Palmitoylation is the switch that assigns calnexin to quality control or ER Ca²⁺ signaling Free

The main functions of the endoplasmic reticulum (ER) are the production of secretory and membrane proteins as well as lipids, and the storage of Ca²⁺. An integral requirement for these functions is the interaction of the ER with mitochondria, which occurs on the mitochondria-associated membrane (MAM), a subdomain of the ER that makes close contacts between the ER and the mitochondria (Raturi and Simmen, 2013; Simmen et al., 2010). Here, the ER exchanges Ca²⁺ with mitochondria through the ER Ca²⁺-handling inositol (1,4,5)-triphosphate receptors (IP3R) and the sarcoendoplasmic reticulum (SR) Ca²⁺ transport ATPase (SERCA) (Rizzuto et al., 2009). Under resting conditions, Ca²⁺ delivery from the ER to the mitochondrial matrix is needed to activate the mitochondrial enzyme pyruvate dehydrogenase (PDH), which drives the tricarboxylic acid (TCA) cycle (Cárdenas et al., 2010). Given the crucial role that mitochondrial metabolism plays in death and survival of the cell (Glancy and Balaban, 2012; Zhivotovsky and Orrenius, 2011), ER–mitochondria Ca²⁺ exchange has to be regulated at the source. Various ER chaperones and oxidoreductases perform this function by dictating Ca²⁺ channels and pumps to open or close dependent on ER redox and Ca²⁺ homeostasis (Simmen et al., 2010). Examples are Ero1α, which activates IP3Rs (Li et al., 2009), and ERp44, which inhibits IP3Rs (Higo et al., 2005).

Calnexin is another key ER chaperone that binds to nascent glycoproteins (Rutkevich and Williams, 2011). This binding slows down client protein folding and prevents their aggregation owing to retention of folding intermediates in the ER. Upon release from calnexin, glycoproteins can follow one of three pathways: they might fold rapidly and correctly, in which case they will be exported to the Golgi apparatus, or if they are not yet correctly folded they can be re-glucosylated and returned to the calnexin cycle (Ruddock and Molinari, 2006). Proteins that have been trapped in the calnexin cycle for a prolonged period are eventually subject to extensive mannose trimming by mannosidase I and ER-associated degradation enhancing α-mannosidase-like protein (EDEM), followed by retrotranslocation from the ER to the cytosol and degradation by the proteasome, a process known as ER-associated degradation (ERAD) (Lederkremer, 2009). All along, calnexin works in partnership with ERp57, an oxidoreductase that is a member of the protein disulfide isomerase (PDI) family (Coe and Michalak, 2010; Oliver et al., 1997; Zapun et al., 1998). Studies have shown that a certain subset of disulfide-bonded heavily glycosylated proteins such as the low-density lipoprotein receptor (LDLR) depend specifically on this partnership for efficient folding and subsequent trafficking through the secretory system (Jessop et al., 2007; Rutkevich and Williams, 2011).

Surprisingly, our laboratory and others have determined that calnexin is not only found on the rough ER, but is often enriched on the MAM (Hayashi and Su, 2007; Myhill et al., 2008; Wieckowski et al., 2009). This finding further reinforces the notion that calnexin is a central player for the regulation of ER Ca²⁺ signaling, and is also consistent with the fact that calnexin interacts with the SERCA Ca²⁺ pump (Roderick et al., 2000). Interestingly, the role of calnexin in Ca²⁺ signaling could depend on ER homeostasis, given that ER stress results in the relocation of calnexin towards the ER quality control compartment (ERQC) and its association with slowly folding proteins (Frenkel et al., 2004).

Therefore, we have examined whether the targeting of calnexin to different ER subdomains correlates with an ER-stress-dependent role in ER Ca²⁺ signaling towards mitochondria. We further examined whether such a role correlated with the extent of binding of calnexin to its known partners ERp57 and SERCA2b. Calnexin targeting to the MAM depends on a juxtamembrane palmitoylation motif (Lynes et al., 2012) that is implicated in mediating the interaction of calnexin with the translocon (Lakkaraju et al., 2012) and with lipid rafts on the plasma membrane (Ferrera et al., 2008). In addition to calnexin, palmitoylation is known to modify several ER proteins, including heme oxygenase-1 and ORAI1, but the consequences of this modification are currently unknown (Dowal et al., 2011; Kang et al., 2008). We had shown earlier that the interference with palmitoylation relocates heme oxygenase-1 (Lynes et al., 2012), but ER-associated palmitoylation might have multiple diverse effects (Lakkaraju et al., 2012). To investigate the functional significance of palmitoylation-dependent targeting of calnexin, we characterized the functional consequences of the palmitoylation state of calnexin for the cell in terms of cellular Ca²⁺ signaling and quality control of the known calnexin substrates LDLR and asialoglycoprotein receptor (ASGPR).

