ER–Mitochondria contact sites: A new regulator of cellular calcium flux comes into play

Ca²⁺ flux from the ER to mitochondria is a major determinant of several mitochondrial processes. Basal Ca²⁺ oscillations drive mitochondrial metabolism for the production of ATP and mitochondrial substrates used in anabolic processes (Cárdenas et al., 2010; Fig. 1). In contrast, mitochondrial Ca²⁺ overload can lead to mitochondrial permeability transition pore opening and subsequent cell death (Rizzuto et al., 2012). Dysregulation of Ca²⁺ fluxes is involved in several human disorders (Rizzuto et al., 2012; Marchi et al., 2014; Krols et al., 2016), and its tight regulation is therefore crucial. This regulation is achieved by local modulation of Ca²⁺ transport systems, the tightness of the association between ER and mitochondrial membranes and the ER Ca²⁺ load. The ER Ca²⁺ load is determined by ER Ca²⁺ uptake systems (sarco-ER Ca²⁺ transport ATPases [SERCAs] with the ubiquitously expressed SERCA2b as the housekeeping isoform) and intraluminal Ca²⁺-binding proteins like calreticulin, calnexin (CNX), and GRP78/BiP, which maintain high Ca²⁺ levels in the ER. Activation of the intracellular Ca²⁺ release channels inositol 1,4,5-trisphosphate receptors (IP₃Rs) on the ER membrane leads to stimulation-induced Ca²⁺ release from the ER. At membrane contact sites between the ER and mitochondria (Fig. 1), where the ER and mitochondrial membranes are held in close proximity through protein tethers (Krols et al., 2016), specialized ER microdomains known as the mitochondria-associated membranes (MAM) are enriched in IP₃Rs. IP₃Rs are physically connected to voltage-dependent anion channels type 1 (VDAC1) in the outer mitochondrial membrane via the chaperone Grp75 (Szabadkai et al., 2006; Fig. 1). As a consequence, the local Ca²⁺ concentration rises to higher levels within these microdomains than in the cytosol, creating Ca²⁺ hotspots needed to overcome the low-affinity Ca²⁺ uptake properties of the mitochondrial Ca²⁺ uniporter complex (Rizzuto et al., 2012).

The activity of Ca²⁺-transport systems like SERCA and IP₃Rs is controlled by several interactors. The oxidoreductases Ero1α and ERp44 directly interact with IP₃Rs and modulate its Ca²⁺-flux properties in a redox-sensitive manner (Higo et al., 2005; Anelli et al., 2012; Fig. 1). In addition, the Sigma1 receptor, which is enriched at MAMs, is released from the Ca²⁺-dependent chaperone BiP/GRP78 and promotes prolonged ER calcium release through the stabilization of IP₃R3 when ER Ca²⁺ content is low (Hayashi and Su, 2007; Fig. 1). SERCA2b activity is modulated by CNX, which is targeted to the MAM in a palmitoylation and phosphofurin acidic cluster sorting protein 2–dependent manner (Roderick et al., 2000; Lynes et al., 2012, 2013). Similar to IP₃Rs, SERCA2b is sensitive to the redox state of the ER lumen. SERCA2b harbors cysteines in its fourth luminal loop (L4), and oxidation of these cysteines by ERp57 inhibits SERCA2b activity (Li and Camacho, 2004; Fig. 1), though this conclusion has been challenged (Appenzeller-Herzog and Simmen, 2016). Moreover, the redox-sensitive protein SEPN1 was recently shown to bind to the SERCA2b L4 cysteines and to enhance its ER Ca²⁺ uptake activity, which protects cells against the reactive oxygen species (ROS) produced during oxidative protein folding (Marino et al., 2015; Fig. 1) In this issue, Raturi et al. describe thioredoxin-related transmembrane protein 1 (TMX1) as a novel SERCA2b-inhibiting protein at ER–mitochondria membrane contact sites, thereby providing new insights into SERCA2b regulation (Raturi et al., 2016; Fig. 1).

