Mitochondrial calcium exchange in physiology and disease

Abstract

The uptake of calcium into and extrusion of calcium from the mitochondrial matrix is a fundamental biological process that has critical effects on cellular metabolism, signaling, and survival. Disruption of mitochondrial calcium (_mCa²⁺) cycling is implicated in numerous acquired diseases such as heart failure, stroke, neurodegeneration, diabetes, and cancer and is genetically linked to several inherited neuromuscular disorders. Understanding the mechanisms responsible for _mCa²⁺ exchange therefore holds great promise for the treatment of these diseases. The past decade has seen the genetic identification of many of the key proteins that mediate mitochondrial calcium uptake and efflux. Here, we present an overview of the phenomenon of _mCa²⁺ transport and a comprehensive examination of the molecular machinery that mediates calcium flux across the inner mitochondrial membrane: the mitochondrial uniporter complex (consisting of MCU, EMRE, MICU1, MICU2, MICU3, MCUB, and MCUR1), NCLX, LETM1, the mitochondrial ryanodine receptor, and the mitochondrial permeability transition pore. We then consider the physiological implications of _mCa²⁺ flux and evaluate how alterations in _mCa²⁺ homeostasis contribute to human disease. This review concludes by highlighting opportunities and challenges for therapeutic intervention in pathologies characterized by aberrant _mCa²⁺ handling and by summarizing critical unanswered questions regarding the biology of _mCa²⁺ flux.

CLINICAL HIGHLIGHTS

Mitochondrial Calcium Homeostasis Is Perturbed in Human Disease

• Primary mutations in genes encoding mitochondrial calcium-handling proteins including MICU1, MICU2, and LETM1 cause neuromuscular disorders.

• Altered expression of mitochondrial calcium-handling genes and changes in the balance between mitochondrial calcium uptake and efflux are noted in acquired conditions such as heart failure, stroke, neurodegenerative disease, diabetes, and cancer.

• Excessive mitochondrial calcium uptake is clearly detrimental in ischemia-reperfusion injury following myocardial infarction or stroke, and contributes to cell death and subsequent organ-level dysfunction.

• Diminished mitochondrial calcium content can impair cellular bioenergetics and may contribute to organ-level dysfunction in chronic diseases such as heart failure.

• Chronic neurodegenerative conditions such as Alzheimer’s disease and Parkinson’s disease are characterized by mitochondrial calcium overload.

Therapeutic Applications

• Blocking mitochondrial calcium uptake through the mitochondrial calcium uniporter complex or enhancing mitochondrial calcium efflux through NCLX is protective in mouse models of ischemia-reperfusion injury.

• NCLX has emerged as a promising target for neurodegenerative diseases such as Alzheimer’s disease and Parkinson’s disease.

• There is growing interest in manipulating either mitochondrial calcium uptake or mitochondrial calcium efflux for the treatment of cancer. However, more research is needed in this area, because increasing net mitochondrial calcium accumulation has variably been shown to promote either prosurvival signaling or apoptosis in tumor cells.

Current Challenges to Targeting Mitochondrial Calcium Exchange in Human Disease

• Development of compounds that can specifically activate or inhibit the mitochondrial calcium uniporter complex or NCLX.

• Ensuring tissue-specific drug delivery to avoid detrimental effects of altering mitochondrial calcium homeostasis in tissues that are not affected in a given disease.

• Need for greater understanding of the temporal changes in mitochondrial calcium handling throughout the progression of chronic illness, to match the appropriate therapeutic strategy to each stage of the disease.

1. INTRODUCTION

The earliest observations of calcium (Ca²⁺) uptake by isolated mitochondria were reported by Slater and Cleland in the early 1950s (1) in studies examining the instability of cardiac mitochondrial or “sarcosome” preparations used to investigate oxidative phosphorylation. Slater and Cleland found that, after isolation, the mitochondria contained the vast majority of all Ca²⁺ originally distributed throughout the cardiac tissue, and further demonstrated in vitro that mitochondria isolated in Ca²⁺-free conditions, with the Ca²⁺ chelator ethylenediaminetetraacetic acid (EDTA), were capable of taking up nearly all the Ca²⁺ from a bath solution containing 1 µM CaCl₂ (1). These early investigations found Ca²⁺ chelation with EDTA to have a stabilizing effect on oxidative phosphorylation (OXPHOS), as it prevented the gradual decline in the activity of OXPHOS reactions over the course of several hours of experimentation. Although this finding suggested the potential for mitochondrial Ca²⁺ (_mCa²⁺) uptake to inhibit OXPHOS, likely due to deleterious _mCa²⁺ overload, it stood in stark contrast to previous reports of Ca²⁺-dependent stimulation of OXPHOS (2). Slater and Cleland had the insight that _mCa²⁺ may play dual roles, both to stimulate OXPHOS and, in some cases, to destroy or disrupt the OXPHOS machinery (1). Thus, this early work hinted at the biphasic effects of _mCa²⁺ calcium and foreshadowed research into the roles of _mCa²⁺ in metabolism and mitochondrial permeability transition (PT) and cell death that has continued ever since.

Subsequent in vitro studies by DeLuca and Engstrom (3) and by Vasington and Murphy (4) revealed rapid uptake of radiolabeled ⁴⁵Ca²⁺ by isolated rat kidney mitochondria. These experiments demonstrated the requirement for an oxidizable carbon substrate such as succinate, α-ketoglutarate, glutamate, or malate; ATP or ADP + P_i; and magnesium (Mg²⁺) to permit _mCa²⁺ uptake. They also revealed that electron transport chain (ETC) inhibition (with Antimycin A, dicumarol, or 2,3-dimercaptopropanol) attenuated _mCa²⁺ uptake, but uncoupling or inhibition of ATP synthesis with dinitrophenol or oligomycin A did not (3). Vasington and Murphy (4) concluded that the activity of the ETC is required to drive _mCa²⁺ uptake and could drive the import of as much as 2 µmol of Ca²⁺ per milligram of mitochondrial protein. They proposed a stoichiometric relationship between Ca²⁺ uptake and electron transport, and indeed, this robust capacity for _mCa²⁺ uptake was later shown to depend on the establishment of a highly electronegative potential (ΔΨ_m, approximately −180 mV) across the inner mitochondrial membrane (IMM). As Peter Mitchell’s elegant work on chemiosmotic theory established, this mitochondrial membrane potential is generated as protons (H⁺) are pumped out of the mitochondrial matrix by the ETC (5, 6). Both DeLuca and Engstrom and Vasington and Murphy noted the eventual loss of Ca²⁺ from mitochondria upon prolonged time after isolation, removal of ADP, or addition of ETC inhibitors or uncouplers. These observations strengthened the notion that a functional ETC that maintains ΔΨ_m is required for _mCa²⁺ sequestration and indicated that there are also controlled pathways for _mCa²⁺ efflux.

The 1970s brought a growing appreciation that the release of Ca²⁺ from the mitochondrial matrix could occur in a regulated manner. In experiments investigating the mechanisms of excitation-contraction coupling in cardiac muscle, Carafoli and colleagues (7, 8) discovered that incubation with NaCl or LiCl, but not KCl, could stimulate the release of Ca²⁺ from rat heart mitochondria. This stimulated Ca²⁺ release occurred independent of respiratory function, suggesting the presence of an exchanger protein capable of extruding matrix Ca²⁺ in exchange for sodium (Na⁺) or lithium (Li⁺). Subsequent studies by Pozzan et al. (9) supported the idea that _mCa²⁺ efflux could be coupled directly or indirectly to H⁺ leak into the matrix. This phenomenon, when coupled to cooperative activation of _mCa²⁺ influx (10–12), would minimize _mCa²⁺ accumulation under steady-state conditions despite the highly electronegative ΔΨ_m favoring _mCa²⁺ influx. Thus, the net mitochondrial Ca²⁺ content would tend to increase only under conditions of increased cytosolic Ca²⁺ concentration. This pattern of _mCa²⁺ accumulation could allow the mitochondria to buffer changes in cytosolic Ca²⁺ and in doing so act as a signal to couple changes in cellular function to coordinated responses in the mitochondria.

Direct proof of the principle that changes in cytosolic Ca²⁺ concentration drive changes in mitochondrial Ca²⁺ within intact cells was provided in 1992 in groundbreaking work by Pozzan’s group. In this study, Rizzuto et al. fused the Ca²⁺-sensitive photoprotein aequorin to the mitochondrial targeting sequence of subunit VIII of human cytochrome-c oxidase, to mediate mitochondrial import of the Ca²⁺ reporter to the inner mitochondrial membrane (IMM) (13). This approach allowed the first simultaneous live-cell measurements of both mitochondrial Ca²⁺ and cytosolic Ca²⁺, which was measured by the Ca²⁺-sensitive dye Fura-2. In the 30 years since, a wide array of reagents have been developed for the measurement of mitochondrial Ca²⁺ flux in intact cells. For an assessment of the tools currently available for assessing mitochondrial Ca²⁺ exchange in isolated mitochondria or cellular systems, and the strengths and limitations of these approaches, the reader is directed to two recent articles (14, 15). Advances in molecular biology have revealed conservation of _mCa²⁺ flux across all tissues and taxa. The mitochondrial research community has also begun to identify the specific proteins that transport Ca²⁺ into and out of the mitochondria and that regulate this Ca²⁺ exchange. This review begins by summarizing the pathways responsible for raising and lowering cytosolic Ca²⁺ concentration. We then focus on the current understanding of the identity, function, and molecular regulation of the proteins that transport Ca²⁺ across the outer and inner mitochondrial membranes, and summarize the cellular processes directly impacted by mitochondrial Ca²⁺ signaling as well as the emerging functions of the mitochondrial Ca²⁺ microdomain. We conclude with an examination of the role of altered _mCa²⁺ exchange in human disease (TABLE 1), a discussion of potential therapeutic strategies based on manipulation of _mCa²⁺ handling, and a summary of the major gaps in the current understanding of _mCa²⁺ exchange.

Table 1.

Human genetic diseases linked to _mCa²⁺ handling proteins

Reference	Genetic Defect	Disease/Major Phenotype(s)
Logan et al. (199)	Loss of function mutations in MICU1	Excessive mCa2+ uptake, proximal myopathy, learning defects, and a progressive extrapyramidal movement disorder
Lewis-Smith et al. (200)	Homozygous, 2,755-bp deletion in MICU1	Lethargy, fatigue, and muscle weakness presenting in childhood
O’Grady et al. (201)	Loss of function mutation in MICU1	Congenital muscular dystrophy
Musa et al. (202)	Homozygous c.553C>T (p.Q185*) mutation in MICU1; predicted to cause complete loss of protein function	Muscle weakness, fatigue; in some instances accompanied by extrapyramidal signs, learning disability, nystagmus, and cataracts
Mojbafan et al. (203)	C1295delA mutation in exon 13 of MICU1, causing a frameshift and protein truncation	Myopathy with extrapyramidal signs
Wilton et al. (573)	Two distinct heterozygous MICU1 mutations in the same patient, one on the maternal allele and one on the paternal allele: • c.161 + 1G>A in MICU1, likely disrupting splicing and gene function • c.386G>C in MICU1, resulting in R129P missense mutation	Myopathy with extrapyramidal signs, also acute encephalopathy and developmental brain abnormalities
Shamseldin et al. (233)	Homozygous c.42G>A:P.W14* mutation in MICU2, causing an early stop codon and protein truncation	Neurodevelopmental disorder with severe cognitive impairment, spasticity, and white matter involvement. Patient fibroblasts have excessive mCa2+ uptake.
Endele et al. (341)	Deletion of regions of the short arm of chromosome 4 (4p16.3), including LETM1	Wolf–Hirschhorn syndrome: impaired growth, developmental delay, microcephaly, and mental defects. Sometimes involves impaired muscle tone, seizures, and congenital heart defects.
South et al. (371)	Microdeletion in 4p16.3, including LETM1	Developmental delays and delayed growth, unique facial features distinct from those of Wolf–Hirschhorn syndrome; variable involvement of seizures
Cyr et al. (375)	Microduplication of 4p16.3, including LETM1	Macrocephaly, normal growth, irregular iris pigmentation, delayed development, dysmorphic features, and seizures
Roselló et al. (376)	Submicroscopic duplication in 4p16.2, including LETM1	Phenotype intermediate between Wolf–Hirschhorn syndrome and 4p trisomy; facial dysmorphism, delayed gross motor development

See glossary for abbreviations.

2. INTRACELLULAR Ca²⁺ CYCLING

The fundamental stimulus for mitochondrial Ca²⁺ uptake is an increase in the local cytosolic Ca²⁺ concentration. Although a full accounting of the mechanisms controlling cytosolic Ca²⁺ flux is beyond the scope of this article, the key points are highlighted below and the reader is directed to the excellent reviews cited throughout this section for further details.

2.1. Ca²⁺ Transport Across the Plasma Membrane

Resting cytosolic calcium Ca²⁺ concentration in eukaryotic cells is typically held around 100 nM, ∼10,000–20,000 times lower than the extracellular Ca²⁺ concentration of ∼1 mM (16, 17). This large Ca²⁺ gradient represents a massive thermodynamic driving force for Ca²⁺ entry across the plasma membrane and is maintained by the action of the plasma membrane calcium ATPase (PMCA), which uses the energy of ATP hydrolysis to pump Ca²⁺ out of the cell (16). The ATP consumed to pump Ca²⁺ “uphill” against its electrochemical gradient represents an energetic investment by the cell to establish a strong driving force for subsequent, regulated Ca²⁺ entry into the cell. The electrical potential across the plasma membrane (ΔΨ, approximately −70 mV inside the cell relative to outside) is established by the electrogenic Na⁺-potassium (K⁺)-ATPase pumping 3 Na⁺ ions out of the cell in exchange for the import of 2 K⁺ ions, contributing to the net driving force favoring influx of positively charged Ca²⁺ across the plasma membrane (18). The plasma membrane potential and the Na⁺ gradient [∼140 mM Na⁺ extracellular, ∼8–12 mM Na⁺ cytosolic (19–21)] also drive Na⁺/Ca²⁺ exchange through the electrogenic plasma membrane Na⁺/Ca²⁺ exchanger, NCX. NCX exports 1 Ca²⁺ ion out of the cell in exchange for the entry of 3 Na⁺ ions, which helps to maintain the Ca²⁺ gradient across the plasma membrane at the expense of the net entry of 1 positive charge into the cell (16). Similarly, electrogenic plasma membrane Na⁺/Ca²⁺-K⁺ exchangers (NCKXs) use the driving force for Na⁺ entry established by the Na⁺-K⁺-ATPase to export 1 Ca²⁺ ion and 1 K⁺ ion in exchange for the entry of 4 Na⁺ ions (17). Ca²⁺ is also sequestered in intracellular stores, such as the endoplasmic reticulum (ER)/sarcoplasmic reticulum (SR), which contributes to the low resting cytosolic Ca²⁺ concentration. The accumulation of Ca²⁺ in intracellular stores is achieved via the action of the sarco(endo)plasmic reticulum Ca²⁺-ATPase (SERCA), which pumps Ca²⁺ against its concentration gradient into the ER/SR (17).

When plasma membrane Ca²⁺ channels open, the potential energy stored across the plasma membrane allows Ca²⁺ to rapidly enter the cell, flowing down its electrochemical gradient. The resultant elevation of intracellular Ca²⁺ (_iCa²⁺) is an early event in many intracellular signaling cascades including cAMP-response element binding protein (CREB)-dependent transcriptional responses, calcium/calmodulin-dependent signaling, calcineurin-NFAT signaling, synaptic vesicle release, and calcium-induced Ca²⁺ release that triggers myofilament cross-bridge cycling (reviewed in Refs. 22–26). Another fundamental consequence of elevated cytosolic Ca²⁺ concentration, and a major focus of this review, is that it drives mitochondrial Ca²⁺ uptake, which has numerous implications for cellular metabolism and viability. In this way, the cell can use the energy invested in the extracellular-intracellular Ca²⁺ gradient to effect rapid and reliable intracellular signaling.

The specific proteins that mediate Ca²⁺ influx across the plasma membrane vary across cell types but in general include voltage-gated Ca²⁺-permeant channels, including the L-type and T-type Ca²⁺ channels and transient receptor potential channels, which open in response to plasma membrane depolarization (for instance, during an action potential), and ligand-gated Ca²⁺-permeant channels including certain purinergic receptors, nicotinic receptors, and N-methyl-d-aspartate (NMDA) receptors, which open upon binding to extracellular signaling molecules such as ATP and neurotransmitters (17, 23, 24).

2.2. Release of Ca²⁺ From Intracellular Stores

A key feature of cellular Ca²⁺ signaling is that the activation of plasma membrane receptors can initiate intracellular signaling cascades that ultimately cause release of Ca²⁺ from the ER/SR. These pathways serve as another mechanism to raise _iCa²⁺ concentration and can allow for localized release of Ca²⁺ at more discrete regions within the interior of the cell, a concept that is addressed in further detail in sect. 5. G_q-coupled receptors at the plasma membrane are activated by extracellular ligands such as histamine, 5-hydroxytryptamine, ATP, and angiotensin II. They then activate phospholipase Cβ, which hydrolyzes phosphatidylinositol-4,5-bisphosphate to generate diacylglycerol and inositol-(1,4,5)-trisphosphate (IP₃) (27) (also reviewed in Refs. 28–32). Several additional phospholipase C isoforms are activated via other signaling mechanisms and likewise generate IP₃ (31). IP₃ then binds and activates the IP₃ receptor (IP₃R), a Ca²⁺ release channel on the ER membrane, allowing Ca²⁺ to flow from the ER lumen (∼0.5 mM Ca²⁺ concentration) to the cytosol (∼100 nM resting Ca²⁺ concentration) (33, 34).

In excitable cell types such as muscle and neurons, influx of extracellular Ca²⁺ through voltage-gated Ca²⁺ channels at the plasma membrane in response to an action potential also triggers release of Ca²⁺ from the SR/ER by activating the ryanodine receptor (RyR), in a process known as Ca²⁺-induced Ca²⁺ release (23, 24, 35, 36). It should be noted that in skeletal muscle intracellular Ca²⁺ release through the RyR does not strictly require the influx of extracellular Ca²⁺, as the predominant isoform expressed in this cell type (RyR1) can be activated in response to plasma membrane depolarization through its physical interaction with the voltage-sensitive dihydropyridine receptor that is located on the t-tubule membrane (36). Ca²⁺-induced Ca²⁺ release from the ER/SR can amplify the cytosolic Ca²⁺ signal in response to a relatively smaller influx of Ca²⁺ across the plasma membrane and give rise to global cytosolic Ca²⁺ transients such as those responsible for muscle contraction.

Another important point is that the ER/SR often closely associates with the mitochondria, allowing for precise positioning of intracellular Ca²⁺ release in close proximity to the mitochondria, at sites termed “mitochondria-associated membranes” (MAMs) (34). This organization creates local, spatially constrained cytosolic microdomains in which Ca²⁺ released from the ER/SR can accumulate at concentrations high enough to activate _mCa²⁺ uptake, without requiring robust cell-wide increases in Ca²⁺ levels. The functional implications of this special arrangement for ER/SR-mitochondrial Ca²⁺ transfer and cellular function are explored in greater depth in sects. 3 and 5.

3. THE MOLECULAR MACHINERY CONTROLLING _mCa²⁺ FLUX

Upon stimulation, cytosolic Ca²⁺ can rise because of the influx of extracellular Ca²⁺ and/or the release of Ca²⁺ from the ER/SR via the mechanisms introduced above (reviewed in Refs. 16, 17, 31, 37–39). As local concentrations of Ca²⁺ rise above ∼400 nM, as likely occurs at sites of ER-mitochondria contact, the mitochondria begin to rapidly take up Ca²⁺ from the cytosol. The highly electronegative inner mitochondrial membrane potential (ΔΨ_m, approximately −180 mV) provides the thermodynamic driving force for the flow of Ca²⁺ into the matrix, where the resting free Ca²⁺ concentration is kept low, ∼100 nM, due to buffering of Ca²⁺ ions as calcium phosphate (21, 40). The high buffering capacity of the mitochondrial matrix serves as a “sink” for cellular Ca²⁺ storage and favors net _mCa²⁺ uptake when cytosolic Ca²⁺ concentration rises. Early studies noted that the kinetics of _mCa²⁺ uptake follow a sigmoidal curve, with the rate of _mCa²⁺ uptake accelerating rapidly as extramitochondrial Ca²⁺ concentration increases, until plateauing at a cytosolic or bath Ca²⁺ concentration around 200 µM (10–12). This behavior reflects robust Ca²⁺ regulation of the _mCa²⁺ uptake machinery, which is a consequence of the numerous proteins that modulate the activity of the _mCa²⁺ uptake channels.

3.1. Ca²⁺ Transport Across the Outer Mitochondrial Membrane

The outer mitochondrial membrane (OMM) is generally thought to impose minimal to no resistance to Ca²⁺ transport, with Ca²⁺ moving from the cytosol to the mitochondrial intermembrane space (IMS) through the 30- to 35-kDa voltage-dependent anion channel (VDAC), also referred to in the older literature as “porin” because of its similarity with pore-forming proteins in the membranes of bacteria (41). VDAC proteins are present in all metazoans, with a high degree of conservation among eukaryotes. Three related VDAC isoforms (VDAC1–3) are present in chordates (42), and VDAC1 is the most abundant isoform (43). The reader is referred to an excellent review by Shoshan-Barmatz et al. (42) for a comprehensive discussion of the genetic and structural diversity of VDAC variants across taxa and the phylogenetic relationships between VDAC isoforms.

The VDAC channel is made up of a 19-stranded β-barrel comprising the channel pore and a NH₂-terminal α-helix that lies within the pore (44–46). VDAC is readily permeable to ions and small molecules < 5 kDa such as phosphate, chloride, and the adenine nucleotides, with a greater conductance for anions than for cations of similar size (47, 48). VDAC displays voltage sensitivity, in that it switches to “closed” states of diminished conductance as transmembrane voltage increases beyond ±30–40 mV (47, 48). In this way, VDAC behavior at the OMM may be influenced by the electrochemical gradient of H⁺ in the IMS that is established by the activity of the ETC at the inner membrane, which in turn influences the potential across the OMM (negative on the cytosolic side; positive on the IMS side) (47, 49, 50). Interestingly, the “substate” conformations of VDAC exhibit selectivity to cation transport (50, 51). VDAC contains Ca²⁺ binding sites and when constituted into lipid bilayers or liposomes is permeable to Ca²⁺ (52–54). Experiments in liposomes indicate that Ca²⁺ itself regulates VDAC conductance, with increased Ca²⁺ concentration favoring increased conductance of the closed or subconductance states of the channel (55). Ca²⁺ may also increase ATP transport across the OMM, indicating that Ca²⁺ can control the overall small molecule permeability of VDAC as well as its ionic conductance (55). The Ca²⁺ permeability of specific VDAC isoforms is further regulated by interaction with other proteins such as the antiapoptotic protein Bcl-xL (56). A number of posttranslational modifications of VDAC have been identified by mass spectrometry, including phosphorylation of multiple residues and lysine acetylation of VDAC1–3 (reviewed in Ref. 57). However, the consequences of such modifications for overall VDAC permeability or Ca²⁺ transport remain to be determined.