To test whether ER stress influences the function of calnexin in ER Ca²⁺ signaling and chaperoning, we first examined the localization of calnexin following ER stress by biochemical fractionation. We decided to separate ER membranes into light and heavy membranes. Using this technique, we detected a shift of calnexin towards light membranes that started after 1 h of tunicamycin treatment and increased at 4 h (Fig. 1A). No such shift was observed for PDI or BAP31 (data not shown), two other ER proteins. To more precisely determine the origin and destination of calnexin during short-term ER stress, we used a biochemical fractionation technique that we had developed to separate MAM markers from other ER markers (Myhill et al., 2008). HeLa cells were treated with two different ER stressors, thapsigargin and tunicamycin, for 4 h, then homogenized and fractionated using Optiprep density gradient medium (Fig. 1B; supplementary material Fig. S1, Fig. S2A). The general ER marker PDI and MAM-localized long-chain acyl-CoA ligase 4 (FACL4) were used to control for the sedimentation of the MAM compartment. Under control conditions, close to 40% of calnexin co-fractionates with the MAM marker FACL4 and with the mitochondria at the bottom of the Optiprep gradient, as previously shown by our laboratory and others. However, after treatment with ER stressors, the association of calnexin, but not of PDI or FACL4 with MAM fractions was diminished to less than 20%. The relocation of calnexin did not lead to increased targeting to the plasma membrane (data not shown). Immunofluorescence microscopy similarly suggested reduced apposition between calnexin and mitochondria (Fig. 1C). After 4 h of treatment with tunicamycin, a decrease in the overlap of the calnexin and mitotracker signals was observed (yellow signal, inset box) in HeLa cells labeled with antibodies against calnexin and PDI, as well as mitotracker. This result is consistent with a decrease in the localization of calnexin at the MAM. At first glance, the relocation of calnexin away from mitochondria during ER stress appears surprising, given earlier reports by us and others that this condition increases visible contacts between ER and mitochondria (Bravo et al., 2011; Csordás et al., 2006). At the same time, however, this movement of calnexin away from the MAM suggests a specific function that is not due to known structural changes in the ER.

Next, we aimed to identify the mechanism that could lead to this relocation of calnexin. Because the phosphorylation state of calnexin is important for its localization within the ER and its binding to PACS-2 (Myhill et al., 2008), we tested the status of phosphorylation for each known site during an ER stress timecourse. Our results ruled out that a change in phosphorylation is behind the relocation of calnexin, given that only long-term, but not short-term, stress led to a reduction of calnexin phosphorylation on serine 565, which controls PACS-2 binding, and serine 584 (Fig. 2A), which controls the interaction of calnexin with ribosomes and SERCA2b (Chevet et al., 1999; Roderick et al., 2000). Another explanation for the change in localization of calnexin during short-term ER stress could be that calnexin loses its palmitoylation, known to be responsible for its MAM targeting (Lynes et al., 2012). Thus, we used a ‘click chemistry’ approach to compare the calnexin palmitoylation signal from untreated cells to that of cells treated with tunicamycin or DTT for 4 h to induce a short-term ER stress (Fig. 2B; supplementary material Fig. S2B). We observed that both ER stress situations indeed led to calnexin depalmitoylation. Therefore, in contrast to long-term ER stress (Delom et al., 2007), short-term ER stress results in calnexin depalmitoylation, but not dephosphorylation, together with a loss of MAM enrichment.

Next, we aimed to test whether the relocation of calnexin within the ER that occurs in parallel with its depalmitoylation also affects its key interactors, because such an observation could give clues as to what palmitoylation of calnexin is necessary for. Because calnexin interacts with SERCA2b (Roderick et al., 2000), we decided to first test whether treatment with ER stressors also affects SERCA2b. Thus, we examined the distribution of SERCA2b on our Optiprep gradient, as well as on heavy versus light membranes. Whereas calnexin moved away from the MAM and heavy membranes of the ER, we could, however, not detect any movement of SERCA2b under these conditions (Fig. 3A). This suggested that calnexin and SERCA2b undergo different fates during ER stress.

