The role of mitochondria in cancer was long debated. In the 20s of the last century, Otto Warburg made the pivotal observation that cancer cells mostly rely on glycolysis for ATP production even in the presence of optimal pO₂. Warburg proposed that mitochondrial dysfunction was the primary cause of preferential aerobic glycolysis. In addition, only cells able to increase glycolytic ATP production during periods of respiratory failure would progress toward cancer. Thus, cancer would be a consequence of progressive and irreversible defect in oxidative metabolism. However, experimental evidence revealed the weaknesses of this model. First, in some tumors, cells have normal respiratory capacity and/or are able to reactivate oxidative phosphorylation in case of impaired glycolysis. Second, in alternative to glucose metabolism, certain cancers utilize glutamine as an energy source (1). In addition, Warburg’s theory did not explain the existence of cancer-related gene mutations and, finally, it was unclear how impaired respiration would lead to uncontrolled cell growth. As a result, for many years, tumor metabolism was overlooked, in favor of a more accepted view of cancer as genetic disease. Recently, the importance of metabolism in cancer progression has been reconsidered. Mutations in genes encoding mitochondrial proteins and mitochondrial dysfunctions are linked to cancer progression and respiration defects contribute to genome instability. In addition, damage in respiration underlies genome instability which, in turn, sustains respiratory impairment and cancer progression (2). Mitochondrial Ca²⁺ signaling plays different roles in pathophysiology. On one hand, by buffering cytosolic [Ca²⁺], it regulates cell processes controlled by intracellular [Ca²⁺], like those regulated by Ca²⁺-binding proteins. On the other side, within mitochondria, Ca²⁺ regulates three TCA cycle enzymes, thus exerting a positive role on oxidative metabolism. However, excessive mitochondrial Ca²⁺ entry consequent to stress stimuli causes opening of the mitochondrial permeability transition pore (mPTP) and release of proapoptotic factors which eventually lead to cell death (3). On this basis, for a long time, the study of mitochondrial Ca²⁺ signaling in cancer biology has been primarily focused on the Ca²⁺-dependent activation of the proapoptotic axis as a putative therapeutic strategy. However, once stimulated with an appropriate agonist, cancer cells display large mitochondrial Ca²⁺ peaks, raising the question whether effective mitochondrial Ca²⁺ accumulation might play a permissive role in cancer progression. Thus, the scenario is quite complex, and recent literature suggests that mitochondrial Ca²⁺ homeostasis plays different roles in cancer, depending on the type and stage of the tumor and on whether primary tumors or metastasis are being analyzed.

Once the electrophysiological properties of the mitochondrial calcium uniporter (MCU), the highly selective channel responsible for mitochondrial Ca²⁺ uptake, were described (4), the genes encoding both the pore and the regulatory subunits have been eventually identified (3). Therefore, studies aimed at further delving into the role of mitochondrial Ca²⁺ homeostasis in pathophysiology, including cancer progression, were conducted based on the molecular modulation of the MCU complex components.

Genetic modulation of MCU expression confirmed the previously recognized role of mitochondrial Ca²⁺ overload in triggering cell death. Thus, the old paradigm of excessive Ca²⁺ uptake being permissive for apoptotic cell death still holds true. However, other studies highlighted novel and still incompletely defined aspects of MCU in cancer, indicating that mitochondrial Ca²⁺ uptake plays multiple and complex roles in cancer biology. This review will summarize the molecular structure of MCU, the regulatory mechanisms underlying its activity, and will uncover its role in pathophysiology, with particular attention to cancer biology.

Voltage-dependent anion channels (VDACs) are the most abundant proteins of the outer mitochondrial membrane. VDAC1, VDAC2, and VDAC3 are permeable to <5 kDa metabolites and are characterized by a weak anion selectivity in the open channel conformation, while they show strong cation preference in the closed states (5). The three VDAC isoforms share comparable channel properties in Ca²⁺ transport from the cytosol to the intermembrane space (6); however, they differ in specific functions. In particular, VDAC1, by interacting with the InsP3R located at the ER, specifically senses low amplitude proapoptotic Ca²⁺ signals (7). In addition, isoform-specific interactions of VDAC with apoptotic regulators contribute to the selective control of cell fate (3). VDACs are involved in many other functions, including metabolism regulation and energy production. For further discussion on VDACs, the reader is referred to specialized publications (8–11).