Some models propose that VDAC specifically regulates OMM Ca²⁺ transport at discrete spatial domains where the OMM is in proximity with Ca²⁺ release sites on the ER (discussed in greater detail in sect. 5). For instance, overexpression of rat VDAC1 in HeLa cells potentiates _mCa²⁺ uptake in response to treatment with agonists that stimulate ER Ca²⁺ release (58). However, VDAC1 overexpression does not alter the extent or kinetics of _mCa²⁺ uptake when permeabilized cells are exposed to increasing concentrations of bath Ca²⁺ (58). This finding indicates that VDAC does not increase OMM permeability to Ca²⁺ in a general, global fashion but instead predominantly increases OMM Ca²⁺ permeability at sites of ER-mitochondria contact. Fitting with this notion, silencing of VDAC1, 2, or 3 in HeLa cells attenuates and overexpression of VDAC1, 2, or 3 increases _mCa²⁺ uptake that occurs after histamine-induced release of ER Ca²⁺ (59). These effects of VDAC on _mCa²⁺ uptake occur independent of any alteration in agonist-induced ER Ca²⁺ release (59). Intriguing work suggests that the IP₃R and VDAC interact in a physical complex that spans the ER and OMM, and that physical interaction between VDAC and the IP₃R stimulates Ca²⁺ uptake through VDAC (60). This can explain why VDAC overexpression specifically potentiates _mCa²⁺ uptake after agonist-induced ER Ca²⁺ release rather than after a global elevation of cytosolic or bath Ca²⁺, because VDAC activity may only increase at sites of agonist-induced ER Ca²⁺ release, where it is subject to positive regulation by the IP₃R. However, given that Ca²⁺ permeation even occurs when VDAC is in the “closed” confirmation, it is unlikely that VDAC is a real-time regulator of _mCa²⁺ flux; rather, it is permissive for Ca²⁺ transport across the OMM.

In addition to its roles in Ca²⁺ transport across the OMM, VDAC has also been implicated in cell death. VDAC can associate with the IMM adenine nucleotide translocator and form a large, nonspecific channel that spans both the outer and inner mitochondrial membranes and is believed to represent a component of the mitochondrial permeability transition pore (mPTP) (reviewed extensively in Refs. 42, 61 and discussed further below). However, the strict requirement for VDAC in mPTP opening and necrotic cell death is called into question by experiments in which genetic disruption of the three mammalian VDAC isoforms either alone or in combination fails to prevent mitochondrial permeability transition and cell death (62, 63).

A small number of case studies describe human patients with genetic deletion of VDAC1. The first was reported by Huizing et al. in a study of patients with mitochondrial myopathy in which no defect in mitochondrial metabolic enzymes could be identified, despite signs of impaired substrate oxidation and ATP production in muscle tissue (64). The affected patient exhibited psychomotor retardation and increased blood lactate in response to an intravenous glucose loading test, pointing to a mitochondrial defect (65). VDAC content was reduced 10-fold in skeletal muscle but was normal in fibroblasts from this patient, pointing to potential tissue-specific expression or modification of this protein (65). De Pinto and colleagues also observed reduced VDAC content in a muscle biopsy from a patient who likely suffered from Pearson’s disease due to a large mitochondrial DNA deletion, although it is possible that the reduction in VDAC expression was simply a consequence of gross mitochondrial dysfunction in this individual (42, 66).

Animal models with disruption of the various VDAC isoforms support a role for the channel in shaping mitochondrial metabolism and suggest a fair amount of functional redundancy among the three isoforms. Mouse embryonic stem cells with deletion of VDAC 1, VDAC2, or VDAC3 are viable but have diminished oxygen consumption, consistent with a role for VDAC in delivering metabolites such as phosphate and ADP needed for OXPHOS to the mitochondria (67). This study did not differentiate any potential effects of metabolite delivery versus Ca²⁺ transport to the mitochondria as potential causes for these deficits, but as this impaired metabolic phenotype is observed under basal conditions, rather than in response to any Ca²⁺-mobilizing stimulation, the authors’ interpretation that the phenotype is largely driven by reduced mitochondrial metabolite availability is likely correct.

Constitutive knockout (KO) of either Vdac1 or Vdac3 in mice yields viable animals (68, 69), but deletion of Vdac2 results in embryonic lethality (70). Vdac1^−/− striated muscle has no change in basal respiration, despite altered mitochondrial sensitivity for ADP [K_m(ADP)] in heart and skeletal muscle (68). Additionally, Vdac1 deletion causes ultrastructural changes including altered cristae structure in the mitochondria of the gastrocnemius and soleus muscles and, to a lesser extent, the heart (68). Vdac3^−/− mice are healthy, but the males are infertile because of reduced sperm motility attributed to axoneme instability (69). Vdac3^−/− animals also have enlarged mitochondria and diminished respiratory chain activity in skeletal muscle (69), consistent with a requirement for VDAC3 in normal skeletal muscle metabolism. Finally, mice lacking VDAC1 or VDAC3 alone or in combination exhibit disrupted fear conditioning and spatial learning, which are associated with defects in synaptic plasticity in the hippocampus (71). mPTP inhibition with cyclosporine A recapitulates the impaired synaptic plasticity observed in VDAC-deficient mice, suggesting that this defect is related to loss of VDAC-dependent mPTP opening and potential alterations to _mCa²⁺ homeostasis in hippocampal neurons (71). Notably, treatment of Vdac1^−/−/Vdac3^−/− mouse embryonic fibroblasts (MEFs) with siRNA against Vdac2 in order to eliminate all cellular VDAC expression is insufficient to prevent mitochondrial permeability transition and cell death upon stimulation with H₂O₂, ionomycin, staurosporine, or TNF-α (62). Although this study did not specifically investigate the metabolic phenotype of these VDAC-deficient MEFs, the fact that these cells are viable suggests that alternative pathways are sufficient, at least under basal conditions, to supply adequate metabolites and Ca²⁺ to the mitochondria to support homeostatic metabolic function. It should also be noted that there was no difference in the ability of VDAC1/2/3-null cells to take up _mCa²⁺, arguing against its requirement for OMM Ca²⁺ permeation (62). Functional redundancy and compensation among the three mammalian VDAC genes themselves and with other Ca²⁺- and metabolite-transporting pathways may also explain why pathological disruption of VDAC has so rarely been detected in the clinical setting. For example, the presence of VDAC2 appears sufficient to mostly compensate for the absence of VDAC1 and VDAC3 (68–71); at the very least, the loss of VDAC1 and/or VDAC3 is better tolerated than the loss of VDAC2 in terms of overall viability. These studies also relied on constitutive deletion of the various VDAC isoforms, so it is possible that there are robust compensatory mechanisms during development that minimize the detrimental effects of VDAC1 and/or VDAC3 deletion and result in relatively mild phenotypes.

3.2. Ca²⁺ Uptake Across the Inner Mitochondrial Membrane

After crossing the outer mitochondrial membrane via VDAC, Ca²⁺ must be transported from the intermembrane space across the inner mitochondrial membrane to enter the mitochondrial matrix. Two Ca²⁺ uptake mechanisms with distinct kinetics have been described for this process, the Ca²⁺ uniporter and the rapid mode of Ca²⁺ uptake (RaM). However, it has not yet been determined whether these two Ca²⁺ uptake modes are mediated by the same or different molecular machinery (72).

The more familiar _mCa²⁺ uptake activity, described above and in Refs. 10–12, in which Ca²⁺ influx is driven by ΔΨ_m (approximately −180 mV on the matrix side of the IMM relative to the IMS; favoring influx of positively charged Ca²⁺), occurs without direct coupling to the transport of another ion and shows sigmoidal kinetics as a function of cytosolic Ca²⁺ concentration. This mode of _mCa²⁺ uptake has been referred to as the “calcium uniporter” since the 1970s and 1980s (72–75). _mCa²⁺ uptake via the uniporter can be inhibited by the hexavalent cation ruthenium red (76) and by its derivative Ru360 (77). Conversely, uniporter-mediated _mCa²⁺ uptake can be activated by taurine (78); by several plant flavonoids including kaempferol, quercetin, genistein, and genistin (79); and, at low cytosolic Ca²⁺ concentrations, by spermine and spermidine (80).

The alternative, rapid mode of Ca²⁺ uptake (RaM) was first described in liver mitochondria by Sparagna et al. (81) as a _mCa²⁺ uptake mechanism that functions at the beginning of a Ca²⁺ pulse and so allows mitochondria to accumulate Ca²⁺ even if the duration of the cytosolic Ca²⁺ pulse is insufficient for activation of the uniporter. The RaM functions at cytosolic Ca²⁺ concentrations below 200 nM that are too low for activation of the uniporter (81) but begins to inactivate as the surrounding Ca²⁺ concentration increases beyond 100–150 nM (72). Mallilankaraman et al. (82) hypothesize that this effect could arise from changes in MICU-dependent regulation of the uniporter as local cytosolic Ca²⁺ concentration changes. That is, at low Ca²⁺ concentrations the RaM may be mediated by an “ungated” form of the uniporter channel that is not subject to regulation by the Ca²⁺-binding protein MICU1, a concept that is discussed in detail below. As local Ca²⁺ concentration rises beyond 100–150 nM, increased binding of Ca²⁺ to MICU1 may enable MICU1 to “gate” the uniporter, thus terminating the RaM current, despite persistence of the thermodynamic driving force for _mCa²⁺ uptake.

After inactivation, the RaM can be reset in response to a brief decrease in bath Ca²⁺ concentration (81). This behavior effectively allows for substantial _mCa²⁺ sequestration via the RaM in response to a series of Ca²⁺ pulses. Thus, the RaM may provide a mechanism for _mCa²⁺ accumulation in response to continuous, low-amplitude cytosolic Ca²⁺ oscillations that may occur physiologically. Sparagna et al. (81) interpret this rapid activation/inactivation feature of the RaM as a potential mechanism to tune _mCa²⁺ accumulation based on the frequency of cytosolic Ca²⁺ oscillations. Subsequent studies detected the RaM in heart mitochondria and fast, high-capacity _mCa²⁺ uptake consistent with the RaM in rat chromaffin cells (83), motor neurons (84), and mast cells (85). However, the properties of RaM activation and inhibition in the heart differ from the RaM observed in liver mitochondria. The time required for cytosolic Ca²⁺ to remain low in order to reset the RaM after its inactivation is longer in the heart than in the liver (86), and the relative sensitivity of the RaM to activation or inhibition by adenine nucleotides differs between the two tissues (86). Buntinas et al. (86) propose that these tissue-specific differences in the RaM may be matched to differences in patterns of cytosolic Ca²⁺ oscillations in liver (triggered by an external stimulus like a hormone; infrequent but of long duration), versus heart (continuous; frequent but of short duration), and may permit sufficient _mCa²⁺ uptake to stimulate mitochondrial metabolism while at the same time preventing detrimental _mCa²⁺ overload and permeability transition.

The pharmacological sensitivity of the RaM generally resembles that of the _mCa²⁺ uniporter. The RaM can be blocked by ruthenium red, although this requires more than an order of magnitude greater concentration of ruthenium red than is required to inhibit the uniporter (81). Just as spermine activates the uniporter, spermine activates the RaM in liver mitochondria and, to a lesser extent, in heart mitochondria as well (81, 86). However, one point of distinction is that slow _mCa²⁺ uptake (i.e., uniporter activity) is inhibited by cytosolic Mg²⁺ but rapid _mCa²⁺ uptake is not (87).

3.3. Composition of the Mitochondrial Calcium Uniporter Complex

The molecular composition of the mitochondrial calcium uniporter channel began to be pieced together in the 2010s, as a number of research groups used evolutionary genetics and bioinformatic approaches to identify the mitochondrial inner membrane proteins required for or capable of modulating mitochondrial Ca²⁺ uptake. The multiprotein channel responsible for the uniporter activity described above has been termed the “mitochondrial uniporter channel” or “mtCU” (FIGURE 1).

FIGURE 1.

Ca²⁺ influx and efflux pathways of the inner mitochondrial membrane. The mitochondrial calcium uniporter channel (mtCU) exists as a channel composed of a total of 4 MCU subunits and up to 4 EMRE subunits. For simplicity, only a single EMRE subunit is shown here. In some cases, the dominant-negative MCU paralog, MCUB, replaces MCU subunits within the complex. The channel is gated by dimers of MICU1/2 or, in some tissues, dimers of MICU1/3. The accessory subunit MCUR1 also binds to and regulates the channel. Two uniporter channels can dimerize, and the function of these dimers is regulated by interactions between the MICU2 subunit associated with each channel. The driving force for Ca²⁺ entry through the mtCU is the highly negative (approximately −180 mV) potential established across the inner mitochondrial membrane (IMM) by the electron transport chain. Other proposed routes of mitochondrial Ca²⁺ (_mCa²⁺) uptake may include the mitochondrial ryanodine receptor (mRyR) or reverse-mode operation of NCLX and LETM1. The Na⁺/Ca²⁺ exchanger NCLX is a major route for _mCa²⁺ efflux, and the Ca²⁺/H⁺ exchanger LETM1 may contribute to _mCa²⁺ efflux in some tissues. Transient opening of the mitochondrial permeability transition pore (mPTP), termed “flickering,” is proposed as another mechanism by which Ca²⁺ leaves the mitochondrial matrix under physiological conditions. See glossary for abbreviations.

This channel is composed of the channel-forming IMM protein, mitochondrial calcium uniporter (MCU); the integral scaffolding protein, essential mitochondrial response element (EMRE); and the regulatory proteins mitochondrial calcium uptake 1, 2, and 3 (MICU1/2/3) that project into the intermembrane space. The MCU paralog, MCUB, also contributes to the mtCU pore under certain conditions. Additional regulatory proteins including the IMM protein mitochondrial calcium uniporter regulator 1 (MCUR1) also form part of the mtCU and are reported to alter channel activity. These mtCU constituents, and their spatial arrangement within the uniporter complex, are discussed in detail below. The reader is directed to several articles (88–96) discussed throughout this section for excellent illustrations of the evolutionary relationships between mtCU components. The specific identity of the protein(s) that mediates the RaM remains unknown but may include one or more of the components of the mtCU.

3.3.1. MCU.

In the early 2000s, the Clapham group patch-clamped mitoplasts derived from the inner mitochondrial membrane of COS-7 cells and ascribed mitochondrial calcium uniporter activity to a single Ca²⁺-selective, inwardly rectifying ion channel with high Ca²⁺ affinity (97). This work provided evidence that the uniporter is a Ca²⁺ channel rather than a carrier protein and demonstrated that the properties of this channel, such as sensitivity to ruthenium red, match those of the activity of the theoretical uniporter postulated from historical studies of Ca²⁺ uptake in intact mitochondria (10–12, 72–76).

The genetic identity of the core pore-forming subunit of the mitochondrial calcium uniporter channel was determined in 2011 in a series of elegant papers from the Mootha and Rizzuto laboratories. These studies used a combination of phylogenetic profiling, RNA expression analysis, and mitochondrial protein expression analysis to identify genes coexpressed and functionally related to MICU1, the first uniporter complex component to be identified (98–100). These analyses identified a ubiquitously expressed transmembrane protein, CCDC109A, as a potential MICU1-interacting mitochondrial protein consisting of two transmembrane regions connected by a loop enriched with acidic residues (98, 99). CCDC109A coimmunoprecipitates with MICU1 and is localized to the inner mitochondrial membrane (98, 99). Knockdown of CCDC109A in HeLa cells or mouse liver attenuates rapid _mCa²⁺ uptake, and mutational analysis of this protein further demonstrated the requirement of a conserved region that links the protein’s two transmembrane domains, termed the DIME motif, for Ca²⁺ permeation of the channel (98, 99). Mutation of a serine residue adjacent to the DIME motif diminishes the protein’s sensitivity to inhibition by Ru360, a ruthenium red derivative, providing further evidence that this protein forms part of the mitochondrial calcium uniporter channel (99). Recombinant CCDC109A exhibits ruthenium red-sensitive Ca²⁺ conductance when reconstituted into lipid bilayers, confirming its function as an inner membrane Ca²⁺ channel (98). As such, CCDC109A was renamed “MCU” for “mitochondrial calcium uniporter.” Additional experiments by the Mootha laboratory assessed the function of MCU in a more natural setting by using voltage clamp of whole mitoplasts. These studies confirmed MCU’s identity as the pore-forming subunit of the mitochondrial inner membrane Ca²⁺ channel that is responsible for the classically defined uniporter current (101).

3.3.1.1. mcu phylogenetic conservation.

MCU and its homologs are present in all branches of eukaryotes (88). MCU homologs appear in virtually all plants and metazoa, although these genes have been lost from some protozoan and fungal lineages, notably yeast fungi such as Saccharomyces cerevisiae (88). Most species have at least one or two MCU homologs, such as MCUB (discussed below), whereas some have as many as three or four homologs (88). Trypanosome parasites such as Trypanosoma cruzi and T. brucei have four MCU homologs (TcMCU as well as TcMCUb–d), all of which can diminish or enhance _mCa²⁺ uptake when knocked out or overexpressed, respectively (102–105). Putative homologs for MCU have been detected in several species of bacteria (Prevotella oris, Chlorobium phaeobacteroides, and Cytophaga hutchinsonii), raising the possibility that the uniporter predates the emergence of Eukarya (88).

3.3.1.2. mcu structure.

MCU has a predicted molecular mass of ∼40 kDa but is processed to ∼34 kDa upon cleavage of an amino-terminal mitochondrial targeting sequence before incorporation into the inner mitochondrial membrane (99). MCU consists of a large NH₂-terminal domain followed by a coiled-coil domain, a transmembrane domain, a short linker region, a second transmembrane domain, and a second coiled-coil domain (106). The NH₂ and COOH termini of MCU face the mitochondrial matrix, whereas the loop that links the two transmembrane α-helices projects into the intermembrane space (99, 107). MCU can be detected as part of a large complex that migrates at ∼450 kDa when subjected to electrophoresis under nondenaturing conditions, and MCU is capable of oligomerizing with itself (99). Cryo-electron microscopy (cryo-EM) and X-ray crystallography of fungal MCUs support a model in which individual MCU proteins come together as a symmetric homotetramer consisting of a “dimer of dimers,” with fourfold symmetry of the transmembrane domains and twofold symmetry of the NH₂-terminal domains (106, 108, 109) (FIGURE 2).

FIGURE 2.

Assembly and stoichiometry of the mitochondrial calcium uniporter channel. Single MCU subunits assemble into tetramers with the assistance of the accessory mtCU subunit, MCUR1. In the absence of EMRE, these MCU tetramers are not capable of conducting Ca²⁺ across the inner mitochondrial membrane (IMM). EMRE associates with MCU at a likely 1-to-1 ratio in functional Ca²⁺ channels. The regulatory protein MICU1 binds to MCU and EMRE within the intermembrane space (IMS). In conjunction with its binding partners MICU2 or MICU3, MICU1 acts as a gatekeeper for the mtCU at low cytosolic Ca²⁺ concentrations and facilitates cooperative activation of the mtCU as cytosolic Ca²⁺ concentration rises. Current models suggest a stoichiometry of 1 MICU dimer (MICU1/1, MICU1/2, or MICU1/3):4 MCU:up to 4 EMRE. Binding of MICU1 to the channel is proposed to displace the scaffolding factor MCUR1. MCUB, a dominant-negative paralog of MCU, can replace MCU subunits within the channel pore. Incorporation of MCUB into the mtCU impairs Ca²⁺ conductance through the channel by disrupting the structure of channel pore and by displacing MICU dimers from the channel. See glossary for abbreviations.

Nuclear magnetic resonance and electron microscopy investigations of Caenorhabditis elegans MCU indicate that this protein also forms homooligomers in higher organisms (110). Recent data suggest that in zebrafish MCU also organizes into tetramers, although in these zebrafish tetramers the MCU NH₂-terminal domains are arranged in an asymmetric fashion (111). The current structural models of MCU indicate that the conserved DIME motif (W-D-X-X-E-P-V-T-Y, where X indicates hydrophobic residues; corresponding to residues 260–268 of human MCU) that lies within the linker region forms part of the pore entrance and selectivity filter of the channel (106, 108–111). This model is strengthened by the observation that mutation of the DIME motif abrogates MCU-dependent _mCa²⁺ uptake in vitro (99).

Although not part of the transmembrane MCU pore, the large NH₂-terminal domain is thought to regulate MCU activity. Truncated human MCU lacking the NH₂-terminal domain is incapable of restoring mitochondrial Ca²⁺ uptake when expressed in MCU-knockdown HeLa cells, despite correct assembly into complexes containing MICU1 and MICU2 (112). NH₂-terminal deleted MCU also exerts dominant-negative effects on the function of wild-type (WT) MCU, suggesting that the NH₂-terminal domain is critical for appropriate permeation of the channel. The crystal structure of the NH₂-terminal domain of human MCU reveals the presence of a β-grasp-like fold in residues 72–189, which is capable of homooligomerization (113). Several acidic residues (D131, D142, D147, D148, and D166) within this region form a negatively charged patch that can bind divalent cations including Ca²⁺ and Mg²⁺. Such cation binding, or mutation of the acidic residues to eliminate their negative charge, disrupts the oligomerization of the β-grasp-like fold and inhibits MCU activity (113). The association of independent MCU proteins into functional multimeric channels is likewise disrupted by Ca²⁺ and Mg²⁺ (113). Furthermore, either loading of the mitochondria with Mg²⁺ or inhibition of _mCa²⁺ efflux to increase _mCa²⁺ concentration suppresses the rate of acute _mCa²⁺ uptake, independent of changes in ΔΨ_m (113). These findings indicate that the NH₂-terminal domain of MCU is important for oligomerization of MCU into functional mtCU channels and provide a potential mechanism by which matrix Mg²⁺ and possibly matrix Ca²⁺ inhibit MCU activity. Though hypothesized in the literature (114), the effect of certain concentrations of matrix Ca²⁺ to inhibit the uniporter remains controversial. It has been investigated further in recent years by the Foskett laboratory and may involve complex interactions between MCU (in particular, Ca²⁺ binding at the acidic residues D131 and D147 of MCU’s NH₂-terminal domain), EMRE, and MICU1/2 (115, 116).

3.3.1.3. properties and function of mcu.