To test this hypothesis, we immunoprecipitated calnexin from control HeLa cells and HeLa cells treated with either DTT or tunicamycin for 4 h. Under these ER stress conditions, we found that the interaction of calnexin with SERCA2b decreased significantly (Fig. 3B). The replacement of the two cysteine residues of calnexin that are subject to palmitoylation with alanine residues (cysteine 503 and 504 in dog calnexin, CCAA) resulted in a similar reduction of the SERCA2b–calnexin interaction (Fig. 3C). Calnexin phosphorylation on serine 584 had previously been implicated in the regulation of this interaction with SERCA2b (Roderick et al., 2000). However, we could not detect reduction of phosphorylation on this residue within the time frame that led to reduction of the SERCA2b–calnexin interaction. Consistent with these findings, we could also not detect reduced phosphorylation of calnexin in a mutant where we had replaced the two membrane-proximal cysteine residues with alanine residues (CCAA), resulting in non-palmitoylatable calnexin (Lynes et al., 2012) (supplementary material Fig. S2C). Taken together, our results demonstrate that short-term ER stress results in the depalmitoylation of calnexin, but not its dephosphorylation. Calnexin depalmitoylation subsequently results in reduced association of calnexin with SERCA2b.

Our results suggested that palmitoylated calnexin might regulate ER Ca²⁺ signaling. If this were correct we would expect palmitoylated calnexin to regulate the flux of Ca²⁺ between the ER and mitochondria through its role on SERCA (Roderick et al., 2000). Thus, we performed an extensive series of Ca²⁺ assays using calnexin-knockout mouse embryonic fibroblasts (MEFs) (Kraus et al., 2010) that we transfected with wild-type and non-palmitoylatable calnexin. First, we inhibited SERCA Ca²⁺ pumps with thapsigargin and measured the increase of cytosolic Ca²⁺ with Fluo8. We were unable to detect any differences in the accumulation of cytosolic Ca²⁺ with this assay, regardless of whether palmitoylated or non-palmitoylated calnexin was present or not (Fig. 4A). No change in the accumulation of cytosolic Ca²⁺ was observed using FURA-2 either (supplementary material Fig. S3A). These results were consistent with earlier reports (Zuppini et al., 2002). However, we decided to measure ER Ca²⁺ at the source using ER-targeted aequorin (Alvarez and Montero, 2002). Surprisingly, our measurements showed ∼80% more Ca²⁺ in MEFs transfected with wild-type calnexin, but not with calnexin that cannot be palmitoylated (Fig. 4B). Because these cells did not show altered accumulation of cytosolic Ca²⁺ from ER stores, our findings suggested that mitochondrial Ca²⁺ uptake could also be different in cells expressing calnexin compared with that in calnexin-knockout MEFs (Arnaudeau et al., 2001; de Brito and Scorrano, 2008). Therefore, we tested whether mitochondria show differences in Ca²⁺ uptake following the inhibition of ER Ca²⁺ pumps. Indeed, we found that mitochondria in wild-type MEFs take up significantly less Ca²⁺ than calnexin-knockout MEFs (supplementary material Fig. S3B). Moreover, our results with the mitochondrial Ca²⁺ indicator dye Rhod2 in the presence of thapsigargin show that the presence of wild-type calnexin dampened mitochondrial Ca²⁺ uptake following thapsigargin administration, when compared with that in calnexin-knockout MEFs (Fig. 4C). Importantly, this effect was not seen using non-palmitoylated calnexin. Because our results indicate that the presence of palmitoylated calnexin reduced the ability of mitochondria to import Ca²⁺ from the ER, we next tested whether this was also true under the condition of ER stress, when cells normally increase mitochondrial Ca²⁺ import from ER sources (Bravo et al., 2011; Csordás et al., 2006). Our results show that calnexin-knockout MEFs were unable to boost the import of IP3R-released Ca²⁺ into mitochondria after a 4 h tunicamycin stress. However, the transfection of wild-type calnexin, but not of non-palmitoylatable calnexin rescued this deficiency (Fig. 4D), thus making calnexin-knockout MEFs stress-responsive again. Taken together, our results suggest that calnexin must be palmitoylatable to control the activity of SERCA and thus the rate of Ca²⁺ import into mitochondria.

Given that efficient Ca²⁺ transfer from the ER to mitochondria is a prerequisite for the mitochondrial membrane potential (Cárdenas et al., 2010), we next tested the hypothesis that calnexin-knockout cells could exhibit an abnormally high proton gradient across the mitochondrial membranes owing to their high constitutive Ca²⁺ transfer from the ER to mitochondria. By labeling calnexin-knockout cells and their counterparts transfected with wild-type calnexin and non-palmitoylatable calnexin with tetramethylrhodamine methyl ester (TMRM), we were indeed able to confirm this hypothesis by demonstrating that only wild-type calnexin transfection resulted in a specific reduction of the mitochondrial membrane potential (supplementary material Fig. S3C). This was also the case in the presence of KCl, which equilibrates plasma membrane uptake of TMRM (data not shown). Throughout our experiments, we verified that expression of the two calnexin constructs was even (Fig. 4E) and that our transfections did not result in the induction of the unfolded protein response (supplementary material Fig. S3D). Therefore, ER–mitochondria Ca²⁺ transfer is more efficient in the absence of calnexin, but the boosted Ca²⁺ transfer observed under ER stress requires palmitoylated calnexin and increases the ER Ca²⁺ content.

