyc were cloned in the pREP41 and pREP42 vectors and were shuffled against wild-type calnexin (strains SP8075R and SP8160R, respectively). The wild-type calnexin and the viable mutants on pREP41 vector were used to study ER stress induced by tunicamycin (strains SP7951R, SP8490R, SP8488R, and SP8085R).
The construction of mutants mini_cnx1 and lumenal_cnx1 was previously described (Elagöz et al., 1999 ). The mutant lumenalTM_cnx1 was obtained by PCR amplification with the following primers: A2: 5′-AAACATATGAAGTACGGAAAGGTATCT-3′ and L: 5′-AAAGGATCCTTAAGCAAAGAAATAAAAGTAACA-3′. These two primers contained restriction sites for NdeI and BamHI to allow the cloning of the PCR product into the pREP1 and pREP41. For the mutant C-termTM_cnx1_cmyc, two independent fragments were first amplified by PCR. One fragment corresponded to the signal peptide (SP) of the wild-type calnexin and was amplified with the primers A2 and F2: 5′-AGGATCAGCAAGTGATCCCCG-3′. The second fragment corresponding to the transmembrane domain with the cytosolic tail in fusion with a c-myc tag was obtained with primers M2: CTTGCTGATCCTATTGGGATTGCAATTGTTGCCGTT-3′ and c-mycSTOPcnx1: 5′-GGATCCTTACATGGCATTCAAGTCCTCTTCAGAAATGAGCTTTTGCTCCATGTCTTCATTCTTCGCAGT-3′. These two PCR products were mixed together to perform an overlap PCR amplification with the oligonucleotides primers A2 and c-mycSTOPcnx1. These two oligonucleotide primers contain restriction sites (NdeI for A2 and BamHI for c-mycSTOPcnx1), allowing their cloning into the pREP vectors family. The final product was digested with the restriction enzymes NdeI and BamHI, as it was done for the lumenalTM_cnx1 mutant and cloned into the pREP1 and pREP41 vector. The homologue of HMGB1 in S. pombe is Hmg1/2 that is encoded by the open reading frame (ORF) SPBC28F2.11, which was amplified from genomic DNA by PCR with the following primers: HMG1/2 NdeI forward: 5′-AAACATATGGCTCAAAACTCAACCC-3′ and HMG1/2 SalI reverse: 5′-AAAGTCGACTCAATTAGCAACTTTGGC-3′, and cloned into the cloning vector pDRIVE (PCR Cloning Kit; QIAGEN, Chatsworth, CA) following the manufacturer's recommendations. The gene hmg1/2 was then cut out from pDRIVE-HMG1/2 with the restriction enzymes NdeI and SalI and cloned into pREP42 and pREP2 vectors. The cDNA encoding human calnexin was kindly provided by Dr. Michael Brenner (Brigham Women's Hospital, Boston, MA; Rajagopalan et al., 1994 ) The gene was amplified with the following oligonucleotide primers: cnxh1-NdeI-F: 5′-AAACATATGGAAGGGAAGTGGTTGCTG-3′ and cnxh1-BamHI-R: 5′-AAAGGATCCTCACTCTCTTCGTGGCTT-3′ and cloned into the cloning pDRIVE vector (QIAGEN PCR Cloning Kit) following the manufacturer' recommendations. The cDNA encoding human calnexin was then digested with restriction enzymes NdeI and BamHI and cloned into the pREP41 and pREP1.
To test the viability of new calnexin mutants, S. pombe (cnx1Δ + pcnx1+) cells containing a deletion of genomic calnexin (cnx1Δ) and a plasmid bearing a wild-type copy of the calnexin gene (cnx1+) were transformed with plasmids bearing the calnexin mutants to be tested (strains SP8075R and SP8160R). S. pombe strains bearing the two plasmids were grown for 7 d at 30°C in 5 ml liquid nonselective media (MM supplemented with adenine and appropriate supplements, i.e., uracil and/or leucine). Cells were plated on the same solid nonselective media, and auxotrophy for uracil and/or leucine was analyzed after 5–7 d by toothpicking on selective plates. The viability of the calnexin mutants was determined by examining the cell's capacity to conserve only the plasmid encoding mutant calnexin without the pcnx1+ plasmid.
Microsomal membranes were prepared as previously described (Elagoz et al., 1999 ). S. pombe microsomal membranes were treated for 15 min at 4°C by mixing 1 vol of either 1 M NaCl or 0.2% SDS. Membrane lysates were spun at 80,000 × g for 1 h at 4°C in a L8–70M Sorvall ultracentrifuge (Newton, CT), and then the pellet from this spin was resuspended in 0.1 ml of 3× Laemmli's sample buffer (P fraction). Proteins in the supernatant fraction were treated for 30 min at 4°C in the presence of 6% ice-cold trichloroacetic acid (TCA) and spun at 2000 × g for 45 min at 4°C. The pellet was washed twice in ice-cold 80% acetone and dissolved in 0, 1 ml 3× Laemmli's sample buffer (S fraction). Before SDS-PAGE, samples were boiled for 5 min.
Immunoprecipitations from cells (the equivalent of OD595 0.8–1.0), treated or not with 10 μg/ml tunicamycin for 24 h, were carried out as previously described (Jannatipour et al., 1998 ).
Protein extractions were carried out as previously described (Elagoz et al., 1999 ) Protein extracts were migrated on a 10 or 12% (wt/vol) SDS-PAGE gel. Proteins were transferred onto nitrocellulose membrane according to the manufacturer's instructions. Immunoblotting to detect Cnx1p and mutants of calnexin was carried out with the anti-Cnx1p rabbit polyclonal antibody (LAR223), dilution 1:30,000 or with the anti-cmyc mouse mAb 9E10, dilution 1:500. Immunoblotting to detect BiP was performed with the anti-BiP rabbit polyclonal antibody (LAR311), dilution 1:30,000 (Collin et al., 2004 ). When necessary, the intensity of the immunoblotting bands was quantified with Bio-Rad Quantity One 4.6.5 Basic program (Richmond, CA).