Patch-clamp experiments exploring the biophysical properties of mtCU complexes were performed by Kirichok’s laboratory following the genetic identification of MCU. Patch-clamp measurements of whole mitoplasts derived from mouse heart, skeletal muscle, liver, kidney, or brown adipose tissue revealed distinct differences in MCU current densities among these tissues, ranging from 58 pA/pF in skeletal muscle down to 2.1 pA/pF in the heart (117). Further analysis of the electrophysiological properties of the MCU current in heart and skeletal muscle mitoplasts confirmed that the same molecule transports this current in both tissues but that the overall activity of heart MCU is reduced compared with skeletal muscle MCU (117). The measured differences in MCU current across tissues likely also reflect tissue-specific differences in the incorporation of regulatory components of the mtCU into the intact uniporter channel. The observation that MCU activity is approximately fivefold higher in neonatal versus adult mouse heart mitoplasts (117) suggests that overall mtCU activity is regulated in a developmental manner as well.

3.3.1.4. genetic diseases associated with mcu.

In contrast to other mtCU components (TABLE 1), no primary mutations in MCU have yet been causatively linked to human disease. However, a recent report of increased MCU transcription and increased _mCa²⁺ uptake occurring downstream of mutations in leucine-rich repeat kinase 2 (LRRK2), which is linked to late-onset familial Parkinson’s disease, supports a role for a secondary increase in _mCa²⁺ uptake in the neurodegeneration observed in this disease (118). Such secondary alterations in MCU expression or function may likewise contribute to the development or progression of a wide variety of genetic diseases, but no causal relationships have yet been established.

3.3.1.5. animal models of mcu and roles of mcu in physiology.

3.3.1.5.1. Germline MCU deletion.

As the pore-forming subunit of the mitochondrial calcium uniporter complex, MCU is absolutely critical for the rapid _mCa²⁺ uptake traditionally described as “uniporter” activity. This fact is supported by the phenotypes of numerous animal models in which MCU is knocked out or disrupted (TABLE 2). The first genetic mouse model for MCU disruption was published in 2013 by the Finkel and Murphy laboratories and employed a gene trap strategy to constitutively delete MCU from all tissues (119) (also reviewed in Refs. 120, 121). Somewhat surprisingly, these animals are viable on a hybrid background and show no obvious defects in mitochondrial structure or number. However, rapid Ca²⁺ uptake over a range of 0.5–5.7 mM bath Ca²⁺ is ablated in isolated MCU^−/− skeletal muscle mitochondria and cardiac mitochondria. The same lack of rapid _mCa²⁺ uptake is noted in intact mouse embryonic fibroblasts (MEFs) prepared from MCU^−/− embryos and in isolated MCU^−/− adult mouse cardiomyocytes. Despite lowering resting _mCa²⁺ concentration, constitutive loss of MCU has little effect on basal metabolic function of MEFs or purified mitochondria (119). Rather, the main metabolic consequence of constitutive MCU ablation is loss of the ability to increase matrix Ca²⁺ concentration to stimulate mitochondrial respiration (119).

Table 2.

Animal models for _mCa²⁺ handling proteins of the inner mitochondrial membrane

Reference	Model	Major Phenotype(s)
Pan et al. (119)	Mcu gene trap mouse (Mcu-null)	Ablation of rapid mCa2+ uptake; reduced skeletal muscle PDH activity; impaired capacity for treadmill running; reduced grip strength. Loss of Ca2+-induced PT, but cells are not protected from death with other stimuli. No protection from cardiac I/R injury.
Holmstrom et al. (122)	Mcu gene trap mouse (Mcu-null)	Impaired Ca2+-induced respiration but no cardiac functional defects basally; no change in heart’s response to chronic pressure overload or acute isoproterenol stress
Wu et al. (124)	αMHC-DN-MCU mice [constitutive expression of dominant-negative mutant (DIME→QIMQ) MCU in cardiomyocytes]	Impaired chronotropic response to isoproterenol; aberrant cellular Ca2+ handling related to impaired ATP production
Rasmussen et al. (125)	αMHC-DN-MCU mice [constitutive expression of dominant-negative mutant (DIME→QIMQ) MCU in cardiomyocytes]	Increased basal oxygen consumption, impaired inotropic and lusitropic contractile responses to increased cardiac pacing frequency; no protection against cardiac I/R injury
Luongo et al. (126)	Mcuflfl × αMHC-MCM mice; tamoxifen-inducible deletion of Mcu in adult cardiomyocytes	Reduced mPTP activation upon acute Ca2+; stress protection from cardiac I/R injury; diminished contractile response to acute β-adrenergic stimulation
Kwong et al. (127)	Mcuflfl × αMHC-MCM mice; tamoxifen-inducible deletion of Mcu in adult cardiomyocytes	No basal cardiac phenotype out to 1 yr of age, diminished NCLX expression and activity; reduced mPTP activation upon acute Ca2+, stress protection from cardiac I/R injury; diminished contractile response to acute β-adrenergic stimulation; impaired treadmill running that could be overcome with an exercise warm-up period; no protection from chronic pressure overload
Mammucari et al. (128)	8-wk AAV-mediated MCU overexpression in mouse hindlimb skeletal muscle	Increased skeletal muscle fiber size; increased mitochondrial content; increased PGC-1α4 and IGF1-Akt/PKB signaling
Mammucari et al. (128)	8-wk AAV-mediated MCU knockdown in mouse hindlimb skeletal muscle	Decreased skeletal muscle fiber size; decreased mitochondrial content; attenuated PGC-1α4 and IGF1-Akt/PKB signaling
Altamimi et al. (130)	Mcuflfl × αMHC-MCM mice; tamoxifen-inducible deletion of Mcu in adult cardiomyocytes	Increased fatty acid oxidation supporting increased contractility at baseline and after isoproterenol
Kwong et al. (129)	Mcuflfl × MyoD-Cre mice; constitutive deletion of Mcu in skeletal muscle	Impaired mCa2+ uptake an attenuation of Ca2+-stimulated mitochondrial respiration; impaired treadmill running capacity that could be overcome with a warm-up period; impaired glucose oxidation and impaired entry of carbon substrates into the TCA cycle; increased capacity for fatty acid metabolism.
Kwong et al. (129)	Mcuflfl × skeletal muscle α actin-MCM mice; tamoxifen-inducible deletion of MCU in adult skeletal muscle	Impaired mCa2+ uptake but no effect on skeletal muscle growth
Flicker et al. (132)	Mcuflfl × UCP1-Cre mice; constitutive deletion of Mcu in brown adipose tissue	Little effect on cold tolerance, diet-induced obesity, or transcriptional responses to cold exposure, despite effective ablation of acute mCa2+ uptake
Drago and Davis (134)	Drosophila with RNAi-mediated silencing of MCU homolog (GC18769) in mushroom body neurons	Defects in memory, but not learning, in adult flies. This effect was caused by silencing of MCU specifically during pupal development, which led to altered structure of mushroom body neurons.
Drago and Davis (134)	Drosophila with RNAi-mediated global silencing of MCU homolog (GC18769)	Developmentally lethal
Choi et al. (133)	Drosophila with loss-of-function mutant of MCU homolog (GC18769)	No gross defects, but flies are protected from oxidative stress. impaired mCa2+ uptake in response to caffeine.
Hutto et al. (237)	Zebrafish with cone photoreceptor-specific MCU overexpression	Increased mCa2+ uptake and mitochondrial swelling, but this is tolerated through late adulthood. Compensatory downregulation of MICU3.
Choi et al. (133)	Drosophila with muscle-specific MCU overexpression	Lethal, but lethality could be blocked by simultaneous knockdown of IP3R; lethality not blocked by simultaneous knockdown of IP3R and Sod1
Choi et al. (133)	Drosophila with eye-specific overexpression of EMRE	No defective phenotype unless combined with overexpression of MCU; combined overexpression of MCU and EMRE is lethal.
Choi et al. (133)	Drosophila with knockdown of EMRE	Impaired mCa2+ uptake; phenocopies deletion of MCU.
Liu et al. (164)	Mice with constitutive deletion of Emre	Homozygous Emre−/− mice are born less frequently than expected and are small but viable. Resistance to Ca2+-induced permeability transition but no protection from cardiac I/R injury.
Liu et al. (163)		Heterozygous Emre deletion rescues homozygous deletion of Micu1.
Antony et al. (206)	Mice with constitutive deletion of Micu1 (Micu1flfl × germline Cre-eIIa)	Perinatal lethality, trend toward reduced neuron density in nucleus ambiguus and nucleus facialis, possible defect in respiration. MEFs show excessive mCa2+ at low bath Ca2+ concentration.
Antony et al. (206)	Mice with inducible deletion of Micu1 in the liver (Micu1flfl injected with AAV8-Cre under a hepatocyte-specific thyroxine-binding globulin promoter)	Tolerated at baseline, but increased susceptibility to liver injury and impaired liver regeneration following experimental stress with partial hepatectomy
Liu et al. (163)	Mice with CRISPR/Cas9 mediated constitutive deletion of Micu1	Perinatal lethality, small body weight and developmental delay, brain abnormalities, ataxia, and muscle weakness. Gradual compensatory downregulation of EMRE with age is associated with partial normalization of body mass and brain histology.
Debattisti et al. (207)	Mice with constitutive deletion of Micu1 in skeletal muscle (Micu1flfl × Creatine kinase-Cre)	Impaired mCa2+ uptake during twitch and tetanic muscle contraction, and sarcolemmal repair defect leading to muscle weakness and wasting
Drago and Davis (134)	Drosophila with RNAi-mediated silencing of MICU1 homolog (CG4495) in mushroom body neurons	Defects in memory, but not learning, in adult flies
Drago and Davis (134)	Drosophila with RNAi-mediated global silencing of MICU1 homolog (CG4495)	Developmentally lethal
M’Angale and Staveley (179)	Drosophila with inducible RNAi-mediated silencing of MICU1 homolog (GC4495) in neurons	Reduced survival and early loss of locomotor function
Xue et al. (209)	siRNA knockdown of Micu1 via lentiviral injection in the mouse heart	Exacerbated mCa2+ overload and myocardial injury, with exacerbated cardiac dysfunction, upon myocardial I/R
Tufi et al. (195)	Drosophila with homozygous null mutation of MICU1 homolog	Lethal; lethality is not rescued by simultaneous knockout of MCU or EMRE, or by ubiquitous overexpression of MICU3.
Bick et al. (234)	Micu2 gene trap mice with constitutive, global knockout of Micu2	Little baseline phenotype, but diastolic dysfunction leads to left atrial enlargement by 16–18 mo of age. Accelerated cardiac decompensation upon chronic angiotensin II infusion; increased susceptibility to abdominal aortic aneurysm with chronic angiotensin II infusion
Tufi et al. (195)	Drosophila with CRISPR/Cas9-induced disruption of MICU3 homolog	Viable, but have a modest reduction in life span, climbing defect (interpreted as neurological defect) in both young and older flies. No effect on basal mitochondrial respiration in fly head tissue.
Puente et al. (236)	Mice with germline Micu3 deletion	No basal cardiac phenotype but protection from chronic isoproterenol-induced mCa2+ overload, contractile dysfunction, and left ventricular dilation; protection from ex vivo I/R injury
Lambert et al. (242)	CAG-CAT-MCUB transgenic × αMHC-MerCreMer mice (inducible adult cardiomyocyte-specific MCUB overexpression):
	• 5-day cardiac MCUB overexpression	Acute 5-day overexpression: diminished acute mCa2+ uptake; diminished baseline cardiac contractility and diminished contractile response to acute β-adrenergic stimulation; reduced mitochondrial metabolism; increased mortality during cardiac ischemia
	• 1-mo cardiac MCUB overexpression:	Chronic 1-mo overexpression: diminished acute mCa2+ uptake; normal cardiac contractile response to acute β-adrenergic stimulation, normal mitochondrial metabolism; reduced myocardial infarct size upon I/R
Huo et al. (241)	Tetracycline-off model of cardiomyocyte MCUB overexpression (αMHC-tTA × TRE-MCUB mice, constitutively kept off DOX to allow constitutive MCUB overexpression	No detrimental effect on baseline cardiac function; protection of heart from I/R injury and mPTP activation
Huo et al. (241)	Mice with Mcub knockout first allele, for constitutive whole body disruption of Mcub	No detrimental effect on mCa2+ uptake, oxygen consumption, or cardiac function at baseline; exacerbated myocardial injury, cardiac remodeling, and contractile function following cardiac I/R
Tomar et al. (253)	Mice with constitutive, endothelial cell-specific deletion of Mcur1 (Mcur1flfl × VE-Cad-Cre)	Impaired endothelial cell mCa2+ uptake and reduced basal and agonist-induced increase in ATP; diminished cell proliferation and migration but increased autophagy. Mice are normal but have increased body heat dissipation associated with increased UCP2 expression in Mcur1-KO endothelial cells.
Tomar et al. (253)	Mice with constitutive, cardiomyocyte-specific deletion of Mcur1 (Mcur1flfl × αMHC-Cre)	Born at expected ratios but are small and die within 3 wk of birth; decreased cardiomyocyte mCa2+ uptake and mtCU currents, but normal cardiomyocyte mCa2+ content; increased autophagy in Mcur1-KO cardiomyocytes
Beutner et al. (275)	Mice with constitutive RyR1 knockout (RyR1−/−)	Lethal birth defect; animals die immediately after birth; trend toward elevated basal oxygen consumption in neonatal RyR1−/− heart homogenates; loss of Ca2+-stimulated increase in mitochondrial respiration
Luongo et al. (166)	Mice with tamoxifen-inducible, cardiomyocyte-specific deletion of NCLX (Slc8b1fl/fl × αMHC-MerCreMer)	Reduced cardiomyocyte mCa2+ efflux, cardiomyocyte mCa2+ overload, increased ROS production, and cardiomyocyte necrosis; left ventricular dilation and impaired contractility, cardiac hypertrophy and fibrosis; 87% lethality within 2 wk of cardiomyocyte NCLX deletion; rescued by simultaneous deletion of the mPTP component Cyclophilin D
Luongo et al. (166)	Mice with constitutive, cardiomyocyte-specific NCLX deletion (Slc8b1flfl × αMHC-Cre)	Normal viability but reduced mCa2+ efflux; compensatory reduction in mCa2+ uptake; normal cytosolic Ca2+ handling
Luongo et al. (166)	Mice with cardiomyocyte-specific, doxycycline-controlled transgenic NCLX overexpression (TRE-NCLX × αMCH-tTA)	Increased cardiomyocyte mCa2+ efflux; increased resistance to permeability transition, protection against cardiac I/R injury; protection from LV dilation, contractile dysfunction, hypertrophy, fibrosis, and inflammation after myocardial infarction
Jadiya et al. (323)	Neuronal-specific, constitutive deletion of NCLX in 3xTg-AD Alzheimer’s disease mouse model (3xTg-AD × Slc8b1fl/fl × Camk2a-Cre)	Accelerated impairment in spatial working memory, contextual recall, and cued recall; increased amyloid burden and tau pathology
Jadiya et al. (323)	Neuronal-specific, doxycycline-controlled overexpression of NCLX in 3xTg-AD Alzheimer’s disease mouse model (3xTg-AD × TRE-NCLX × Camk2a-tTA)	Protection against age-related cognitive decline, reduced amyloid plaque burden and tau pathology; reduced susceptibility to permeability transition; reduced Alzheimer’s disease-associated increase in brain superoxide production and lipid peroxidation
Pathak et al. (325)	Mice with constitutive, global CRISPR/Cas9-meditated disruption of NCLX (Slc8b1−/−)	Viable and develop fewer and smaller tumors when subjected to the colitis-associated colorectal cancer model
Sharma et al. (326)	C. elegans with null mutation in the NCLX-like gene ncx-9	Developmental defects in the left/right projection patterning of the GABAergic motor neuron circuit
Hasegawa and van der Bliek (343)	C. elegans with null mutation in letm-1	Homozygous animals arrest in L3 larval stage and are small and infertile. Heterozygous animals are normal.
Hasegawa and van der Bliek (343)	C. elegans with RNAi-mediated knockdown of letm-1	Delayed development and small body size at adulthood; swollen and disorganized mitochondria
Hasegawa and van der Bliek (343)	C. elegans with transgenic expression of LETM-1 under the control of the myo-3 promoter	Crimping of the mitochondrial matrix, swelling of outer mitochondrial membrane, occasional detachment of outer membrane from the inner mitochondrial membrane and matrix
McQuibban et al. (353)	Drosophila with ubiquitous LETM1 (CG4589) knockdown using the tub-GAL4 driver, the da-GAL4 driver, or the act-GAL4 driver	Lethal during development
McQuibban et al. (353)	Drosophila with muscle-specific LETM1 (CG4589) knockdown using the mef2-GAL4 driver	Larvae are small and have reduced physical activity; most flies arrest growth in the pupal stage. Flies that progress to adulthood are small, weak, and unable to fly.
McQuibban et al. (353)	Drosophila with eye-specific LETM1 (CG4589) knockdown using the ey-GAL4 driver	Small eye facets surrounded by scar tissue; ommatidia exhibit swollen mitochondria.
McQuibban et al. (353)	Drosophila with nervous system-specific LETM1 (CG4589) knockdown using the elav-GAL4 driver or s-nyb-GAL4 driver	Reduced speed of locomotion and increased time spent immobile; reduced neurotransmitter release following nerve stimulation
Jiang et al. (372)	Mice with homozygous deletion of Letm1 via gene trap	Lethal by day 6.5 of embryogenesis
Jiang et al. (372)	Mice with heterozygous deletion of Letm1 via gene trap	∼50% die by day E13.5. Surviving mice are relatively normal and have normal mitochondrial morphology, but have impaired brain ATP concentration and reduced PDH activity. Have increased seizures in response to kainic acid.
Zhang et al. (377)	Rats with lentiviral knockdown of LETM1 in the hippocampus and dentate gyrus	Mitochondrial swelling and reduced mitochondrial gene expression; increased susceptibility to seizures in a pilocarpine-induced epilepsy model. Increased seizure susceptibility is not corrected by treatment with the K+/H+ ionophore nigericin.
Elrod et al. (414)	Mice with constitutive, global ablation of Cyclophilin D (Ppif−/−)	Loss of mPTP activity; impaired cardiac mCa2+ efflux; increased cardiac TCA cycle flux and shift toward glucose rather than fatty acid metabolism; impaired cardiac contractility in response to acute isoproterenol infusion; exaggerated pressure overload-induced cardiac dysfunction and hypertrophy; exaggerated cardiac dysfunction and death upon chronic exercise

See glossary for abbreviations.

Pan et al. (119) report that germline MCU-null animals have reduced grip strength, forelimb muscle strength, and maximal work capacity when running on an inclined treadmill, despite no changes in skeletal muscle fiber type composition. These findings indicate that rapid _mCa²⁺ uptake through the mtCU may be required for acute increases in muscle force production and/or power output, likely by supporting increased mitochondrial ATP production required to fuel increased muscle contraction (FIGURE 3).

FIGURE 3.

Ca²⁺ signaling in the parallel activation of ATP-consuming processes and ATP production. Ca²⁺ serves as a second messenger that coordinates the activation of both ATP-consuming and ATP-generating processes within the cell. This process is particularly important for tissues with a high energy demand such as the heart. The parallel activation of ATP consumption and ATP generation within cardiomyocytes is illustrated stepwise as follows: 1) Extracellular Ca²⁺ enters the cardiomyocyte through the L-type calcium channel (LTCC) that opens with each action potential. 2) Ca²⁺ that entered the cardiomyocyte through the LTCC activates the ryanodine receptor (RyR2), triggering calcium-induced calcium release from the sarcoplasmic reticulum (SR). 3) Increased cytosolic Ca²⁺ stimulates the ATP-consuming process of myofilament cross-bridge cycling and cardiac contraction. At the same time, increased cytosolic Ca²⁺ concentration drives increased mitochondrial Ca²⁺ uptake that stimulates mitochondrial ATP production. 4) This mechanism of Ca²⁺-dependent parallel activation allows the heart to respond to an increased workload with a sustained increase in mitochondrial ATP production that matches the increase in ATP consumption by the contractile apparatus. See glossary for abbreviations.

Additional experiments by Pan et al. argue for a role for MCU-dependent _mCa²⁺ uptake in mitochondrial permeability transition. Both liver and heart MCU^−/− mitochondria are remarkably resistant to Ca²⁺-induced mPTP opening (119). Somewhat unexpectedly, this protection against Ca²⁺-induced mPTP opening does not translate to any measurable protection of constitutive MCU-null hearts from ex vivo, global ischemia-reperfusion (I/R) injury (119). The reasons for this discrepancy are incompletely understood but may involve roles for elevated cytosolic, rather than mitochondrial, Ca²⁺ in cell death signaling during I/R injury or upregulation of mPTP-independent cell death pathways such as necroptosis in Mcu^−/− mice (119, 120).

A later study by the same group used the MCU gene trap mouse model to examine the specific consequences of germline MCU deletion on cardiac physiology. Mcu^−/− cardiac mitochondria exhibit reduced resting matrix Ca²⁺ content and an impaired respiratory response to Ca²⁺ (122). EMRE levels are also reduced in Mcu^−/− hearts. However, constitutive loss of MCU has no detectable effect on baseline cardiac function out to nearly 2 yr of age, on cardiac structural or functional phenotype after 8 wk of transverse aortic constriction (TAC), or on contractile responses to acute in vivo isoproterenol stimulation (122). The lack of a significant defect in acute functional responses to β-adrenergic stimulation in MCU-null hearts, despite a loss of mitochondrial Ca²⁺ uptake (119), cast some doubt on the link between acute _mCa²⁺ signaling and the increase in mitochondrial ATP synthesis required to fuel an increase in cardiac contractility. However, in light of findings in conditional MCU models discussed below, this negative result may simply be the result of the constitutive germline deletion strategy allowing for significant compensatory mechanisms to be activated during both development and aging, thereby masking any role for MCU in these processes (121, 123).

3.3.1.5.2. Conditional MCU models.

A series of conditional MCU mutant mouse models published in 2015 offer further insight into the role of MCU in cardiac physiology and disease. Mark Anderson’s group (124) developed transgenic mice in which the cardiac-specific α-myosin heavy chain (α-MHC) promoter drives expression of a dominant-negative MCU (DN-MCU) shortly after birth. In this construct, the aspartic acid and glutamic acid residues of the critical DIME motive are mutated to glutamine (i.e., DIME→QIMQ), resulting in ablation of MCU activity, presumably due to the displacement of endogenous MCU from the uniporter channel. DN-MCU mice have a normal basal heart rate but have a blunted increase in heart rate upon in vivo isoproterenol administration or increased physical activity, which correlates with diminished atrial ATP content. Administration of exogenous ATP in vitro to isolated cardiac sinoatrial node cells expressing DN-MCU rescues the isoproterenol-induced increase in action potential frequency, suggesting a link between _mCa²⁺ uptake and cellular ATP production required for chronotropic responsiveness (124). DN-MCU sinoatrial cells also exhibit diminished SR Ca²⁺ content and reduced diastolic Ca²⁺ release. Wu et al. (124) hypothesize that this effect results from impaired ATP-dependent SERCA activity required for Ca²⁺ reuptake into the SR. They propose that MCU is specifically required for the increase in sinoatrial node activity and heart rate in response to stress, because _mCa²⁺ uptake through MCU enhances OXPHOS activity to increase ATP levels sufficiently to power ionic homeostasis and cellular Ca²⁺ cycling.