A 480 kDa inner mitochondrial membrane (IMM) multimeric complex, i.e., the MCU, comprising channel pore-forming subunits and regulatory elements, ensures Ca²⁺ entry into mitochondria (Figure 1). Long before the electrophysiological properties and the molecular structure of MCU were disclosed, cell Ca²⁺ signaling was already extensively characterized. The driving force for Ca²⁺ entry into mitochondria is represented by the electrochemical gradient across the IMM, which is negative in the mitochondrial matrix (about −180 mV). However, despite this huge driving force, resting mitochondrial [Ca²⁺] is in the nanomolar range, indicating that the affinity of the channel for Ca²⁺ is low. In addition, the mitochondrial Na⁺/Ca²⁺ exchanger ensures efficient extrusion of Ca²⁺ from mitochondria (12). Upon stimulation with InsP3-generating agonists, cytosolic Ca²⁺ transients reach the micromolar range and are paralleled by sharp mitochondrial [Ca²⁺] increases. This apparent conundrum is explained by the fact that ER and mitochondria share sites of near proximity, where the distance between the two organelles is extremely close (13, 14). These sites are named mitochondrial associated membranes (MAMs). At MAMs, Ca²⁺ released from the ER generates microdomains of high [Ca²⁺], which are sensed by mitochondria and which are responsible for the significant increase in mitochondrial [Ca²⁺]. The amount of Ca²⁺ uptake by mitochondria follows a non-linear relationship related to the amount of Ca²⁺ present at microdomains. Ca²⁺ uptake is inefficient not only at low [Ca²⁺], for the reasons explained above, and for the newly described role of MCU modulators (see below), but also in the presence of too high [Ca²⁺] at which the inhibitory effect of Ca²⁺ on InsP3R prevails. Thus, efficient mitochondrial Ca²⁺ uptake is warranted by the prompt removal of Ca²⁺ from microdomains exerted by mitochondria.

In 2004, Clapham and colleagues, by patch-clamping mitoplasts preparations, unequivocally identified a Ca²⁺-selective mitochondrial channel whose current had the properties of that mediated by MCU (4). Subsequent characterization demonstrated that different tissues display different MCU activity, according to the specific physiological properties of Ca²⁺ signaling in each tissue, suggesting that tissue-specific regulation of the channel occurs (15).

The idea that mitochondrial Ca²⁺ signaling regulates crucial hallmarks of cancer finds its roots in seminal studies investigating cell functions regulated by mitochondrial Ca²⁺ uptake. In particular, one of the first observations was that mitochondrial Ca²⁺ overload sensitizes cells to apoptotic stimuli, and this has been always viewed as a promising strategy to eliminate aberrant cells, including cancer cells, which would have otherwise escaped apoptotic death (3). In recent years, the role of mitochondrial Ca²⁺ uptake has evolved in parallel to novel hallmark of cancer (69), including the renewed interest for the deregulation of cellular energetics. On this topic, one of the conundrums that continue to puzzle researchers in the field is related to the so-called Warburg effect, i.e., the fact that cancer cells mainly rely on glycolysis-produced ATP despite normal pO₂ and despite being provided with functional mitochondria, in principle capable of efficient oxidative phosphorylation. The matter is even more intricate if one considers the fact that cancer cell lines (e.g., HeLa, MDA-MB-231, and others) promptly respond to InsP3-generating stimuli with high mitochondrial Ca²⁺ peaks. Clearly, efficient mitochondrial Ca²⁺ uptake in these cells is not primarily utilized to stimulate TCA cycle and oxidative phosphorylation and should have other functions, if any. Partial answers derive from studies of mitochondria metabolism that have unraveled important roles for mROS production, regulation of NADH/NADPH, and utilization of metabolic intermediates in cancer.