Briefly, cells overexpressing calnexin were transformed with the S. pombe pURSP1 genomic bank (Barbet et al., 1992 ) containing insertions between 2.5 and 10 Kb (the generous gift of Dr. Anthony Carr, University of Sussex, United Kingdom). About 150,000 clones were screened and 50 potential suppressors were selected. All 50 potential candidate strains were subjected to a secondary screening by Western blotting, to verify that the incoming genomic plasmid did not affect the levels of overexpression of calnexin. After verification, only one candidate was a real suppressor of cell death by overexpression, without it affecting the levels of calnexin. The ORF-identified SPBC28F2.11 encoding Hmg1/2p, the homologue of the mammalian nuclear protein HMGB1, was cloned into the pREP2 overexpression vector.
Cells overexpressing wild-type or mutants of calnexin and control strains (see Table 1) were cultured for 43 h until saturation in MM containing the required supplements in the presence of 5 μg/ml thiamine to repress the nmt1 promoter of the pREP1 or the pREP2 vector. At saturation, the cells were diluted to OD595 = 0.3 in the same media and cultured for 19 h. Then, to induce the nmt1 promoter, an aliquot containing 5 × 107 cells was taken, and the cells were washed twice with 1 ml of media without thiamine, after which the cells were resuspended in an appropriate volume and 1 × 107 cells were inoculated in 3 ml of media without thiamine. Maximal levels of wild-type or mutants of calnexin were reached after 18 h of culture in MM without thiamine. In all experiments, the time point zero was defined as the time point where the thiamine is removed from the medium.
Cells only expressing full-length calnexin or viable calnexin mutants at basal levels (see Table 1) were cultured for 43 h until saturation in MM with the required supplements. To obtain exponentially growing cells, the cells were diluted and cultured overnight until they reached OD595 = 0.3–0.8. A volume of 10 ml of culture was adjusted to OD595 0.3, and cells were treated with 10 μg/ml tunicamycin for 40 h at 30°C. Tests for metacaspase activation and nuclear fragmentation [DAPI (4′,6-diamidino-2-phenylindole) staining] were performed at this time.
The survival of cells was measured by two different techniques: by serial 10-fold dilutions (drop tests on plates) and by cytometry with the vital fluorescent dye Phloxin B. For serial-dilution spotting experiments, an equivalent of OD595 nm = 0.8 was taken from exponentially growing cells in media containing thiamine. These cells were washed twice in media free of thiamine and adjusted to an equivalent of OD595 nm = 0.5. The cells were serially diluted (10−1–10−4), spotted on solid media with or without thiamine, and incubated for 7 d at 30°C. Viability assays with the Phloxin B fluorescent vital dye were carried out as previously described (Roux et al., 2006 ).
Culture samples were taken at time points 32, 40, 48, 56, and 72 h after gene induction in the media depleted of thiamine or after 40 h of tunicamycin treatment. Aliquots containing 1 × 107 cells were washed once in 1 ml of 1× PBS, pH 7.4 (136 mM NaCl, 25 mM KCl, 12 mM NaHPO4, 18 mM KH2PO4) and resuspended in 150 μl of 1× PBS, pH 7.4, containing 10 μM FITC-VAD-FMK (valyl-alanyl-aspartyl-[O-methyl]-fluoromethylketone; CasPACE, Promega, Madison, WI). After incubation for 20 min at 30°C, cells were washed once in 1× PBS, pH 7.4, and resuspended in 100 μl 1× PBS, pH 7.4, to be analyzed by fluorescence-activated cell sorting (FACS).
TUNEL (Terminal uridine deoxynucleotidyl transferase dUTP nick end labeling) assay was performed with the APO-BRDU TUNEL Kit (Phoenix Flow Systems, San Diego, CA), essentially following the manufacturer's recommendations. After 96 h of gene induction by culturing cells in media without thiamine, 1.4 × 107 cells were taken and fixed with 1 ml of 3.7% formaldehyde. After fixation, the cell wall was digested by resuspending the cell pellet in 200 μl of sorbitol buffer (1.2 M sorbitol, 0.5 mM MgCl2, potassium phosphate, pH 6.8) containing 5 mg/ml lysing enzymes (Sigma, Oakville, ON, Canada), and incubating for 90 min at room temperature, followed by 30-min incubation at 37°C. The cell pellet was resuspended in 500 μl of permeabilization solution (0.1% Triton in 0.1% sodium citrate) and kept on ice for 2 min, washed twice with 400 μl of WASH solution, and incubated in 50 μl of TUNEL solution for 30 min at 30°C. After incubation, the cells were washed twice in WASH solution and incubated 30 min at room temperature in the dark with 100 μl antibodies solution (anti-BrdU antibodies). Staining of the cells was analyzed by flow cytometry (FACS).
Annexin V staining was performed with the Annexin V-FITC Apoptosis Detection Kit (BioVision Research Products, Mountain View, CA) essentially following the manufacturer's recommendations. After 48 h of gene induction by culturing cells in media without thiamine, 1.4 × 107 cells were centrifuged, resuspended, and incubated for 2 h at room temperature in 200 μl of sorbitol buffer (1.2 M sorbitol, 0.5 mM MgCl2, potassium phosphate, pH 6.8) containing 10 mg/ml Zymolyase 20T (Seikagaku, , Tokyo, Japan) in order to digest the cell wall. The cell pellet was washed with 400 μl of binding buffer (10 mM HEPES/NaOH, 140 mM NaCl, 2.5 mM CaCl2, and 1.2 M sorbitol) and resuspended in 35 μl of binding buffer containing 5 μl of Annexin V-FITC and incubated at room temperature in the dark for 20 min. After incubation, the cells were washed once in 400 μl of binding buffer and resuspended in 500 μl of 1× binding buffer (Annexin V-FITC Apoptosis Detection Kit, BioVision Research Products). Staining of the cells was analyzed by flow cytometry (FACS).