In addition to its interaction with SERCA2b, calnexin interacts with the oxidoreductase ERp57 in the ER to mediate protein folding. This role of calnexin is particularly important for slow-folding substrates (Frenkel et al., 2004). Thus, we decided to also test whether the interaction between calnexin and ERp57 is similarly affected by ER stress. First, we examined the distribution of calnexin and ERp57 by Optiprep gradient and fractionation of cellular membranes into heavy and light membranes. In contrast to SERCA2b (Fig. 3A), the distribution of ERp57 closely matched that of calnexin and both turned out to move from heavy to light membranes following an ER stress (Fig. 5A), suggesting that these two proteins might share functions under ER stress. Indeed, we observed that when cells are treated with either DTT or tunicamycin for 4 h, the amount of interaction between calnexin and ERp57 increased (Fig. 5B). ER stress could lead to the misfolding of calnexin itself, which could then trigger its association with other chaperones. To test this possibility, we expressed FLAG-tagged calnexin and the mutant calnexin that cannot be palmitoylated. As shown in Fig. 5C, the calnexin CCAA mutant that cannot be palmitoylated indeed interacted with ERp57 almost twice as well. Likewise, the binding of calnexin to ERp57 was approximately doubled in the presence of 2-bromopalmitate, an inhibitor of palmitoylation (supplementary material Fig. S4A).

Our results suggested that the change in the interaction between calnexin and ERp57 that was dependent on palmitoylation is of a functional nature. We therefore sought to identify a folding substrate that is particularly dependent on the calnexin–ERp57 folding pathway. For this purpose, we first chose the LDLR, which is known to undergo protein folding by forming mixed disulfides with ERp57 and depend on calnexin folding (Jessop et al., 2007). To determine whether LDLR maturation proceeds differently when calnexin is palmitoylated or not, we used the calnexin-knockout MEFs that we transfected with wild-type calnexin or the calnexin CCAA mutant. Despite the expression level of the LDLR in cell lysates being unaltered (Fig. 5D), we found that the amount of LDLR on the cell surface increased when wild-type calnexin was expressed, but decreased when CCAA calnexin was expressed (Fig. 5D), suggesting that non-palmitoylated calnexin is indeed more stringent in ER quality control retention than the at least partially palmitoylated wild-type calnexin.

Efficient intracellular retention of LDLR by non-palmitoylated calnexin could indicate that this form of calnexin interacts with unfolded domains of proteins that await exit from the ER or, alternatively, ER-associated degradation (ERAD). Thus, we analyzed the interaction of calnexin with an ERAD substrate, the uncleaved precursor of asialoglycoprotein receptor (ASGPR) H2a, which is normally completely retained in the ER (Kamhi-Nesher et al., 2001; Shenkman et al., 1997). Calnexin CCAA interacted much more robustly than wild-type CNX with H2a (Fig. 6A). Calnexin CCAA also caused a slight increase in total H2a, probably because of decreased targeting of this protein for ERAD. Pulse-chase analysis showed that dissociation of H2a from calnexin followed a different pattern when we transfected nonpalmitoylatable calnexin compared with upon transfection of wild-type calnexin (Fig. 6B). Whereas wild-type calnexin dissociated efficiently with time from the ERAD substrate H2a, non-palmitoylatable CCAA calnexin reassociated and remained in a complex with H2a for extended periods of time. For the pulse samples, two bands can be seen for H2a precursor molecules, the lower one corresponding to an underglycosylated species (one of the glycosylation sites unoccupied). As we had shown previously (Frenkel et al., 2004), the fully glycosylated species shifts progressively to a faster migration because of the trimming of mannose residues on its N-glycans, whereas the underglycosylated lower species is quickly degraded.

We next tested whether palmitoylation influences calnexin targeting to the pericentriolar ERQC, where calnexin and ERAD substrates like H2a accumulate upon inhibition of their degradation or under ER stress (Frenkel et al., 2004; Kamhi-Nesher et al., 2001; Kondratyev et al., 2007). Thus, we compared the localization of the non-palmitoylatable mutant with wild-type calnexin. Whereas in untreated cells, both proteins showed a disperse ER pattern, upon short-term proteasomal inhibition for 3 h, calnexin CCAA showed a more pronounced relocation to the juxtanuclear ERQC and colocalization with H2a linked to monomeric RFP (H2a–RFP), according to the Manders coefficient (Fig. 6C–F). Taken together, our results demonstrate the preferential interaction of non-palmitoylated calnexin with ERAD substrates, which extends to the ERQC, where dissociation from the misfolded glycoprotein would require the ability of calnexin to undergo palmitoylation.