To verify whether correlations between MCU complex components expression, tumor progression, and prognosis exist, different approaches have been used. In a study by Hall et al., the Breastmark algorithm, which integrates many DNA microarray experiments for which clinical data are available, was interrogated to determine the relationship between expression of MCU and MICU1 and survival (70). The results of this study indicate that combined overexpression of MCU and underexpression of MICU1 is a feature of patients with poor prognosis, while overexpression of MICU1 and of MCUb, alone or in combination, correlates with better prognosis. A similar analysis was performed by interrogating the Oncomine microarray database bank. In this case, MCU expression was analyzed in different breast cancer subtypes in comparison to normal tissue. The analysis revealed that MCU is overexpressed in breast cancer and also that invasive ductal breast cancer expresses more MCU compared to ductal carcinoma in situ. Finally, MCU expression correlates with lymph node invasion (71). When different breast cancer subtypes where compared, a higher expression of MCU was highlighted in estrogen receptor-negative patient samples, especially in basal-like breast cancers compared to luminal A and B subtypes (72). A fourth study correlated mRNA levels of MCU and related proteins (MCUb, MICU1–3, and EMRE) from the TCGA breast cancer dataset with clinical stages (73) demonstrating that MCU expression increases with tumor size and lymph node infiltration, while MCUb expression decreases. Expression of related proteins, i.e., MICU1, MICU2, MICU3, and EMRE did not show significative correlation with tumor stages. In head and neck squamous cell carcinoma, MICU1 expression correlates with the expression of EZH2, which is a negative prognostic factor. EZH2 as well as MICU1 inhibition triggers cell cycle arrest and apoptosis. The EZH2 chemical inhibitor DZNep reduced MICU1 protein levels, suggesting that MICU1 could be involved in EZH2-regulated cell-death pathway (74). Finally, in pancreatic cancer MCU and MICU1 genes undergo loss of heterozygosity (75), although whether MCU and MICU1 are oncogenes or tumor suppressor genes still needs to be clarified. Further studies are required to clarify the expression profile of MCU components in a wider array of cancers.

The metabolic and biological effects of MCU complex modulation have been investigated in different cancers, although the heterogeneity intrinsic of tumor development suggests that much still needs to be explored. In addition, analyses of the expression profiles of the uniporter components in cancer specimens have unravel additional information on the role of MCU on cancer progression.

One of the first reports investigating the role of MCU in cancer progression identified miR-25 as a cancer-related MCU-targeting miRNA family. In detail, in colon adenocarcinoma, miR-25 is overexpressed and, accordingly, MCU expression is inhibited. In agreement with the role of mitochondrial Ca²⁺ uptake in controlling cell survival and death, both in prostate and in colon carcinoma cells, miR-25 overexpression reduces mitochondrial Ca²⁺ uptake and sensitization to apoptotic stimuli, while expression of an “antagomiR” against miR-25 had a deleterious effect on cells that where stressed with an apoptotic challenge (60).

Celastrol, a triterpene extracted from the plant Tripterygium wilfordii, was reported to induce mitochondrial Ca²⁺ uptake and ER stress in cancer cells. Celastrol treatment induced paraptosis of these cells, a form of programmed cell-death morphologically and biochemically distinct from apoptosis (76). Intriguingly, MCU silencing partially decreased celastrol-induced cancer cell death, supporting the view of a deleterious effect of Ca²⁺ uptake in stress conditions.

Besides resistance to apoptosis, a fundamental step toward tumor initiation and progression is escape from senescence, a stable proliferation arrest status that exerts a protective function against many pathologies, including cancer. MCU was identified, together with ITPR2, as a novel regulator of both replicative and oncogene-induced senescence (OIS) (77). During OIS, ITPR2 activity triggers Ca²⁺ release from the ER, which is followed by mitochondrial Ca²⁺ accumulation, loss of membrane potential, increased ROS production, and senescence. Vice versa, loss of MCU enables escape from OIS.

In a particular cancer model, i.e., triple negative breast cancer (TNBC), the role of MCU has been extensively investigated both in terms of cell viability and on migration/invasion properties (Table 1). Based on the expression data reported above, suggesting increased MCU expression in basal-like breast cancer, Curry et al. hypothesized a survival advantage triggered by MCU (72). Accordingly, MCU was silenced in MDA-MB-231 cell line and cell viability and resistance to stress stimuli was measured. MCU silencing per se did not affect at all cell proliferation and viability. Also, when cells were treated with the BCL-2 inhibitor ABT-263, caspase-dependent cell death was unaffected. However, caspase-independent death triggered by treatment with the Ca²⁺ ionophore ionomycin was potentiated by MCU knockdown, suggesting that MCU inhibition could in principle be viewed as a therapeutical strategy against TNBC.

