Pharmacological modulation of mitochondrial calcium homeostasis

Abstract

Mitochondria are pivotal organelles in calcium (Ca²⁺) handling and signalling, constituting intracellular checkpoints for numerous processes that are vital for cell life. Alterations in mitochondrial Ca²⁺ homeostasis have been linked to a variety of pathological conditions and are critical in the aetiology of several human diseases. Efforts have been taken to harness mitochondrial Ca²⁺ transport mechanisms for therapeutic intervention, but pharmacological compounds that direct and selectively modulate mitochondrial Ca²⁺ homeostasis are currently lacking. New avenues have, however, emerged with the breakthrough discoveries on the genetic identification of the main players involved in mitochondrial Ca²⁺ influx and efflux pathways and with recent hints towards a deep understanding of the function of these molecular systems. Here, we review the current advances in the understanding of the mechanisms and regulation of mitochondrial Ca²⁺ homeostasis and its contribution to physiology and human disease. We also introduce and comment on the recent progress towards a systems‐level pharmacological targeting of mitochondrial Ca²⁺ homeostasis.

Introduction

In virtually all eukaryotic cells, mitochondria are primed organelles for the regulation of intracellular Ca²⁺ signalling. They can take up and release Ca²⁺ ions through the concerted action of influx and efflux pathways to shape the spatio‐temporal dynamics of cytosolic Ca²⁺ transients (Nicholls, 1978; Nicholls & Scott, 1980; Zoccarato & Nicholls, 1982). In addition, mitochondrial Ca²⁺ entry is coupled to an increase in energy production (Denton & McCormack, 1980) to meet the rising energy demand of signalling cells (Szabadkai & Duchen, 2008). Mitochondrial Ca²⁺ homeostasis has been extensively studied for decades and functionally linked to the regulation of diverse physiological processes such as energy metabolism, tissue growth and development, neurotransmitter release, muscle contraction, autophagy, and cell death. More than half a century ago, it was shown that Ca²⁺ uptake by isolated mammalian mitochondria was a rapid, electrogenic and relatively selective process that was driven by the negative potential (∼−180 mV) across the inner mitochondrial membrane (IMM) and inhibited by lanthanide cations and ruthenium red (RuR) (Deluca & Engstrom, 1961; Vasington & Murphy, 1962; Chance, 1965; Brand et al. 1976). At the same time, it was demonstrated that the presence of Na⁺ in the medium and changes in pH were able to stimulate the efflux of Ca²⁺ accumulated in isolated mitochondria (Carafoli et al. 1974). It was thus concluded that both mitochondrial Ca²⁺ influx and mitochondrial Ca²⁺ efflux occurred through specific carriers located in the IMM, a putative and ubiquitous mitochondrial Ca²⁺ uniporter and either a Na⁺/Ca²⁺ or a H⁺/Ca²⁺ exchanger, in excitable and non‐excitable tissues, respectively. However, although much progress has been made to elucidate the biophysical properties of mitochondrial Ca²⁺ entry and release, the molecular identities of these transporters remained elusive until 2010. Then, the mitochondrial Ca²⁺ uniporter 1 (MICU1; Perocchi et al. 2010) was discovered as the founding member of the mitochondrial Ca²⁺ uniporter complex and the Li‐permeable Na⁺/Ca²⁺ exchanger (NCLX; Palty et al. 2010) as a new member of the Na⁺/Ca²⁺ exchanger superfamily, opening a new era in the mitochondria and Ca²⁺ signalling fields. These findings allowed targeted genetic loss‐ and gain‐of‐function analyses in vitro and in vivo towards a comprehensive assessment of the physiological and pathophysiological role of mitochondrial Ca²⁺ homeostasis in cells, tissues and whole organisms (De Stefani et al. 2016). Moreover, the identification of MICU1 loss‐of‐function mutations in human patients with neuromuscular disorders (Logan et al. 2014; Lewis‐Smith et al. 2016) has confirmed that modulating mitochondrial Ca²⁺ homeostasis might constitute a promising therapeutic approach.

This review attempts to provide an overview of the latest knowledge on mitochondrial Ca²⁺ homeostasis and its modulation for therapeutic intervention in human diseases. First, we summarize the findings that led to the identification and characterization of the molecular components involved in mitochondrial Ca²⁺ uptake and efflux. We then discuss the relevance of mitochondrial Ca²⁺ homeostasis in physiology and disease and the so far available pharmacological approaches to modulate mitochondrial Ca²⁺ uptake and release pathways. Finally, we address recent progresses on pharmacological targeting of mitochondrial Ca²⁺ homeostasis and present a newly established high‐throughput chemical screening approach for the systematic identification of modulators of the mitochondrial Ca²⁺ uniporter (Arduino et al. 2017).