The results presented herein uncover a mechanistic description of the role of calnexin palmitoylation during ER stress. We demonstrate that reversible palmitoylation assigns calnexin to the interaction with SERCA2b and a role in the regulation of Ca²⁺ transfer to mitochondria, whereas non-palmitoylated calnexin tends to interact with ERp57 to fulfill its role in protein folding and quality control (Fig. 7). Our results therefore illustrate for the first time how palmitoylation can restrict calnexin to either role, and we propose that palmitoylation is a key mechanism of ER stress signaling. Our observations identify palmitoylation as a quick signal that reassigns calnexin to specific tasks within the ER, unlike phosphorylation, which determines calnexin interaction with the ribosome, and another ER chaperone, BAP31, in a less dynamic way. Changes in calnexin phosphorylation require significantly longer exposure to ER stress when compared to the changes in palmitoylation (Fig. 2). Moreover, whereas the calnexin phosphorylation state results in intra-ER relocations mediated by the cytosolic sorting protein PACS-2 (Myhill et al., 2008), palmitoylation might regulate the calnexin localization by allowing (or not) its interaction with its effectors SERCA2b and ERp57, which are localized to sites of Ca²⁺ signaling and protein folding, respectively. Interestingly, calnexin appears to follow a targeting pattern that is probably relatively unique, since ER and mitochondria move closer to each other during ER stress, suggesting an enrichment of ER proteins at the MAM or a tightening of ER–mitochondria contacts (Bravo et al., 2011; Csordás et al., 2006).

So far, the interaction of calnexin with its substrates has been explained from the substrate's point of view, given that calnexin binds exclusively monoglucosylated substrates (Lederkremer, 2009). This property, together with the enzymatic action of UDPGlc:glycoprotein glucosyltransferase and glucosidases I and II leads to a cyclic interaction of folding substrates with calnexin until proper folding is reached or ERAD is triggered. Our new data now indicate that calnexin itself undergoes a palmitoylation-dependent cycle within the ER and shuttles back and forth from regulating Ca²⁺ signaling at or close to the MAM and mediating protein folding and quality control at or close to the rough ER, including the targeting of unfolded or misfolded proteins to the ERQC (Leitman et al., 2013). Interestingly, these results could help explain the anti-stress function of 2-bromopalmitate (2BP) treatment, which attenuates the induction of ER stress transcription factors and the progression of apoptosis (Baldwin et al., 2012). In this scenario, calnexin would become more efficient as a chaperone and would hence mitigate the effects of any additionally introduced ER stressor. Consistent with such an idea, we observed that the inhibition of palmitoylation with 2BP leads to a 2-fold increase in the binding of ERp57 with calnexin (supplementary material Fig. S4A), increases the association with H2a and reduces the amount of mature LDLR on the surface (supplementary material Fig. S4B). These findings, together with the decreased amount of surface LDLR upon expression of CCAA calnexin, suggest that palmitoylation facilitates the dissociation of calnexin from folding intermediates.

Accordingly, palmitoylated calnexin appears to be less important for protein folding, consistent with its predominant localization to the MAM (Lynes et al., 2012). However, our results also reinforce the close relationship between ER folding assistants and the control of mitochondrial metabolism (Simmen et al., 2010). In our specific example, calnexin expression and its reversible palmitoylation are required to allow the ER to properly signal via Ca²⁺ towards mitochondria. It is apparently owing to the presence of palmitoylated calnexin that the ER can signal a state of stress to mitochondria, probably to alleviate the accumulation of unfolded proteins within the ER (Bravo et al., 2011). However, this function is also crucial for mitochondria metabolism, as suggested by the abnormally high mitochondria membrane potential in the absence of palmitoylated calnexin. Surprisingly, our results contradict previous reports that calnexin acts as an inhibitor of SERCA2b in Xenopus oocytes: direct measurement of ER Ca²⁺ with aequorin suggests that the basis of our effects is an increased activity of SERCA2b in the presence of palmitoylatable calnexin, but not of the non-palmitoylated calnexin CCAA mutant (Fig. 4B). A potential explanation for this discrepancy could lie in special properties of Xenopus oocytes or in the determination of the influence of calnexin on SERCA2b via the measurement of cytosolic Ca²⁺ waves following calnexin overexpression, as used in the earlier study (Roderick et al., 2000), rather than direct measurements of ER Ca²⁺ or assays of ER–mitochondria Ca²⁺ cross-talk.