For DAPI staining, samples containing 1.4 × 107 cells were taken after 96 h of induction or after 40 h of tunicamycin treatment. Cells were fixed for 10 min in a solution of 3.7% formaldehyde and washed once in 1, PBS, pH 7.4, containing 1% Nonidet P-40 and twice in 1× PBS, pH 7.4. The cells were resuspended in 100 μl 1× PBS, pH 7.4, to a final concentration of 5 × 107–1 × 108 cells/ml. Suitable quantities of cells were applied to a poly-lysine–coated coverslips, washed, and let dry. The slides were mounted with a DAPI-containing mounting media (1 μg/ml DAPI, 1 mg/ml p-phenylenediamine, 90% glycerol). Microscopic analysis was performed using a fluorescence inverted microscope Nikon TE2000U (Melville, NY). Images were acquired using a motion-picture camera CCD coolSnapFX M 12 bit (Roper Scientific, Tucson, AZ) and treated with the UIC Metamorph software (Universal Imaging, West Chester, PA).
The accumulation of reactive oxygen species (ROS) was determnined essentially as previously described (Roux et al., 2006 ). Culture samples were taken after 24 h of induction. Samples containing 1–2 × 107 cells were incubated with 30 μM dihydrorhodamine123 (Sigma), for 30 min at 30°C. A suitable quantity of cells were applied to a poly-lysine-coated coverslip, washed with 1× PBS, pH 7.4, and let dry. Fluorescence was observed with a fluorescence microscope Nikon E800. Images were acquired using a motion-picture camera CCD coolSnapFX M 12 bit and treated with UIC Metamorph software.
Cells were stained with Phloxin B, FITC-VAD-FMK or with the TUNEL assay as described above. Flow cytometry analyses were performed using a FACS Calibur (BD Biosciences, Mountain View, CA) device, on 10,000 cells. Emission from the argon laser was at 488 nm; emission settings were 515–545 nm (filter FL-1) for FITC-VAD-FMK and fluorescein staining or 560–600 nm (filter FL-2) for Phloxin B staining. The percentage of positive stained cells was determined as the population of fluorescent cells with a higher fluorescent intensity than a stained negative control. Parameters of the stained negative control were adjusted with an unstained negative control. Each experiment was repeated three times.
The significance of the variations of results among strains was determined by a global ANOVA as described in http://www.physics.cbsju.edu/stats/anova.html. In certain figures, the significance to variation with respect to controls or Cnx1p was evaluated by a Student's t test.
We have previously demonstrated that although calnexin is an essential chaperone playing key roles in protein folding and quality control in the ER, its vital function in S. pombe is not its chaperone activity (Elagoz et al., 1999 ; Marechal et al., 2004 ). Because the overexpression of a gene generally leads to a gain of function (Ramer et al., 1992 ), to obtain clues as to the essential cellular role of calnexin we overexpressed cnx1+ in the fission yeast. Interestingly, we observed that the overexpression of calnexin resulted in a severe decline in the viability of the culture. To characterize this phenotype, we assayed the viability of cells overexpressing calnexin by measuring their capacity to grow on inducing media, i.e., without thiamine (Figure 1B). When cells containing the pREP1cnx1+ plasmid were spotted on media free of thiamine, we observed a dramatic reduction in the ability to form colonies compared with the control strain containing the pREP3X vector. By contrast, no difference with the control was observed when cells were spotted on thiamine-containing media, which represses the overexpression of calnexin (Figure 1B). The same phenotype was observed when cells were stained with Phloxin B, a fluorescent vital dye that accumulates within dead cells (Roux et al., 2006 ). At 48 h after induction, ~50% of cells overexpressing calnexin were stained with Phloxin B compared with nearly 0% for cells harboring the empty vector control (Figure 1C). To assess whether this death phenotype was specific to calnexin and was not due to an overloading of the ER lumen, we overexpressed Pdi1p, a soluble lumenal ER foldase involved in the formation of disulfide bonds of newly synthesized polypeptides (Maattanen et al., 2006 ). No cell death was observed by spotting on media free of thiamine or by Phloxin B staining when cells overexpressed Pdi1p (Figure 1, B and C). However, the strain overexpressing Pdi1p appeared to grow more slowly than the control cells containing the empty vector (Figure 1B). Because, calnexin is an ER integral-membrane protein, it remained possible that the cell-death phenotype could be the result of saturating the ER membrane and not due to a specific effect of a calnexin function. To investigate this possibility, we overexpressed the Sec61β subunit of the translocon (Sbh1p; see Table 1), which is an ER-membrane protein (Romisch, 1999 ). As depicted in Figures 1, B and C, the overexpression of Sec61β did not cause cell death. Finally, to ascertain that overexpression of calnexin does not disrupt the ER membrane, the ER membranes were extracted and treated with Tris buffer alone (mock) as control, with a high concentration of NaCl to disrupt ionic interactions, and with SDS to break all interactions. This experiment showed that overexpression of calnexin does not alter the integrity of the membrane because no proteins were detected in the supernatant of the mock treatment (see Supplemental Figure S1). Together, these observations demonstrate that the death phenotype of cells overexpressing calnexin is specifically inherent to this ER chaperone and not due to an artifact caused by the overloading of the capacity of the ER membrane and/or lumen.