The mitochondrial Ca²⁺ uniporter

Many attempts have been made to identify the components of the mitochondrial Ca²⁺ uniporter, for example by purifying Ca²⁺‐binding proteins from isolated mitochondria and then reconstituting Ca²⁺ uptake in vitro. The group of Lehninger reported on the purification of a soluble heat‐labile protein of over 150 kDa from rat liver mitochondria subjected to osmotic shock (Lehninger, 1971). This factor was able of binding Ca²⁺ with high affinity and it was sensitive to lanthanides. Prompted by the serendipitous finding that mitochondria were stained by the polysaccharide stain RuR and that this completely inhibited mitochondrial Ca²⁺ uptake, Sottocasa and colleagues focused on the extraction of glycoproteins as possible active centres of Ca²⁺ binding and transport (Sottocasa et al. 1972). From ox and rat liver mitochondria they isolated a major band with a molecular mass of approximately 30 kDa by PAGE, which was stained by murexidine, a Ca²⁺ dye, and whose high‐affinity binding (K _d of approximately 100 nm) was affected by both lanthanides and RuR. This proteinaceous factor, named ‘calvectin’, was found primarily in the intermembrane space in a free soluble form, but could become membrane bound in a Ca²⁺‐dependent manner. Calvectin was able to increase Ca²⁺ conductance in artificial lipid bilayers and its antisera specifically inhibited Ca²⁺ transport by rat liver mitoplasts (mitochondria stripped of their outer membrane) without affecting respiration. This protein was therefore suggested to represent a major component of the mitochondrial Ca²⁺ uniporter. Other laboratories also suggested that the components responsible for Ca²⁺ transport in mitochondria were not of protein but rather of lipid nature. A lipid moisture isolated from the Ca²⁺‐transported glycoprotein was shown to form channels in lipid bilayers with an electroconductivity between 100 and 200 times higher in the presence of Ca²⁺ and the ability to transport Ca²⁺ was lost in delipidated mitochondria (Medvedev et al. 1982). Similarly, Sokolove and Brenza (1983) demonstrated that cardiolipin, an abundant lipid of mitochondrial membranes, binds Ca²⁺ with an apparent affinity of 700 nm in the absence of other competitive divalent cations such as Mn²⁺, Zn²⁺ and Mg²⁺. Calcium binding to cardiolipin was also inhibited by lanthanides and RuR. After these pioneering reports, various Ca²⁺ binding glycoproteins and peptides were isolated, all of which were able to bind Ca²⁺ and were inhibited by lanthanides and RuR (Mironova & Utesheva Zh, 1989; Zazueta et al. 1998). Several of these factors have been dismissed as components of the Ca²⁺ uniporter, as they were then found to be either cytosolic or microsomal contaminants and the majority of those observations faded away with lack of follow‐up studies.

Much later, electrophysiological measurements of Ca²⁺ currents in single mitoplasts demonstrated that the mitochondrial Ca²⁺ uniporter is an inward rectifying channel, highly selective for Ca²⁺, and with a remarkable high capacity for Ca²⁺ transport (half‐saturation at ∼20 mm) (Kirichok et al. 2004), and several proteins, such as the mitochondrial type 1 ryanodine receptor (mRyR1) (Beutner et al. 2005) and the isoforms 2 and 3 of the ion‐transporting uncoupling protein UCP (UCP2 and UCP3) (Trenker et al. 2007), were initially proposed to form this channel. However, the direct involvement of these proteins in mediating RuR‐sensitive Ca²⁺ uptake has been questioned. Thus, even after so much effort spanning several decades of research and when the majority of the new genomics and proteomics approaches were already available, the molecular nature of the uniporter had remained a mystery to molecular biologists, biochemists and physiologists alike. In 2010, Perocchi et al. identified MICU1 as the first genuine member of the uniporter channel based on the conceptual integration of physiological and biophysical clues from 50 years of literature on mitochondrial Ca²⁺ homeostasis in different organisms with comparative genomics and mitochondrial proteomics analyses (Perocchi et al. 2010). Analyses performed during the 1960s and ’70s had shown that uniporter‐dependent Ca²⁺ uptake is not a prerogative of all eukaryotic organisms, as it can be measured in mitochondria of mammals and kinetoplastids but not in fungi. Therefore, human genes encoding core components of the uniporter should be conserved in the mitochondrial proteome of organisms that are competent for Ca²⁺ uptake but not in other species devoid of such ability. Of the thousand plus human proteins localized to mitochondria, the authors narrowed down the search to just 18 that fulfilled the evolutionary signature of the uniporter, were associated with the IMM and were widely expressed in mitochondria from several mouse tissues. RNA interference (RNAi) against these 18 genes highlighted one, CBARA1, which was thereafter renamed MICU1 (Perocchi et al. 2010). Silencing MICU1 did not disrupt mitochondrial respiration or membrane potential but impaired mitochondrial Ca²⁺ entry in intact and permeabilized cells, as well as in mitochondria isolated from mouse liver. It also attenuated the metabolic coupling between cytosolic Ca²⁺ transients and stimulation of energy metabolism. MICU1 had two predicted EF hands that were found to be essential for regulating the activity of the uniporter, suggesting a role in Ca²⁺ sensing. Building on this discovery, the same authors and the group of Rizzuto (Baughman et al. 2011; De Stefani et al. 2011) identified a 40 kDa coiled‐coiled protein, known as CCDC109A, that coevolved with MICU1 and had two transmembrane domains, which they proved to be the bona fide, pore‐forming subunit of the mitochondrial calcium uniporter and threrefore renamed it mitochondrial Ca²⁺ uniporter (MCU). First, RNAi‐mediated silencing of MCU was found to strongly inhibit mitochondrial Ca²⁺ uptake in cultured cells and in purified mouse liver mitochondria, whereas MCU overexpression enhanced RuR‐sensitive mitochondrial Ca²⁺ uptake in intact and permeabilized cells. Single‐point mutations of conserved acidic residues (E257A, D261A and E263A) within the short loop linking the two transmembrane domains of MCU on the intermembrane space side, termed the DIME motif, abrogated the ability of mitochondria to take up Ca²⁺, indicating these residues were required for Ca²⁺ transport (Baughman et al. 2011; De Stefani et al. 2011; Chaudhuri et al. 2013). Expression of S259A (Baughman et al. 2011; Chaudhuri et al. 2013) and D261A MCU (Arduino et al. 2017) mutants conferred reduced RuR sensitivity, highlighting a possible role in RuR‐dependent MCU inhibition. Second, expression of pure MCU in planar lipid bilayers was shown to be sufficient to reconstitute ion channel activity, generating a single channel Ca²⁺ current consistent with that of the uniporter (De Stefani et al. 2011). Finally, heterologous expression of the MCU orthologous protein from Dictyostelium dyscodium was sufficient to reconstitute mitochondrial Ca²⁺ uptake capacity in the yeast Saccharomyces cerevisiae, which lacks all uniporter components (Kovacs‐Bogdan et al. 2014). Altogether, these observations confirmed that MCU is the Ca²⁺‐conducting pore of the long‐sought mitochondrial uniporter. Moreover, recent studies on the structural biology (Oxenoid et al. 2016), coordination chemistry and molecular dynamics (Cao et al. 2017) of MCU from Caenorhabditis elegans suggested that a functional channel results from the pentameric assembly of MCU multimers stabilized by a coiled‐coil motif protruding into the mitochondrial matrix. In this structural model, the DIME motif forms the selectivity filter of the channel, whereby ion permeation and inhibitor binding occur. These insights have been recently exploited for in silico modelling of MCU and binding of mitoxantrone, a novel inhibitor of the uniporter channel (Arduino et al. 2017).