Despite this rather marked difference in ER Ca²⁺ content, we have, however, not been able to detect differences in the induction of the unfolded protein response that are dependent on the presence or absence of calnexin (supplementary material Fig. S3D). This is consistent with earlier findings on calnexin-knockout cells (Coe et al., 2008). Rather, our findings confirm the paramount role the ER plays in the regulation of mitochondria metabolism, as elegantly demonstrated by Foskett and colleagues (Cárdenas et al., 2010).

Although the enzyme that palmitoylates calnexin has been identified as DHHC6 (Lakkaraju et al., 2012), our results predict that depalmitoylation enzymes of the thioesterase family are probably also important regulators of the ER stress response (Baekkeskov and Kanaani, 2009). However, because the APT inhibitor palmostatin B (Rusch et al., 2011) did not influence our observations (data not shown), it has to be yet unknown thioesterases that mediate calnexin shuttling between the MAM and the ER. Given the demonstrated efficacy of interference with the ER stress response in various clinical applications (Wang and Kaufman, 2012), interference with ER protein palmitoylation now provides a new therapeutic avenue with new characteristics and opportunities.

All chemicals were from Sigma (Oakville, ON) except Optiprep (Axis Shield, Norton, MA) and lactacystin (EMD Millipore, Billerica, MA). Rainbow [¹⁴C]-labeled methylated protein standards were obtained from GE Healthcare (Little Chalfont, Buckinghamshire, UK). Promix cell labeling mix ([³⁵S]Met plus [³⁵S]Cys), (>1000 Ci/mmol) was from PerkinElmer Life and Analytical Sciences (Boston, MA). Protein-A–Sepharose was from Repligen (Needham, MA). The antibodies have been purchased as follows: mouse anti-ERp57 (StressMarq, Victoria, BC), mouse and rabbit anti-FLAG (Rockland, Gilbertsville, PA; Sigma, Oakville, ON), mouse anti-SERCA2b (EMD Millipore, Billerica, MA), mouse anti-tubulin (Sigma, St Louis, MO), rabbit anti-LDLR (Biovision, Milpitas, CA), goat anti-FACL4 (Abcam, Cambridge, UK), mouse anti-PDI (Thermo-Pierce, Rockford, IL), mouse anti-complex-2 (Mitosciences, Eugene, OR), rabbit anti-calnexin antibody (Lynes et al., 2012). The anti-phospho-calnexin antibodies were from Abcam (P563 dog, P583 human, respectively) or generated by 21st Century Biochemicals (Marlboro, MA) on behalf of us (P534, P544 dog). Goat anti-mouse IgG conjugated to agarose was from Sigma. Goat anti-mouse and anti-rabbit secondary fluorescent antibodies were from Life Technologies (Carlsbad, CA). The rabbit polyclonal anti-H2a carboxyterminal antibody was as used in earlier studies (Tolchinsky et al., 1996). Goat anti-mouse IgG antibody conjugated to Cy2 was from Jackson Labs (West Grove, PA).

HeLa, mouse embryonic fibroblasts (MEFs) and human embryonic kidney (HEK) 293 cells were grown in DMEM plus 10% fetal calf serum (FCS) and NIH 3T3 cells in DMEM plus 10% new born calf serum. All cells were grown at 37°C under an atmosphere of 5% CO₂. HeLa cells were transiently transfected with Metafectene (Biontex, Martinsried, Germany). Transient transfection of NIH 3T3 cells was performed using an MP-100 Microporator (Digital Bio) according to the manufacturer's instructions. Transient transfection of HEK 293 cells was performed according to the Ca²⁺ phosphate method. The experiments were performed 24–48 h after transfection.

Cells were washed twice with cold PBS with Ca²⁺ and Mg²⁺ (PBS++) and then incubated with 0.3 mg/ml EZ Link Sulfo-NHS-LC-Biotin (Thermo-Pierce, Rockford, IL) in cold PBS++ for 30 minutes on ice at 4°C. The biotinylation reaction was then quenched for 5 minutes with 50 mM glycine in PBS++, and rinsed twice with cold PBS. The cells were then lysed in mRIPA buffer with Complete protease inhibitors (Roche, Basel, Switzerland) and scraped into microcentrifuge tubes. Post-nuclear supernatants were obtained by centrifuging the lysates at 800 g for 5 minutes at 4°C. In parallel, lysates of non-biotinylated cells from each experimental condition were prepared as above. The biotinylated samples were then incubated at 4°C overnight on a rocker with 25 µl of 40% streptavidin–agarose beads (Sigma, St Louis, MO) prepared in PBS. The beads were washed once with PBS, resuspended in Laemmli buffer and heated at 100°C for 5 minutes. The lysates of the non-biotinylated cells were denatured in a similar manner. All samples were then analyzed by SDS-PAGE and western blotting.