A similar hypothesis was tested by Hall et al. (70). As reported above, their expression study convincingly correlated combined overexpression of MCU and underexpression of MICU1 with poor patient survival. Thus, MCU and MICU1 were silenced in MDA-MB-231 cells, and the effects on survival of cells exposed to irradiation, chemotherapeutic agents, or nutrient deprivation were measured. However, while cervical cancer HeLa cells and non-cancerous breast epithelial HMEC cells showed reduced survival, MDA-MB-231 cells were insensitive to changes in MCU activity. In addition, also overexpression of MCU was ineffective in altering cell response to ceramide-induced toxicity. Thus, the role of mitochondrial Ca²⁺ uptake in cancer cell survival depends on cancer type and on stress stimuli, not to mention the role that cancer niche and extracellular environment exert. In this context, the role of MCU is still obscure.

A different issue is the role that MCU plays in modulating motility and invasion potential of cancer cells. Two reports agreed in indicating MCU as an essential modulator of MDA-MB-231 cell migration (71, 73); however, different mechanisms were proposed. According to Tang et al., MCU silencing inhibited serum- or thapsigargin-induced store-operated Ca²⁺ entry (SOCE) which, in turn, was required for proper cell migration. Tosatto et al. confirmed the data that MCU is required for MDA-MB-231 migration and extended this observation to other triple-negative breast cancer cell lines. However, despite the consistent role on cell motility, the effect of MCU silencing on SOCE was cell line dependent, indicating that different mechanisms are involved, likely directly dependent on the strong reduction in mitochondrial Ca²⁺ uptake, rather than on a secondary effect on global Ca²⁺ homeostasis. MCU inhibition caused a strong reduction in mROS production and HIF-1α expression. In addition, restoring HIF-1α levels was sufficient to overcome the decrease in motility. The essential role of MCU in TNBC progression was demonstrated by in vivo experiments in which MDA-MB-231 cells were injected into the fat pad of immune-deficient SCID mice. Deletion of MCU markedly retarded primary tumor formation. In addition, at equal primary tumor size, absence of MCU impaired lung and lymph node colonization (73).

These studies indicate that the MCU complex plays different roles in different cancer types and stages. Research is still needed to extend the analysis to in vivo models and to additional cancers and to clarify whether mitochondrial Ca²⁺ uptake impinges on the tumor microenvironment.

An important amount of work has been done to disclose the pathophysiological role of MCU in different tissues (Figure 2).

On this issue, one of the first studies after the identification of MCU was carried out in neonatal cardiomyocytes ex vivo, demonstrating that MCU expression regulates the buffering of cytoplasmic Ca²⁺ during systole. In detail, cytoplasmic Ca²⁺ peaks are increased or reduced by MCU silencing or overexpression, respectively (78). However, the first in vivo analysis of MCU^−/− hearts gave unexpected results that are still debated. MCU^−/− hearts show no sign of protection against I/R injury (30). Interestingly, while wild-type hearts were protected from I/R injury when treated with CsA, which inhibits Ca²⁺-dependent cell death mediated by the opening of mPTP, MCU^−/− hearts lacked CsA protection. This result suggests that a Ca²⁺-independent death pathway occurs in the absence of MCU. Experiments on isolated heart mitochondria showed impaired mitochondrial Ca²⁺ accumulation and reduced matrix Ca²⁺ levels in MCU^−/− strain (79), consistent with the data previously reported on skeletal muscle isolated mitochondria (30). However, basal ATP levels were similar between MCU^−/− and wild-type hearts, suggestive of preserved basal metabolic function. In addition, an accurate study of heart morphology by echocardiographic analysis of 12- and 20-month-old mice revealed similar stroke volume, ventricular volumes, and wall thickness between MCU^−/− and wild-type controls, indicating that MCU^−/− mice maintain basal cardiac function over at least 20 months of life (79). Upon physiological stress induced by isoproterenol stimulation, left ventricular cardiac output and heart rate were undistinguishable from those of wild-type controls even though MCU^−/− cardiomyocytes did not accumulate Ca²⁺. Also upon chronic stress, induced by transverse aortic constriction (TAC), MCU^−/− and wild-type genotypes display similar decline in ejection fraction, fractional shortening, hypertrophy and fibrosis.