MCU was initially found to reside in the IMM as a large complex of roughly 450 kDa (Baughman et al. 2011). Over the past few years, the macromolecular nature of this channel was confirmed and several additional subunits have been characterized. MCU can form hetero‐oligomers with its paralogue, MCUb, a 33 kDa protein with very similar amino acid sequence and topological features (Raffaello et al. 2013). Co‐expression studies of MCU and MCUb in planar lipid bilayers and intact cells demonstrated that the presence of MCUb decreases MCU‐dependent Ca²⁺ uptake activity, suggesting that MCUb is a dominant‐negative pore‐forming component of the uniporter. The regulation of MCU activity by extramitochondrial Ca²⁺ is dictated by a duet of EF‐hand containing proteins, MICU1 and its paralogue EFHA1 (termed MICU2), interacting with MCU from the mitochondrial intermembrane space (Plovanich et al. 2013; Patron et al. 2014; Liu et al. 2016; Kamer et al. 2017). A related protein termed MICU3 was also proposed to be a member of the uniporter based on its sequence similarity to MICU2, but neither its mitochondrial localization nor its involvement in Ca²⁺ uptake have been confirmed so far. Following a similar strategy to Perocchi et al. (2010), Mallilarankaraman and colleagues identified the mitochondrial Ca²⁺ uniporter regulator 1 (MCUR1), formerly known as CCDC90A, as an essential membrane component of the uniporter (Mallilankaraman et al. 2012; Vais et al. 2015). Interestingly, MCUR1 was shown to have an opposite membrane topology to MCU, whereby both N‐ and C‐termini project into the intermembrane space, while the majority of the protein, including its coil‐coil domain, faces the mitochondrial matrix, suggesting that unlike MCU and MCUb, MCUR1 does not act as a pore‐forming subunit. In this regard, MCUR1 was recently suggested to function as a scaffold factor for the assembly of the uniporter channel (Tomar et al. 2016), although other independent roles have also been ascribed to it, such as the assembly of complex IV (Paupe et al. 2015) and the regulation of Ca²⁺‐induced mitochondrial permeability transition (Chaudhuri et al. 2016). Finally, a 10 kDa integral IMM protein termed essential MCU regulator (EMRE) was found to be indispensable for uniporter‐mediated Ca²⁺ uptake and together with MCU was sufficient to reconstitute Ca²⁺ uptake in yeast mitochondria (Sancak et al. 2013; Kovacs‐Bogdan et al. 2014). EMRE is thought to be a multifunctional element of the uniporter that is required for inwardly rectifying Ca²⁺ currents, bridging of MCU and its regulators MICU1 and MICU2 (Sancak et al. 2013; Kovacs‐Bogdan et al. 2014; Petrungaro et al. 2015; Tsai et al. 2016) and sensing of mitochondrial matrix Ca²⁺ (Vais et al. 2016). In contrast to its essential role in higher eukaryotes, EMRE expression is absent in some organisms where MCU is conserved and functional, such as D. discoideum (Kovacs‐Bogdan et al. 2014). To date, the precise role of this protein in both function and regulation of the uniporter remains to be clearly elucidated. In addition to this plethora of structural and regulatory subunits, both uniporter activity and composition can also vary across tissues through the expression of tissue‐specific components. For example, MICU2 and MICU3 were found to be mostly expressed in visceral organs and central nervous system, respectively (Plovanich et al. 2013; Murgia & Rizzuto, 2015), whereas a recently found MICU1 alternative splice variant, MICU1.1, is skeletal muscle‐specific (Vecellio Reane et al. 2016). The discovery of the major molecular players of mitochondrial Ca²⁺ homeostasis constitutes a great accomplishment for the mitochondrial biology and Ca²⁺ fields, and additional elements might be on the research horizon.

Mitochondrial Ca²⁺ efflux

Outward Ca²⁺ flux from mitochondria is mainly mediated by two transport systems, mitochondrial Na⁺/Ca²⁺ (mNCLX) and H⁺/Ca²⁺ (mHCX) exchangers. By a genome‐wide siRNA screen in Drosophila cells, the leucine zipper and EF‐hand containing transmembrane protein 1 (LETM1) was found to regulate RuR‐sensitive mitochondrial Ca²⁺ uptake at cytosolic concentrations below, but not above, 1 um and to couple the movement of Ca²⁺ in exchange for H⁺ both in intact cells and in proteoliposomes (Jiang et al. 2009). Thus, LETM1 was proposed to function as the mitochondrial electroneutral H⁺/Ca²⁺ antiporter. However, previous studies had characterized the same protein as a member of the mitochondrial K⁺/H⁺ exchanger and shown that it was essential to control mitochondrial volume (Nowikovsky et al. 2004; McQuibban et al. 2010). The hypothesis that LETM1 could mediate both extrusion and uptake of Ca²⁺ independently of K⁺ was also put forward (Jiang et al. 2013; Tsai et al. 2014) and later challenged based on evidence for electroneutral exchange of H⁺ with K⁺ in vivo, thus disclaiming the contribution of LETM1 to mitochondrial Ca²⁺ efflux (De Marchi et al. 2014; Nowikovsky & Bernardi, 2014). Despite the renowned relevance of mitochondrial outward shuttling mechanisms for cell physiology, the molecular identity of mHCX still remains unclear.

