The Trans Golgi Region is a Labile Intracellular Ca2+ Store Sensitive to Emetine

The Golgi apparatus (GA) is a bona fide Ca2+ store; however, there is a lack of GA-specific Ca2+ mobilizing agents. Here, we report that emetine specifically releases Ca2+ from GA in HeLa and HL-1 atrial myocytes. Additionally, it has become evident that the trans-Golgi is a labile Ca2+ store that requires a continuous source of Ca2+ from either the external milieu or from the ER, to enable it to produce a detectable transient increase in cytosolic Ca2+. Our data indicates that the emetine-sensitive Ca2+ mobilizing mechanism is different from the two classical Ca2+ release mechanisms, i.e. IP3 and ryanodine receptors. This newly discovered ability of emetine to release Ca2+ from the GA may explain why chronic consumption of ipecac syrup has muscle side effects.

Ca 2+ uptake is catalyzed by SERCA and, in the medial-Golgi, also by SPCA 17 . The trans-Golgi region appears not to express IP 3 Rs or SERCAs and Ca 2+ uptake involves the SPCA. In neonatal cardiomyocytes, the trans-Golgi appears to express RyRs and accordingly it releases Ca 2+ in response to caffeine 16 . Indirect evidence suggests that the Ca 2+ level within the GA lumen modulates vesicular trafficking and the correct sorting of proteins 22 . While using emetine to study the well-established role of Sec61 translocon in the ER Ca 2+ leak 23 , we found that this alkaloid, which is also present in ipecac syrup 24 , was able to reduce the Ca 2+ content of an undefined Ca 2+ store in HeLa cells. Ipecac syrup is an emetic agent 25 and its chronic consumption results in reversible myopathy and cardiomyopathy 26 . At the molecular level, this alkaloid acts as an inhibitor of protein synthesis 27 , by targeting the small subunit of ribosomes 28 , without actually detaching the ribosomal complex from the translocon 29 .
Using a variety of probes and experimental approaches, here we demonstrate that emetine is capable of mobilizing Ca 2+ from the medial-and trans-regions of the GA, without affecting the ER Ca 2+ content. Emetine may represent the first member of a chemical library that will enable studies on the mechanism of GA Ca 2+ homeostasis and its role in Ca 2+ pathophysiology.

Results
Emetine decreases the [Ca 2+ ] L in an intracellular store different from the ER. To investigate the effect of emetine on Ca 2+ homeostasis, we carried out simultaneous recordings of the signals from two fluorescent indicators; Fura-2 and Mag-Fluo-4, in HeLa cells. While Fura-2 was homogeneously distributed in the cytosol and nucleus (and thus it is a bona fide cytosolic and nuclear Ca 2+ indicator, [Ca 2+ ] c ), Mag-Fluo-4 fluorescence was not distributed in a homogenous manner. Rather, a faint signal was observed in the region corresponding to the nuclear membrane, as well as reticular structures in the cytoplasmic region, while the nucleus itself was completely dark ( Supplementary Fig. 1A). Vesicular structures some of which were located in the perinuclear region and colocalized with a GA marker, also displayed strong fluorescence ( Supplementary Fig. 1C). The majority of the Mag-Fluo-4 fluorescence signal (75%) colocalized with both the ER and the GA ( Supplementary Fig. 1B,D), therefore we consider this fluorescence signal as an indicator of luminal Ca 2+ concentration ([Ca 2+ ] L ). In the presence of external [Ca 2+ ], the combination of ATP (an IP 3 -generating agonist) and thapsigargin (Tg, an irreversible SERCA pump inhibitor) induced a transient increase in the [Ca 2+ ] c and this was associated with a sustained reduction in the [Ca 2+ ] L signal (Fig. 1A, black trace). Ten minutes later, the addition of emetine (50 μM) resulted in a further reduction in the [Ca 2+ ] L without any change in the [Ca 2+ ] c . Reversing the order of addition, i.e., emetine first followed by ATP and Tg, resulted in an initial decrease in the [Ca 2+ ] L , followed by a further reduction of the signal upon addition of ATP and Tg (Fig. 1A, red trace). The changes in the [Ca 2+ ] c were qualitatively similar using either protocol, i.e. no rise upon addition of emetine and a transient increase in response to ATP and Tg. However, the peak [Ca 2+ ] c elicited by ATP and Tg was significantly smaller when they were added after emetine (Fig. 1B, upper panel), while the reduction in the [Ca 2+ ] L induced by emetine was significantly larger when this alkaloid was added as the first stimulus (Fig. 1B, lower panel, second red bar).