DN-MCU hearts also have higher rates of oxygen consumption than controls across multiple frequencies of electrical pacing, despite a diminishment in left ventricular inotropic and lusitropic responses to electrical pacing (125). This effect on oxygen consumption rate (OCR) is not observed under low-Ca²⁺ conditions in isolated cardiomyocytes or mitochondria, suggesting that MCU inhibition causes extramitochondrial adaptations leading to increased OCR during Ca²⁺ stress. Such adaptations could arise as a result of the increased energetic demand needed to maintain cytosolic Ca²⁺ homeostasis in the absence of MCU-dependent _mCa²⁺ uptake. Similar to the MCU gene trap hearts (122), DN-MCU hearts are not protected from I/R-induced cell death (125). This observation reinforces the notion that chronic disruption of MCU activity may drive compensatory adaptations to maintain critical cellular processes (cell death signaling, etc.) in the absence of rapid _mCa²⁺ uptake. Indeed, this study reports increased expression of the cell death gene Bax in DN-MCU hearts, which may explain why DN-MCU hearts are not protected from I/R injury despite abolished rapid _mCa²⁺ uptake (125).

The question of the acute role for MCU in cardiac stress physiology, separable from compensatory effects allowing adaptation to chronic knockout or inhibition of MCU, was clarified by the generation of MCU floxed mice by our group (126) and the Molkentin laboratory (127), allowing for tissue-specific and temporally controlled MCU ablation when crossed to inducible Cre drivers. The major findings from these studies are presented here; for a more thorough examination and a discussion of how we can reconcile conflicting results from different genetic MCU models, the reader is directed to recent reviews (121, 123). Rapid _mCa²⁺ uptake is ablated in cardiomyocytes isolated from tamoxifen-treated Mcu^fl/fl × αMHC-MerCreMer (MCM) mice. However, no differences in basal cardiomyocyte _mCa²⁺ are observed between controls and Mcu^fl/fl × αMHC-MCM mice (126, 127). This finding is consistent with the Molkentin laboratory’s observation of no baseline phenotype in MCU conditional knockout hearts at up to 1 yr of age (127). Decreased expression and activity of the mitochondrial sodium/calcium exchanger, NCLX, may help to maintain relatively normal basal matrix Ca²⁺ content in these hearts (127).

Acute cardiomyocyte Mcu deletion blunts the increase in cardiac contractility upon adrenergic stimulation; this is associated with an impaired increase in _mCa²⁺ content, OCR, and NADH production (126, 127). Kwong et al. (127) also note a diminished maximal treadmill running capacity in mice with acute cardiomyocyte Mcu deletion, presumably related to impaired matching of cardiac energetic demand and cardiac energy production. However, this exercise defect can be overcome if these mice are allowed a prolonged exercise warm-up period before the running test (127). Furthermore, when subjected to prolonged isoproterenol stimulation, the OCR, cardiac contractility, and _mCa²⁺ content of MCU-knockout cardiomyocytes or hearts eventually attain values matching those of control mice (127). These results strengthen the idea that MCU is specifically required for acute increases in mitochondrial Ca²⁺, to mediate acute adaptations to the bioenergetic stress of increased contractile demand (FIGURE 3). Given sufficient time to take effect, though, additional MCU-independent pathways can increase _mCa²⁺ content and enable mitochondrial and/or extramitochondrial energetic pathways to provide cardiomyocytes with sufficient ATP to support contractility.

In contrast to findings in mice with germline MCU deletion (119) or constitutive cardiomyocyte expression of DN-MCU (125), acute loss of MCU from adult cardiomyocytes in Mcu^fl/fl × αMHC-MCM mice protects against myocardial I/R injury and preserves contractile function after reperfusion (126, 127). Acute MCU deletion also reduces mitochondrial swelling and the frequency of mPTP opening in response to Ca²⁺ stress, indicating that the mtCU is responsible for acute _mCa²⁺ overload that triggers permeability transition and cardiomyocyte death upon intracellular Ca²⁺ overload during I/R injury (126, 127). Finally, a small study testing the consequences of adult cardiomyocyte-specific MCU knockout on the heart’s response to chronic hemodynamic stress resulting from 8 wk of transverse aortic constriction (TAC) showed no effect of MCU loss on pathological cardiac remodeling or contractile function (127). These data would suggest that MCU is dispensable for the cardiomyocytes’ adaptations to sustained hemodynamic or energetic stress. However, the length of time (18 wk) between the start of tamoxifen administration and the end point of the TAC study warrants caution in this interpretation, as this may have been sufficient time for alternative pathways to compensate for the lack of MCU and mask any beneficial or detrimental effects that might have been observed with more acute Mcu deletion.

Despite the discrepancies in the above models of constitutive versus conditional MCU disruption, several consistent themes regarding the physiological role of MCU emerge from these animal models: 1) MCU is required for rapid _mCa²⁺ uptake, which may be most physiologically relevant in situations requiring acute increases in cellular energetic demand (i.e., during exercise or sympathetic stimulation) rather than required for basal cellular function. 2) Additional, slower routes of _mCa²⁺ entry independent of MCU and/or the mtCU complex may exist to support residual matrix Ca²⁺ levels in MCU-deficient mitochondria, with the caveat that accurate measurement of basal matrix Ca²⁺ levels is tricky business (126) and so any additional routes of _mCa²⁺ entry may be moot in these experiments. 3) Excessive Ca²⁺ uptake via the mtCU complex contributes to _mCa²⁺-dependent mPTP opening. 4) Over time, cells may remodel to compensate for the loss of MCU to sustain critical processes such as energy metabolism and cell death versus survival signaling via alternative, MCU-independent mechanisms. Such adaptive mechanisms may help to explain why disruption of MCU in adult animals has little apparent effect on the basal, unstressed phenotype of tissues such as the heart.

Additional animal models of MCU manipulation have yielded further insights into the physiological roles of MCU in tissues beyond the heart. Adeno-associated virus (AAV)-mediated MCU overexpression in mouse skeletal muscle increases fiber size in both slow and fast muscles, whereas MCU knockdown decreases fiber size in both muscle types (128). Skeletal muscle MCU overexpression is associated with increased mitochondrial content, increased expression of PGC-1α4 (a regulator of mitochondrial biogenesis), and increased downstream signaling by prohypertrophic insulin, IGF1-Akt/PKB, and mTOR pathways (128). In contrast, MCU knockdown attenuates these pathways (128). Mammucari et al. (128) demonstrate that MCU overexpression is sufficient to attenuate muscle atrophy following denervation caused by sciatic nerve section, suggesting that MCU-dependent _mCa²⁺ uptake can counteract pathological atrophy as well as stimulate hypertrophy. These authors propose that MCU overexpression triggers muscle growth by altering the anabolic versus catabolic balance, rather than exclusively via a metabolic effect, since aerobic metabolism is not altered by MCU levels and hypertrophic effects of MCU overexpression are observed in both oxidative and glycolytic muscles.

Constitutive deletion of MCU in skeletal muscle with a MyoD-Cre driver in floxed MCU mice does not affect muscle size or AKT signaling (129). In contrast to the findings by Mammucari et al. (128), genetic deletion of MCU in adult skeletal muscle likewise has no effect on muscle weight or pathology (129). However, the disparity in these results may reflect differences in MCU knockout versus knockdown strategy or the sex or genetic background of the mice used [CD1 males (128) vs. C57BL/6 males and females (129)]. MCU-knockout skeletal muscle exhibits impaired glucose oxidation, consistent with limited _mCa²⁺-dependent stimulation of pyruvate dehydrogenase (PDH) activity, but an increased capacity for fatty acid metabolism (129). A similar phenotype of increased cardiac fatty acid metabolism is observed in mice with adult cardiomyocyte-specific loss of MCU when examined 3 mo after MCU deletion (130). In this study, ex vivo working hearts with MCU deletion (Mcu^fl/fl × αMHC-MCM mice) unexpectedly show enhanced contractile function compared with Mcu^fl/fl controls, both at baseline and after acute isoproterenol challenge. Hearts of both genotypes increase glucose oxidation in response to isoproterenol, but only hearts with Mcu deletion increase fatty acid oxidation after isoproterenol administration. Thus, MCU-null striated muscle may adapt to impaired _mCa²⁺-regulated carbohydrate oxidation by upregulating pathways for fatty acid metabolism as an alternative mechanism to access energy reserves during periods of stress. Altamimi et al. (130) propose that such increased capacity for fatty acid utilization explains the enhanced contractile function observed in MCU-deleted hearts. Unfortunately, αMHC-MCM controls were not included in this study (130), thus precluding any discrimination of any specific effects of MCU loss alone on cardiac function and metabolism, as opposed to effects of cardiac Cre expression, which is noted to generate significant phenotypes (131). Despite these limitations, the studies discussed above support the hypothesis that acute _mCa²⁺ uptake through MCU supports the use of carbohydrates by striated muscle tissue, with particularly important acute effects on muscle energetics and work capacity, but that with sufficient time, muscle can functionally compensate for the loss of MCU by upregulating fatty acid oxidation. That is, MCU-dependent Ca²⁺ uptake is likely not a direct regulator of fatty acid oxidation, but rather increased fatty acid oxidation is a downstream compensatory adaptation to diminished _mCa²⁺ uptake.

Elucidation of the role of MCU in the physiology of nonmuscle tissues is ongoing. A third Mcu^fl/fl mouse on the C57BL/6 background was independently generated by Vamsi Mootha’s group and crossed to mice expressing Cre under the control of the uncoupling protein 1 (UCP1) promoter, allowing for constitutive deletion of Mcu in brown adipose tissue (BAT) (132). Despite evidence for large mtCU currents in BAT (117), ablation of acute _mCa²⁺ uptake via MCU knockout has little effect on cold tolerance, diet-induced obesity, or cold-induced transcriptional responses (132). Thus, MCU appears largely dispensable for bioenergetics in BAT (132), in contrast to the effects described above for striated muscle. These findings may reflect diversity in the physiological relevance of MCU-mediated _mCa²⁺ uptake across different tissue types. It is important to note, however, that in this constitutive MCU-knockout model compensatory adaptations akin to those discussed for the original MCU gene trap mice (119–121, 123) may have obscured any physiological roles for MCU within BAT.

Research in Drosophila offers additional evidence for the importance of this protein to cellular function. Deletion of the Drosophila MCU homolog, GC18769, effectively ablates rapid _mCa²⁺ uptake but has no effect on viability or metabolic phenotype (133). MCU-mutant flies exhibit increased resistance to oxidative stress (H₂O₂) compared with wild-type flies, as indicated by increased survival, reduced TUNEL staining, and reduced expression of proapoptotic markers. This phenotype is attributed to a reduction in _mCa²⁺ uptake otherwise triggered by oxidative stress. Drosophila have also been used to investigate MCU in the brain. Drago and Davis (134) describe flies with knockdown of CG18769 via RNA interference (RNAi) in different neuronal populations of the olfactory system and show that silencing of MCU in mushroom body neurons impairs memory without affecting learning. Mutation of the MCU DIME motif in mushroom body neurons yields the same effect, indicating that functional MCU within this cell population is required for memory. Remarkably, memory seems to rely predominantly on the appropriate expression of MCU during development. Silencing of MCU specifically during the pupal stage affects memory in adult flies, whereas silencing of MCU earlier in development or in adult flies produces no memory defect (134). The loss of MCU during pupal development reduces synaptic vesicle number and/or size in the horizontal lobes of the mushroom body neurons and increases the length and field volume of the axons of αβ mushroom body neurons. Thus, MCU function is required during neuronal development to establish the appropriate neural circuitry or neuronal structure and function that underlies adult memory (134).

3.3.1.6. regulation of mcu.

The mouse models of MCU deletion discussed above suggest that MCU may be an attractive therapeutic target for pathologies involving cellular death caused by acute _mCa²⁺ overload, such as ischemia-reperfusion injury following treatment for a myocardial infarction. The molecular mechanisms controlling MCU expression and activity are thus a topic of active research. In vitro experiments in chicken DT40 B lymphocytes and HeLa cells reveal that the Ca²⁺-regulated transcription factor cyclic adenosine monophosphate response element-binding protein (CREB) binds directly to the MCU promoter and stimulates transcription of MCU (135). This mechanism is thought to increase MCU expression in order to increase _mCa²⁺ uptake in response to cytosolic Ca²⁺ signaling, perhaps as a way for the cell to buffer excess cytosolic Ca²⁺ and shape Ca²⁺-regulated signaling within the cell. A recent study in rats indicates that increased phospho-CREB-dependent MCU expression in the spinal cord dorsal horn contributes to the development of morphine tolerance (136). This work also detected binding of cytoplasmic polyadenylation element binding protein 1 (CPEB1) to the 3′ untranslated region of MCU mRNA, and showed that knockdown of CPEB1 decreases MCU protein expression, suggesting that CPEB1 facilitates translation of MCU mRNA into protein (136). Although CREB-dependent MCU transcription may occur in some cell types, one caveat to this hypothesis, which may preclude it from representing a robust mode of physiological regulation, is that the proposed signaling would put the cell in peril of mitochondrial Ca²⁺ overload and cell death.

MCU expression is also regulated by microRNAs (miRs). Zaglia et al. (137) find that miR-1 inhibits the translation of MCU mRNA and is inversely correlated with increasing MCU levels observed in both mouse and human hearts during physiological and pathological cardiac hypertrophy. These authors propose that chronic β-adrenergic receptor activation in pathological hypertrophy represses miR-1 expression and allows for increased cardiac MCU content. Fitting with this model, pharmacological β-adrenergic receptor blockade prevents the decline in miR-1 levels and associated increase in MCU protein expression observed in the hearts of mice subjected to transverse aortic constriction (137). miR-25 likewise downregulates MCU expression, which may protect cardiomyocytes against the damaging effects of oxidative stress by preventing _mCa²⁺ overload (138). miR-25 is also linked to downregulation of MCU in colon cancer and cancer cell lines, possibly functioning as a mechanism that prevents apoptosis and promotes the survival of cancer cells (139). In line with these findings, enhanced proliferation, migration, and apoptosis resistance of pulmonary artery smooth muscle cells in pulmonary arterial hypertension are associated with increased miR-138 expression and a trend toward increased expression of miR-25. These miRs together downregulate MCU expression, reduce _mCa²⁺ levels, and increase cytosolic Ca²⁺ concentration (140). However, a conflicting report suggests that, in breast cancer, increased MCU expression instead promotes cell migration, invasion, and metastasis by increasing glycolysis and the Warburg effect (141). In this study, increased expression of miR-340, another negative regulator of MCU expression, reduces the metastatic potential of breast cancer cell lines injected into nude mice. Considered together, these results indicate that numerous miRs can negatively regulate MCU expression, although the specific consequences of altered cellular MCU protein levels for cellular survival, growth, and proliferation can vary across different cell types and across distinct pathologies. However, it should be noted that a full review of the literature suggests minimal to no changes in MCU expression in response to physiological or pathological stimuli. Indeed, transcriptional modulation of the mtCU appears much greater for the mtCU regulators MICU1, EMRE, and MCUB.

MCU protein is subjected to a variety of posttranslational modifications that alter overall MCU activity. Acute regulation of MCU by a kinase, CaMKII, was first reported by Mark Anderson’s laboratory in 2012 (142). Genetic inhibition of CaMKII activity attenuates MCU current and protects the heart against pathologies associated with _mCa²⁺ overload, including I/R injury, myocardial infarction, and injury due to acute isoprenaline stimulation (142). This study identified serines 57 and 92 of MCU as potential CaMKII phosphorylation sites and demonstrated that phospho-null mutation of these sites to alanine is sufficient to block a CaMKII-dependent increase in MCU current in vitro in mitoplasts prepared from human embryonic kidney (HEK) cells. It should be noted, though, that concrete evidence of CaMKII at the IMM remains elusive. The idea that CaMKII signaling acutely regulates MCU current in the heart is challenged by Kirichok’s group, who question the identity of the large currents that Joiner et al. (142) identified as I_MCU, on the basis of the currents being two orders of magnitude greater than MCU currents measured in the Kirichok laboratory; discrepancies in Ru360 sensitivity of the measured current; and failure to replicate Joiner et al.’s finding of increased mitoplast MCU current upon addition of constitutively active CaMKII in vitro (143). However, other data indicate that S92A mutation of MCU impairs MCU activity in vitro (112), strengthening the idea that phosphorylation or other modification of this residue can modulate mtCU function and _mCa²⁺ uptake. The possible functional consequences of MCU S92 phosphorylation by CaMKII or other kinases remains a topic of active research. Conflicting studies variously indicate a role for CaMKII in phosphorylating S92 to promote MCU activity, mitochondrial mobility, and the migration of vascular smooth muscle cells (144) or, instead, little consequence of CaMKII deletion on _mCa²⁺ uptake, mitochondrial redox state, respiration, or reactive oxygen species (ROS) production in cardiac mitochondria (145). A possible explanation for these disparate findings is that the role of CaMKII in MCU regulation at S92 or other sites is highly dependent on the particular tissue of study. The identification of AMP-activated protein kinase (AMPK) as another potential regulator of MCU S57 phosphorylation and uniporter activity (146) emphasizes that multiple signaling cascades likely converge to affect multiple regulatory residues on MCU, with the ultimate functional consequences for MCU activity depending on the particular combination of posttranslational modifications present on the protein at any given time.

MCU is also phosphorylated at as yet undefined tyrosine residue(s) by the Ca²⁺-sensitive protein tyrosine kinase, proline-rich tyrosine kinase 2 (PYK2), leading to an increase in _mCa²⁺ uptake capacity (147). Pyk2-dependent tyrosine phosphorylation and stimulation of MCU may underlie the increase in _mCa²⁺ uptake observed upon treatment of H9c2 cardiac myoblasts and neonatal rat cardiomyocytes with α-adrenergic agonists such as phenylephrine. Interestingly, Pyk2-dependent MCU phosphorylation is associated with increased oligomerization of MCU, suggesting that tyrosine phosphorylation of MCU may promote _mCa²⁺ uptake by facilitating the organization of individual MCU monomers into functional multisubunit channels (147).

Additional forms of posttranslational modification beyond phosphorylation have been detected for MCU as well. The Madesh laboratory identified a conserved residue, C97, as a key residue of human MCU that is subject to oxidative modification (148). Upon oxidative stress, C97 undergoes S-glutathionylation, which enhances _mCa²⁺ uptake via a mechanism that likely involves structural changes in the MCU NH₂-terminal domain that allow for increased oligomerization of MCU into high-molecular weight, functional uniporter complexes. Thus, Dong et al. (148) propose that cellular oxidative stress promotes _mCa²⁺ uptake through MCU, which may stimulate further production of ROS by the mitochondria and set up a positive feedback loop of increasing _mCa²⁺ uptake and ROS production that could predispose cells to death. Similar to the feedback look reported for transcriptional regulation of MCU, this mechanism may contribute to pathophysiology but is likely not a physiological regulator of the mtCU.

3.3.1.7. pharmacological modulation of mcu.

The mtCU can be inhibited by ruthenium red and its derivatives, such as Ru360. Early mutagenesis studies revealed that Ru360 blocks MCU activity by binding at or near the MCU DIME motif (99). Later studies suggest that Ru360 specifically binds the DIME motif at the carboxylate ring of D240, which is exposed to the intermembrane space at the entrance of the MCU selectivity filter (149). Despite the value of ruthenium red and Ru360 as experimental tools to study _mCa²⁺ uptake and its biological consequences in vitro, their potential therapeutic utility is limited by their poor cell permeability and off-target effects (150) (also reviewed in Ref. 151). Woods et al. (152) developed a Ru360 analog, termed Ru265, that exhibits improved cell permeability over Ru360 and that inhibits MCU in a manner that is not affected by MCU S259A mutation, indicating that its site of action is distinct from that of Ru360. Instead, Ru265 inhibits MCU by binding to the NH₂-terminal domain and favoring MCU aggregation in a configuration that is distinct from typical MCU oligomerization and inhibits MCU Ca²⁺ conductance. Recent small-molecule screens have also identified a number of additional MCU inhibitors that may be amenable to clinical use. The chemotherapeutic drug mitoxantrone inhibits MCU by interacting with acidic residues of the selectivity filter (153). The cell-permeant compound DS16570511 also inhibits _mCa²⁺ uptake in HEK293A cells and in isolated perfused rat hearts (154). The exact site(s) of DS16570511 action within MCU or the uniporter complex has yet to be identified, although its inhibition of _mCa²⁺ uptake may involve effects on MCU and/or MICU1 (154).

A number of chemical MCU activators have been described as well. MCU is reversibly activated by several plant flavonoids, the most potent of which is kaempferol, which triggers a nearly twofold increase in _mCa²⁺ uptake rate in HeLa cells at as little as 1 µM concentration (79). Proof of concept for the therapeutic application of MCU activation by kaempferol was demonstrated in 2017 as Schweitzer et al. (155) showed that treatment of a mouse model of catecholaminergic polymorphic ventricular tachycardia with kaempferol attenuates isoproterenol-induced arrhythmogenic cytosolic Ca²⁺ waves in cardiomyocytes in vitro, and attenuates caffeine + epinephrine-induced ventricular tachycardia in mice in vivo. These findings suggest that pharmacological manipulation of MCU activity may be a feasible approach not only for preventing _mCa²⁺ overload but also for correcting pathogenic defects in cytosolic Ca²⁺ handling.

3.3.2. EMRE.

The transmembrane protein EMRE (Essential MCU Regulator; gene name: Single-pass membrane protein with aspartate-rich tail 1, mitochondrial, SMDT1) was identified as an mtCU component in a proteomics screen in HEK293T cells performed by the Mootha laboratory (89). Knockdown of EMRE with RNA interference (RNAi) in HEK293T or HeLa cells phenocopies MCU knockdown and abrogates rapid _mCa²⁺ uptake, indicating that EMRE is required for MCU function in human cells. Furthermore, rapid _mCa²⁺ uptake is not rescued by overexpression of MCU in EMRE-deficient cells (89), indicating that such Ca²⁺ flux requires the simultaneous expression of both MCU and EMRE (FIGURE 2). Subsequent analysis revealed that EMRE is broadly expressed across mammalian tissues, with both EMRE RNA (89) and protein (156) detected in all 14 tissues examined by Mootha’s group. This ubiquitous expression of mammalian EMRE likely reflects its essential role in mtCU function in higher organisms.

3.3.2.1. emre phylogenetic conservation.