In a seminal study by Palty et al. (2010), it was found that the gene FLJ22233 (also known as SLC8B1) encodes an IMM‐located protein that mediates both Li⁺‐ and Na⁺‐dependent Ca²⁺ clearance from the mitochondrial matrix. The authors named this protein mNCLX and demonstrated that it constitutes the mitochondrial Ca²⁺ antiporter. Similarly to MCU‐dependent mitochondrial Ca²⁺ uptake, the action of mNCLX is electrogenic, with 3 Na⁺/1 Ca²⁺ ions per transport cycle. Consequently, mNCLX constitutes the main pathway for Na⁺ influx into mitochondria, and its activity is finely tuned to sense changes in intracellular Na⁺ concentrations within the physiological range. The discovery of the molecular identity of mNCLX (Palty et al. 2010) was instrumental for understanding the molecular mechanisms of mitochondrial Ca²⁺ homeostasis. Accordingly, a functional crosstalk between mNCLX and other Na⁺ and Ca²⁺ transporters located on both mitochondrial and plasma membranes has been described (Nita et al. 2015; Ben‐Kasus Nissim et al. 2017), underscoring the significant physiological role of mNCLX for the regulation of both Ca²⁺ and Na⁺ signalling pathways.

Physiopathological role of mitochondrial Ca²⁺ homeostasis

Genetic and biochemical strategies modulating mitochondrial Ca²⁺ homeostasis have already demonstrated that privileged functional relationships exist between mitochondrial Ca²⁺ homeostasis, cell bioenergetics and cell fate‐determination pathways. Calcium accumulation in functional mitochondria, mediated by the MCU complex (MCUC) and modulated by mNCLX, regulates mitochondrial ATP generation as well as cytosolic NAD⁺/NADH metabolism, thus sustaining the energy requirements of the cell. Ca²⁺ loading in the mitochondrial matrix stimulates mitochondrial generation of NADH, which can be shuttled to the cytosol where it initiates retrograde signalling, resulting in inhibition of sirtuins activity and alterations in protein acetylation (Marcu et al. 2014). Moreover, Ca²⁺ constitutively released from the ER to mitochondria is essential for efficient oxygen consumption, maintenance of cell bioenergetics and inhibition of autophagy (Cardenas et al. 2010). Conversely, mitochondrial Ca²⁺ overload, or perturbations in mitochondrial Ca²⁺ homeostasis, can sensitize cells to distinct modes of cell death induced by different stimuli, and in a cell type‐specific way (Curry et al. 2013; Qiu et al. 2013; Liao et al. 2015; Cardenas et al. 2016).

A large body of evidence from experimental models and human subjects corroborates the notion that mitochondrial Ca²⁺ deregulation is a relevant feature of several human disorders. As a remarkable example, its role in cardiovascular pathologies has been supported by several reports demonstrating that in heart failure, elevated cytosolic levels of Na⁺, which stimulates mitochondrial Ca²⁺ efflux via the mNCLX, reduce mitochondrial bioenergetic responses and promote mitochondrial oxidative stress in cardiomyocytes (Maack et al. 2006). In contrast, mitochondrial Ca²⁺ overload caused by sarcoplasmic reticulum (SR) Ca²⁺ leakage elicits mitochondrial dysfunction and contributes to impaired cardiac function in postischaemic heart failure (Santulli et al. 2015). Also in this pathological context, loss‐of‐function mutations and genetic ablation of the major players in mitochondrial Ca²⁺ influx and efflux pathways in in vivo models have demonstrated that mitochondrial Ca²⁺ homeostasis is a determining factor in cardiac physiology. Cardiac‐specific deletion of mNCLX was recently found to elicit severe heart failure, predisposing mice to sudden death (Luongo et al. 2017). Moreover, adult inducible cardiac‐specific deletion of MCU and transgenic expression of a dominant‐negative MCU (DN‐MCU) isoform impaired the ability of the heart to adapt to certain stresses that require increases in mitochondrial metabolism, for example the heart's ‘fight‐or‐flight’ stress response associated to an increased cardiac contractile performance (Kwong et al. 2015; Luongo et al. 2015; Wu et al. 2015). In addition, inducible MCU ablation conferred significant protection against ischaemia–reperfusion injury in adult heart (Kwong et al. 2015), as expected from the anticipated role of mitochondrial Ca²⁺ overload on induction of cell death. In contrast to these findings, mice constitutively lacking MCU or expressing the DN‐MCU did not exhibit marked cardiac deficits, but a pathological response to ischaemia–reperfusion injury and only a slightly impaired skeletal muscle metabolism and peak performance (Pan et al. 2013; Holmstrom et al. 2015; Rasmussen et al. 2015). The discrepancies observed between the phenotypes of different MCU knockout mouse models as well as between in vivo and in vitro cellular models may result from both timing and duration of the genetic perturbation (e.g. acute versus chronic MCU deletion). Moreover, with regard to the mouse models, while MCU deletion at the whole‐organism level is embryonically lethal in a pure C57BL/6 mouse strain, an MCU knockout mouse model could, however, be established in a mixed C57BL/6xCD1 line (Pan et al. 2013). This paradox can be explained by an increase in the genetic variability of the outbred C57BL/6xCD1 strain that may result in compensatory mechanisms for the lack of MCU, and thus in milder phenotypes. However, the nature of those compensatory mechanisms is still unknown. Moreover, whole‐body deletion of MICU1 in mice was perinatally lethal, though no major anatomical deficiencies were observed. Instead, conditional MICU1 knock down in liver resulted in extensive inflammation, tissue damage and failure of the liver's regeneration ability (Antony et al. 2016). In an independent mouse model established in a different strain with a slightly different background, homozygous MICU1 deletion caused a significant but partial postnatal mortality (Liu et al. 2016). Similar to human patients carrying loss‐of‐function MICU1 mutations, viable MICU1 knockout mice revealed a strong phenotype, including atypical mitochondrial morphology, reduced muscle ATP levels and features of neuromuscular disorders such as ataxia (Logan et al. 2014). Reminiscent of these features, silencing of MCU and MICU1 in Drosophila melanogaster during development in a brain region essential for memory resulted in impaired memory formation in adulthood without affecting the learning ability (Drago & Davis, 2016). In contrast to the lethality phenotype observed for the MICU1‐null mice, constitutive loss of MICU2 in pure C57BL/6 background mice was recently found to be compatible with life and development, although leading to alterations in cardiovascular homeostasis associated with diastolic heart failure (Bick et al. 2017). All the above‐mentioned findings together with further evidence from other studies (Table 1) suggest that mitochondrial Ca²⁺‐handling proteins are relevant for drug targeting.