To exclude any effect of emetine on Ca 2+ influx across the PM (e.g., capacitative Ca 2+ entry, CCE), we carried out the same experiment in the absence of external [Ca 2+ ] (Fig. 1C). Qualitatively, the results were similar to those obtained in medium containing 1.8 mM CaCl 2 , however the amplitude of the [Ca 2+ ] c peaks were substantially reduced (Fig. 1D, upper panel) and the reductions in the [Ca 2+ ] L caused by ATP and Tg and by emetine were larger, suggesting that the absence of external Ca 2+ facilitated store depletion. Similar to what was observed in presence of external Ca 2+ , the ATP and Tg-triggered [Ca 2+ ] c response was decreased by the presence of emetine (Fig. 1C, red trace). This effect cannot simply be explained by the cells being maintained in the absence of external Ca 2+ for a prolonged period of time, since this 10 minute period had no effect on the amplitude of agonist-induced [Ca 2+ ] c response ( Supplementary Fig. 2). The finding that emetine had no effect on the [Ca 2+ ] L responses induced by ATP and Tg and vice versa (Fig. 1D, lower panel) suggests that emetine releases Ca 2+ from a compartment that does not contain SERCA or IP 3 receptors, and is thus distinct from the ER.
Emetine did not increase the [Ca 2+ ] c at any of the concentrations tested ( Supplementary Fig. S3A). However, the reduction in the [Ca 2+ ] L response showed a concentration dependence ( Supplementary Fig. 3B) between 10 and 20 μM (Supplementary Fig. 3C). In the absence of external [Ca 2+ ], the emetine-induced [Ca 2+ ] L response was larger and kinetically displayed two phases ( Supplementary Fig. 3D, black trace). Isoemetine (50 μM), which is an isomer of emetine that does not block protein synthesis 30 , also reduced the [Ca 2+ ] L , although its effect was slower and of smaller amplitude ( Supplementary Fig. 3D, red trace).
Emetine releases Ca 2+ from an acidic intracellular Ca 2+ store. It has been demonstrated that cells contain two different types of intracellular Ca 2+ stores, the first is represented primarily by the ER/SR and the other is characterized by having an acidic lumen, and is collectively named "acidic Ca 2+ stores" 31 . A typical characteristic of the acidic Ca 2+ stores is that they can be rapidly depleted by the combination of ionomycin (1 μM) and a substance that can collapse the acidic pH gradient, such as monensin (10 μM) 32 . In the experiments presented in Fig. 2, we have used the combination of ATP and Tg to deplete the ER, followed 10 min later by the combination of ionomycin and monensin. The latter treatment, resulted in a small rise in [Ca 2+ ] c and a large reduction in the [Ca 2+ ] L ( Fig. 2A, lower panel). More importantly, the emetine-induced [Ca 2+ ] L decrease was inhibited by depletion of the acidic Ca 2+ stores (Fig. 2B). This data suggests that emetine was targeting acidic Ca 2+ stores to mobilize Ca 2+ .