TMX1 belongs to the family of protein disulfide isomerases and consists of an ER luminal domain that harbors a CXXC reductase active site, a single transmembrane domain, and a cytosolic stretch that contains a palmitoylation motif, required for targeting and/or retention of TMX1 at the MAM (Lynes et al., 2012). Raturi et al. (2016) show that loss of TMX1 in cultured cells increases ER Ca²⁺ uptake and enhances cytosolic Ca²⁺ clearance. Interestingly, the impact of TMX1 on SERCA2b was partially antagonized by the SERCA2b regulator CNX (Roderick et al., 2000; Lynes et al., 2013). Consequently, TMX1–SERCA2b complexes were more prominent in CNX-deficient cells, and vice versa. Yet, it remains unclear how each of these proteins modulates SERCA2b activity and how they influence each other. At the functional level, TMX1 was proposed to inhibit SERCA2b based on the fact that loss of TMX1 led to an increased ER Ca²⁺ uptake rate and a corresponding increase in ER Ca²⁺-store content, measured with genetically encoded ER-targeted Ca²⁺-sensitive biosensors. However, further work is needed to assess TMX1’s effect on SERCA2b activity, by using more direct SERCA2b activity assays like ATPase measurements in microsomes of SERCA2b-overexpressing cells expressing or lacking TMX1, or liposome-based systems using purified SERCA2b and TMX1, if feasible. Alternatively, changes in ER Ca²⁺-leak pathways and/or intraluminal Ca²⁺-buffering proteins could partially account for the observed TMX1 effects. Strikingly, cells lacking TMX1 displayed a reduced increase in cytosolic [Ca²⁺] and “relative” drop in ER Ca²⁺ levels in response to extracellular agonists, despite a higher ER Ca²⁺ store content. This was explained by the enhanced ability of the ER to retain Ca²⁺ through the activation of SERCA2b. A challenging aspect of this model is the difference of several orders of magnitude in the kinetics of Ca²⁺ transport between fast Ca²⁺ channels like IP₃Rs and slow Ca²⁺ pumps like SERCA. It is possible that this imbalance is compensated for by a high local concentration of SERCA2b pumps over IP₃R channels, e.g., in ER–mitochondria microdomains, or by very short opening times of IP₃Rs.

As both TMX1 and CNX are enriched at the MAM and both their interactions with SERCA2b depend on their MAM-targeting palmitoylation motif (Lynes et al., 2013; Raturi et al., 2016), Raturi et al. (2016) hypothesized that TMX1 locally regulates SERCA2b, thereby impacting ER–mitochondrial Ca²⁺ flux. Indeed, cells overexpressing wild-type TMX1 showed an augmented flux of Ca²⁺ to the mitochondria after IP₃-dependent stimulation of the cells (Fig. 1 A), whereas cells with lowered TMX1 levels displayed a diminished mitochondrial Ca²⁺ flux (Fig. 1 B). Though the modulation of ER–mitochondrial Ca²⁺ flux by TMX1 could be related to its effects on SERCA2b activity, other factors are likely involved as well. Indeed, Raturi et al. (2016) demonstrated that the ER and mitochondria in TMX1-deficient cells show a lower degree of tight association, which is a well-established determinant of ER–mitochondrial Ca²⁺ transfer (Csordás et al., 2006). How TMX1 affects ER–mitochondrial coupling is currently unclear. TMX1 may be involved in a protein complex tethering both organelles or may impact the expression of ER–mitochondria tethering proteins. An intriguing alternative could be that it is the ER–mitochondria membrane association itself that is sensitive to ER Ca²⁺ levels or local [Ca²⁺] in the ER–mitochondria interspace. This may negatively impact the formation of IP3R–GRP75–VDAC1 complexes, which are responsible for efficient ER–mitochondrial Ca²⁺ transfer (Fig. 1 B). Loss of TMX1 leads to increased ER Ca²⁺ content, thereby increasing the likelihood for proapoptotic Ca²⁺ transfers into the mitochondria under certain conditions (Rizzuto et al., 2012). Therefore, it is possible that a decrease in the overall tight ER–mitochondria contacts constitutes a compensatory mechanism that is needed for cell survival in the absence of TMX1.

The augmented Ca²⁺ flux in cells overexpressing TMX1 was abolished upon expression of the palmitoylation-deficient or thioredoxin mutants of TMX1, which fail to interact with SERCA2b at the MAM (Raturi et al., 2016). Consistent with the essential role of the thioredoxin motif of TMX1 for SERCA2b binding, the interaction of TMX1 with SERCA2b was sensitive to the ER lumen redox state (Raturi et al., 2016). Chemically induced hyperoxidation promoted TMX1 targeting to the MAM and the TMX1–SERCA2b association (Raturi et al., 2016), whereas reducing conditions had the opposite effects. Hence, it is possible that TMX1, like SEPN1, interacts with the L4 cysteines in SERCA2b. However, the individual contributions of the SERCA2b interaction partners, including SEPN1 and CNX, and their interplay at steady-state or in varying redox-state conditions remain unclear.