HeLa cells transfected for 48 h with FLAG–calnexin were treated with tunicamycin (10 µM) or DTT (2 mM) for 4 h. Next, cells were labeled with 100 µM alkynyl-palmitate or palmitate conjugated to BSA for 3 h. Cells were harvested in 0.1% SDS-RIPA buffer (containing CPI protease inhibitor, EDTA-free, Roche, Laval, QC) and calnexin was immunoprecipitated using the anti-FLAG antibody (Rockland). The ‘click’ reaction was then carried out on the immunoprecipitated labeled proteins by incubating them for 30 minutes at 37°C with 2 mM TBTA, 50 mM CuSO₄, 50 mM TCEP and 2 mM Biotin-azide. Duplicate samples were then separated by SDS-PAGE, and transferred onto PVDF membranes, which were washed with either 0.1 M Tris-HCl pH 7.0 or 0.1 M KOH. This alkali treatment removes fatty acids incorporated into proteins via thioester bonds but not via amide bonds and contributes to ensuring the specificity of the signal. Palmitoylation was detected by probing both membranes with HRP-conjugated Neutravidin using ECL.

HeLa cells were treated with tunicamycin (10 µM) or thapsigargin (1.5 µM) for 4 h, and immunofluorescence microscopy was performed using the indicated primary antibodies, according to the protocol described previously (Gilady et al., 2010). LDLR surface binding was detected by binding of a 1∶100 dilution of anti-LDLR antibody in DMEM, 10% FBS and 1% BSA. Intracellular FLAG-tagged calnexin was detected after permeabilization with 1% Triton X-100. The procedures employed with NIH 3T3 cells were as described previously (Avezov et al., 2008; Kamhi-Nesher et al., 2001). Confocal microscopy was performed on a Zeiss laser scanning confocal microscope (LSM 510; Carl Zeiss, Jena, Germany) as described previously (Avezov et al., 2008). Colocalization analysis (Manders) was performed using ImageJ and Imaris software.

Cells were washed twice with PBS++ and incubated for 30 minutes at room temperature with 2 mM Dithiobis (succinimidyl proprionate) (Thermo Scientific, Rockford, IL) in PBS++ to crosslink interacting proteins. The cells were then washed twice more and incubated in 10 mM NH₄Cl in PBS++ for 10 minutes to quench the crosslinking reaction. The cells were then washed a final time in PBS++ and harvested in CHAPS lysis buffer (1% CHAPS, 10 mM Tris pH 7.4, 150 mM NaCl, 1 mM EDTA) containing Complete protease inhibitors (Roche, Basel, Switzerland). Post-nuclear supernatants were obtained by centrifuging the lysates for 5 minutes at 4°C at 800 g, and were subsequently incubated with the indicated antibodies for 1 h at 4°C on a rocker. Protein-A–Sepharose beads were then added and the lysates incubated for a further 1 h. The beads were then washed three times in CHAPS buffer and resuspended in Laemmli buffer and analyzed by SDS-PAGE and western blotting. For immunoprecipitation from HEK 293 cells, cell lysis was performed in 2% sodium cholate for 30 min on ice, and debris and nuclei were pelleted in a microfuge for 30 minutes at 4°C. The samples were immunoprecipitated with anti-FLAG and goat anti-mouse-IgG–agarose overnight, followed by washes and elution by boiling with sample buffer containing β-mercaptoethanol.

Subconfluent (90%) cell monolayers in 100-mm dishes were labeled with [³⁵S]Cys. Co-immunoprecipitation with anti-calnexin antibody was performed as described previously (Frenkel et al., 2004). Briefly, after metabolic labeling, cells were lysed in HBS buffer pH 7.5, containing 2% sodium cholate, cell lysates were immunoprecipitated with anti-FLAG antibody, boiled in 1% SDS, then diluted with 10 volumes of 1% Triton X-100, 0.5% sodium deoxycholate in HBS and reimmunoprecipitated with anti-H2a antibody. SDS-PAGE gels were analyzed by fluorography using 20% 2,5-diphenyloxazole and were exposed to Biomax MS film using a transcreen-LE from Kodak (Vancouver, BC). Quantification was performed in a Fujifilm FLA 5100 phosphorimager (Japan).

Calnexin-knockout cells transfected as indicated were treated with 1.5 µM thapsigargin for 16 h or left as controls. Total RNA was isolated using TRIzol (Life Technologies, Carlsbad, CA) and amplified with primers for Xbp-1 (5′-CCTTGTGGTTGAGAACCAGG-3′ and 5′-CTAGAGGCTTGGTGTATAC-3′) and GAPDH, respectively (5′-AACTTTGGCATTGTGGAAGG-3′ and 5′-ACACATTGGGGGTAGGAACA-3′). PCR products were separated on 7.5% and 1% acrylamide gels, respectively.