Table 1.

Phenotypic analysis of the mitochondrial Ca²⁺ uniporter complex components after genetic manipulation in the context of a whole organism or a specific tissue

Gene	Specie	Organ/tissue	Genetic perturbation	Phenotype	Refs/source
Mitochondrial Ca2+ influx
MCU	M. musculus	Whole organism (C57BL/6)	KO	Embryonic mortality	IMPC
		Whole organism (C57BL/6xCD1)	KO	Small size Intolerance to strenuous work	Pan et al. (2013)
		Whole organism	KO, single allele	Decreased cardiac stroke volume Increased fasted circulating glucose levels	IMPC
		Skeletal muscle	OE	Skeletal muscle hypertrophy	Mammucari et al. (2015)
			KD	Skeletal muscle atrophy
		Heart	DN – MCU expression	Impaired cathecolamine‐induced heart rate increase	Wu et al. (2015)
			KO	Absent responsiveness to acute stress and “fight‐or‐flight” response Protection from ischaemia/reperfusion injury	Luongo et al. (2015) Kwong et al. (2015)
	D. melanogaster	Mushroom body neurons	KD (during pupation)	Impaired adult memory formation	Drago & Davis (2016)
	D. rerio	Whole organism	KD	Alterations during gastrulation	Prudent et al. (2013)
	C. elegans	Whole organism	KO	Impaired wound healing	Xu & Chisholm (2014)
	T. brucei	Whole organism	KD/KO	Defective growth	Huang et al. (2013)
			OE	Decreased infectivity to mice
MCUb	M. musculus	Whole organism	KO	Impaired righting response Absent vibrissae Increased mean corpuscular volume	IMPC
MCUR1	M. musculus	Heart	KO	Partial perinatal mortality	Tomar et al. (2016)
		Vasculature	KO	Small size Postnatal mortality
MICU1	H. sapiens	Whole organism	Loss‐of‐function mutations	Features of neuromuscular disease Cognitive impairment	Logan et al. (2014)
		Whole organism	Deletion (exon 1)	Early‐onset proximal muscle weakness Exercise‐induced lethargy Confusion	Lewis‐Smith et al. (2016)
	M. musculus	Whole organism (C57BL/6)	KO	Perinatal mortality	Antony et al. (2016)
		Whole organism (C57BL/6N or C57BL/6NxJF1)	KO	Partial perinatal mortality Severe neurological and myophatic complications (in surviving mice)	Liu et al. (2016)
		Liver	KD	Extensive liver inflammation and damage Impaired liver regeneration	Antony et al. (2016)
MICU2	M. musculus	Whole organism	KO	Abnormal active cardiac relaxation Lethal abdominal aortic aneurisms (in hypertensive mice)	Bick et al. (2017)
EMRE	M. musculus	Whole organism	KO	Embryonic mortality	IMPC
	M. musculus (MICU1 −/−)	Whole organism	KO, single allele	Survival (rescue of high perinatal mortality)	Liu et al. (2016)
Mitochondrial Ca2+ efflux
NCLX	M. musculus	Heart	KO	Significant postnatal mortality Cardiac hypertrophy (in surviving mice)	Luongo et al. (2017)

C. elegans, Caenorhabditis elegans; D. melanogaster, Drosophila melanogaster; DN‐MCU, dominant negative MCU; D. rerio, Danio rerio; EMRE, essential MCU regulator; H. sapiens, Homo sapiens; IMPC, International Mouse Phenotyping Consortium; KD, knockdown; KO, knockout; MCU, mitochondrial calcium uniporter; MCUb, mitochondrial calcium uniporter b; MCUR1, mitochondrial calcium uniporter regulator 1; MICU1, mitochondrial calcium uptake 1; MICU2, mitochondrial calcium uptake 2; M. musculus, Mus musculus; NCLX, Li‐permeable Na⁺/Ca²⁺ exchanger; OE, overexpression; T. brucei, Trypanosoma brucei.

Pharmacological modulation of mitochondrial Ca²⁺ homeostasis

For more than 30 years, the only approaches and tools available to address the physiopathological importance of mitochondrial Ca²⁺ homeostasis relied on the use of pharmacological compounds that impinge on mitochondrial Ca²⁺ uptake and release processes.