Additional studies were carried out in a transgenic mouse model, expressing a DN-MCU (80, 81). As expected, the heart mitochondria from DN-MCU mice showed loss of mitochondrial Ca²⁺ accumulation (80), although DN-MCU does not completely abolish mitochondrial Ca²⁺ uptake when expressed in HeLa cells (16). DN-MCU mice showed hearts with normal chamber size and heart beat. Increased heart rate triggered by isoproterenol treatment requires enhanced ATP production necessary to support SERCA activity and appropriate Ca²⁺ load of the sarcoplasmic reticulum in the SAN (sinus atrial node) cells. MCU inhibition impaired fight or flight heart rate by decreasing ATP levels, and dialysis of ATP was sufficient to rescue the phenotype in cardiac pacemaking cells (80). Enhanced oxygen consumption was a peculiar feature of DN-MCU perfused hearts, while it was not observed in isolated mitochondria or permeabilized cardiomyocytes of the same animals, suggesting that the increased energetic demand is related to extramitochondrial Ca²⁺ homeostasis. In addition, DN-MCU ventricular cardiomyocytes exhibited increased cytosolic [Ca²⁺], in line with the Ca²⁺ buffering effect of mitochondria, which was partially overturned by ATP dialysis (81). Consistent with results obtained in MCU^−/− mouse (30), DN-MCU expression was not protective against myocardial death induced by I/R injury, despite maintenance of ΔΨm and decreased ROS production (81).

In both mouse models discussed above, embryonic adaptations to MCU modulation could not be excluded. To overcome this issue, in 2015 a new mouse model was generated, with two LoxP sites flanking exons 5 and 6 of MCU gene. These animals were crossed with mice expressing a tamoxifen-inducible Cre recombinase driven by a cardiomyocyte-specific promoter (α-myosin heavy chain promoter). MCU gene deletion was induced in adult mice, and cardiac function was evaluated (82, 83). Similarly to DN-MCU and MCU total knockout mice, basal cardiac function and heart morphology were normal. On the other hand, studies of inducible cardiac-specific MCU^−/− mice confirmed the impairment in the fight or flight response. Cardiomyocytes derived from these mice had normal respiration rate in basal conditions, while decreased oxygen consumption rate after β-adrenergic stimulation suggesting a MCU pivotal role during increase heart workload. However, no differences were observed in fractional shortening, cardiomyocytes area and heart weight upon TAC. Finally, as opposed to constitutive MCU^−/− mice and DN-MCU animals, inducible MCU ablation strongly protected hearts from I/R-induced cell death.

Besides heart, another tissue strictly dependent on oxidative metabolism is skeletal muscle. Skeletal muscle is the most affected tissue in constitutive MCU^−/− mice although at very mild levels. Skeletal muscle mitochondrial matrix [Ca²⁺] was reduced compared to wild-type controls, although not completely abolished. A decrease maximal oxygen consumption rate was detected. In addition, PDH activity was impaired in agreement with the pivotal role of mitochondrial Ca²⁺ accumulation in the control of ATP production. Finally, MCU^−/− animals show defective maximal power output generation. Subsequently, to discriminate the compensatory effects occurring during embryonic development, MCU modulation was achieved after birth by means of adeno-associated viral vectors for both overexpression and silencing (84). Modulation of mitochondrial Ca²⁺ accumulation was shown to contribute to skeletal muscle trophism. Indeed, MCU overexpression and downregulation trigger muscle hypertrophy and atrophy, respectively, both during postnatal development and in adulthood. Most importantly, MCU overexpression protects from denervation-induced muscle atrophy caused by sciatic nerve excision suggesting a possible therapeutic role of mitochondrial Ca²⁺ uptake in muscle atrophy. Surprisingly, these effects are independent from the control of aerobic metabolism as demonstrated by different evidence. First, PDH activity, although defective in MCU silenced muscles, was unaffected in MCU overexpressing muscles. Second, hypertrophy was comparable in both oxidative and glycolytic muscles and, finally, analyses of aerobic metabolism revealed no major alterations. Conversely, the control of skeletal muscle mass by mitochondrial Ca²⁺ modulation is due to the role of two major hypertrophic pathways of skeletal muscle, PGC-1α4 and IGF1-AKT/PKB. Taken together, these results demonstrate the existence of a Ca²⁺-dependent mitochondria-to-nucleus signaling route that clearly links organelle physiology to the control of muscle mass. As mentioned above, loss-of function mutations in MICU1 genes cause a disease which includes neuromuscular defects and learning difficulties (43), and MICU1 deletion in mice is either lethal (44) or causing similar phenotype to the human disease (32). Finally, a recent work carried out by Zampieri and coworkers reported increased skeletal muscle MCU protein levels in human seniors subjected to muscle training, suggesting that MCU represents a potential pharmacological target to counteract sarcopenia (85).