ER Ca²⁺ was measured with pHSVerAEQ (gift from Javier Alvarez, Valladolid, Spain). Cells were transfected with this plasmid (and constructs expressing calnexin as indicated) and processed for measurements 24 h following transfection as published (Alvarez and Montero, 2002). Following the depletion of ER Ca²⁺ with 0.5 mM EGTA and 1 µM 2,5-di-tert-butyl-benzohydroquinone (Sigma), aequorin was reconstituted with 10 µM coelenterazine hcp (Sigma). After leaving the cells for 1 h at room temperature in the dark, cells were perfused with medium containing 1 mM Ca²⁺. Luminescence from this condition and total luminescence following digitonin permeabilization was assayed on a Lumat luminometer (Berthold, Bad Wildbad, Germany). For mitochondrial Ca²⁺, mitochondrially targeted R-GECO1 was used (Zhao et al., 2011). Calnexin wild-type and knockout MEFs were transfected with a plasmid encoding mitochondrial R-GECO1, whose fluorescence was measured on a Olympus Fluoview FV1000 microscope and quantified using the FV10-ASW 3.1 software (Olympus, Richmond Hill, ON). Fluorescence measurements were initialized after establishment of a baseline, followed by the addition of 1 µM thapsigargin.

To construct a plasmid expressing mitochondria-targeted R-GECO1, the gene for R-GECO1 in pTorPE¹ was used as a template. PCR was carried out by using primers: ‘GCaMP_FW_BamH1_mito’ (5′-GAGGATCCAACCATGGTCGACTCATCACGTC-3′) and ‘GCaMP_RV_HindIII’ (5′-CGCAAGCTTCTACTTCGCTGTCATCATTTGTAC-3′). PCR products were purified using 1% agarose gel (Agarose S, Nippon Gene Co.) electrophoresis. DNA was extracted from the appropriate gel slice using the GeneJet DNA Purification Kit (Thermo Scientific). The PCR product was then digested with BamH1 and HindIII (Thermo Scientific), repurified as described above, and ligated into the the CMV-mito-GEM-GECO1¹ vector previously digested with the same restriction enzymes. The resulting ligation products were transformed into DH10B E. coli by electroporation, and the transformed E. coli were grown on agar plates (with 400 µg/ml ampicillin) overnight. E. coli colonies on agar plates were then picked up and cultured in LB medium (with 100 µg/ml ampicillin, 250 rpm shaking) for 12 to 16 h. DNA plasmid purification was performed using the GeneJET Plasmid Miniprep Kit (Thermo Scientific), and the purified mitochondria-targeting R-GECO1 DNA plasmids were verified by sequencing at the University of Alberta Molecular Biology Services Unit.

Cells were loaded with either 1 µM Fluo8 (AAT Bioquest, Sunnyvale, CA), 1 µM Rhod2 (Life Technologies, Carlsbad, CA) or 20 nM TMRM (Sigma, St Louis, MO) and incubated in DMEM (Life Technologies, Carlsbad, CA) for 30 minutes at 37°C. Cells were then harvested in HEPES-buffered saline (0.1% glucose, 0.1% BSA) and subjected to flow cytometry using a FACS Scan cytometer (BD Biosciences, Mississauga, ON) before treatment and 10 (50 µM histamine) or 20 seconds (1.5 µM thapsigargin) after drug administration.

HeLa cells were treated with the indicated stressors and then washed twice in PBS++ and collected in mitochondria homogenization buffer (10 mM HEPES pH 7.4, 250 mM sucrose, 1 mM EDTA, 1 mM EGTA). The cell suspension was passed through an 18 µM clearance ball-bearing homogenizer (Isobiotech, Heidelberg, Germany) 15 times. The cells were subsequently centrifuged for 10 minutes at 800 g at 4°C, to pellet nuclei and unbroken cells. The supernatant was layered over a 10–30% continuous gradient of Optiprep density gradient medium (Axis-Shield, Norton, MA), and was centrifuged in an SW55 Ti rotor (Beckman Coulter, Mississauga, ON) for 3 h at 32,700 rpm. Fractions were taken from the top of the gradient and analyzed by SDS-PAGE and western blotting.

We thank Marek Michalak for helpful discussions. We especially thank Klaus Ballanyi for help with the supplementary material. The plasmid encoding ER-targeted aequorin was a generous gift from Javier Alvarez, Valladolid, Spain.