Since the late 1960s, pioneer reports have demonstrated that lanthanides (particularly La³⁺) and the transition metal derivative RuR are powerful inhibitors of Ca²⁺‐associated responses in mitochondria (Mela, 1968, 1969). Studies by Vasington et al. (1972) and Reed & Bygrave (1974) established RuR and lanthanides as tightly binding inhibitors of Ca²⁺ transport in mitochondria, competing for different binding sites, which differ from the Ca²⁺‐transport site. Years later, an oxygen‐bridged dimeric Ru amine complex, purified from RuR, which absorbs light at 360 nm (named Ru360), was reported to exert a more specific inhibition of mitochondrial Ca²⁺ uptake in vitro (Ying et al. 1991; Matlib et al. 1998; Zazueta et al. 1998) and in situ in intact cardiomyocytes (Matlib et al. 1998). Other compounds that directly target additional mitochondrial functions, by dissipating the ΔΨ_m (e.g. uncouplers such as carbonyl cyanide‐p‐trifluoromethoxyphenylhydrazone (FCCP)/carbonyl cyanide m‐chlorophenyl hydrazone (CCCP) or dinitrophenol), inhibiting the ability to synthetize or transport ATP (oligomycin), and blocking the respiratory chain complexes (e.g. rotenone, antimycin A, potassium cyanide), have also an inhibitory effect on mitochondrial Ca²⁺ uptake. Several compounds are also known to inhibit mitochondrial Ca²⁺ extrusion through the mNCLX. Among those are benzothiazepines and benzodiazepines, including diltiazem, clonazepam and amiloride, as well as related compounds such as semotiadil, verapamil, trifluoroperazine, tetraphenilphosphonium and cyclosporine A (Jurkowitz et al. 1983; Matlib et al. 1983; Wolkowicz et al. 1983; Wingrove & Gunter, 1986; Wei et al. 2011). Of this class, the benzothiazepine CGP‐37157 was shown to be the most potent and selective inhibitor of mNCLX (Matlib et al. 1983; Chiesi et al. 1988), with a 10‐fold higher affinity than any other inhibitor.

The inhibitory properties of these compounds on mitochondrial Ca²⁺ dynamics are especially relevant in pathological situations where mitochondrial Ca²⁺ overload has been shown to be detrimental. In fact, protective effects of approaches that moderate mitochondrial Ca²⁺ accumulation have been demonstrated for different disease contexts. Chemical inhibition by RuR and Ru360 improved the functional recovery of hearts after ischaemia and suppressed arrhythmias and haemodynamic dysfunction elicited by reperfusion in vivo (Garcia‐Rivas Gde et al. 2006). Similarly, modest depolarization of ΔΨ_m, by transient pharmacological uncoupling of mitochondrial oxidative phosphorylation, was also found to be a potential therapeutic strategy for several human disorders that involve metabolic and mitochondrial oxidative stress, including Parkinson's and Alzheimer's diseases (Wu et al. 2011; Geisler et al. 2017), cerebral ischaemia (Korde et al. 2005), heart failure (Brennan et al. 2006), and metabolic diseases such as diabetes and obesity (Parascandola, 1974; Perry et al. 2013; Tao et al. 2014). Pharmacological inhibition of mNCLX with CGP‐37157 conferred protection in hippocampal slices against veratridine‐induced Ca²⁺ and Na⁺ overload by regulating oxidative stress and p38 mitogen‐activated protein kinase‐linked activation of cell death (Nicolau et al. 2010). In addition, by enhancing mitochondrial oxidative metabolism, CGP‐37157 promoted ATP generation and increased glucose‐stimulated insulin secretion in rat islets and INS‐1 cells (Lee et al. 2003). These initial results illustrated the potential utility of drugs that inhibit mitochondrial Ca²⁺ release as neuroprotective tools and insulin secretagogues, underscoring mNCLX as a novel drug target in neurological and metabolic disorders.

However, all of the pharmacological approaches and tools to modulate mitochondrial Ca²⁺ homeostasis listed above have major drawbacks when applied in vivo. Classical MCU and mNCLX inhibitors (1) lack specificity, as they also inhibit other channels and transporters in the cell; (2) show several side effects, interfering with other organelle functions and/or several extra‐mitochondrial targets; and (3) reveal suboptimal cellular targeting, as most of them are poorly or not cell‐permeant. The most commonly used mitochondrial Ca²⁺ uptake inhibitors, RuR, Ru360 and its derivatives (Nathan et al. 2017), are potent and effective on isolated mitochondria, but their use in intact cells or in vivo is inadequate due to their limited permeance across the plasma membrane (Hajnoczky et al. 2006). In addition, RuR, due to its affinity for ryanodine receptors, was also shown to inhibit Ca²⁺ release from the SR (Chamberlain et al. 1984). Moreover, positive and negative inotropic responses to RuR were observed in isolated rat hearts in a dose‐dependent manner. These effects were attributed to its concomitant capacity to inhibit SR Ca²⁺ release or sarcolemmal Na⁺/Ca²⁺ exchange (Gupta et al. 1988).

Mitochondrial Ca²⁺ uptake inhibition is often indirectly achieved by depolarizing the ΔΨ_m. Nevertheless, in most cells types, compounds that inhibit the electron transport chain (ETC) also increase mitochondrial reactive oxygen species production (rotenone, antimycin, myxothiazol) and inhibit the plasma membrane Na⁺/K⁺‐ATPase (oligomycin) (Plesner & Plesner, 1991). In addition, the protonophore activity of uncouplers (FCCP/CCCP or dinitrophenol) is not limited to mitochondria, as they are also able to dissipate proton gradients across the plasma membrane and other membranous organelles at concentrations comparable to the ones used for mitochondria. Therefore, these chemical strategies lead to numerous undesired side effects, including plasma membrane depolarization, inhibition of ATP production, alterations in intracellular pH, and ultimately cytotoxicity (Juthberg & Brismar, 1997; Buckler & Vaughan‐Jones, 1998; Park et al. 2002).