In contrast to MCU, which is found in nearly all eukaryotes, EMRE arose more recently in evolutionary history. EMRE is found only in metazoans and not in plants, protozoa, or fungi (89). EMRE is required for Ca²⁺-transporting activity of the mtCU in animal lineages from Drosophila to human (133). However, the absence of EMRE in lower taxa such as the Amoebozoa slime mold Dictyostelium discoideum (157) and the plant Arabidopsis thaliana (158) does not disrupt uniporter activity, suggesting that evolutionary changes in MCU or overall uniporter complex structure and function in animals necessitated the coordinated development of EMRE to maintain MCU in a functional state. Indeed, the expression of MCU from species that lack endogenous EMRE is sufficient to restore rapid, Ru360-sensitive _mCa²⁺ uptake in mammalian MCU/EMRE double-knockout cells, whereas expression of MCU from other metazoans such as C. elegans or Drosophila is not (158).

3.3.2.2. emre structure.

EMRE is a single-pass, transmembrane protein ∼10 kDa in size, with a predicted mitochondrial targeting sequence [amino acid (AA) 1–47], a transmembrane region (AA 65–84), and a conserved, aspartate-rich COOH terminus (AA 85–107) (89, 159). The membrane topology of EMRE remained a matter of debate for some time after its identification, with Vais et al. (115) initially asserting that the COOH terminus of EMRE faces the mitochondrial matrix based on patterns of proteinase digestion assays. However, this study tested only truncated NH₂- or COOH-terminal EMRE constructs or COOH-terminal FLAG-tagged EMRE, which may not have appropriately incorporated into the inner mitochondrial membrane or may have displayed alternative secondary structure. The current consensus, based on multiple antibody binding and proteinase protection assays comparing NH₂- or COOH-terminal FLAG-tagged EMRE, is that the NH₂ terminus of EMRE faces the mitochondrial matrix, whereas the aspartate-rich COOH terminus extends into the IMS (158, 159). This orientation is supported by the finding by Tsai et al. (158) that only cysteines engineered into the COOH terminus, and not the NH₂ terminus, of EMRE can be labeled by addition of polyethylene glycol maleimide to the IMS side of isolated mitoplasts. More recently, cryo-EM resolution of EMRE structure has confirmed that its NH₂ terminus is located in the matrix (160). Fitting with this model for EMRE orientation is the finding that the COOH terminus of EMRE binds to MICU1, which associates with the mtCU in the IMS. This interaction occurs via an interaction between a polybasic sequence in MICU1 (KKKR) that likely binds to the polyaspartate tail of EMRE (EDDDDDD) (158).

P60 and the region encompassing S85 to A90 are critical for EMRE function. Mutation of these residues of EMRE disrupts the MCU-EMRE interaction and abrogates mtCU Ca²⁺ uptake activity when EMRE is coexpressed with MCU in yeast (159). Deletion of EMRE residues 48–53 or 54–57 partially reduces MCU-dependent Ca²⁺ uptake, indicating that these residues may also alter the MCU-EMRE interaction (159). Tsai et al. further support the notion that the transmembrane helix of EMRE, rather than its NH₂ or COOH terminus, is critical for mtCU activity. They show that the GxxxS motif spanning residues G81–S85 within EMRE’s transmembrane helix is crucial for MCU-EMRE binding and mtCU-dependent Ca²⁺ uptake (158, 161). They propose that this region of EMRE’s transmembrane helix interacts with a region of MCU’s first transmembrane helix (158).

3.3.2.3. properties and function of emre.

The initial characterization of EMRE by Mootha’s laboratory indicated that MCU activity is lost in human HEK293T cells with shRNA knockdown or TALEN-mediated deletion of EMRE (89). This finding supports the conclusion that EMRE is required for mtCU channel activity. However, interpretation of the precise role or strict requirement for EMRE in mtCU function is complicated by the prior observation by De Stefani et al. (98) that His-tagged mouse MCU expressed in Escherichia coli or a cell-free transcription/translation system conducts calcium Ca²⁺ when reconstituted in vitro in lipid bilayers. Possible explanations for this discrepancy are structural changes in MCU induced by addition of the His tag or the absence of additional endogenous inhibitory binding partners or normal inhibitory posttranslational modifications on MCU that would not have been added to the construct during expression in bacteria or in vitro. Recent studies concur with the early assertion that EMRE is required for metazoan MCU to conduct Ca²⁺ (157–159, 161, 162). Furthermore, expression of mouse EMRE alone in yeast, in the absence of MCU, does not confer yeast mitochondria with uniporter activity, demonstrating that EMRE itself is not a Ca²⁺ transporter (159).

Examination of uniporter function across taxa indicates that EMRE arose evolutionarily in metazoan lineages in concert with structural changes to MCU that rendered MCU incapable of conducting Ca²⁺ on its own. When incorporated into yeast, human MCU requires coexpression of EMRE to conduct Ca²⁺, whereas MCU from the slime mold D. discoideum (DdMCU) is competent to transport Ca²⁺ in the absence of EMRE (157). Furthermore, DdMCU is sufficient to restore uniporter activity in human HEK293T MCU- or EMRE-knockout (KO) cells (157). Wu et al. report similar distinctions concerning the requirement for EMRE between taxa. When reconstituted in lipid bilayers, fungal MCU from Pyronema omphalodes (pMCU) conducts Ca²⁺ by itself, but human MCU requires coexpression of EMRE to transport Ca²⁺ (162). The structural features of MCU that may determine the differences in its dependence on EMRE in metazoans versus other groups are discussed below.

3.3.2.4. interaction of emre with the metazoan mtcu complex.

EMRE protein is stabilized by association with MCU and possibly other components of the mtCU complex. Loss of MCU in HEK293T cells causes a coordinate decrease in EMRE protein level despite normal levels of EMRE mRNA (89). Selective loss of EMRE does not affect the protein expression or stability of other uniporter components, although it does shift the molecular mass of immunoprecipitated, MCU-containing complexes from ∼480 kDa to ∼300 kDa, similar to the shift in molecular mass caused by loss of MICU1 (89). Indeed, the presence of EMRE is required for the interaction of MICU1/2 with the uniporter complex, although MICU1/2 and MCUB are not required for EMRE to interact with MCU (89). Thus, Sancak et al. propose a model in which EMRE binds to MICU1/2 in the intermembrane space and with MCU in the inner mitochondrial membrane. These authors speculate that these interactions allow EMRE to couple the Ca²⁺-sensing abilities of MICU1/2 to MCU’s ability to open and allow _mCa²⁺ influx in response to an increase in extramitochondrial Ca²⁺ concentration (89).

Kevin Foskett’s laboratory additionally proposed that EMRE functions as a sensor of matrix Ca²⁺ content and helps to inhibit MCU in response to rising matrix Ca²⁺ levels as a mechanism to prevent _mCa²⁺ overload (115, 116). In this model, the mtCU complex has Ca²⁺-sensing regulatory functions on both the matrix side (dependent on MCU’s NH₂ terminus) and the IMS side (conferred by MICU1/2). EMRE functionally couples these two Ca²⁺-sensing functions via its interactions with MCU and MICU1, resulting in maximal matrix-dependent inhibition of the mtCU at a matrix Ca²⁺ concentration of ∼400 nM. EMRE binds MICU1 via its polyaspartate COOH-terminal tail (158), and EMRE-KO cells rescued with EMRE COOH-terminal mutants display normal _mCa²⁺ uptake capacity but fail to shut off _mCa²⁺ uptake upon a drop in cytosolic or bath Ca²⁺ concentration (115). This behavior resembles the phenotype of MICU1-null cells (described below) and suggests that the COOH-terminal tail of EMRE is required for the functional interaction of MICU1/2 and EMRE, which transmits the sensing of IMS Ca²⁺ concentration by MICU1/2 into functional changes in the structure of MCU.

Fitting with a model in which the COOH terminus of EMRE mediates EMRE/MICU1 binding, Tsai et al. (158) report that COOH-terminal deletion of EMRE results in increased _mCa²⁺ uptake at low bath Ca²⁺ concentrations, likely because of disruption of MICU1-EMRE interaction and the loss of MICU1/2 from the uniporter complex. This study also identified the first transmembrane helix of MCU as the region that interacts with the transmembrane helix of EMRE (158). Thus, the authors proposed that EMRE is critical for the normal function of the mtCU via two main effects: first, EMRE activates MCU or is permissive for its Ca²⁺-transporting activity, and second, EMRE localizes MICU1 in proximity to MCU in order to allow Ca²⁺-sensitive regulation of MCU activity.

A major advance in understanding the mechanism by which EMRE controls MCU function was the recent report of a 3.6-Å-resolution cryo-EM structure of human MCU and EMRE complexed together (160). This study confirmed a tetrameric assembly of human MCU and also defined a 1:1 stoichiometry of MCU and EMRE subunits within the mtCU complex (FIGURE 2). Each EMRE subunit is located at the periphery of the MCU transmembrane pore and interacts with two neighboring MCU subunits at a total of three points of contact. First, the central portion of the EMRE transmembrane helix sits at a 45° angle against the first transmembrane helix of one MCU subunit, consistent with previous reports of these two regions interacting (158, 159, 161). Second, the NH₂-terminal portion of the EMRE transmembrane helix contacts the COOH terminus of the second transmembrane helix of the neighboring MCU subunit. Third, the NH₂ terminus of EMRE forms a β-hairpin that nestles into a groove between coiled-coil domain helices (CC2a and CC2b) of the neighboring MCU subunit. Wang et al. (160) report that a six-residue juxtamembrane loop (JML) that connects CC2a and the second transmembrane helix of each MCU subunit is highly ordered in human MCU and sits parallel to the lipid bilayer and blocks the matrix side of the MCU transmembrane domain. In fungal MCU, the structure of the JML is much less ordered. The JML of human MCU is linked to EMRE via hydrogen bonds between R297 in CC2a of MCU, T285 in the JML of MCU, and the carbonyl group of EMRE V61 (160). They propose that the MCU JML forms a luminal gate, specifically in metazoan MCUs, that may require EMRE for an open conformation and Ca²⁺ permeation. In a series of elegant experiments, Wang et al. demonstrate that substitution of the JML of human MCU with the JML from either D. discoideum or the fungus Neosartorya fischeri, but not the JML from Drosophila, allows the modified human MCU to conduct Ca²⁺ in the absence of EMRE. Thus, it appears that the JML of MCU in metazoans versus lower organisms dictates the uniporter’s functional dependence on EMRE. In the absence of EMRE, the coiled-coil domains of MCU lie closer to the central axis of the pore, potentially bringing the JMLs of neighboring MCU subunits into close proximity to block Ca²⁺ flow through the channel. In the Wang et al. model, the NH₂ terminus of EMRE pulls the MCU CC2a domain away from the central axis of the channel, along with coiled-coil domain helix1, CC2b, and the JML, to maintain the MCU luminal gate in an open conformation for Ca²⁺ permeation (160). In the absence of EMRE, the MCU NH₂-terminal domains undergo a conformational change that disrupts MCU dimerization, indicating that EMRE also promotes MCU oligomerization (160).

3.3.2.5. genetic diseases associated with emre.

As for MCU, no mutations in EMRE have yet been reported to cause human disease.

3.3.2.6. animal models of emre.

The first EMRE-knockout mouse model was generated by Toren Finkel’s and Elizabeth Murphy’s laboratories in 2016 (163) (TABLE 2). This was prompted by the observation that a decrease in EMRE expression tends to correlate with a less severe phenotype of MICU1-null mice that survive the perinatal period (discussed in greater detail below). Heterozygous deletion of EMRE rescues the perinatal lethality and excessive _mCa²⁺ uptake, and corrects defects in motor performance and skeletal muscle strength in Micu1^−/− mice. These data suggest that the reduction in EMRE levels reduces overall uniporter activity to compensate for the loss of MICU1’s gatekeeping function, which normally minimizes _mCa²⁺ uptake at low Ca²⁺ concentrations. Interestingly, mice with homozygous deletion of EMRE alone are viable on a mixed genetic background and have no obvious neurological or skeletal muscle defects under basal conditions, similar to previous findings for a gene-trap MCU-knockout model on a mixed genetic background (119, 163). Although mice with constitutive Emre deletion have impaired _mCa²⁺ uptake and resistance to Ca²⁺-induced permeability transition, they are not protected from cardiac I/R injury (164). This lack of obvious phenotypic consequence for constitutive EMRE deletion, despite a strict requirement of EMRE for mtCU function in animals, may reflect the initiation of pathways that compensate for the lack of _mCa²⁺ uptake throughout development and/or may reflect the biological robustness of the mixed genetic background.

In Drosophila, knockdown of EMRE via RNA interference (RNAi) impairs _mCa²⁺ uptake, as expected (133). Overexpression of EMRE, specifically in the Drosophila eye, has no obvious consequences, possibly because the addition of EMRE alone has little functional effect in the absence of concurrent MCU overexpression. Consistent with this idea, simultaneous overexpression of MCU and EMRE is lethal (133). Further studies of EMRE-null animal models under stressed conditions, and the development of mice with conditional rather than constitutive, global deletion of EMRE, will further clarify the in vivo requirements for EMRE in _mCa²⁺ handling and associated biological functions.

3.3.2.7. regulation of emre.

The predominant endogenous mode of regulation of EMRE is the control of its protein level via proteolytic degradation by mitochondrial proteases. MCU^−/− HEK293T cells have reduced levels of EMRE protein expression, despite normal levels of EMRE mRNA (89), a phenomenon confirmed with acute deletion of Mcu in MEFs (165). Moreover, acute deletion of NCLX in adult cardiomyocytes results in a rapid reduction in EMRE protein, suggesting that proteolysis of EMRE may be a mechanism to downregulate mtCU function in the context of _mCa²⁺ overload (166). A proteomic screen of the neuronal interactome of mitochondrial m-AAA proteases consisting of homooligomers of AFG3L2 or heterooligomers of AFG3L2 and SPG7 revealed that the m-AAA protease binds to the matrix protein C2ORF47, which was renamed m-AAA protease interacting protein 1 (MIAP1) (167). Knockdown of either MAIP1 or m-AAA protease subunits in cells reduces the accumulation of an 11-kDa form of EMRE, without affecting protein levels of other mtCU components or _mCa²⁺-handling proteins (167). The 11-kDa form corresponds to full-length EMRE, before cleavage of the NH₂-terminal mitochondrion-targeting sequence (MTS); postprocessed EMRE = 7 kDa. Protease protection assays indicated that preprocessed EMRE is localized to the intermembrane space, whereas the MTS-cleaved EMRE is fully incorporated into the IMM. Additional experiments suggest that MAIP1 binds to nascent EMRE in the intermembrane space and protects it from degradation by the i-AAA protease, YME1L (167).

Once incorporated into the IMM, EMRE participates in the assembly of MCU and MICU1/2 into the mtCU complex as described above. Loss of the m-AAA protease promotes mtCU assembly into high-molecular weight complexes by stabilizing EMRE expression (167). Later work corroborated the role of SPG7 or AFG3L2 in EMRE degradation (168). EMRE in low-molecular weight complexes, smaller than the mtCU, is especially susceptible to degradation by the m-AAA protease, leading Konig et al. (167) to conclude that the protease preferentially degrades EMRE that is not already incorporated into the mtCU, and so may limit the assembly of additional functional mtCU complexes. Indeed, the expression of mutant forms of EMRE or MCU that abolish the EMRE-MCU interaction accelerate the turnover of EMRE, supporting the notion that binding to MCU protects EMRE from proteolytic degradation (168). The above-mentioned data obtained with acute deletion of NCLX (166) suggest that disaggregase activity may precede EMRE proteolysis as a mechanism to downregulate functional, high-molecular weight mtCU activity. The relevance of the m-AAA protease for EMRE stability, mtCU formation, and mitochondrial Ca²⁺ homeostasis was validated in vivo in mice with neuronal-specific knockout of the AFG3L2 subunit of the protease (167). Loss of the protease results in an accumulation of MCU-EMRE complexes ∼400 kDa in size, which lack MICU1/3 regulatory subunits that are normally part of the mtCU in the brain. Consistent with a loss of mtCU gatekeeping function, these neurons exhibit increased _mCa²⁺ uptake and increased sensitivity to mPTP opening. m-AAA protease-deficient neurons also display increased levels of an ∼100-kDa complex likely containing MICU1, MICU3, and EMRE, which might represent an intermediate step in mtCU assembly. Konig et al. (167) propose that the MICU subunits titrate EMRE away from MCU during uniporter assembly and that the m-AAA protease degrades any noncomplexed EMRE. According to this model, the presence of the m-AAA protease protects the cell from _mCa²⁺ overload by preventing the formation of “ungated” EMRE-MCU uniporter channels (i.e., channels lacking MICU1/2/3) that might mediate unchecked _mCa²⁺ uptake. One challenge to this idea is a recent report that expression of a protease-resistant EMRE mutant (P76A) does not cause excess _mCa²⁺ uptake unless MICU1 is simultaneously depleted by ∼35% (168). This result suggests that cells normally have sufficient MICU1 to bind to and gate all EMRE-MCU complexes and that there is little accumulation of ungated EMRE-MCU complexes. The discrepancy between these two studies may arise from different extents of EMRE stabilization and accumulation in the two model systems (point mutant of EMRE vs. m-AAA protease depletion), or differences in the availability of endogenous MICU1 between the different cell types used for the experiments (neurons and HeLa cells vs. HEK293 cells). Despite these inconsistencies, the studies agree that accumulation of EMRE has the potential to increase the number of ungated EMRE-MCU complexes, which may result in excess _mCa²⁺ uptake at low cytosolic Ca²⁺ levels, and both highlight the importance of EMRE expression and stability in mtCU regulation.

3.3.3. MICU proteins: the gatekeepers of the mtCU.

A distinct property of the mtCU is that channel activation by cytosolic or IMS Ca²⁺ differs across distinct tissues and can change during development or with disease. This sensitivity of mtCU activation to changes in Ca²⁺ concentration is largely determined by the interactions of three regulatory mtCU components, the MICU proteins. Current models suggest that MICU1 binds directly to the core of the mtCU (MCU + EMRE) and has dual functions to regulate the channel: MICU1 functions as the gatekeeper that prevents Ca²⁺ flux through the channel when cytosolic/IMS Ca²⁺ concentration is low, but also facilitates the cooperative activation of the mtCU as cytosolic/IMS Ca²⁺ concentration rises. MICU2 and MICU3 do not bind to the core mtCU (MCU + EMRE) directly, but rather interact with the channel complex via dimerization with MICU1. The dimerization of MICU1 with MICU2 versus MICU3, or versus another molecule of MICU1, modulates the Ca²⁺ responsiveness of mtCU gatekeeping and cooperative activation. The molecular interactions between the MICU proteins that allow for this complex regulation of mtCU function are examined in the following sections.

3.3.3.1. micu1.

The peripheral membrane protein MICU1 [Mitochondrial Calcium Uptake Protein 1; gene name: MICU1 (or CBARA1, calcium-binding atopy-related autoantigen 1)] was the first mtCU component to be identified, using an integrative screen designed to identify genes involved in _mCa²⁺ uptake on the basis of comparative physiology, evolutionary genomics, and mitochondrial proteomics (100). In this study, Perocchi et al. (100) showed that knockdown of MICU1 in HeLa cells greatly reduces _mCa²⁺ uptake in response to histamine and that MICU1 is required for the coupling of increased cytosolic Ca²⁺ content to an increase in mitochondrial bioenergetics. The observation that silencing of MICU1 reduces rather than completely ablates rapid _mCa²⁺ uptake suggested that MICU1 is a regulator of the mitochondrial calcium uniporter complex rather than a necessary component of its Ca²⁺-conducting pore. This was known by the Mootha group at the time of publication, as their identification of MCU, the pore-forming component, was published only a few months later. MICU1 contains two canonical, Ca²⁺-binding EF-hand domains that are critical for _mCa²⁺ uptake, suggesting a role for MICU1 in Ca²⁺-dependent regulation of the uniporter (100). MICU1 is broadly expressed and is present in the mitochondria of at least 11 different mammalian tissues (156). Within the mitochondrion, MICU1 is located within the IMS, where it is associates with the inner mitochondrial membrane (IMM) (100, 158, 169, 170). Tagging of MICU1 with the peroxidase reporter APEX2 and examination of MICU1-APEX2 localization via electron microscopy indicated that MICU1 is present both within the intercristae IMM, as well as within the cristae themselves (170). This was later confirmed in another study examining MICU1 submitochondrial localization (171).

MICU1 has distinct effects on uniporter function depending on the concentration of cytosolic or IMS Ca²⁺ to which it is exposed. MICU1 limits or inhibits _mCa²⁺ uptake under basal conditions, despite the strong driving force for _mCa²⁺ entry created by the highly electronegative mitochondrial membrane potential (82). Knockdown of MICU1 in HeLa cells elevates resting _mCa²⁺ concentration, although in contrast to initial work on MICU1 (100), Mallilankaraman et al. (82) observe no effect of MICU1 knockdown on _mCa²⁺ uptake in response to an agonist-induced increase in cytosolic Ca²⁺. Permeabilized MICU1 knockdown cells readily take up Ca²⁺ even at low bath Ca²⁺ concentrations (0.5–1 µM Ca²⁺) that are insufficient to trigger rapid _mCa²⁺ uptake in control cells. This finding suggests that MICU1 acts as a gatekeeper to prevent _mCa²⁺ uptake through the uniporter when cytosolic and IMS Ca²⁺ concentration is low (82). This “gatekeeping” function is lost upon mutation of the EF-hands of MICU1, indicating that a Ca²⁺-sensing function of MICU1 is necessary for its inhibition of the channel under resting, low-cytosolic Ca²⁺ conditions (82), although the requirement of MICU1 EF-hands in this function has since been challenged, as EF-hand mutants are capable of restoring gating (172, 173). Similar effects of MICU1 knockdown on uniporter function are found when Ca²⁺ currents of HeLa mitoplasts are examined by patch clamp (174). Experiments in primary mouse hepatocytes and HeLa cells verified that MICU1 acts as a gatekeeper to set the threshold for Ca²⁺ activation of uniporter _mCa²⁺ uptake, with MICU1 knockdown increasing _mCa²⁺ uptake at extramitochondrial Ca²⁺ concentrations over the range of ∼0.3–2 µM (169). The slope of the linear fit of the double logarithmic plot of initial _mCa²⁺ uptake rate versus cytosolic Ca²⁺ concentration is reduced in MICU1-knockdown cells, leading Csordás et al. (169) to propose that MICU1 also contributes to cooperative activation of MCU-dependent _mCa²⁺ uptake (i.e., greater rates of _mCa²⁺ uptake when the mitochondria are exposed to higher cytosolic or IMS Ca²⁺ concentration). However, a caveat of this interpretation is that it is mostly based on the greater rate of uptake observed in MICU1-knockdown cells at low cytosolic Ca²⁺, as the uptake rates for WT and MICU1-knockdown cells at high cytosolic Ca²⁺ in this study appear similar in magnitude (169). Current models propose that MICU1 modulates uptake in a biphasic fashion by both inhibiting the mtCU at low cytosolic/IMS Ca²⁺ concentration (i.e., the gatekeeping function that “sets the Ca²⁺ threshold” for mtCU activation) as well as facilitating cooperative activation of the channel as Ca²⁺ concentration rises (175) (FIGURE 4).