Raturi et al. (2016) further underscore the emerging concept that Ca²⁺ signaling and the luminal redox state of the ER are intertwined, especially at the MAMs. At the ER luminal side, ER Ca²⁺ release and uptake mechanisms are directly modulated by redox-sensitive chaperones and oxidoreductases. In contrast, ER Ca²⁺ levels regulate the activity of BiP, CNX, and calreticulin, proteins involved in oxidative protein folding. In addition, a close interconnection exists between Ca²⁺ and ROS at the ER–mitochondrial junction. Using an H₂O₂ sensor targeted to this interface, H₂O₂ has been shown to be transferred from the mitochondrial intermembrane space to the ER–mitochondrial interface (Booth et al., 2016). H₂O₂ release is stimulated by ER Ca²⁺ release, and, in turn, these H₂O₂ transients are able to modulate ER Ca²⁺ discharge, as blocking H₂O₂ release abolished normal Ca²⁺ oscillations (Booth et al., 2016; Fig. 1 A). It is currently unclear how mitochondria-derived H₂O₂ influences ER Ca²⁺ release. Although new proteins and mechanisms regulating Ca²⁺ release emerge, the redox-dependent regulation of Ca²⁺ channels is not fully understood. Ca²⁺ transfer to the mitochondrial matrix via ER–mitochondria contact sites is important for ATP production (Cárdenas et al., 2010), so ROS-sensitive proteins at the MAM could serve as sensors for ER protein folding capacity that use Ca²⁺ to communicate the local ATP requirement to the juxtaposed mitochondria. As oxidative protein folding within the ER is a major consumer of mitochondria-derived ATP, such communication may be pivotal for normal cell proliferation.

Several proteins that regulate Ca²⁺ and ROS signaling at the MAMs are implicated in diseases, such as cancer (Marchi et al., 2014) and neurodegeneration (Krols et al., 2016), emphasizing the need for further exploration of the mechanisms controlling Ca²⁺ and ROS signaling. Raturi et al. (2016) demonstrate that TMX1 protein levels influence tumor growth. In cell culture, reduced TMX1 protein levels correlated with a higher rate of cell death and a reduced multiplication rate. These findings are in agreement with a recent study showing the dependence of cancer cells to constitutive ER–mitochondrial Ca²⁺ transfer (Cárdenas et al., 2016). Nevertheless, the conclusions of both studies diverge at the in vivo level. Whereas transplanting TMX1-depleted HeLa cells in mice led to increased tumor growth (Raturi et al., 2016), pharmacological inhibition of IP₃Rs (through the injection of the drug Xestospongin B in the tumor) resulted in a significant reduction of tumor size (Cárdenas et al., 2016). A possible explanation for this discrepancy relates to the different mechanisms by which ER–mitochondrial Ca²⁺ transfer is impaired. Indeed, a strong inhibition of IP₃R channels by injection of the drug Xestospongin B directly impacts ER–mitochondrial Ca²⁺ signaling, likely leading to a prominent and severe inhibition in ER–mitochondrial Ca²⁺ fluxes. Because tumor cells depend on low-level Ca²⁺ for their metabolic requirements, loss of ER–mitochondrial Ca²⁺ transfer results in impaired tumor growth and reduced viability (Cárdenas et al., 2016). In contrast, partial depletion of TMX1 could indirectly impact ER–mitochondrial Ca²⁺ signaling by activating SERCA2b, leading to a moderate reduction of Ca²⁺ flux toward the mitochondria. This Ca²⁺ flux could promote tumor growth in vivo by contributing to the Warburg effect (the switch to aerobic glycolysis in tumor cells) and possibly by preventing the occurrence of excessive, proapoptotic mitochondrial Ca²⁺ signals, although in vitro data argue for increased cell death in the absence of TMX1 (Raturi et al., 2016). As the full range of TMX1 functions is unknown and the contribution of Ca²⁺ transfer to the enhanced tumor growth upon TMX1 depletion was not studied, other mechanisms might be involved in promoting tumorigenesis upon TMX1 depletion. These seemingly opposite effects of lack of TMX1 in vitro (with increased cell death and decreased proliferation) and in vivo (with increased tumor growth) ought to be investigated further.

Overall, it has become clear that ER–mitochondrial Ca²⁺ transfer plays a central role in oncogenesis and in the response of cancer cells to chemotherapy (Marchi et al., 2014). Other tumor suppressors, including p53, have been proposed to influence contact site formation and Ca²⁺ transfer at the MAMs (Bittremieux et al., 2016). FATE1 was identified as an ER–mitochondria uncoupler, whereby it diminishes Ca²⁺ flux and promotes chemotherapy resistance in adrenocortical carcinoma cells (Doghman-Bouguerra et al., 2016). Hence, advancing our knowledge on the composition, function, and regulation of these contact sites and on how all these different players cooperate to control interorganelle communication will be of utmost importance in the search for new treatments for diseases such as cancer, diabetes, and neurodegenerative disorders (Marchi et al., 2014; Krols et al., 2016).