Major concerns were also raised regarding the use of the known inhibitors of mNCLX owing to their additional interaction with many Ca²⁺ channels and transporters in the cell. The benzodiazepine and benzothiazepine analogues diltiazem and clonazepam were found to inhibit the L‐type Ca²⁺ channel in the plasma membrane (Koidl et al. 1997) and other receptor proteins in mitochondria (Kinnally et al. 1993). Even the classical blocker CGP‐37157, at low micromolar concentration, was more recently shown to affect Ca²⁺ fluxes through the ryanodine receptor and block the sarco/endoplasmic reticulum Ca²⁺‐ATPase pump in striated muscle (Neumann et al. 2011). Many other studies also demonstrated the pleiotropic nature of this compound in different cell types. Resembling the effects of the other described inhibitors, CGP37157 also inhibits the L‐type Ca²⁺ channel in β‐cells at concentrations considered specific for mNCLX (Luciani et al. 2007), and modulates intracellular Ca²⁺ entry through voltage gated Ca²⁺ channels in cortical neurons, attenuating NMDA‐induced cytosolic and mitochondrial Ca²⁺ overload (Ruiz et al. 2014). As these channels constitute an integral part of the cellular machinery that regulates Ca²⁺ homeostasis, this constitutes a confounding factor to inhibition of mitochondrial Ca²⁺ efflux mechanisms. Furthermore, this evidence further compromised the role of mNCLX in the regulation of insulin secretion and prevention of neuronal cell death.

Taken together, these comprehensive studies demonstrated that caution should be taken regarding the use of these chemical inhibitors of mitochondrial influx and efflux pathways, especially in an in vivo context. Therefore, over the past years there has been an urge to find drugs and therapeutic strategies that directly and specifically modulate mitochondrial Ca²⁺ homeostasis.

Novel drug‐screening strategies to identify direct modulators of mitochondrial Ca²⁺ homeostasis

The molecular characterization of mitochondrial Ca²⁺ pathways has opened the possibility of systematic and large‐scale drug screenings to identify novel means to target MCUC and mNCLX. The right approach to drug discovery depends on the availability of a robust, affordable and highly selective assay compatible with high‐throughput screening (HTS) (Walters & Namchuk, 2003). Common methods used to monitor dynamic changes in mitochondrial Ca²⁺ levels, such as cell‐based Ca²⁺ imaging assays and patch‐clamp recordings in mitoplasts, have not been optimized so far for HTS. Furthermore, the close functional interconnection between intracellular Ca²⁺ signalling, energy production, ΔΨ_m and Ca²⁺ uptake and release constitute a major challenge in the design of robust and effective high‐throughput assays for the discovery of specific modulators of MCUC and mNCLX. As mentioned above, the entry and exit of Ca²⁺ in mitochondria is dependent on the same ΔΨ_m used to produce ATP and, therefore, any compound with inhibitory effects on mitochondrial bioenergetics and ion gradients will indirectly affect mitochondrial Ca²⁺ homeostasis. Moreover, MCUC and mNCLX are intracellular targets and their activities depend on increases of cytoplasmic ionic concentrations by signalling events upstream of mitochondria. Accordingly, there is the potential in cell‐based assays for numerous false‐positive hits.

Very recently, a novel systematic chemical biology approach was developed by our group to identify small molecule drugs that directly modulate MCU (Arduino et al. 2017). We used mitochondria isolated from the yeast S. cerevisiae, wherein mitochondrial Ca²⁺ uptake activity was reconstituted by heterologous expression of the human MCU and its essential regulator EMRE, and MCU‐mediated Ca²⁺ uptake was quantitatively measured by the Ca²⁺‐sensitive photoprotein aequorin, stably expressed in the mitochondrial matrix. This approach takes advantage of several key properties of yeast mitochondria. First, they do not possess any intrinsic mitochondrial Ca²⁺ permeability (Carafoli & Lehninger, 1971). Second, they have a simplified OXPHOS system lacking complex I but including a d‐lactate:cytochrome c oxidoreductase (DLD), which is sensitive to d‐lactate and enables the direct transfer of electrons to cytochrome c oxidase (complex IV). Thus, when mitochondria are energized with d‐lactate there is a bioenergetic shunt pathway that bypasses the majority of the ETC complexes and NADH generating pathways while it is still sufficient to build up ΔΨ_m (Gregolin & D'Alberton, 1964; Pajot & Claisse, 1974). In our study, Ca²⁺ uptake in mitochondria fuelled with succinate remained sensitive to blockers of all the ETC complexes, whereas the use of d‐lactate as a respiratory substrate rescued the inhibitory effects of ETC blockers upstream of complex IV (Arduino et al. 2017). Likewise, yeast mitochondria developed and maintained a CCCP‐insensitive ΔΨ_m when assayed in an isosmotic mannitol–sucrose medium. In contrast, in a nearly isotonic KCl‐based medium, CCCP treatment resulted in a dramatic reduction of mitochondrial Ca²⁺ uptake (Arduino et al. 2017). Finally, yeast mitochondria reconstituted with coelenterazine can be frozen without losing the ability to generate a ΔΨ_m that is sufficiently high to allow Ca²⁺ fluxes as well as other mitochondrial processes such as, for example, protein import (Hartl & Neupert, 1990; Koll et al. 1992; Izawa & Unger, 2017). Altogether, these properties make the use of reconstituted yeast mitochondria a powerful high‐throughput drug screening strategy to identify specific and direct modulators of MCU activity, by minimizing false discovery rate due to confounding effects of drug‐mediated inhibition of ΔΨ_m and bioenergetics, ETC and signalling events upstream of mitochondria.