Next we investigated whether the death of S. pombe cells overexpressing calnexin is due to an apoptotic process or whether these cells undergo unspecific necrotic death. Cells undergoing apoptosis display typical morphological and biochemical markers, which necrotic cells do not exhibit (Kerr et al., 1972 ; Jin and Reed, 2002 ; Madeo et al., 2004 ). Typical markers of apoptosis include phosphatidyl serine externalization, caspase activation, chromatin breakage, and nuclear fragmentation. Cells entering in the apoptotic cascade usually show phosphatidyl serine externalization as a first phenotype associated with this kind of death. About 45% of cells overexpressing Cnx1p were stained with Annexin V-FITC compared with 0% for the control cells (overexpressing Pdi1p or Sec61β, or harboring the empty vector; Figure 2A). This demonstrates that phosphatidyl serine is specifically externalized due to calnexin overexpression. Fluorescence microscopy with DAPI revealed nuclear fragmentation in cells overexpressing calnexin, whereas the nuclei of control cells (overexpressing Pdi1p or Sec61β or harboring the empty vector) remained intact for the same time point (Figure 2B). To assess if nuclear fragmentation was due to chromatin breakage, the TUNEL assay was performed and quantified by FACS analysis. Cells overexpressing calnexin exhibited a significant TUNEL-positive phenotype compared with the controls (cells overexpressing Pdi1p or Sec61β or harboring the empty vector), which is consistent with the DAPI phenotype (Figure 2C). We and others (Hauptmann et al., 2006 ; Roux et al., 2006 ) have previously shown that caspase-like (metacaspase; Uren et al., 2000 ) activation can be assessed in S. cerevisiae and in S. pombe by the permeable fluorescent maker FITC-VAD-FMK. FACS analysis revealed that ~40% of cells overexpressing calnexin were metacaspase positive at 72 h after induction (Figure 2D). In contrast, no FITC-VAD-FMK fluorescent cells were observed in the control cultures of the strains overexpressing Pdi1p or Sec61β or harboring the empty vector. High levels of ROS are associated with apoptosis due to ER stress in mammalian and yeast cells (Carmody and Cotter, 2001 ). We measured this apoptotic marker by staining cells with dihydrorhodamine 123 (DHR123), which is oxidized to fluorescent rhodamine by ROS. As shown by fluorescent microscopy analysis, only cells overexpressing calnexin produced ROS, thus indicating that overexpression of Cnx1p may cause ER stress (Figure 2E). Collectively, these observations show that overexpression of calnexin in S. pombe leads to cell death with the typical markers associated with apoptosis.
To identify molecular determinants of calnexin involved in the triggering apoptotic death, we overexpressed cnx1 truncation mutants. We began our analysis with lumenal_Cnx1p and mini_Cnx1p, two viable mutants previously characterized (Elagoz et al., 1999 ). The lumenal_cnx1 mutant encodes the entire lumenal domain of calnexin (Figure 3A), which was shown to have the same level of chaperone activity as wild-type calnexin (Elagoz et al., 1999 ; Marechal et al., 2004 ). The mini_cnx1 mutant encodes a truncated calnexin protein spanning the last 52 amino acids (aa) of the lumenal domain, the TM and the cytosolic tail (Figure 3A). This mutant exhibits no detectable chaperone activity (Elagoz et al., 1999 ; Marechal et al., 2004 ).
Remarkably, overexpression of lumenal_Cnx1p did not provoke cell death, as measured by spotting on inducing media or by Phloxin B (Figure 3, C and D). In agreement with these results, cells overexpressing lumenal_Cnx1p did not exhibit apoptotic markers (nuclear fragmentation, DNA breakage, and metacaspase activation; Figure 4, A–C). In contrast, the overexpression of mini_Cnx1p resulted in significant cell death (Figure 3, C and D). Overexpression of mini_Cnx1p also provoked apoptotic phenotypes similar to those caused by overexpression of wild-type calnexin (Figure 4, A–C). Therefore, removing the TM and the cytosolic tail completely abolished the apoptotic effect triggered by overexpression of calnexin.
To further map the determinants involved in the induction of apoptosis by overexpression of calnexin, we constructed two additional mutants (Figure 3A). One of them is called lumenalTM_cnx1 and encodes the lumenal domain of calnexin and includes the TM anchor. The other mutant is C-termTM_cnx1_cmyc and encodes the TM and the cytosolic tail of calnexin with a c-myc tag to facilitate immunodetection (Figure 3B). The mutant lumenalTM_Cnx1p conferred viability, whereas C-termTM_Cnx1p_cmyc could not sustain viability in a Δcnx1 strain (Figure 3A). That the encoded mutant proteins were indeed correctly integrated into the ER membrane was verified using cell fractionation and detergent solubilization, as we have previously described for wild-type calnexin and mini_Cnx1p (Elagoz et al., 1999 ; see Supplemental Figure S1).
As for wild-type calnexin, the overexpression of lumenalTM_Cnx1p and C-termTM_Cnx1p_cmyc mutants resulted in apoptotic cell death (Figure 3, C and D), as judged by the typical markers nuclear fragmentation, DNA breakage, and metacaspase activation (Figure 4, A–C). However, in the case of the mutant lumenalTM_Cnx1p, the apoptotic death phenotype was stronger than with wild-type calnexin, whereas in the case of the C-termTM_Cnx1p_cmyc mutant the apoptotic death phenotype was weaker. For instance, cells overexpressing lumenalTM_Cnx1p were not able to form colonies on media without thiamine, whereas cells overexpressing C-termTM_Cnx1p_cmyc formed colonies, albeit not as efficiently as the control strains (Figure 3C). Moreover, ~70% of the cells overexpressing the lumenalTM_Cnx1p mutant were dead as judged by Phloxin B staining, whereas only 21% of dead cells were observed for the strain overexpressing C-termTM_Cnx1p_cmyc (Figure 3D). Cells overexpressing these two mutants showed distinctive nuclear fragmentation, DNA breakage, and metacaspase activation (Figure 4, A–C). The strongest difference in apoptotic markers was observed for metacaspase activation measurements because 70% of cells overexpressing lumenalTM_Cnx1p were positive after 72 h of induction, compared with ~27% for C-termTM_Cnx1p_cmyc (Figure 4C). Together these experiments indicate that the TM is required for the induction of apoptotic cell death by calnexin overexpression and suggest that some calnexin sequences on either or both sides of the ER membrane might be necessary for this lethal phenotype. Importantly, as the mini_Cnx1p and C-termTM_Cnx1p_cmyc mutants do not exhibit chaperone activity, it clearly appears that the role of calnexin in apoptosis is distinct from its chaperone function in the ER lumen.