To further corroborate that the emetine-sensitive intracellular Ca 2+ release is from the acidic Ca 2+ stores, the proton gradient of these compartments was collapsed using two approaches: incubation with either NH 4 Cl (18 mM) or bafilomycin (baf, 100 nM). This NH 4 Cl concentration was selected to avoid having an effect on the IP 3 -induced Ca 2+ release, while the baf concentration was low enough to be specific for vacuolar H + -ATPase 33 . The rise in [Ca 2+ ] c and the associated drop in [Ca 2+ ] L induced by ATP and Tg were unaffected by pre-incubation with NH 4 Cl for 15 minutes (Fig. 3A, red trace); whereas 15 minutes pre-incubation with baf significantly increased  Supplementary Fig. 4A,B). These effects can be explained by the fact that baf increases the [Ca 2+ ] L ( Supplementary Fig. 4A, inset) as we have previously reported 34 . These results further point to the acidic Ca 2+ stores as being the target of emetine.
To determine whether lysosomes are the target of emetine; HeLa cells were pre-incubated with the dipeptide glycyl-L-phenylalanine-2-naphthylamide (GPN). This is a substrate of the intralysosomal protease cathepsin C, and accumulation of the hydrolysis product therein will osmotically disrupt the lysosomes 35 . Surprisingly, GPN caused a dramatic inhibition of both the [Ca 2+ ] c rise and the [Ca 2+ ] L decrease induced by ATP and Tg (Fig. 3C, red trace), suggesting that this drug has a strong effect on the ER, in addition to affecting lysosomes. The emetine-induced [Ca 2+ ] L drop was clearly enhanced by GPN pre-treatment (Fig. 3D, lower panel), indicating that lysosomes are not the target for emetine. As shown in Supplementary Fig. 4C Fig. 4C, red trace). Since GPN has non-specific effects on the ER ( Supplementary Fig. 4D), we searched for a more specific treatment to release Ca 2+ from the lysosomal compartment. To this end, we treated HeLa cells with NAADP/AM, a membrane permeable NAADP analogue believed to be a specific lysosomal Ca 2+ mobilizer via TPC2 channels 36 . The application of 10 nM NAADP-AM resulted in a slow but significant drop in the [Ca 2+ ] L without any effect on the [Ca 2+ ] c (Fig. 3E, black trace). The addition of BZ194 (an inhibitor of the NAADP receptor) or Ned19 (a blocker of TPC2), inhibited the effect of NAADP-AM on the [Ca 2+ ] L (Fig. 3E, red and blue traces). However, none of these reagents modified the emetine-induced reduction in the [Ca 2+ ] L (Fig. 3F). Collectively, these data support the idea that neither lysosomes nor TPC2 channels are targeted by emetine.
The Ca 2+ store responding to emetine is labile and localized in the perinuclear region. To gain an insight into the identity of the Ca 2+ store responding to emetine, we performed confocal microscopy experiments in HeLa cells loaded with Mag-Fluo-4, in the presence of external [Ca 2+ ]. The perinuclear and cytosolic regions of the cells that were labelled with Mag-Fluo-4 ( Fig. 4A) were then examined during the application of emetine. Addition of emetine resulted in a net decrease in the Mag-Fluo-4 fluorescence, but only in the perinuclear region; there was no significant change in the cytosolic region ( Fig. 4B,C). The mean Fura-2 signals, that were obtained as shown in Figs 1-3, represent the combined response of half a million cells in suspension in a cuvette, therefore, small localized increases in cytosolic Ca 2+ cannot be detected by this approach. To investigate whether emetine was producing a localized increase in the [Ca 2+ ] c , cells were loaded with another cytosolic Ca 2+ indicator, Fluo-4, and analyzed by confocal microscopy. Unlike the results obtained with Fura-2 in the cell populations, emetine was able to trigger a localized and small [Ca 2+ ] c response, 4-5 fold smaller than that induced by ATP (Fig. 4D). Noteworthy, this rise in perinuclear [Ca 2+ ] c was observed only when emetine was added within 45 seconds after ATP addition (Fig. 4E). As observed in the experiment presented in Fig. 1, pre-treatment with emetine decreased the amplitude of the ATP-induced [Ca 2+ ] c response (Fig. 4F). These data indicate that emetine induces a small and localized [Ca 2+ ] c response in the perinuclear region; but only shortly after the ER has released its Ca 2+ , suggesting that the emetine-sensitive intracellular compartment is a labile Ca 2+ store because emetine can only increase the [Ca 2+ ] c in a short time window.