In a screen of ∼700 Food and Drug Administration (FDA)‐approved drugs, our group identified mitoxantrone as a specific inhibitor of MCU (Arduino et al. 2017). We confirmed that mitoxantrone inhibits mitochondrial Ca²⁺ uptake in mammalian cells without affecting mitochondrial bioenergetics. In addition, patch‐clamp electrophysiology in human mitoplasts treated with mitoxantrone showed a direct and reversible inhibition of MCU‐dependent Ca²⁺ currents when the drug was applied on the cytosolic but not on the matrix side. Although Ca²⁺ channel blockers are often reported to have off‐target effects on other ion channels, mitoxantrone exhibited an exquisite selectivity for MCU when tested on ER and plasma membrane channels. Based on our structure ‐ activity relationship (SAR) analysis, we found that the antineoplastic and anti‐MCU properties of mitoxantrone are not interrelated. Accordingly, the quinizarin moiety, which is common to other related chemotherapeutic drugs and mediates DNA intercalation and topoisomerase II binding, is dispensable for the inhibition of MCU. Instead, the positively charged side chains of mitoxantrone at positions 5 and 8 played a key role as they mediate the binding to highly conserved aspartate residues in the selectivity filter of the uniporter channel (Arduino et al. 2017).

Altogether, these results validate the yeast mitochondria‐based screening strategy as a tool to discover drug molecules directly targeting MCUC. Indeed, the assay represents a flexible, cost‐effective HTS solution, which can be exploited to reconstitute and target other subunits of the uniporter, such as tissue‐specific regulators (Fig. 1). This would address the need to tailor the modulation of MCUC activity to the tissue's physiology as well as to pathological consequences of loss or gain of function mutations. Additionally, drug delivery should ensure specificity to an intended tissue or cell type while simultaneously minimizing cytotoxicity by reducing side effects due to undesired drug accumulation on peripheral healthy tissues. This is normally achieved by active targeting, i.e. targeting unique receptors in specific cell types or tissues of interest by ligands present on the surface of drug delivery nanocarriers (e.g. polymeric nanoparticles or liposomes) (Cheng et al. 2015). Nowadays, the application of these drug delivery platforms constitutes a promising approach of demonstrated efficacy even in disease contexts where pharmacological targeting of MCUC has been proven to be beneficial, such as myocardial infarction and ischaemia–reperfusion injury (Takahama et al. 2009; Magruder et al. 2017).

Figure 1. Workflow of the drug screen assay in reconstituted yeast mitochondria.

The yeast S. cerevisiae constitutes a versatile system that can be reconstituted with different components of the MCUC machinery (e.g. MCU and EMRE and additionally MCUb, MICU1 and/or MICU2, as represented with dashed lines). When D‐Lactate (D‐Lac) is supplied to mitochondria as the energy source, it provides a bioenergetic shunt pathway that minimizes the detection of false‐positive hits. This drug screen platform allows the quantification of mitochondrial Ca²⁺ uptake kinetics based on mitochondria‐targeted‐aequorin luminescence emitted at 469 nm. MCUC modulators are accurately identified based on their effects on mitochondrial Ca²⁺ uptake kinetics. MAS, mannitol‐sucrose buffer; DLD, D‐lactate:cytochrome c oxidoreductase; TCA, tricarboxylic acid cycle; Q, coenzyme Q; Cytc, cytochrome c; II, succinate dehydrogenase; III, coenzyme Q:cytochrome c‐oxidoreductase; IV, cytochrome c oxidase.

In the near future, in order to enhance the translational value of differently regulated proteins involved in mitochondrial Ca²⁺ homeostasis as biomarkers or disease‐targets, it is essential to integrate hints from different disciplines. Those can comprise results from drug screenings, detailed information from clinical studies (e.g. disease endophenotypes) and prediction of novel drug–target interactions. This strategy will not only allow the identification of novel drug molecules but also provide a broader knowledge on their application to treat or halt human diseases in which mitochondrial Ca²⁺ homeostasis deregulation and mitochondrial dysfunction are pathological hallmarks.

Additional information

Competing interests

None of the authors have any conflict of interests to declare.

Author contributions

Both authors have approved the final version of the manuscript and agree to be accountable for all aspects of the work. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

Funding

This work was supported by the German Research Foundation (DFG) under the Emmy Noether Programme (PE 2053/1‐1) and the Bavarian Ministry of Sciences, Research and the Arts in the framework of the Bavarian Molecular Biosystems Research Network (D2‐F5121.2‐10c/4822) to F.P. and D.M.A.

Biographies

Daniela Arduino obtained her PhD in ‘Biomedical Sciences’ from CNC‐UC, Portugal. Her studies in the laboratories of Prof. Cardoso and Prof. Cuervo uncovered a main role for mitochondrial metabolism in the regulation of autophagy–lysosomal pathways in Parkinson's disease. She is currently a postdoctoral fellow in Fabiana Perocchi's laboratory, where she has been developing novel high‐throughput chemical screening approaches to identify specific modulators of mitochondrial Ca²⁺ homeostasis as well as characterizing mitochondrial signalling networks and elucidating their potential roles in cell physiology and disease.

Fabiana Perocchi is group leader at LMU and HMGU in Munich. Throughout her graduate studies at EMBL and postdoctoral training at Harvard, she became an expert in the area of functional genomics applied to the study of mitochondrial biogenesis and calcium signalling. Her research interests have been on the development of integrative strategies that combine systematic approaches with genetic, biochemical and physiological studies of mitochondrial function.