Because its overexpression induces apoptotic death, we wanted to investigate whether calnexin is involved in apoptosis provoked by ER stress. Tunicamycin is an inhibitor of N-glycosylation that strongly induces apoptosis in mammalian cells and in the budding yeast (Perez-Sala and Mollinedo, 1995 ; Hacki et al., 2000 ; Hauptmann et al., 2006 ; Hauptmann and Lehle, 2008 ). We have previously reported that tunicamycin efficiently inhibits protein N-glycosylation and induces ER stress in S. pombe (Jannatipour and Rokeach, 1995 ; Jannatipour et al., 1998 ; Beaulieu et al., 1999 ). As depicted in Figure 5A, S. pombe cells treated with tunicamycin displayed typical apoptotic makers, such as nuclear fragmentation and metacaspase activation. To investigate if calnexin is involved in apoptosis induced by the ER stressor tunicamycin, we constructed S. pombe strains solely expressing the mutant version of calnexin at the basal level, in a Δcnx1 genetic background. The genes coding for the calnexin mutants to be tested were cloned into the pREP41 vector, which expresses calnexin at the basal endogenous level (Jannatipour et al., 1998 ). It is important to note that because calnexin is essential for viability in S. pombe, only viable calnexin mutants can be tested in the Δcnx1 background. The mutants tested were mini_cnx1, lumenal_cnx1, and lumenalTM_cnx1. After treatment with tunicamycin, ~35% of wild-type calnexin cells were positive for metacaspase activation as measured by staining with the fluorescent marker FITC-VAD-FMK (Figure 5B). The same level of metacaspase activation was observed with the strains mini_cnx1 and lumenalTM_cnx1 mutants (Figure 5B). Interestingly, only 17% of the lumenal_cnx1 mutant cells treated with tunicamycin stained positive for metacaspase activation (Figure 5B). This diminution in metacaspase activity corresponds to a reduction of 50% of the total metacaspase-positive wild-type calnexin cells.
BiP anchoring to the ER membrane was reported in mammalian cells subjected to ER stress (Rao et al., 2002 ; Reddy et al., 2003 ). In previous reports, we have shown that calnexin and BiP associate in a complex, (Jannatipour et al., 1998 ; Elagoz et al., 1999 ; Marechal et al., 2004 ). To investigate whether the BiP-Cnx1p association is affected in the presence of tunicamycin, we immunoprecipitated Cnx1p or lumenal_Cnx1p and verified the presence of BiP in the precipitate. The treatment with tunicamycin increased the interaction between the two partners for both, wild-type calnexin and the lumenal_Cnx1p mutant, with a higher ratio of BiP for the latter (Figure 5C).
Ire1p is a mediator of ER stress that is conserved from yeast to mammals (Patil and Walter, 2001 ; Breckenridge et al., 2003 ). Therefore, we wanted to investigate if apoptosis induced by calnexin overexpression involves Ire1p. We overexpressed calnexin in the Δire1 background and measured cell death and metacaspase activation. Indicating the involvement of Ire1p, we observed ~50% diminution in the levels of cell death and metacaspase activation induced by calnexin overexpression in the Δire1 background (Figure 5D).
To sum up, apoptosis induced by the ER-stressor tunicamycin is strongly reduced in the background of lumenal_Cnx1p, the only mutant tested that does not provoke apoptosis by overexpression. This observation demonstrates the involvement of calnexin in ER-mediated apoptosis and support the notion that the TM and/or the anchoring of calnexin to the ER membrane are important for the proapoptotic function of calnexin in S. pombe. That apoptosis induced by calnexin overexpression is significantly reduced in the absence of Ire1p suggests that this mediator plays a role in the apoptosis pathway involving calnexin.
S. pombe encodes several homologues of known actors of the apoptotic pathway (Uren et al., 2000 ; Madeo et al., 2002 ; Lim et al., 2007 ). So far, only one caspase-like protein was identified in yeast, and the dependence of apoptosis on this metacaspase is variable, depending of the apoptotic inducer (Madeo et al., 2002 ; Bettiga et al., 2004 ; Fannjiang et al., 2004 ; Herker et al., 2004 ; Wadskog et al., 2004 ; Ivanovska and Hardwick, 2005 ; Reiter et al., 2005 ; Liang et al., 2008 ; Mazzoni and Falcone, 2008 ). To test whether the apoptosis induced by calnexin overexpression is dependent of the metacaspase Pca1p, death was assayed by spotting cells on inducing media and by staining cells with Phloxin B in the Δpca1 strain containing an empty vector (Vector) or overexpressing calnexin. No reduction in the death levels was observed in the Δpca1 background (Figure 6, A and B). Metacaspase activation was analyzed with the fluorescent probe FITC-VAD-FMK. We observed a slower kinetics of caspase-like activation in Δpca1 cells; however the level of final activation was similar to the pca1+ strain (Figure 6C). Therefore, apoptosis induced by calnexin overexpression is not absolutely dependent on Pca1p. Interestingly, these results demonstrate the presence of other, yet uncharacterized, caspase-like activities in S. pombe.