Emetine does not release Ca 2+ from the ER nor does it affect mitochondrial Ca 2+ handling. The data presented above provide strong, but still indirect evidence that the intracellular Ca 2+ pool sensitive to   5C) was clearly reduced by the application of emetine (Fig. 5D). Interestingly, in the absence of external [Ca 2+ ] (EGTA), the application of emetine significantly reduced the [Ca 2+ ] n (Fig. 5D). Finally, we studied HeLa cells expressing 4mtD3cpv, a mitochondrial matrix localized Ca 2+ probe. As expected, His and CPA caused a rapid increase in mitochondrial [Ca 2+ ]; however, the prior application of emetine did not affect the amplitude of the agonist-induced mitochondrial Ca 2+ response (Fig. 5E,F).
Emetine releases Ca 2+ from the Golgi apparatus. The experiments described so far, which have been carried out using both chemical Ca 2+ indicators and GECIs, suggested that the GA might be the main target for emetine. The GA is an organelle preferentially localized in the perinuclear region 39 , its luminal [Ca 2+ ] is high 16 and its lumen is slightly acidic 19 . It needs to be stressed that it has been shown previously that the GA sub-compartments have quite distinct Ca 2+ uptake and release characteristics and accordingly we have used two different GECIs targeted to the medial (medial-GoD1cpv) 17 and the trans compartments (GoD1cpv) 16 of the GA.
Application of emetine to cells transfected with GA probes caused a strong reduction in the ΔR/R 0 in both the medial- (Fig. 6A, red trace) and the trans-Golgi region (black trace). Unlike the ER (Fig. 5A), [Ca 2+ ] L in the trans-Golgi was more sensitive to reducing the external [Ca 2+ ] with EGTA (Fig. 6B, black trace), in this condition, ] adds to the notion that this GA region is a labile Ca 2+ store. A potential artifact with the GA targeted cameleons is that alkalization of the lumen will mimic a drop in the [Ca 2+ ] 16 . To address whether emetine affects the luminal pH in the GA, we have used the same protocol as Lissandron et al. to investigate the pH changes in the trans-Golgi 16 . Specifically, we directly excited the acceptor of the cameleon at 510 nm and recorded the fluorescence changes at 540 nm. Under this situation, any change in the acceptor fluorescence signal is independent of the Ca 2+ level and is due to changes in the luminal pH 16 . Supplementary Fig. 5A shows that treatment with 1 μM monensin (black trace), caused a large increase in the acceptor fluorescence; while the addition of 50 μM emetine also elevated the acceptor fluorescence (red trace), but only by about 1/3 of that induced by 1 μM monensin. The effect of emetine was similar to that induced by 250 nM monensin (blue trace). Importantly, the application of emetine resulted in a reduction in both the GA [Ca 2+ ] and an increase in the luminal pH (observed as parallel fluorescence increase in the FRET donor and acceptor) in the medial-Golgi (Supplementary Fig. 5B) and trans-Golgi ( Supplementary Fig. 5C). The application of EGTA, to reduce external [Ca 2+ ], resulted in antiparallel fluorescence changes in the trans-Golgi (Supplementary Fig. 5D) confirming that this action has changed the [Ca 2+ ] without altering the pH. However, although the effect of emetine on the trans-GA region pH is similar to 250 nM monensin, the changes in the ΔR/R 0 induced by emetine were faster and larger than those induced by 250 nM monensin (Fig. 6D). Even higher concentrations of monensin (1 μM) had a smaller effect on the ΔR/R 0 than emetine (Fig. 6E). Taken together, the data shown in Fig. 6 and Supplementary Fig. 5 demonstrate that emetine specifically targets the GA and that this alkaloid has the dual effect of mobilizing Ca 2+ and increasing the luminal pH.