FIGURE 4.

Gating of the mitochondrial calcium uniporter channel and consequences of changes in MICU composition. Clockwise from top left: Classical mtCU: MICU1/2 dimers bind to the mtCU and regulate its activation by Ca²⁺ within the intermembrane space (IMS). The stoichiometry suggested by structural studies of the mtCU is 4 MCU:3 or 4 EMRE:1 MICU1:1 MICU2, although the proportion of the total mtCUs that are regulated by MICUs is determined by the relative MICU expression within each tissue. At low cytosolic/IMS Ca²⁺ concentration ([Ca²⁺]), the MICU1/2 dimer acts as a gatekeeper to keep the channel impermeable to Ca²⁺. As cytosolic/IMS [Ca²⁺] rises, the MICU1/2 dimer facilitates the cooperative activation of the channel, resulting in a sigmoidal relationship between mtCU-dependent _mCa²⁺ uptake and cytosolic/IMS [Ca²⁺]. Excess MICU1: Increased expression of MICU1 may displace MICU2 from existing MICU1/2 dimers, resulting in MICU1 homodimers. Displacement of MICU2 results in loss of its ability to tune MICU1-dependent mtCU gating and cooperative activation, shifting the relationship between mtCU-dependent _mCa²⁺ uptake and cytosolic/IMS Ca²⁺ concentration to the left. Excess MICU3: Increased expression of MICU3 may displace MICU2 from existing MICU1/2 dimers, resulting in MICU1/3 dimers. The properties of MICU3-dependent regulation of MICU1 differ somewhat from those of MICU2 (see text for details); thus, replacement of MICU2 with MICU3 also shifts the relationship between mtCU-dependent _mCa²⁺ uptake and cytosolic/IMS Ca²⁺ concentration to the left. Excess MICU1/2: Increased expression of MICU1/2 results in more stringent regulation of Ca²⁺ flux through the mtCU, with enhanced channel gatekeeping at low cytosolic/IMS [Ca²⁺], but greater activation of the channel at higher [Ca²⁺]. This behavior may result from a greater number of MICU1/2 dimers binding to each mtCU, or from an increase in the number of mtCUs within the tissue being gated by MICU1/2. Loss of MICU1: MICU2 and MICU3 are incapable of binding to and regulating the mtCU in the absence of MICU1. Loss of MICU1 therefore results in loss of all MICU-dependent mtCU regulation. The absence of MICU-dependent channel gatekeeping results in increased _mCa²⁺ uptake at low cytosolic/IMS [Ca²⁺], while the absence of MICU-dependent cooperative activation makes further mtCU activation less responsive to increases in cytosolic/IMS [Ca²⁺]. See glossary for abbreviations.

These properties allow MICU1, when forming heterodimers with MICU2, to act as Ca²⁺-sensitive “on-off switch” for the mitochondrial calcium uniporter channel (173, 176). As noted above, the ability for MICU1 to inhibit the uniporter within the range of typical resting cytosolic Ca²⁺ levels (∼100–150 nM) may also contribute to inactivation of the RaM current over this range of Ca²⁺ concentrations (82). The consequences of the multifaceted functions of MICU1—both preventing and promoting _mCa²⁺ uptake—are discussed in further detail below.

3.3.3.1.1. MICU1 phylogenetic conservation.

Most species of Eukarya that express MCU also express MICU1, and the few species that lack MCU typically also lack MICU1 (88). A notable exception is that fungi that contain MCU generally lack MICU1 and its homologs, indicating that these lineages may rely on alternative, MICU1-independent mechanisms for regulation of the uniporter channel (88). In fungal species that do express MICU1 homologs, it appears that these homologs are not involved in mtCU regulation. For example, the fungus Aspergillis fumigatus contains a putative MCU homolog named McuA, which is required for acute _mCa²⁺ uptake (177). A. fumigatus also contains a putative MICU1 homolog named AgcA, which contains two EF-hands and thus is predicted to bind Ca²⁺. Although AgcA deletion improves resistance to azole and oxidative stress, it does not modulate acute _mCa²⁺ uptake, suggesting that the predominant function of AgcA is independent of the mtCU (177). An additional layer of complexity is seen in trypanosomatids, where the mtCU-gatekeeping and mtCU-cooperative activation functions of MICU1 are separable. MICU1 does not prevent _mCa²⁺ uptake at low, basal Ca²⁺ concentrations in the parasite T. cruzi as it does in mammalian cells, suggesting that MICU1 does not function in channel gatekeeping in this species (94). Rather, disruption of MICU1 in T. cruzi ablates _mCa²⁺ uptake at both low (1.5 µM) and high (up to 50 µM) bath or cytosolic Ca²⁺ concentration (94). This result is consistent with the purported role of MICU1 in facilitating the cooperative activation of the mtCU.

In vertebrate species, the MICU1 and MCU genes are located next to each other on the same chromosome and share a potential bidirectional promoter (88). Bick et al. (88) propose that this arrangement allows for highly coordinated expression of MICU1 and MCU proteins within these taxa. However, this genetic organization does not preclude the possibility that MICU1 protein could have both mtCU-dependent and mtCU-independent functions within these species. Despite the apparent diversity of mtCU-dependent and/or mtCU-independent MICU1 functions across evolution, the Ca²⁺-coordinating residues within its two EF-hands remain highly conserved from protozoa to humans (100). This degree of conservation indicates that a fundamental function of MICU1 across species is to serve as a mitochondrial Ca²⁺ sensor.

3.3.3.1.2. MICU1 structure.

The human MICU1 gene encodes a 476-amino acid protein of ∼54 kDa comprised mostly of α-helices (178). The NH₂ terminus consists of a presequence (residues 1–33) followed by a predicted transmembrane domain (residues 34–52; numbers throughout this paragraph correspond to residues within 476-amino acid human MICU1 isoform 2) and a mitochondrial targeting sequence with a cleavage site at residue 115 (100, 179). A second transmembrane domain is predicted around residues 194–216 (179). The predominant feature of MICU1 is a pair of two highly conserved, classical Ca²⁺-binding EF-hands (“EF1” and “EF2”) in the COOH-terminal half of the protein. The EF-hands are located at residues 218–253 and 408–443 and are separated by a long α-helical spacer (100, 178, 179). A conserved cysteine residue (C463) allows MICU1 to form disulfide bridges with other MICU proteins, resulting in MICU1 homodimers or MICU1/2 or MICU1/3 heterodimers (96, 180).

3.3.3.1.3. Alternative splice/transcript variants of MICU1.

Several transcript variants for human MICU1 are reported in the NCBI Gene database, including the canonical transcript encoding the classical 476-AA MICU1 protein (MICU1 isoform 2). Another encodes a 482-AA protein (MICU1 isoform 4) due to inclusion of an additional alternative exon, which introduces four amino acids (DFWQ) after residue 179.

Six protein-coding splice variants for mouse Micu1 are annotated on ENSEMBL. The expression and function of these various MICU1 protein isoforms remain largely unknown except for the mouse isoform that corresponds to human MICU1 isoform 4. This variant (named “MICU1.1”) is characterized by the insertion of four residues (EFWQ) after residue 181 of mouse MICU1, a location that is relatively far from the two EF-hand domains (181). MICU1.1 heterodimerizes with MICU2, and upon Ca²⁺ binding MICU1.1/2 heterodimers facilitate more net _mCa²⁺ uptake than MICU1/2 heterodimers. Such MICU1.1/2 dimers still act as a gatekeeper for the mtCU at low cytosolic Ca²⁺ concentrations but shift the threshold for channel activation to a lower Ca²⁺ concentration compared with MICU1/2 dimers (181). Despite its distance from the EF-hand domains, the 4-AA insertion increases the Ca²⁺-binding affinity of the MICU1.1 isoform (181). This change likely accounts for the shift in the Ca²⁺ threshold for activation of uniporter channels containing MICU1.1. Micu1.1 mRNA expression is mostly restricted to skeletal muscle, where the Micu1.1-to-Micu1 ratio is 8:1 (181). Micu1.1 mRNA is also detected in the brain, at a 1-to-1 ratio with Micu1 (181). Vecellio Reane et al. (181) propose that MICU1.1 allows skeletal muscle to be highly sensitive to changes in cytosolic Ca²⁺, such as occur during muscle stimulation, and thus allows sufficient _mCa²⁺ uptake through the mtCU to increase ATP production for use in muscle contraction.

3.3.3.1.4. Properties and function of MICU1.

Early work in HEK293 cells confirmed the initial findings (82, 169) that MICU1 inhibits _mCa²⁺ uptake at low cytosolic or IMS Ca²⁺ concentrations < 2 µM, but promotes _mCa²⁺ uptake as extramitochondrial Ca²⁺ concentration rises past ∼4 µM (114, 158). MICU1 modulation of _mCa²⁺ uptake is sigmoidal, with the rate plateauing at higher extramitochondrial Ca²⁺ concentrations ∼30 µM, at which point _mCa²⁺ uptake in MICU1-knockdown cells is similar to that in wild-type cells (158). Initial structural studies indicated that when extramitochondrial Ca²⁺ concentration is low MICU1 forms hexamers that may inhibit MCU. As the local Ca²⁺ concentration rises, MICU1 binds Ca²⁺ with high affinity (∼15–20 µM) and undergoes conformational changes, particularly at the EF-hands, and forms multiple lower-order oligomers to allow Ca²⁺ permeation of MCU (182). Such MICU1 rearrangement is specifically controlled by changes in cytosolic or IMS Ca²⁺ concentration, rather than changes in _mCa²⁺ content, and occurs regardless of the presence of EMRE and MCU (183). Although the hexameric structure of MICU1 was determined in the absence of MCU (182), MICU1 expressed in eukaryotic cells can be detected on blue native-PAGE gels at molecular weights corresponding to MICU1/2 hexamers, suggesting that higher-order MICU oligomers are capable of forming in intact systems (181). However, subsequent studies of the mtCU holocomplex, discussed below, indicate that it is MICU dimers, rather than hexamers, that are functionally relevant for regulating individual uniporter channels (184–186).

The relative ratio MICU1:MCU within a cell confers an additional layer of regulation on mtCU function because it controls how much of the uniporter channel population is subject to gatekeeping and regulation, and how responsive the tissue’s mitochondria will be to changes in cytosolic Ca²⁺ (FIGURE 4). For example, liver exhibits a high MICU1-to-MCU ratio, and therefore is effectively gated, but also has enhanced cooperative activation of the channel (187). This means that liver has low Ca²⁺ uptake at low cytosolic Ca²⁺ levels, because of a high threshold for Ca²⁺ activation set by the presence of MICU dimers on the mtCU, but also has profound _mCa²⁺ uptake at high cytosolic Ca²⁺ levels. Tissues such as the heart and skeletal muscle that have a low ratio of MICU1:MCU have a greater portion of “ungated” mtCUs and exhibit a lower Ca²⁺ threshold (i.e., greater _mCa²⁺ uptake at low Ca²⁺ levels), but relatively lower _mCa²⁺ uptake at high cytosolic Ca²⁺ concentrations due to fewer channels exhibiting cooperative activation (187) (FIGURE 4). Paillard et al. (187) hypothesize that regulation of the MICU1-to-MCU ratio allows tissues to tune how responsive the uniporter is to cytosolic Ca²⁺ concentration, as a mechanism to modulate the Ca²⁺ sensitivity of oxidative metabolism. In support of this view, Payne et al. (188) propose an average stoichiometry of two MICU1/2 dimers associated with two or three EMRE subunits per MCU tetramer in HEK293T cells and demonstrate that an increase in the relative MICU1-to-MCU ratio enhances gatekeeping at low cytosolic Ca²⁺ concentrations (i.e., increases the threshold cytosolic Ca²⁺ concentration required to relieve channel inhibition). The existence of tissue-specific regulation of the Ca²⁺ sensitivity of mitochondrial metabolism is not surprising. Different tissues are exposed to very different frequencies and amplitudes of cytosolic Ca²⁺ transients or spikes (31). For instance, the heart is constantly exposed to cyclic changes in cytosolic Ca²⁺ concentration and thus has more background “noise” to filter out before modifying steady-state _mCa²⁺ content, compared with more quiescent tissues that may only undergo Ca²⁺ transients in response to discrete, infrequent physiological signals.

MICU1 also appears to facilitate _mCa²⁺ uptake by overriding the eventual inhibition of mtCU activity that otherwise occurs as matrix Ca²⁺ content rises upon rapid _mCa²⁺ uptake (114). MCU current is partially inhibited at a matrix Ca²⁺ concentration of ∼400 nM, with greater MCU current observed at higher or lower _mCa²⁺ concentrations (115). The Foskett group proposes that an inhibitory matrix Ca²⁺ sensor within the MCU NH₂-terminal domain (including residues D131 and D147) allows the channel to close as matrix Ca²⁺ concentration rises up to this level, even despite persistent Ca²⁺ binding by MICU1 and MICU2 in the IMS (116). Such matrix Ca²⁺-dependent inhibition of the mtCU is lost upon mutation of EMRE’s COOH terminus (115), the portion of EMRE that normally helps anchor MICU1 to the mtCU (158), or in the absence of MICU1 (115). These results suggest that the inhibitory mechanism is responsive to Ca²⁺ concentration within the IMS, in addition to the matrix, or at least somehow is sensed by MICU1 (115). Vais et al. (116) therefore reason that the mechanisms of Ca²⁺ sensing and changes in protein function that control Ca²⁺ flux through the mtCU depend on both MICU1/2 and MCU, and are coupled to each other across the IMM. With continued Ca²⁺ influx, mtCU inhibition is relieved as _mCa²⁺ concentration rises past ∼400 nM (115). This behavior may help account for the phenomenon of cooperative activation of the uniporter and may enable appropriate loading of the mitochondria with Ca²⁺ in response to prolonged, moderate increases in cytosolic Ca²⁺ concentration. The finding that MICU1 and MICU2 mediate the gradual activation of rapid _mCa²⁺ uptake that occurs upon small but sustained increases in cytosolic Ca²⁺ concentration (189) supports the possibility that MICU1/2 are directly involved in the relief of matrix Ca²⁺-dependent MCU inhibition. This behavior, specifically the relief of partial MCU inhibition as matrix Ca²⁺ rises past 400 nM, also enables continued _mCa²⁺ uptake to the point of depolarization, as observed in Ca²⁺ retention capacity experiments or mitochondrial swelling assays.

Finally, MICU1 lets the mtCU distinguish between Ca²⁺ and manganese (Mn²⁺) and allows the channel to favor the transport of Ca²⁺ over other ions (190). Kamer et al. (190) attribute this discriminatory function to the fact that Ca²⁺ elicits a conformational change in MICU1 but Mn²⁺ does not. These findings agree with the current structural models (discussed below) in which MICU1 physically blocks Ca²⁺ permeation through the mtCU pore under basal conditions and in which this inhibition is relieved only upon a Ca²⁺-induced conformational shift in MICU1 that alters its interaction with the channel pore. Loss of MICU1 therefore makes human cells more susceptible to Mn²⁺ toxicity (190), because it no longer blocks the channel pore to prevent Mn²⁺ permeation. Conversely, expression of MICU1 is sufficient to protect against mtCU-dependent Mn²⁺ toxicity (191). This protection likely results from MICU1’s physical blocking of the channel pore in low-Ca²⁺ conditions rather than any direct action of MICU1 to bind Ca²⁺ or Mn²⁺, because it is preserved even with mutation of MICU1’s EF-hands (191).

3.3.3.1.5. Interaction of MICU1 with the metazoan mtCU complex.

MICU1 association with the mtCU requires an interaction with either MCU or EMRE: MICU1 binds directly to MCU independent of EMRE but also binds directly to EMRE independent of MCU (158) (FIGURE 2). MICU1 binds to MCU via an NH₂-terminal polybasic domain (residues 99–110 of MICU1) that contains a series of positively charged lysine residues (174). These residues interact with the two coiled-coil domains of MCU and anchor MICU1 to the mtCU, independent of changes in local Ca²⁺ concentration (174). The same polybasic sequence of MICU1 (KKKKR) mediates its electrostatic interaction with the COOH-terminal polyaspartate tail of EMRE. This MICU1-EMRE interaction is required to maintain MICU1 near the uniporter pore, where it can function in channel gatekeeping (158). Two positively charged arginine residues of MICU1 (R440 and R442, comprising a “DIME interacting domain”) also bind directly to negatively charged residues within the MCU DIME motif (D261 and E264), allowing MICU1 to block the pore of the uniporter by competing for the same site on MCU where the pharmacological inhibitor Ru360 binds (192).

MICU1 dimerizes with MICU2 (180) or MICU3 (96) and can also form homodimers or higher-order hexamers (trimers of dimers) (174, 176, 182). Binding of MICU2 or MICU3 to the uniporter is dependent upon association with MICU1, i.e., neither protein directly tethers to MCU or EMRE independent of MICU1 (96, 172, 180). Structural studies using cryo-EM reveal how the MICU1/2 heterodimer gates the mtCU. Under conditions of no cytosolic/IMS Ca²⁺, the uniporter complex exists predominantly in a “uniplex” form with individual uniporter channels with a stoichiometry of 4 MCU:4 EMRE:1 MICU1:1 MICU2 (184) (FIGURE 2). In this conformation, the polybasic sequence of a single MICU1 molecule interacts directly with a ring of D261 residues formed by the four MCU subunits to seal the entrance of the channel pore. A single molecule of MICU2 binds to MICU1, without directly interacting with the MCU subunits (184).

Wang et al. (185) propose an alternative low-Ca²⁺ (<1 nM) stoichiometry of 4 MCU:3 EMRE:1 MICU1:1 MICU2, where the MICU1/2 heterodimer displaces one EMRE subunit from the complex. However, these authors agree that, at low Ca²⁺, MICU1 blocks the IMS entrance of the uniporter pore via its interaction with the D ring created by the MCU subunits (185). At high Ca²⁺ concentrations (∼2 mM Ca²⁺), a greater proportion of the uniporter channels exist as V-shaped dimers of the individual uniplex channels formed by MCU + EMRE, with two MICU1-MICU2 heterodimers bridging the two uniplexes, although it should be noted that dimerization is not strictly required for mtCU channel function (184, 186). A portion of the dimers lose MICU1/2 (184), but the functional consequence of MICU1/2 loss specifically from the dimerized uniplex has not yet been tested. Channel dimerization is mediated by MCU’s NH₂-terminal domain (160, 184). Rearrangement of the uniporter under high-Ca²⁺ conditions does not appreciably alter the structure of the channel pore, indicating that the mechanism for channel gating occurs elsewhere within the complex (184). In the high-Ca²⁺ channel dimer conformation, the MICU1/2 dimers have bound Ca²⁺ and the MICU1 subunits have rotated such that the uniporter interaction domains no longer cover the pores of their respective channels, potentially allowing for Ca²⁺ permeation (184–186). MICU1 remains connected to the channel in this high-Ca²⁺ state via its contacts with the COOH terminus of EMRE (184).

It is not immediately clear whether the disruption of the MICU-pore interaction occurs independent of channel dimerization, due to conformational changes within MICU1/2 themselves, or rather depends on dimer formation and the association between the bridge of MICU1/2 heterodimers, which may pull MICU1 away from the pore. Wang et al. (186) suggest that the two MICU1/2 heterodimers that bridge the dimerized uniplexes become more compact in structure upon binding Ca²⁺. This conformational shift in the MICU1/2 heterodimer likely involves a transition from interactions between E242 in MICU1 and R352 in MICU2 in the Ca²⁺-free state to an interaction dependent on MICU1 F383 and MICU2 E196, I333, and M337 in the Ca²⁺-bound state (193). The MICU1/2 heterodimer then changes position relative to the channel pore, rotating out of a state in which each MICU1/2 dimer faces the center of its corresponding uniplex channel and blocks the pore, and into a state where the MICU1/2 dimer is pointed toward the periphery of the uniplex channel and can contact the MICU1/2 dimer from the adjacent uniplex channel (186). Consistent with this view, the interaction between MICU1 and MCU is favored in the absence of Ca²⁺, whereas Ca²⁺ binding to the MICU1/2 heterodimer favors the interaction between MICU1 and EMRE (194). A shift in MICU1/2 positioning from the center to the periphery of each uniplex channel upon Ca²⁺ binding agrees with the finding that Ca²⁺ induces a conformational change in the MICU1 bound to a single mtCU from a more condensed arrangement to a more spatially distributed arrangement (183).

Fan et al. (184) report that the bridge structure composed of two MICU1/2 heterodimers is formed via back-to-back contacts between the two MICU2 subunits and that this bridge structure may allow for cooperative activation of the two dimerized channels. According to this model, Ca²⁺ would be required to bind to both of the MCU1/2 heterodimers in order to elicit reciprocal conformational changes in both of the dimerized uniplex channels and fully open both pores (184). It is also possible that the bridge structure formed by the two MICU1/2 heterodimers facilitates _mCa²⁺ uptake simply by stabilizing the open-pore conformation of the mtCU.

These elegant structural studies present an attractive model for how MICU1/2 control mtCU activity. MICU1 gates the channel by acting like a rotating lid that either covers or uncovers the pore depending on Ca²⁺ binding to MICU1/2, thus prohibiting or permitting Ca²⁺ permeation, respectively. In turn, contacts between MICU2 subunits of adjacent, dimerized uniplex channels help to hold each channel’s “lid” in the open conformation, away from the channel pore.

3.3.3.1.6. mtCU-independent functions of MICU1.

Genetic experiments in Drosophila indicate that MICU1 has important mtCU-independent functions distinct from its classical role in regulating _mCa²⁺ uptake. Deletion of MICU1 is lethal to flies and is associated with deleterious _mCa²⁺ overload, whereas deletion of MCU or EMRE abolishes _mCa²⁺ uptake but has little impact on viability (195). Surprisingly, simultaneous loss of MICU1 with either MCU or EMRE in order to ablate _mCa²⁺ uptake through the mtCU is insufficient to rescue lethality (195). These results suggest that loss of MICU1 has additional harmful consequences besides _mCa²⁺ overload. The putative mtCU-independent functions of MICU1 are the subject of active investigation. CBARA1/MICU1 was detected in a genome-wide screen for genetic adaptation to high altitude (196) and may be involved in hypoxia-inducible factor-1 (HIF-1)-dependent regulation of mitochondrial metabolism in low-oxygen environments (197). Whether such a role for MICU1 is related to its regulation of the mtCU or to some other cellular function remains to be determined. A recent study using structured illumination microscopy detected MICU1 at the inner boundary membrane of the IMM, where it localizes to mitochondrial cristae junctions (171, 198). In contrast, MCU and EMRE are distributed all throughout the IMM under basal conditions and become associated with MICU1 at the inner boundary membrane only upon an increase in IMS Ca²⁺ concentration (171). This intriguing result suggests a novel mechanism by which changes in cytosolic or IMS Ca²⁺ concentration may regulate mtCU complex assembly and _mCa²⁺ uptake.