In mammalian cells, calnexin was demonstrated to associate with the transmembrane ER protein Bap31. Bap31 mediates apoptosis after its cleavage by caspase 8, and calnexin was proposed to act as a scaffold in this process (Delom et al., 2006, 2007 ). To assess if the homologue of Bap31 in S. pombe is required for apoptosis induction by calnexin overexpression, we measure death rate and metacaspase activation in cells deleted for the homologue of Bap31 (Δdam1/bap31). Overexpression of calnexin in Δdam1/bap31 strain caused death levels similar to that of the wild-type strain (Figure 6, A and B). Here again, a delay in metacaspase induction was observed in the Δdam1/bap31 strain, reaching the final levels similar that of the wild-type strain (Figure 6C). Therefore, Bap31 is not crucial in S. pombe to mediate apoptosis induced by calnexin overexpression.
To identify cellular factors involved in cell death mediated by calnexin, we carried out a genetic screen to isolate suppressors of the lethality caused by the overproduction of calnexin. Briefly, cells overexpressing calnexin were transformed with the pURSP1 S. pombe genomic bank, and 150,000 transformants were screened for survival, i.e., the ability to form colonies. Of 50 potential candidates isolated in the primary screen, one was an authentic suppressor (Figure 7A). This suppressor clone encoded Hmg1/2p, the homologue of the mammalian nuclear protein HMGB1. This protein is part of an HMG family that interacts with DNA for replication, transcription, and DNA repair. Importantly, human HMGB1 was reported as an antiapoptotic protein in mammalian and in fission yeast cells (Brezniceanu et al., 2003 ).
To investigate the suppressor effect of Hmg1/2p on death by calnexin overexpression, we synthesized its encoding gene (SPBC28F2.11) by PCR amplification and cloned it into an overexpression vector. Simultaneous co-overexpression of Hmg1/2p with calnexin significantly reduced the levels of cell death, as judged by spotting on inducing media and by Phloxin B staining (Figure 7, A and B). About 50% of cells stained positively with Phloxin B when calnexin was overproduced alone, compared with 20% of cells when was Hmg1/2p overproduced in concert (Figure 7B). Examination of the metacaspase activation revealed that overexpression of Hmg1/2p reduced the levels of induction of apoptosis (Figure 7C). About 45% of the cells overexpressing solely wild-type calnexin were positive for metacaspase activation, whereas only 10% of the cells co-overexpressing Hmg1/2p stained positively with the fluorescent marker FITC-VAD-FMK (Figure 7C). Western blotting showed that the suppression of cell death caused by the simultaneous overexpression of Hmg1/2p was not due to a reduction in the calnexin levels (data not shown). These observations indicate that Hmg1/2p has antiapoptotic function in S. pombe and that this protein may have a regulatory role in apoptosis involving calnexin.
S. pombe calnexin is a close sequence homologue of its human counterpart, especially when compared with the S. cerevisiae molecule (Jannatipour and Rokeach, 1995 ). Recent studies in mammalian cells have demonstrated the involvement of calnexin in the apoptosis induced by ER stress in mammals (Zuppini et al., 2002 ; Takizawa et al., 2004 ; Tomassini et al., 2004 ; Delom et al., 2006; Groenendyk et al., 2006 ; Delom et al., 2007 ). Because several studies reported interchangeability between mammalian and fission yeast proteins in apoptotic processes (Komatsu et al., 2000 ), we asked whether human calnexin could also induce apoptosis when overexpressed in S. pombe.
Indeed, the overproduction of human calnexin in S. pombe provoked cell death in ~30% of cells, as measured by Phloxin B staining (Figure 8A). Moreover, ~30% of cells overexpressing human calnexin exhibited metacaspase activation as measured with the fluorescent probe FITC-VAD-FMK. Although these levels of death and caspase activity are lower than those observed with wild-type S. pombe calnexin, these results argue for the conservation of the mechanisms of calnexin-mediated apoptotic cell death, at least in part, between human and S. pombe cells.
Cell survival under ER-stress conditions is largely dependent on the UPR pathway (Patil and Walter, 2001 ; Schroder and Kaufman, 2005 ; Marciniak and Ron, 2006 ; Zhao and Ackerman, 2006 ; Lai et al., 2007 ; Ron and Walter, 2007 ). However, if the overload caused by ER stressors cannot be counteracted, the UPR response switches from its prosurvival function to the signaling of apoptotic death. A major current question regarding the cell's response to ER stress is how the switch from prosurvival to prodeath is mechanistically decided. According to currents models, the mammalian UPR senses intralumenal stress through the three upstream ER transducers: IRE1, PERK, and ATF6; signaling via these stress receptors is regulated by the ER molecular chaperone BiP (Lee et al., 1999 ; Li and Lee, 2006 ; Ni and Lee, 2007 ). Nevertheless, it is unclear whether other upstream regulators of the mechanism ER-stress apoptosis do exist, and much remains to be elucidated about the downstream factors and interactions involved in the transduction of the cell-death signal.
Recent work showed that mammalian calnexin is required for the cleavage of Bap31 and thus for the generation of the proapoptotic p20 fragment under tunicamycin stress; these authors proposed that calnexin acts as a scaffold for Bap31 processing by caspase 8 (Zuppini et al., 2002 ; Delom et al., 2006; Groenendyk et al., 2006 ; Delom et al., 2007 ). These observations raise the possibility that calnexin could be involved in the early steps relaying the signal toward apoptotic death initiated by overwhelming ER stress.
Here, we demonstrate that the overexpression of calnexin in S. pombe induces cell death with typical apoptotic features including early death, phosphatidyl serine exposure, metacaspase activation, ROS production, nuclear fragmentation, and DNA breakage. This effect is specific to calnexin and is not due to overloading of the ER capacity, because overproduction of the ER proteins PDI or Sec61β did not induce the apoptotic-death phenotypes and is not due to the loss of ER membrane integrity, as confirmed by microsome analysis.