Emetine releases Ca 2+ from the trans-Golgi of heart HL-1 cells. Since the main side effect associated with chronic emetine consumption is reversible myopathy and cardiomyopathy 26 , we wondered whether emetine specifically targets the RyR (the main receptor expressed in the SR of cardiac and muscle cells). To this end, we ] n when perfused either before or after caffeine (Fig. 7A). Furthermore, caffeine caused a decrease in the SR [Ca 2+ ] of cells transfected with the SR targeted cameleon, D4ER, whereas emetine was totally without effect on the SR [Ca 2+ ] either when applied before (Fig. 7B, black trace) or after caffeine (Fig. 7B, red trace). In HL-1 cells that were transfected with the trans-GA Ca 2+ probe (GoD1cpv), emetine caused a significant reduction in the ΔR/R 0 (Fig. 7C, black trace), and the amplitude was similar to that obtained with caffeine (Fig. 7D). Interestingly, the prior application of emetine or caffeine resulted in a slightly smaller luminal response to the other alkaloid when it was subsequently applied (Fig. 7C,D).

Discussion
Emetine, a well-known inhibitor of protein synthesis 27 also blocks the Ca 2+ leak from the ER via translocon 23 . While we were studying the nature of this ER Ca 2+ leak, we discovered that emetine was able to decrease [Ca 2+ ] L in Mag-Fluo-4 loaded HeLa cells. Our colocalization data indicated that the majority of Mag-Fluo-4 fluorescence signal comes from both the ER and the GA and this situation allowed us to discover that emetine was specifically mobilizing Ca 2+ from the GA and not from the ER. However, three different Ca 2+ indicators, Fura-2, Fluo-4 and H2BD3cpv, did not show any increase in the [Ca 2+ ] c in response to emetine, when it was applied as the first stimulus. Interestingly, we found that releasing Ca 2+ from the ER, appears to have loaded with Ca 2+ the emetine-sensitive store to the extent that the application of this alkaloid produced a small and transient increase in the [Ca 2+ ] c , which was localized only to the perinuclear region of HeLa cells. Indeed, it has been previously shown that His and CPA transiently increases the trans-GA [Ca 2+ ] L 16 . Furthermore, the emetine-sensitive store can be considered a labile Ca 2+ pool because the Ca 2+ captured from the ER is lost very rapidly. In this regard, FRET experiments showed that the removal of ] L more rapidly than the [Ca 2+ ] L in the ER, supporting the idea that trans-Golgi region is a labile or leaky Ca 2+ store that needs a constant supply of Ca 2+ from either the ER or the external medium. Another critical observation was that emetine reduced the amplitude of the agonist-induced [Ca 2+ ] c rise in HeLa cells. Since emetine neither decreased the Ca 2+ content of the ER nor inhibited Ca 2+ release induced by an agonist, then it is unlikely that emetine is inhibiting the IP 3 Rs or reducing the Ca 2+ content of the ER. Additionally, decreasing the external [Ca 2+ ] after application of emetine resulted in a clear reduction of the resting [Ca 2+ ] n as observed with H2BD3cpv. However, the molecular mechanism behind this effect of emetine is still undefined. We do not think that emetine is increasing plasma membrane-mediated Ca 2+ extrusion, because the time course of the reduction in the agonist-induced [Ca 2+ ] i response was not altered by the presence of this alkaloid. Moreover, emetine did not decrease the amplitude of the caffeine-induced [Ca 2+ ] n response in HL-1 cells, suggesting that the effect of emetine on reducing cytosolic Ca 2+ responses is not generalized.