3.3.3.1.7. Genetic diseases associated with MICU1.

MICU1 is unique among the other mtCU components in that there is substantial evidence that genetic mutation of MICU1 is linked to human disease (TABLE 1). MICU1 mutations associated with excessive _mCa²⁺ uptake are found in patients exhibiting proximal myopathy, learning deficits, and a progressive extrapyramidal movement disorder (involuntary/uncontrolled movement or tremors) (199). Homozygous deletions in exon 1 of MICU1 resulting in loss of protein impair _mCa²⁺ uptake and are associated with fatigue, lethargy, and weakness, although patients have normal muscle biopsies (200). Disruption of either the gatekeeping or cooperative activation functions of MICU1 thus appears sufficient to cause human neuromuscular disease. Additional reports of MICU1 mutations in patients with congenital muscular dystrophy (201) and syndromes characterized by developmental delays and neuromuscular impairments (202, 203) support this conclusion. The mechanisms by which various mutations in MICU1 lead to neuromuscular disease probably vary depending on whether the particular mutation favors or instead diminishes _mCa²⁺ uptake. For instance, loss of MICU1’s gatekeeping function may lead to futile Ca²⁺ cycling that ultimately runs down the mitochondrial membrane potential (ΔΨ_m) via secondary effects through the mitochondrial sodium/calcium exchanger (NCLX) and the mitochondrial sodium/hydrogen exchanger, and/or via mitochondrial Ca²⁺ overload inducing mitochondrial dysfunction and permeability transition. Reduction of ΔΨ_m would limit mitochondrial ATP production and thus place an energetic stress on the cell (204). On the other hand, mutations that impair _mCa²⁺ uptake could impair energy production by limiting Ca²⁺-dependent stimulation of the tricarboxylic acid (TCA) cycle and respiratory complexes. Excess _mCa²⁺ uptake upon loss of MICU1 gatekeeping also sensitizes tissues to cell death (82) and could explain neural as well as skeletal muscle defects in MICU1-mutant patients. Finally, since MICU1 regulates regrowth following axon injury in adult neurons (205), defective neuronal homeostasis may also contribute to neuromuscular defects in these individuals.

3.3.3.1.8. Animal models of MICU1.

Several different genetic knockout strategies have been employed to examine the function of MICU1 in vivo (TABLE 2). The Hajnóczky laboratory crossed Micu1^loxP/loxP mice to germline-expressing E2a-Cre animals to generate a constitutive Micu1-knockout model (206). Heterozygous Micu1^+/− mice are viable and fertile, but homozygous Micu1^−/− mice die within several hours of birth (206). Loss of MICU1 is hypothesized to cause perinatal lethality because it interferes with brain stem control of respiratory function (206). Micu1^−/− MEFs exhibit excessive _mCa²⁺ uptake at low cytoplasmic Ca²⁺ concentrations and diminished _mCa²⁺ uptake at high cytosolic Ca²⁺ concentrations, as expected for loss of gatekeeping and cooperative activation of the mtCU (206). Acute Micu1 deletion alone has little effect on the adult mouse liver but renders it more susceptible to necrosis and microvesicular steatosis following partial hepatectomy and impairs hepatocyte proliferation and liver regeneration after injury (206). The regenerative capacity of Micu1^−/− livers is rescued by the mPTP inhibitor NIM811, suggesting that the tissue damage and impaired proliferation are caused by _mCa²⁺ overload-induced mitochondrial permeability transition and cell death (206).

Constitutive deletion of Micu1 via CRISPR/Cas9 also yields high rates of perinatal mortality in Micu1^−/− pups (163). Micu1^−/− pups that survive the perinatal period are ∼50% smaller than WT littermates at 1 wk of age (163). By 1 mo of age, surviving Micu1^−/− mice show signs of ataxia and have impaired performance on a balance beam and impaired skeletal muscle strength and coordination, much like human patients with MICU1 mutations (163). These neuromuscular defects are associated with abnormal brain development and reduced skeletal muscle fiber size (163). Micu1^−/− animals also have reduced numbers of splenic B cells, indicating that MICU1 exerts important functions in the immune system (163). Liu et al. (163) observe that, despite these defects in early life, the overall appearance, body weight, and elevated resting _mCa²⁺ levels in the brains of surviving Micu1^−/− mice improve as the animals age. These improvements are attributed to a gradual downregulation of EMRE expression that normalizes _mCa²⁺ uptake at low cytosolic Ca²⁺ concentrations and thus minimizes _mCa²⁺ overload (163). These findings highlight the plasticity of mtCU composition and the existence of counterregulatory mechanisms that can normalize _mCa²⁺ handling despite the disruption of a given mtCU component.

Constitutive deletion of Micu1 specifically in skeletal muscle (Creatine kinase-Cre × Micu1^fl/fl) reduces the Ca²⁺ threshold for _mCa²⁺ uptake and results in proximal muscle weakness and atrophy (207). Skeletal muscle knockout (_skmKO) of Micu1 does not alter cytosolic Ca²⁺ transients but reduces _mCa²⁺ accumulation during single-fiber twitches and during tetanic contraction (207). Resting _mCa²⁺ content is elevated in Micu1-_skmKO fibers (207), and this may contribute to the smaller overall amplitude of _mCa²⁺ transients during muscle contraction, as opposed to a direct effect of Micu1 deletion to reduce net _mCa²⁺ uptake. Interestingly, though, Micu1-_skmKO muscles have impaired sarcolemmal repair after acute muscle injury (207), and efficient muscle fiber repair depends on acute _mCa²⁺ uptake (208). These findings suggest that loss of MICU1 impairs skeletal muscle function by two mechanisms: 1) reducing _mCa²⁺ uptake during muscle contraction, leading to impaired force generation, and 2) diminishing the efficiency of sarcolemmal repair after injury to a muscle fiber, thus increasing the likelihood of muscle cell necrosis and progressive loss of muscle mass and strength (207).

These studies in mice with genetic Micu1 disruption demonstrate that the regulation of acute _mCa²⁺ uptake—not simply whether or not acute _mCa²⁺ is able to occur—is sufficient to affect a number of cellular processes that rely on Ca²⁺ handling and that impact physiological function at the level of the whole organism. This conclusion is further supported by in vivo research in Drosophila, where global silencing of the MICU1 homolog CG4495 via RNAi is lethal, and where silencing of CG4495 specifically in mushroom body neurons causes memory deficits that recapitulate the effects of silencing of the MCU homolog CG18769 (134). Likewise, inducible knockdown of GC4495 in Drosophila neurons reduces survival and is associated with early loss of locomotor ability (179). These effects are rescued by overexpression of the antiapoptotic Drosophila Bcl-2 homolog, Buffy, implicating aberrant neuronal death in the phenotype of flies with neuronal GC4495 knockdown (179). The recent observation that acute knockdown of Micu1 in the adult mouse heart aggravates _mCa²⁺ overload, infarct size, cardiac dysfunction, and apoptosis following myocardial ischemia-reperfusion injury (209) provides further evidence that appropriate modulation of uniporter activity by MICU1 is critical for cell survival during stress and has substantial impacts on whole organism physiology and disease.

This interesting result indicates that MICU1 activity is relevant for the regulation of _mCa²⁺ handling within the heart, at least under stressed conditions where cytosolic Ca²⁺ concentration may be elevated beyond normal ranges. That is, MICU1 appears to oppose net _mCa²⁺ accumulation at high cytosolic Ca²⁺ concentrations within the cardiomyocyte. How it may do so is puzzling. Whether MICU1 is still functioning as a mtCU gatekeeper in this context or is exerting additional functions at sites distinct from the mtCU remains to be determined. Recent work by the Lederer group suggests that cardiomyocytes exhibit no Ca²⁺ threshold for mtCU gatekeeping, at least over the range of ∼400 nM to 12 µM extramitochondrial Ca²⁺ (15, 210). It is possible that these experiments simply missed the range of high extramitochondrial Ca²⁺ concentrations where MICU1-dependent mtCU regulation becomes appreciable within the heart. Cytosolic Ca²⁺ can reach ∼2 µM during the cardiomyocyte Ca²⁺ transient (21), and recall that this concentration may be 5–20 times higher in microdomains between the ER/SR and mitochondria. Ca²⁺ concentration in this microdomain may be elevated even more in pathological settings. If the net ratio of MICU1:MCU in the heart is indeed very low compared with other tissues (187) and the majority of mtCUs in the cardiomyocyte are ungated, this could result in the appearance of minimal gatekeeping activity and minimal extramitochondrial/IMS Ca²⁺-dependent mtCU activation over the range of 400 nM to 12 µM Ca²⁺. This is particularly true for experiments considering the overall behavior of a population of mtCUs (188) (FIGURE 4), whether measured within a batch of isolated cardiomyocytes or even within a patch-clamped cardiomyocyte mitoplast.

Further complicating the question of whether MICU1 has an appreciable effect on mtCU function in the heart is the possibility that the susceptibility of MICU1-knockdown hearts to I/R injury reflects the effects of altered mtCU activity and/or the loss of critical mtCU-independent MICU1 functions. To this point, it is interesting to consider how the baseline phenotypes of animal models of MICU1 knockout or knockdown differ from baseline phenotypes of animal models of MCU knockout. In contrast to the effects of MICU1 disruption discussed above, global loss of MCU is well tolerated under basal, unstressed conditions in flies (195) and in mice on a mixed genetic background (119). At least two factors may account for these differences. First, MICU1 disruption generally results in cellular phenotypes associated with _mCa²⁺ overload (tissue damage, permeability transition, necrosis, death of specific cell populations), consistent with a loss of MICU1-dependent mtCU gatekeeping activity. That mPTP inhibition (206) or downregulation of EMRE to limit _mCa²⁺ uptake (163) corrects some of these defects supports this notion. This _mCa²⁺-overload phenotype of MICU1-deficient animals is the opposite of that seen with loss of MCU function, which ablates all measurable rapid _mCa²⁺ uptake and so minimizes the risk of pathogenic _mCa²⁺ overload, which is most relevant during acute cellular Ca²⁺ stress. MCU seems to be less critical for baseline, unstressed function; thus, MCU-knockout animals display a mild baseline/homeostatic phenotype. A second, intriguing hypothesis to explain the discrepancies between animals with loss of MCU versus loss of MICU1 is the possibility that MICU1 exerts functions independent of the mtCU, for instance at cristae junctions (171, 198). Thus, loss of MICU1 may disrupt other Ca²⁺-sensitive mitochondrial processes that are independent of changes in mtCU activity and that cause additional functional defects that cannot be recapitulated with disruption of MCU alone. The generation of vertebrate models in which the effects of MICU1 knockout or overexpression are examined independent of the mtCU (i.e., in a MCU- or EMRE-knockout background) will help to elucidate possible additional functions of MICU1 and their impact on physiology.

3.3.3.1.9. Regulation of MICU1.

MICU1 gene expression is controlled by a number of transcription factors and translational repressors. MICU1 mRNA and protein levels are reduced during hypoxia because of increased expression of the transcription factor Foxd1, which binds to the MICU1 promoter and blocks MICU1 gene expression (211). MICU1 is particularly responsive to profibrotic stimuli such as transforming growth factor (TGF)-β and angiotensin II that acutely increase its gene expression (165), although the specific transcription factors responsible for this pathway have yet to be elucidated. The transcription factor early growth response 1 (EGR1) increases MICU1 expression under conditions of nutrient deficiency and limited TCA cycle substrate availability (212). Nemani et al. (212) demonstrate that such upregulation of MICU1 limits _mCa²⁺ uptake and hypothesize that nutrient-sensitive transcriptional control of MICU1 is a mechanism to prevent _mCa²⁺ overload and cell death during starvation conditions such as ischemia. Several caveats to this work exist, though, limiting confidence in this model of MICU1 regulation (213).

The translation of MICU1 transcripts is repressed by microRNAs (miRs) including miR-195, which is reduced in ovarian cancer, leading to elevated MICU1 expression (214). miR-181c is upregulated in obesity and also indirectly affects MICU1 expression. Increased expression of miR-181c promotes cellular ROS production, causing oxidation and downregulation of the MICU1 transcription factor Sp1 (215), leading to net downregulation of MICU1 (216). This effect of miR-181c is likely due to its direct repression of the mitochondrial gene mt-COX1, which encodes a subunit of complex IV. Downregulation of mt-COX1 and subsequent disruption of complex IV activity increases mitochondrial ROS (217). MICU1 expression is thus regulated by retrograde mitochondria-to-nucleus signaling involving mitochondrial ROS production, which in turn is influenced by changes in _mCa²⁺ concentration (218). These observations suggest the intriguing possibility of a feedback loop involving mitochondrial ROS production that modulates MICU1 expression in response to perturbations in _mCa²⁺ homeostasis. Studies in C. elegans indicate that MICU1 protein translation is also repressed by the mRNA decay factor CAR-1, an ortholog of LSM14 that is downregulated during axon regrowth in conjunction with an increase in MICU1 expression (205).

The import of MICU1 protein into the mitochondria depends on the mitochondrial import complex component Tom70 (209). Experiments in the mouse heart reveal that decreased mitochondrial Tom70 abundance after I/R limits mitochondrial MICU1 content and thereby contributes to _mCa²⁺ overload and myocardial I/R injury (209). MICU1 protein is subject to proteasomal degradation, and its half-life is substantially shorter than that of MCU and MICU2 (219). Matteucci et al. (219) attribute MICU1’s degradation to an interaction between MICU1 and the E3 ubiquitin ligase Parkin. Parkin’s Ubl domain, but not its E3 ubiquitin ligase activity, is required for proteasomal degradation of MICU1, raising the possibility that Parkin acts as a scaffold for the interaction of MICU1 with other ubiquitin ligases rather than ubiquitinating MICU1 itself (219). Valosin-containing protein (VCP) promotes the degradation of ubiquitinated MICU1 in the mouse heart and so inhibits _mCa²⁺ uptake at high cytosolic Ca²⁺ concentrations (220). The specific proteins involved in MICU1 degradation downstream of its interaction with Parkin or VCP remain to be determined.

MICU1 protein function is regulated by a number of posttranslational modifications, the most prominent being redox modification of the cysteine residue that forms disulfide bridges connecting MICU1 homodimers or MICU1/2 or MICU1/3 heterodimers. C55 of the intermembrane space oxidoreductase Mia40/CHCHD4 forms a disulfide bond with MICU1 C465 (C463 according to the numbering scheme noted above) after MICU1 is imported across the OMM (221). Formation of this bond primes MICU1 for subsequent disulfide bond formation and heterodimerization with MICU2 (221). Methylation of R455 by protein arginine methyltransferase 1 (PRMT1) renders MICU1 less sensitive to Ca²⁺ and consequently impairs _mCa²⁺ uptake (222). Preferential binding of uncoupling protein 2 (UCP2) to methylated MICU1 resensitizes MICU1 to Ca²⁺ and restores _mCa²⁺ uptake (222). Uncoupling protein-dependent sensitization of MICU1 is proposed to maintain appropriate _mCa²⁺ cycling in settings of increased cellular PRMT activity such as cancer (222). Finally, S124 within the NH₂ terminus of MICU1 can be phosphorylated by serine/threonine kinase 1 (AKT). Phosphorylation of S124 disrupts MICU1 proteolytic processing and destabilizes MICU1, ultimately causing loss of gatekeeper function, increased basal _mCa²⁺ levels, and increased ROS production (223). This study reports that activated AKT can be detected within isolated mitochondria and localizes to the IMS, where it would have access to MICU1. Multiple investigations support that AKT can translocate from the cytosol to the mitochondrial IMS or matrix (224, 225), which may occur via mechanisms involving the chaperone heat shock protein 90 (226) and/or redox signaling (227). Phospho-null mutation of MICU1 S124 is sufficient to suppress tumor growth in vivo even in the presence of active AKT, indicating that AKT-dependent MICU1 phosphorylation is relevant to cancer biology (223).

3.3.3.1.10. Pharmacological modulation of MICU1.

A recent high-throughput screen designed to identify small molecules capable of modifying mtCU function identified two compounds, named MCU-i4 and MCU-i11, that inhibit _mCa²⁺ uptake (228). Both of these compounds bind within a specific cleft that separates the N lobe and C lobe of MICU1 and fail to inhibit _mCa²⁺ uptake when residues in this cleft (Q302, Q306, and L443) are mutated, suggesting that their inhibitory effect is mediated specifically by MICU1 (228). Both MCU-i4 and MCU-i11 inhibit _mCa²⁺ uptake in skeletal muscle fibers following caffeine-induced SR Ca²⁺ release and are sufficient to attenuate myotube growth in vitro (228). These findings indicate that MCU-i4 and MCU-i11 interfere with cooperative activation rather than the gatekeeping function of MICU1 and therefore are promising tools for the prevention of _mCa²⁺ uptake and _mCa²⁺ overload. An intriguing question for further research is whether these compounds may also be used to regulate mtCU-independent functions of MICU1.

3.3.3.2. micu2.

MICU2 (Mitochondrial Calcium Uptake Protein 2; gene name = MICU2, previously annotated as EFHA1) was identified via bioinformatics as a paralog of MICU1 that arose via gene duplication (93). MICU2 localizes to the mitochondrial IMS and is anchored to the uniporter complex via a physical interaction with MICU1 (FIGURE 2). The interaction between MICU1 and MICU2 is preserved in both high-Ca²⁺ and low-Ca²⁺ conditions (93). siRNA-mediated knockdown of MICU2 or MICU1, either alone or in combination, does not affect basal respiration or oxidative phosphorylation in mitochondria isolated from mouse liver. However, single knockdown of either MICU1 or MICU2 slows the rate of _mCa²⁺ uptake in response to a single large (50 µM) Ca²⁺ bolus and limits the ability of mitochondria to buffer extramitochondrial Ca²⁺ (93). These inhibitory effects on _mCa²⁺ uptake are additive upon simultaneous knockdown of MICU1 + MICU2. However, interpretation of the direct consequences of MICU1 or MICU2 depletion is confounded by the finding that, in certain cell types, knockdown of either of these proteins reduces net MCU protein expression, and is also hindered by the finding that MICU2 knockdown reduces MICU1 protein levels by about half (93). Based on these findings, Plovanich et al. (93) suggest that a major role of the MICU1/2 interaction is for intraprotein stability. Knockdown of MICU1 alone, or in combination with MICU2 knockdown, decreases the overall molecular weight of the mtCU (93), consistent with complete loss of MICU1/2 heterodimers and/or loss of MICU1 homodimers from the uniporter complex. In contrast, knockdown of MICU2 alone shifts only a portion of the existing mtCU complexes to a lower molecular weight (93). This result is consistent with the replacement of MICU1/2 heterodimers by MICU1 homodimers, which have a similar molecular mass, thus preserving the overall mass of a subset of mtCU complexes (∼480 kDa). The portion of mtCUs that do shift to a lower molecular weight upon MICU2 knockdown likely results from a secondary reduction in MICU1 protein abundance, and therefore complete loss of MICU dimers from those particular complexes. Such complex effects of MICU1 and MICU2 on each other’s stability and capacity to regulate the mtCU emphasize the technical challenges in the design and evaluation of experiments aimed at elucidating the role of a single MICU on _mCa²⁺ uptake. This also has important implications for interpretation of experiments using genetic knockout as opposed to partial knockdown of the various MICU proteins (172).

MICU2 contains a pair of highly conserved Ca²⁺-binding EF-hand domains, just like MICU1, that allow it to respond to local changes in Ca²⁺ concentration (93). However, distinct tissue distributions suggest that MICUs have unique roles in cellular physiology. Whereas MICU1 is expressed in the mitochondria of most mammalian tissues, MICU2 expression is less ubiquitous and was detected in mitochondria of only 7 of 14 mouse tissues examined by mass spectrometry (93, 156). Both MICU2 and MICU1 show particularly strong mRNA expression in organs such as the stomach and intestines, but MICU2 expression is enriched over MICU1 in tissues including the heart and prostate (93). Within the mitochondria, MICU2 is closely associated with the outer leaflet of the IMM because of obligate heterodimerization with MICU1 (158, 180).

Early work from Rizzuto’s laboratory suggested that MICU2 inhibits the activity of purified MCU reconstituted in lipid bilayers, leading the group to propose that MICU2 functions as a bona fide negative regulator or “gatekeeper” for the mtCU (180). How the addition of MICU2 could produce this effect on purified MCU when MICU2 is not capable of binding directly to MCU (180) is unclear. Patron et al. (180) show that either MICU1 silencing or MICU1 overexpression in HeLa cells increases _mCa²⁺ uptake, consistent with loss of mtCU gatekeeping. In contrast, they report that knockdown of MICU2 increases whereas overexpression of MICU2 decreases _mCa²⁺ uptake, in broad agreement with their results in lipid bilayers. It is important to note that these experiments leading to the initial view of a MICU2 as a “MCU inhibitor” were not performed over a wide range of experimentally controlled extramitochondrial Ca²⁺ concentrations. Rather, this study examined net _mCa²⁺ transients in response to histamine-induced Ca²⁺ transients in intact cells or _mCa²⁺ uptake in permeabilized cells in response to increasing the bath Ca²⁺ concentration from 0 to the relatively low concentration of 400 nM (180).

Additional studies conducted over a wider range of physiological Ca²⁺ concentrations have refined the field’s understanding of MICU2’s influence on MICU1 behavior and overall mtCU regulation. The Mootha group reports that MICU1 and MICU2 have nonredundant functions but nevertheless work together to prevent _mCa²⁺ uptake through the mtCU when cytosolic Ca²⁺ concentration is low, and to facilitate _mCa²⁺ uptake when cytosolic Ca²⁺ concentration rises (172). Genetic knockout of MICU2 in HEK293T cells increases _mCa²⁺ uptake in response to a pulse of moderate-concentration (1 µM) Ca²⁺, in agreement with Patron et al. (180) and suggesting that MICU2, like MICU1, contributes to overall gatekeeping of the mtCU (172). In this model system, knockout of MICU2 does not reduce MICU1 protein levels, so any effects of MICU2 loss can be specifically attributed to loss of MICU2 function rather than destabilization of MICU1. Mutation of MICU2’s EF-hand domains prevents mitochondria from taking up Ca²⁺ at both moderate (1 µM) and high (40 µM) concentrations, indicating that when MICU2 binds Ca²⁺ it contributes to activation of the mtCU (172). Thus, Kamer and Mootha (172) suggest that MICU1 and MICU2 together gate the mtCU when cytosolic Ca²⁺ concentration is low but then bind Ca²⁺ and alleviate this inhibition as cytosolic Ca²⁺ concentration rises. Loss of MICU2 results in a shift in the Ca²⁺ threshold for this Ca²⁺-dependent activation of _mCa²⁺ uptake (172) (FIGURE 4). Furthermore, since disruption of either MICU1’s or MICU2’s EF-hands ablates _mCa²⁺ uptake, the relief of mtCU channel gatekeeping by rising cytosolic Ca²⁺ concentrations depends upon Ca²⁺ binding to both MICU components of a MICU1/2 heterodimer. Mutation of the MICU2 EF-hands has no effect on _mCa²⁺ uptake in MICU1-knockout cells, indicating that MICU2-dependent channel regulation depends on the presence of MICU1 (172). Indeed, since MICU2 cannot bind to MCU, MICU2 must mediate its effects on mtCU function via its physical interaction with MICU1 (172, 180).