It has been previously reported that the expression of mammalian Bak in the fission yeast is lethal and that Bak requires interactions with S. pombe calnexin to mediate this cell-death phenotype (Torgler et al., 1997 ). The authors proposed that the interaction between Bak and calnexin results in a dominant lethal effect due to the propagation of a death signal or the recruitment of additional interactors to the Bak–calnexin complex (Torgler et al., 1997 ). Our results show that when overexpressed, calnexin induces apoptosis without exogenous factors or ER-stress conditions. In general, overexpression causes a bypass in the regulation of the pathway studied (Ramer et al., 1992 ). This suggests that calnexin overexpression may disturb the control of an apoptotic pathway in which calnexin takes part. A possibility is that the overexpression of calnexin mimics conditions of ER stress in otherwise resting cells.
Truncation of the TM and cytosolic tail of calnexin abolished the apoptotic death caused by overexpression. Using various deletion mutants, we demonstrated that the anchoring of calnexin to the ER membrane is required to induce apoptosis by overexpression. Although the overexpression of the cytosolic tail anchored to the membrane induced cell death, the strongest apoptotic effect was observed by overproducing a calnexin mutant spanning the lumenal domain with the TM. These results may imply that residues on both the intralumenal portion and the cytosolic tail of calnexin could play roles in the induction of apoptosis and that the sequences on either side of the TM may participate in different prodeath interactions. Supporting the importance of the cytosolic tail and the TM, Torgler et al. (1997) reported that expression of human Bak was lethal in an S. pombe strain expressing wild-type calnexin but not in a strain expressing a mutant calnexin lacking both the TM and the cytosolic tail. Importantly, our results demonstrate that sequences within the lumenal domain of calnexin are also involved in proapoptotic interactions. The requirement of the TM to elicit apoptotic death by overexpression suggests that the anchoring of calnexin to the ER membrane may be required for its interaction with some key partner(s) for the assembly of a lethal complex.
Our observations provide first direct evidence that altering the levels of calnexin causes death with apoptotic features. That its overexpression induces programmed cell death is indicative that calnexin could play a role in apoptotic processes induced by “natural” ER stresses. An increase in the calnexin levels due to ER stress may constitute a part of a branch in the mechanism of induction of apoptotic death. Tunicamycin is a potent inhibitor of N-glycosylation that is currently used as an elicitor of ER stress in the study of the apoptosis mechanisms in mammalian cells and in the budding yeast (Perez-Sala and Mollinedo, 1995 ; Hacki et al., 2000 ; Hauptmann et al., 2006 ; Hauptmann and Lehle, 2008 ). We have previously reported that tunicamycin induces ER stress in S. pombe and increases the expression of cnx1+, the gene encoding calnexin (Jannatipour and Rokeach, 1995 ). In this work, we show that also in the case of S. pombe, tunicamycin provokes cell death with typical apoptosis markers.
Using S. pombe strains expressing calnexin mutants at basal levels, we demonstrated that apoptotic cell death induced by tunicamycin is significantly less efficient in cells expressing calnexin without its TM. Cells expressing only the lumenal portion of calnexin exhibited a 50% reduction in apoptotic cell death due to exposure to tunicamycin. Our observations implicate calnexin in apoptosis caused by ER stress. In agreement with our results, calnexin-deficient rodent cells are relatively resistant to apoptosis induced by ER stress (Zuppini et al., 2002 ; Groenendyk et al., 2006 ). However, it has been reported that siRNA inhibition of calnexin expression increases the sensitivity to tunicamycin of apoptosis-resistant human-breast carcinoma MCF-7 cells (Delom et al., 2006). This disparity in the results could be due to differences in the cell lines used in these studies. Nevertheless, transfection of a ΔE calnexin mutant lacking most of the lumenal domain into MCF-7 cells restored the sensitivity to tunicamycin-induced apoptosis (Delom et al., 2006). Together, these observations clearly implicate calnexin in ER-stress–induced apoptosis, in mammalian and in S. pombe.
S. pombe encodes several homologues of proteins characterized for their implication in apoptosis including the thus far only metacaspase identified Pca1p (geneID: SPCC1840.04) and a Bap31 homologue (geneID: SPAC9E9.04). Our results demonstrate that the knockout of these genes does not affect the final level of apoptosis provoked by calnexin overexpression; however, we observed a slower kinetics of induction. Likewise, induction of apoptosis by tunicamycin was not blocked in the Δpca1 and Δdma1/bap31 strains (data not shown). Importantly, our results clearly demonstrate that another, yet uncharacterized, caspase-like activity is involved in this apoptotic process. Similarly, the homologue of Pca1p in S. cerevisiae, Yca1p, is not required for apoptotic death induced by numerous conditions (Madeo et al., 2002 ; Bettiga et al., 2004 ; Fannjiang et al., 2004 ; Herker et al., 2004 ; Wadskog et al., 2004 ; Ivanovska and Hardwick, 2005 ; Reiter et al., 2005 ; Liang et al., 2008 ; Mazzoni and Falcone, 2008 ). More recently, it was reported that Yca1p is not required for apoptosis induced by tunicamycin in S. cerevisiae (Hauptmann and Lehle, 2008 ). These observations suggest the existence of uncharacterized caspase-like activities in both, S. pombe and S. cerevisiae. That cell death occurs by calnexin overexpression in the Δdma1/bap31 background does not exclude the possibility that Bap31 could be part of a complex involving calnexin, but it is indicative that the role of Bap31 is not essential for apoptosis. In this vein, it was shown in mammalian cells that an uncleavable mutant of Bap31 does not abrogate the capacity to the cells to enter FAS-mediated apoptosis but it delays some cytoplasmic apoptotic events (Nguyen et al., 2000 ).