Beside striated muscles, the role of MCU has been studied also in other organs in which mitochondria play a crucial role for tissue-specific functions. One case is represented by endocrine pancreas, specifically β-cells. Soon after MCU discovery, experiments carried on in pancreatic β-cells demonstrated that MCU-mediated mitochondrial Ca²⁺ uptake is essential for cell depolarization-induced increase in ATP/ADP ratio (86). In addition, culture conditions mimicking diabetic milieu delayed mitochondrial Ca²⁺ uptake. Thus, MCU is an essential component of glucose sensing by pancreatic β-cells.

In a totally different context, a recent article reported evidence that mitochondrial [Ca²⁺] could have an impact on macrophage polarization (87). Gu and coworkers demonstrated that MCU^+/- mice are protected from asbestos-induced pulmonary fibrosis. In detail, macrophages of MCU^+/- mice showed attenuated profibrotic polarization after asbestos exposure together with a decrease in ATP production. In addition, histological analysis revealed normal lung architecture without collagen deposition in MCU^+/- mice. These results suggest that the modulation of mitochondrial Ca²⁺ accumulation could be a promising therapeutic target for human fibrosis.

Additional demonstration of the pivotal role of mitochondrial Ca²⁺ uptake in pathophysiology is described by Rimessi and coworkers (88). This study analyzes the Pseudomonas aeruginosa-induced inflammatory process that takes place in cystic fibrosis cells. When inflammation is triggered in a cystic fibrosis cell line model, MCU silencing causes a decrease in inflammasome activation, indicating that mitochondrial Ca²⁺ accumulation is a potential therapeutic target.

Finally, MCU is a target of miR-25, and miRNA-dependent MCU modulation has consequences on different physiopathological settings, as detailed in the Sections “MCU Activity Modulation” and “MCU and Cancer” (60, 61). Also miR-138 has been reported to play a regulatory role on MCU in the context of PAH (62).

The seminal discoveries of cbara1 as the gene encoding MICU1, and of ccdc109a as the one encoding MCU, followed by the identification of the other components of the uniporter, determined the possibility of conducting genetic and molecular studies on mitochondrial Ca²⁺ uptake mechanism and significance. In a simplified view, so far research has progressed in different directions to define: (a) the identity and role of uniporter channel components and interacting subunits, and their mutual regulation; (b) the transcriptional and posttranscriptional mechanisms that contribute to channel activity control; and (c) the pathophysiological consequences of dysregulated mitochondrial Ca²⁺ signaling. For any of these topics, impressive progresses have been achieved in a relatively short time. However, many questions are still unanswered. First, work is still needed to better clarify how the different channel subunits and regulatory components functionally interact with each other and to determine cell-type specificity of these mechanisms. Second, the regulation of uniporter components expression and activity by transcriptional and posttranscriptional events is a still largely unexplored field. Third, the pathophysiological role of MCU in different tissues is still unclear. Specifically on tumor development and progression, the role and significance of mitochondrial Ca²⁺ signaling has been studied only in few cancer types, and nothing is known on the effects of MCU activity in the tumor niche. In addition, although patient-derived samples have been analyzed for the expression levels of MCU components and correlation data between expression and prognosis have been obtained, in vivo tumorigenesis experiments aimed to determine the significance of aberrant mitochondrial Ca²⁺ uptake at different cancer stages have been rarely performed. Research in the next years will uncover these and others aspects of mitochondrial Ca²⁺ signaling.

CM, GG, and RR wrote the manuscript. GG prepared figures.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.