We have observed that the reduction in the [Ca 2+ ] L induced by the combination of ATP and Tg was smaller in the presence than in the absence of external [Ca 2+ ]. We think that this difference resulted from a strong Ca 2+ -dependent inactivation of the IP 3 R, a condition that has been previously reported 41 . Nevertheless, we consider that this combination produced a functional depletion of the IP 3 R-sensitive ER Ca 2+ store. In the presence of external [Ca 2+ ], the [Ca 2+ ] L response to emetine was decreased only when added after the combination of ATP and Tg. This situation could be explained by the activation of CCE upon stimulation with ATP and Tg and this Ca 2+ entry might be captured by the GA, counteracting the Ca 2+ mobilization activity of emetine. Indeed, the [Ca 2+ ] L in the trans-GA region was extremely sensitive to the external [Ca 2+ ], as discussed above. In the absence of external [Ca 2+ ] there will be no CCE and accordingly, the [Ca 2+ ] L response to emetine was not modified by the previous application of ATP and Tg.
Three different approaches, and two different cell types, support the idea that emetine is not releasing Ca 2+ from the ER or the SR. In the absence of external [Ca 2+ ], emetine did not modify the depletion of the agonist-sensitive ER (ATP or histamine) in combination with SERCA pump inhibitors (Tg or CPA). Confocal experiments showed that emetine released Ca 2+ only from the perinuclear region, where 60% of the Golgi marker colocalized with Mag-Fluo-4; a completely different picture was observed for the ER, which is distributed throughout the entire cytosol. We decided then to study whether acidic Ca 2+ stores were targeted by emetine, with the knowledge that Ca 2+ handling in the acidic compartment is dependent on the proton gradient generated across this membrane 31 . We studied the role of the acidic Ca 2+ stores with the combination of ionomycin and monensin. These ionophores produced a smaller and transient [Ca 2+ ] c response compared to the one produced by ATP and Tg, and yet caused a similar reduction in the [Ca 2+ ] L . In addition, these ionophores inhibited the emetine-induced reduction in the [Ca 2+ ] L . To further support the idea that emetine was targeting an acidic Ca 2+ store we employed different approaches to disrupt the proton gradient of these stores, i.e. alkalization with NH 4 Cl 42 and inhibition of the V-type H + −ATPase with bafilomycin 43,44 . This inhibitor increased the agonist-induced Ca 2+ release from the ER, likely by decreasing the Ca 2+ buffering activity of the acidic intracellular stores, in agreement with previous reports 45 ; while NH 4 Cl did not show this effect because cellular alkalization inhibits ER Ca 2+ release by decreasing the activity of SERCA pump 46 . We found that although bafilomycin and NH 4 Cl did not have the same effect on agonist induced Ca 2+ release, they both decreased the emetine-induced [Ca 2+ ] L response by increasing the pH in the intracellular acidic stores. Since these conditions do not deplete acidic Ca 2+ stores within the time frame used here, we think that emetine needs an acidic environment to activate the Ca 2+ mobilization mechanism. Thus, we concluded that an acidic Ca 2+ store was the target of emetine.
The involvement of the endo-lysosomal system was discarded because NAADP or inhibitors of the NAADP receptor did not affect the emetine-induced [Ca 2+ ] L response. Additionally, GPN, a lysosomal disruptor 47  To determine whether the GA was the target for emetine, we transfected cells with either medial-17 or trans-Golgi 16 cameleons as previously reported. Both compartments responded to emetine application with a clear reduction in the [Ca 2+ ] L . Moreover, emetine still decreased the [Ca 2+ ] L in these two GA compartments when cells were maintained in Ca 2+ free extracellular medium containing EGTA, implying that emetine Ca 2+ mobilization activity might involve the activation of an ion channel instead of blocking the Ca 2+ loading mechanism. Since the trans-Golgi region is a labile Ca 2+ store, this might explain why emetine does not produce any increase in the [Ca 2+ ] c as the first stimulus, but only right after Ca 2+ has been released from the ER by an agonist. We have observed that emetine increases the luminal pH as well and to assess the role of pH in the effect of emetine, we titrated pH changes with monensin. These data indicate that only a small part of the effect of emetine on the reduction of the ΔR/R 0 can be explained by its associated increase in the luminal pH, the main effect of emetine is on the reduction in the [Ca 2+ ] L .