The significant reduction in the levels of tunicamycin-induced apoptosis observed with the lumenal_cnx1 mutant could be due to a diminution or a loss of effective interactions with lethal partners as observed for mammalian cells (Zuppini et al., 2002 ; Delom et al., 2006; Groenendyk et al., 2006 ; Delom et al., 2007 ). Ire1p could be taking part of this interaction since its presence is required to attain maximum apoptosis induction by calnexin overexpression. Further experiments are required to dissect the molecular interaction of calnexin with Ire1p and other lethal partners. Such interactions probably require the anchoring of calnexin to the ER membrane to be optimal in transducing the death signal. Interestingly, the lumenalTM_cnx1 strain exhibited the highest levels of apoptosis induced by tunicamycin. These, and our results obtained by overexpression experiments implicate lumenal sequences of calnexin in apoptosis. However, it appears that the chaperone function of calnexin is distinct from its role in apoptosis since the mutant lumenal_Cnx1p is a very effective chaperone (Marechal et al., 2004 ) but a poor effector of tunicamycin-induced apoptosis. Further supporting this point, mini_Cnx1p does not exhibit chaperone activity (Marechal et al., 2004 ) but is highly efficient in inducing apoptosis by overexpression, and mini_cnx1 cells display levels of tunicamycin-induced apoptosis as the wild-type strain.
BiP anchoring to the ER membrane was reported in mammalian cells subjected to ER stress (Rao et al., 2002 ; Reddy et al., 2003 ). We have previously reported that the last 52 residues of the lumenal portion of calnexin are required for viability and do interact with ER chaperone BiP (Jannatipour et al., 1998 ; Beaulieu et al., 1999 ; Elagoz et al., 1999 ; Marechal et al., 2004 ). Because BiP regulates the ER stress receptors IRE1, PERK, and ATF6, it is therefore tempting to speculate that BiP could also play a regulatory role in the proapoptotic function of calnexin (Patil and Walter, 2001 ; Szegezdi et al., 2006 ; Ni and Lee, 2007 ; Ron and Walter, 2007 ). Our coimmunoprecipitation experiments showed that tunicamycin treatment enhances Cnx1p-BiP interaction in both the wild-type calnexin and the mutant lumenal_Cnx1p. However, apoptosis is significantly reduced in the case of the lumenal_Cnx1p mutant. These observations raise the possibility that stronger interaction between Cnx1p and BiP is required for ER-stress–induced apoptosis and that this complex needs to be anchored to the ER membrane to efficiently transduce the apoptotic signal. In such scenario, the diminution of tunicamycin-induced apoptosis in lumenal_Cnx1p cells could be due to a mislocalization of the Cnx1p-BiP complex, because this complex is not localized to the ER membrane in the case of this mutant.
In a genetic screen designed to isolate suppressors of lethality caused by calnexin overexpression we identified Hmg1/2p, the S. pombe homologue of HMGB1. The HMGB1 protein is part of the HMG family that interacts with DNA for replication, transcription, and DNA repair and is a protein overexpressed in human breast carcinoma (Brezniceanu et al., 2003 ; Ulloa and Messmer, 2006 ). Importantly, HMGB1 is an antiapoptotic protein that was identified as an inhibitor of Bak-mediated cell death in mammalian cells and in S. pombe (Brezniceanu et al., 2003 ). In this work we showed that like its mammalian counterpart, Hmg1/2p inhibits the ER-derived apoptotic pathway involving calnexin. Because Bak-induced apoptotic death in S. pombe is dependent on calnexin (Torgler et al., 1997 ) and is inhibited by HMGB1 and because Hmg1/2 represses death by overexpression of calnexin in S. pombe, we hypothesize that the apoptotic pathway involving calnexin are similar in S. pombe and in mammalian cells. Although no sequence homologue of Bak was described in the fission yeast, it is likely that some S. pombe protein plays an analogous function to that of mammalian Bak. The discovery of this lethal partner will be important for the elucidation of the apoptotic processes caused by ER stress in S. pombe and are likely to contribute to our understanding of the mechanism of apoptosis in mammals.
Human calnexin and S. pombe calnexin are very similar in their structure (Jannatipour and Rokeach, 1995 ). Here we report that like S. pombe calnexin, the overexpression of its human orthologue in the fission yeast induces cell death with typical apoptotic features. These results indicate that human calnexin interacts with S. pombe apoptotic machinery and overrides the mechanisms that regulate apoptosis involving endogenous calnexin. Collectively, these observations argue for the conservation between the mammalian and the S. pombe mechanisms of apoptosis involving calnexin and validate S. pombe as a model organism.
Most of our current knowledge on the apoptotic response due to ER stress derives from studies with mammalian cells (Breckenridge et al., 2003 ; Xu et al., 2005 ; Kim et al., 2006 ; Szegezdi et al., 2006 ). Evidence accumulated in the last 10 years has proven that yeasts are interesting models to investigate the mechanism of apoptosis (Ink et al., 1997 ; Madeo et al., 1997 ; Ligr et al., 1998 ; Madeo et al., 1999 ; Madeo et al., 2002 ; Priault et al., 2003 ; Hardwick and Cheng, 2004 ; Madeo et al., 2004 ; Rodriguez-Menocal and D'Urso, 2004 ; Burhans and Weinberger, 2007 ; Frohlich et al., 2007 ; Almeida et al., 2008 ). Although human and S. pombe calnexin share significant sequence similarity (Jannatipour and Rokeach, 1995 ), the S. cerevisiae molecule (Cne1p; Parlati et al., 1995b ) is the most distant known homologue. Thus, the study of the role of calnexin in the mechanisms of ER-stress apoptosis in S. pombe should bring crucial information regarding this cellular process in mammals.
We thank Dr. Michael Brenner for the gift of the plasmid bearing the human calnexin cDNA, and Dr. Anthony car for supplying the pURSP1 S. pombe genomic library. We also thank Dr. Gerardo Ferbeyre and the members of the Rokeach Lab for critical reading of the manuscript. This work was supported by research grants from CIHR (Grant MOP-62703) and NSERC (Grant 171325) to L.A.R. R.G. was the recipient of FES-Département de biochimie studentships.
This article was published online ahead of print in MBC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E08-02-0188) on August 13, 2008.