The effect of emetine was not limited to the GA of HeLa cells. Indeed, similar results have been obtained studying the trans-Golgi region of HL-1 atrial cells, which express RyR2 43 . The trans-Golgi region of HL-1 cells responded to both caffeine and emetine with a similar reduction in the [Ca 2+ ] L . Interestingly, the trans-Golgi Ca 2+ store sensitive to emetine partially overlaps with the region released by caffeine in HL-1 cells. It seems unlikely that the channel activated by emetine is the RyR2.
The SPCA is a Tg-insensitive Ca 2+ pump responsible for Ca 2+ uptake by the trans-Golgi 16 whereas it co-exists with SERCA in the cis/medial-GA 17 . The fundamental importance of SPCA in the physiology of GA has been shown by a number of studies, which have reported that reduction or ablation of SPCA causes morphological alterations in the Golgi structure, as well as the disruption and dysfunction of both vesicle trafficking and protein sorting in the secretory pathway [49][50][51] . However, SPCA has been reported to be expressed in other secretory vesicles, where it is required for Ca 2+ homeostasis in these organelles 52,53 . Our data demonstrate that emetine is directly and specifically affecting GA. However, it is unlikely that emetine is inhibiting SPCA; but it appears to be activing a leak channel in the GA. Further experiments are needed to unveil the molecular nature behind this effect of emetine.
This new role of emetine, as a specific Ca 2+ mobilization agent from GA, might explain the symptoms of cardiac and skeletal myopathies observed in people who are chronically consuming emetine 24,26 . This might lead to a dysfunction in Ca 2+ dynamics in the GA that could result in altered Ca 2+ handling in muscle cells.
Cell culture and transfection. HeLa cells were purchased from ATCC and cultured in Dulbecco's modified Eagle's medium with high D-glucose (4500 mg/l), L-glutamine and sodium pyruvate (110 mg/l) supplemented with 10% fetal bovine serum, and 100 units/mL penicillin and 100 µg/mL streptomycin at 37 °C and 5% of CO 2 in humid conditions. For the culture of HL-1 cells, Claycomb medium (from Sigma-Aldrich) supplemented with norepinephrine (100 µM), L-glutamine (2 mM) and the same concentration of fetal bovine serum and antibiotic as mentioned above was used.

Simultaneous recordings of cytosolic and luminal [Ca 2+ ] in HeLa cell population. HeLa cells were
trypsinized after reaching 80% confluence. The cell suspension was microfuged and the pellet was suspended in extracellular-like saline solution containing in mM: 121 NaCl, 5.4 KCl, 0.8 MgCl 2 , 1.8 CaCl 2 , 6 NaHCO 3 , 25 HEPES and 5.5 glucose [pH 7.3 at room temperature (RT)]. Cell viability was always higher than 95% as determined by trypan blue exclusion assay and 1 × 10 6 cells/ml were loaded with 1 µM of each of the Ca 2+ indicators, i.e. Fura-2/AM and Mag-Fluo-4/AM, to determine changes in cytosolic ([Ca 2+ ] c ) and luminal ([Ca 2+ ] L ) calcium concentrations, respectively. This loading was carried out in the dark and at RT for 2 h. At the end of this time, half a million cells were microfuged twice and the cell pellet was re-suspended in 2.5 ml of saline solution with or without CaCl 2 plus 0.1 mM EGTA, as indicated. Fluorescence signals were recorded at a dwell time of 100 ms and a frequency of 2.7 Hz with excitation wavelengths of 340, 360 and 380 nm for Fura-2 and 495 nm for Mag-Fluo-4 with the same emission wavelength of 515 nm using a DeltaRAM V PTI spectrofluorometer.
The [Ca 2+ ] c calculated from the Fura-2 signals and normalization of the Mag-Fluo-4 signals was carried out as previously described 55 . To discard the participation of lysosomes, cells were incubated with GPN for a time period of 10 min. Since the released naphthylamine (a product of GPN cleavage mediated by cathepsin C) interferes with Fura-2 fluorescence, cells were washed before recording. In the situation where GPN was present, we did not use the 340/380 ratio; only the 380 nm fluorescence signal was used to reflect changes in the [Ca 2+ ] c .
The use of ionomycin and monensin to fully deplete the acidic Ca 2+ stores interfered with the use of digitonin for calibration, as revealed by changes in the Fura-2 Ca 2+ -insensitive fluorescence at 360 nm, precluding the transformation of the 340/380 fluorescence ratio to [Ca 2+ ] c . Thus, we have normalized the 340/380 ratio using the value at time zero (ΔR/R 0 ).

Confocal microscopy experiments.
HeLa cells were cultured onto 21 × 21 mm coverslips until they reached 80% confluence. They were then loaded with 1 µM of either Mag-Fluo-4/AM or Fluo-4/AM in the dark at RT for 2 and 1 hour, respectively. After this time, the coverslip was placed onto the chamber and bathed in extracellular-like saline (composition described above). Fluorescence recordings were obtained using a Carl Zeiss LSM700 confocal inverted microscope with a 63x oil immersion objective (1.4 N.A.). Both Ca 2+ indicators were excited with 488 nm laser line using a pinhole of 45 µm, and fluorescence images were collected every 1.93 s.
Analysis of the Ca 2+ indicator fluorescence signals was carried out with ImageJ (Wayne Rasband, Bethesda, USA). Regions of Interest (ROIs) were drawn in two different regions: perinuclear and cytosolic. Since the Mag-Fluo-4 signal exhibited a constant exponential decrease, the first 20 images (i.e. before the perfusion of any stimulus) were fitted to this equation Y = A 0 exp −x/t + y 0 and this was considered as the basal fluorescence (F 0 ). Changes in the [Ca 2+ ] L were calculated as ∆F/F 0.
Colocalization studies of Mag-Fluo-4 and organelles, either the ER or the GA were carried out by transfecting HeLa cells with 2 µg (4 μL of Lipofectamine 2000) of either mCh-Sec61 beta (gift from Gia Voeltz, Addgene plasmid # 49155) or mCherry-Golgi-7 (gift from Michael Davidson, Addgene plasmid # 55052). Transfected cells were also loaded with Mag-Fluo-4/AM, as described above. Live HeLa cell imaging was carried out with the confocal microscope to visualize the degree of colocalization between Mag-Fluo-4 and the probe for either the ER or the GA. The level of colocalization was quantified using Manders´ coefficient.

FRET experiments for [Ca 2+
] determination using GECIs. FRET experiments were performed as described by Drago et al. 56 . Briefly, cells seeded on coverslips were placed onto an open-topped chamber at 37 °C with modified Krebs-Ringer buffer (mKRB) containing (mM): 135 NaCl, 5 KCl, 10 glucose, 1 Mg 2 Cl, 1.8 CaCl 2 and 10 HEPES (pH 7.4 at 37 °C). Image exposure time was 200 ms and acquisition frequency 0.5 Hz. With these probes, the ratio (R) between the fluorescence intensity emitted by the acceptor (at 540 nm) and by the donor (at 480 nm) fluorophores (upon excitation of the donor) is a function of the FRET efficiency and accordingly of the [Ca 2+ ]. An increase or a decrease in the 540/480 emitted fluorescence ratio thus indicates an increase, or a decrease, in the [Ca 2+ ], respectively. Data are presented as a ∆R/R 0 , where: ∆R is the change of the cpV/CFP emission intensity ratio at any time, and R 0 is the fluorescence emission ratio at the time 0. Statistical Analysis. All the data represent at least three independent experiments. Figures were prepared by GraphPad Prism version 5.0. Averages are expressed as mean ± SEM (n, number of independent experiments) analyzed by unpaired Student´s t test, where *P < 0.05, **P < 0.01 and ***P < 0.001 are statistically significant.