Studies by the Foskett laboratory further clarify MICU2’s role in mtCU regulation. Payne et al. (173) show that genetic knockout of MICU2 in HEK293T cells has no effect on MICU1 expression or MICU1 binding to MCU, and does not alter mtCU gatekeeping at low (100–300 nM) cytosolic Ca²⁺ concentrations or affect basal _mCa²⁺ concentration. That is, MICU2 is not strictly required for mtCU gatekeeping over this Ca²⁺ range. The authors suggest that MICU1, possibly functioning as MICU1 homodimers, is sufficient to gate the mtCU and prevent _mCa²⁺ uptake at low, resting extramitochondrial Ca²⁺ levels. It should be noted, though, that any inferences about specific MICU1 homodimer function in MICU2-knockout cells, where the presence or absence of MICU3 has not been confirmed, remain speculative because such function could instead reflect the consequences of an increase in the relative cellular population of MICU1/3 homodimers in the absence of MICU2.

In contrast to the lipid bilayer results cited above (180), overexpression of MICU2 in MICU1-knockout cells cannot rescue channel gatekeeping and so does not prevent excessive _mCa²⁺ uptake at low cytosolic Ca²⁺ concentrations (173). Again, this fits with the observation that MICU2 does not bind to the mtCU in the absence of MICU1. At higher cytosolic Ca²⁺ concentrations (500 nM to 10 µM), _mCa²⁺ uptake is more sensitive to an increase in Ca²⁺ concentrations in MICU2-knockout cells, and the relationship between _mCa²⁺ uptake rate and cytosolic Ca²⁺ concentration is shifted to the left (173) (FIGURE 4). Thus, Payne et al. propose that MICU2 does not act as an inhibitor of the mtCU but instead modulates the threshold Ca²⁺ concentration at which MICU1’s mtCU gatekeeping activity is relieved. They further suggest that MICU2 reduces the gain of mtCU activation, by opposing MICU1-dependent channel activation at higher Ca²⁺ concentrations via a mechanism that reduces the Ca²⁺ affinity of this activation (173). Thus, the presence of MICU2 shifts the Ca²⁺ dependence of cooperative activation of the channel to the right, making it less responsive to low cytosolic Ca²⁺ concentrations. However, the cooperative nature of mtCU activation by Ca²⁺ is otherwise much the same regardless of whether is MICU2 is present (173). Interestingly, disruption of either of MICU2’s EF-hands disrupts cooperative activation of the mtCU, indicating that MICU2’s ability to bind Ca²⁺ is required both for high Ca²⁺ concentrations to relieve mtCU gating and for cooperative activation of the channel (172, 173).

Considered together, these reports support a model in which the MICU1/2 dimer functions as an “on-off switch” for the uniporter (176) and in which MICU2 specifically modulates the threshold and gain of MICU1’s gatekeeping and cooperative activation functions (173). At low, resting cytosolic Ca²⁺ concentrations, MICU1 gates the mtCU and prevents _mCa²⁺ uptake through the channels; as cytosolic Ca²⁺ concentration rises, MICU1’s gatekeeping function is relieved. Whether the channel is gated by MICU1/1 homodimers or MICU1/2 heterodimers determines the cytosolic/IMS Ca²⁺ concentration that is required to relieve gatekeeping and allow cooperative activation of the channel to occur (FIGURE 4). In the absence of MICU2, the Ca²⁺ sensitivity of mtCU activation is increased, shifting this relationship to the left. This allows for greater _mCa²⁺ uptake at low and moderate Ca²⁺ concentrations, consistent with the initial observations of the effects of MICU2 knockdown (180) and MICU2 knockout (172). When MICU2 is present, it reduces the Ca²⁺ sensitivity of mtCU activation, pushing the curve to the right, and with MICU1 participates in the cooperative activation of the channel at Ca²⁺ concentrations above the new activation threshold. Thus, the Ca²⁺-dependent regulation of the mtCU is mediated by the net effects of MICU1 and MICU2 and by the relative availability of each within the cell.

3.3.3.2.1. MICU2 phylogenetic conservation.

MICU2, along with MICU1, is conserved among vertebrate species, but only a single ancestral MICU isoform is present in plants and protozoa. Sequence analysis suggests that MICU2 arose via gene duplication of the ancestral MICU1-like isoform (93). A MICU2 ortholog is also present in trypanosomatids and C. elegans (93–95), and suggests that the initial gene duplication occurred before divergence of these lineages. MICU2 was subsequently lost from several genera including Leptomona, Crithidia, and Leischmania (94). The observation that the parasite T. cruzi can survive genetic loss of MICU1 or MICU2 but that simultaneous loss of MICU1 and MICU2 is lethal (94) supports the idea that these proteins have nonredundant roles in lower organisms. In contrast to results in HeLa cells (180), MICU2 overexpression in T. cruzi surprisingly does not affect _mCa²⁺ uptake (94). This finding suggests some divergence of MICU2 protein function in parasites versus humans. Another possibility is that T. cruzi already have a high MICU2-to-MICU1 ratio and that the introduction of additional exogenous MICU2 does little to alter mtCU activity. MICU2 knockout in T. cruzi phenocopies MICU1 knockout and limits acute _mCa²⁺ uptake over a range of 1.5–50 µM extramitochondrial Ca²⁺ (94). However, it is not clear whether this finding represents a function of MICU2 that is unique to trypanosomatids or whether impaired _mCa²⁺ uptake upon MICU2 deletion is due to destabilization of MICU1 protein and loss of MICU1-dependent cooperative activation of the mtCU, as occurs with MICU2 knockdown in some mammalian cells.

3.3.3.2.2. MICU2 structure.

MICU2 shares 20–25% protein sequence identity and ∼40% sequence similarity with MICU1, depending on species (93, 94, 180). MICU2 has a molecular mass of 45 kDa (versus ∼50 kDa for MICU1) (180). Much of MICU2’s overall domain architecture is conserved with that of MICU1 and is characterized by two highly conserved Ca²⁺-binding EF-hands separated by a long linker region composed of α-helices (93). The first EF-hand is critical for MICU2 to undergo conformational changes needed to form dimers following binding of Ca²⁺ (229). Like MICU1, MICU2’s NH₂ terminus contains a mitochondrial targeting sequence (93). However, several differences in the NH₂- and COOH-terminal regions of MICU2 and MICU1 contribute to their distinct functional properties. MICU2 has a longer COOH-terminal helix, which is more rigid in structure than the corresponding COOH terminus of MICU1 (230). This COOH-terminal helix is not needed for dimerization between MICU1 and MICU2, but it is required for MICU2-dependent gatekeeping function within cells (230). MICU2’s COOH-terminal helix is stabilized by a short, NH₂-terminal β-strand that is also specific to MICU2 (230).

Dimerization between MICU1 and MICU2 is mediated by disulfide bond formation between C465 of MICU1 and C410 of MICU2. Both of these cysteine residues are critical for heterodimer formation, as mutant C465A MICU1 and C410A MICU2 are incapable of heterodimerization with wild-type MICU2 or MICU1, respectively (180). MICU1/2 heterodimers are further stabilized by the formation of a salt bridge between residues R221 in MICU1 and D330 in MICU2 (229), although this interaction itself is not sufficient to mediate dimerization in the absence of disulfide bond formation. A second salt bridge forms between D231 of MICU1 and R352 of MICU2 (231). Finally, the MICU1/2 heterodimer is stabilized by hydrophobic interactions between M229 of MICU1 and M337 of MICU2 (231). Park et al. (231) suggest that these salt bridges and hydrophobic interactions make the intermolecular association of MICU1/2 heterodimers stronger than the intermolecular association between either MICU1 homodimers or MICU2 homodimers.

As noted above, dimerization of MICU2 with MICU1 is critical for anchoring of MICU2 to the mtCU and for MICU2-dependent regulation of the uniporter channel (172, 180). The three-dimensional structure of the MICU1/2 heterodimer is also responsive to Ca²⁺ binding. In the Ca²⁺-free state, the MICU1/2 heterodimer is characterized by electrostatic interactions between E242 of MICU1 and R352 of MICU2 (193). Upon Ca²⁺ binding, the MICU1/2 heterodimer instead depends on F383 of MICU1 and E196 of MICU2, and is more compact in structure (193). The shift to the compact form of the heterodimer involves the establishment of hydrophobic interactions between F383 of MICU1 and I333 and M337 of MICU2 that are exposed as the EF-hands of MICU1/2 rotate outward after they bind Ca²⁺ (193). The Ca²⁺-induced conformational shift in the MICU1/2 heterodimer is likely responsible for Ca²⁺-dependent changes in mtCU gating and activation. Xing et al. (232) also note that MICU2 forms “back-to-back” homodimers involving residues R107, K121, and D154 that are critical for mtCU gatekeeping and that dissociate in response to Ca²⁺. Thus, contact between multiple MICU2 subunits, possibly involving multiple MICU1/2 dimers present at a given mtCU uniplex, may contribute to MICU2’s regulation of the channel. This model is discussed in further detail below.

3.3.3.2.3. Properties and function of MICU2.

Research since the initial characterization of MICU2 has enhanced our understanding of how MICU2 cooperates with MICU1 to regulate mtCU activity. Matesanz-Isabel et al. (175) agree with the original Rizzuto model and suggest that MICU2 functions as a pure inhibitor of the uniporter, but acknowledge that this apparent inhibitory effect is minimized as cytosolic/IMS Ca²⁺ levels rise past 7 µM. However, this same phenomenon can be explained by a MICU2-dependent decrease in the Ca²⁺ sensitivity of mtCU activation, which reduces channel function at low cytosolic Ca²⁺ concentrations. This function of MICU2 agrees with data indicating that the formation of MICU2 back-to-back dimers is required to suppress _mCa²⁺ uptake at moderate (1 µM) cytosolic Ca²⁺ concentrations and that these dimers dissociate when local Ca²⁺ concentration rises (232). Whether such MICU2 dimer structures within a given uniplex channel influence MICU2’s role in the integration of matrix and IMS Ca²⁺ signals (115, 116) or other functions is undetermined. An interesting concept, though, is the possibility for distinct MICU2 molecules to functionally interact, perhaps influencing the collective behavior of populations of MICU1/2 heterodimers, conferring additional, “emergent” levels of channel regulation beyond that provided by MICU1/1 homodimers alone.

Interesting studies by the Pozzan and Riemer laboratories demonstrate that Ca²⁺-dependent dissociation of the MICU1/2 heterodimer from MCU mediates the gradual activation of the mtCU that occurs in response to sustained, moderate increases in cytosolic Ca²⁺, such as in ischemia-reperfusion (189, 221). Paillard et al. (187) suggest that tissue-specific expression of MICU1 is the limiting factor for gating of the mtCU by MICU1-2/3 heterodimers or MICU1 homodimers, because MICU2 and MICU3 cannot bind to the complex on their own. Thus, even under physiological conditions there is a balance between “ungated” mtCUs that lack MICU dimers and “gated” mtCUs that are regulated by the MICUs, depending on the availability of MICU1 within the tissue (187). In addition to the presence versus absence of MICU dimers on the mtCU (MICU-to-MCU ratio), another key parameter that shapes uniporter regulation is the relative ratio of MICU1:MICU2 at the uniporter. Deletion of MICU2, leaving only the possibility of MICU1 homodimers (or MICU1/3 heterodimers, depending on cell type) increases the ratio of MICU1:MICU2 and shifts the uniporter’s Ca²⁺ activation threshold from ∼600–800 nM down to ∼350 nM (173, 176) (FIGURE 4). The relative ratio of MICU1:MICU2 is particularly low in tissues such as the heart and prostate (93). Assuming that MICU1 abundance is not limiting, a high level of MICU2 expression and corresponding low ratio of MICU1:MICU2 is predicted to replace some MICU1 homodimers with MICU1/2 heterodimers, and so would reduce the Ca²⁺ sensitivity of mtCU activation, thus requiring higher cytosolic Ca²⁺ levels to permit _mCa²⁺ uptake through the channel. Therefore, the relative expression of MICU2 versus MICU1 is another determinant of tissue-specific properties of _mCa²⁺ uptake.

3.3.3.2.4. Interaction of MICU2 with the metazoan mtCU complex.

Binding of MICU2 to the mtCU is mediated by heterodimerization of MICU2 with MICU1 (180), which then interacts with MCU and EMRE and tethers the heterodimer to the complex (158). MICU2 is not capable of binding to MCU directly, in the absence of MICU2 dimerization with MICU1 (172, 180). MICU2 also differs from MICU1 in that it does not interact stably with EMRE (158, 172). The current structural models for the mtCU suggest a “dimer of uniplex channels” configuration (introduced above in sect. 3.3.3.1). This highlights how MICU2 and the MICU1/2 heterodimer cooperate to keep the channel open or closed depending on local Ca²⁺ concentration. Within a single “uniplex” channel (4 MCU:3 or 4 EMRE:1 MICU1:1 MICU2), MICU2 binds to MICU1, which blocks the uniporter pore, but MICU2 does not interact directly with MCU (184, 185). The regions of MICU1’s uniporter interaction domain that directly contact MCU (the α1-helix, P124 in the bend between the α1- and α2-helices, and amino acids that interact with MCU’s D ring) are not conserved in MICU2 (185). So, MICU2 is not capable of blocking the mtCU pore on its own, in the absence of MICU1. Xing et al. (232) propose that contacts between the MICU2 proteins of multiple MICU1/2 heterodimers present on each uniplex channel act as an inhibitory latch that keeps the heterodimers in a position where MICU1 can block the channel pore under low-Ca²⁺ conditions. This is an intriguing idea, although it is at odds with structural models in which each uniplex channel has only a single MICU1/2 heterodimer (184, 185).

As Ca²⁺ concentration increases, favoring dimerization of the individual uniplex channels, the MICU1/2 heterodimers of each uniplex form a back-to-back bridge spanning the distance between the two channels (184, 186). Although the channel activity of the mtCU does not require such dimerization of the uniplex channels (that is, monomeric uniplex channels are fully functional and subject to regulation by the MICU proteins), the pore of each mtCU uniplex does become unblocked in the dimer conformation (184–186). This conformational shift is associated with a change in MICU1 from binding MCU to favoring an interaction with EMRE (194), and so may alter MICU-dependent channel regulation. These observations fit with the finding that the MICU1/2 heterodimer is bound to MCU at low Ca²⁺ concentrations but dissociates from MCU as Ca²⁺ levels rise (221). Both MICU1 and MICU2 can associate with cardiolipin in the IMM (176). It is tempting to speculate that cardiolipin binding stabilizes the Ca²⁺-bound MICU1/2 heterodimer in the “bridging” state where MICU1 releases MCU in favor of EMRE and the heterodimer is pulled away from the mtCU pore. Indeed, structures of the Ca²⁺-bound MICU1/2 heterodimers either in single uniplex channels or in the “bridge” between dimerized uniplex channels place MICU2 in proximity with the outer leaflet of the IMM (185, 186). Thus, cardiolipin binding by MICU2 may stabilize the MICU1/2 heterodimer in the “open,” bridging state.

The particular composition of MICU dimers present on the mtCU uniplex channels will influence how the channels are gated in these structural models. Kamer et al. propose that, in the low-Ca²⁺ state, the relatively inflexible COOH-terminal helix of MICU2 is stabilized by a segment of the MICU2 NH₂ terminus and points toward the center of the mtCU pore, contributing to an “inhibitory latch” for the channel as hypothesized by Xing et al. (230, 232). The COOH-terminal helix of MICU1, in contrast, is much less rigid, which may make it less effective at forming this latch on its own, even though it is predicted to interact with the COOH-terminal helix of MICU2 (230). The unique “inhibitory latch” gating function conferred by MICU2 should then be absent from mtCU channels gated by MICU1/1 homodimers, potentially explaining the increased sensitivity of the mtCU to activation by Ca²⁺ that is observed upon genetic MICU2 disruption. Thus, interactions between the MICU2 constituents of a pair of MICU1/2 dimers present on a single uniplex mtCU may be one mechanism by which the presence of MICU2 decreases the Ca²⁺ sensitivity of mtCU activation.

Another mechanism differentiating the gating behavior of MICU1 homodimers versus MICU1/2 heterodimers is the difference in their affinity for Ca²⁺. MICU1 and MICU2 both bind Ca²⁺ with high affinity, but the Ca²⁺-binding affinity of MICU2 is lower than that of MICU1 (i.e., MICU2 is less sensitive to Ca²⁺) (176). MICU1/2 heterodimers thus have a lower overall Ca²⁺ affinity (∼600 nM, similar to the mtCU gatekeeping threshold observed in wild-type cells) than MICU1 homodimers (172, 176, 231). Park et al. (231) hypothesize that the greater binding affinity of MICU1-MICU2 interactions within heterodimers, versus the binding affinity of MICU1-MICU1 interactions within homodimers, makes it more difficult for MICU1/2 heterodimers to undergo conformational changes required for Ca²⁺ binding. Since Ca²⁺ binds to MICU1 and MICU2 cooperatively and requires conformational changes in both proteins, the net result is that the greater structural stability of the Ca²⁺-free MICU1/2 heterodimer reduces its affinity for Ca²⁺ and makes the mtCU less sensitive to activation at low Ca²⁺ than the MICU1 homodimer does (231). Therefore, a higher cytosolic/IMS Ca²⁺ concentration is required to open mtCU channels gated by MICU1/2 heterodimers than those gated by MICU1 homodimers.

3.3.3.2.5. Genetic diseases associated with MICU2.

Homozygous truncation of MICU2 was recently found in patients with a neurodevelopmental disorder characterized by severe cognitive impairment, spasticity, and white matter involvement (233) (TABLE 1). The mutation (c.42G>A) generates an early stop codon that results in truncation of MICU2 after residue 13 and expression of a shortened, <10-kDa MICU2 protein. Patient skin fibroblasts with this mutation exhibit elevated resting _mCa²⁺ content, as well as diminished cytosolic Ca²⁺ transients and prolonged _mCa²⁺ influx upon stimulation with bradykinin (233). These observations are consistent with the purported role of MICU2 to modulate the Ca²⁺ sensitivity of mtCU gatekeeping and cooperative activation. MICU1 protein levels are normal in MICU2-mutant fibroblasts (233), suggesting that this aberrant Ca²⁺-handling behavior is specifically attributable to loss of MICU2 function, and likely replacement of MICU1/2 heterodimers with MICU1/1 homodimers (173), rather than any secondary effect on MICU1 protein stability. The altered _mCa²⁺ handling in MICU2-mutant cells is associated with a greater sensitivity to oxidative stress in vitro (233). It remains to be determined how increased susceptibility to oxidative stress, and any subsequent neuronal dysfunction or death, may contribute to the neurological phenotypes of individuals with homozygous MICU2 truncation.

3.3.3.2.6. Animal models of MICU2.

Only one genetic animal model to date has been developed to investigate MICU2 function in vivo (TABLE 2). Bick et al. disrupted Micu2 via gene trap to generate mice with constitutive, germline loss of MICU2. Micu2^−/− mice are born at normal Mendelian ratios and have a normal life span (234). That loss of MICU2 is developmentally tolerated much better than loss of MICU1 (163) recapitulates observations in Drosophila, where loss of MICU1 is lethal but loss of MICU3, the only other MICU1 ortholog present in flies, has no impact on viability (195). Bick et al. (234) report that MICU1 and MCU protein levels are reduced in the liver of Micu2^−/− mice but do not comment on other tissues. Loss of MICU2 protein results in more rapid _mCa²⁺ uptake at low Ca²⁺ concentrations, as expected upon loss of MICU2-dependent mtCU regulation (234). However, _mCa²⁺ uptake is slower in response to high Ca²⁺, which the authors attribute to the secondary reduction in MCU expression.

Mitochondria in Micu2^−/− hearts are slightly smaller and more eccentric compared with WT hearts, although cristae structure is normal (234). Micu2^−/− cardiomyocytes exhibit slower relaxation rates due to slowed cytosolic Ca²⁺ clearance, indicative of diastolic dysfunction. As a consequence of this, Micu2^−/− mice develop left atrial enlargement at advanced ages of 16–18 mo (234). Loss of MICU2 does not affect cardiac hypertrophy in response to chronic angiotensin II infusion, but it accelerates angiotensin II-induced decompensation of contractile function (234). MICU2 deletion also increases the animals’ susceptibility to abdominal aortic aneurysm during chronic angiotensin II infusion. This defect is associated with increased expression of inflammatory and extracellular matrix genes, decreased expression of cell-junction genes, and markers of increased ROS generation. Single-cell RNA sequencing revealed that such transcriptional changes occurred in both aortic smooth muscle cells and aortic fibroblasts (234). Collectively, these experiments confirm that MICU2’s modulation of MICU1 function and overall mtCU channel behavior is functionally relevant in vivo and that its expression is particularly important in the adaption of multiple different cell types to stress. An interesting question for future studies will be to determine whether acute and/or tissue-specific Micu2 deletion reveals any further phenotypes that may have been masked by gradual compensation to constitutive deletion of Micu2 in gene trap mice.

3.3.3.2.7. Regulation of MICU2.

Little is known about how MICU2 is regulated at the transcriptional or posttranslational level. As noted above, MICU2 protein is destabilized in the absence of MICU1 (93). This finding suggests the presence of unidentified quality control mechanisms that degrade MICU2 protein that is not appropriately dimerized with MICU1 and anchored to the uniporter complex. Matteucci et al. (219) find that overexpression of the E3 ubiquitin ligase Parkin reduces the stability of MICU2 protein. However, this effect is likely an indirect consequence of Parkin-dependent degradation of MICU1, since, unlike MICU1, mature MICU2 protein within the mitochondria is not rapidly degraded by the ubiquitin proteasome system and persists even with Parkin overexpression (219).

3.3.3.3. micu3.

MICU3 (Mitochondrial Calcium Uptake Protein 3; gene name = MICU3 or EFHA2) was identified as a second paralog of MICU1 at the time of MICU2 discovery (93). Just like MICU1 and MICU2, MICU3 contains two conserved EF-hands that bind Ca²⁺ and allow it to serve as a Ca²⁺ sensor (93, 96). MICU3 protein is detected in the mitochondria of 6 out of 14 mouse tissues with high expression in excitable tissues including the brain, spinal cord, and skeletal muscle (93, 156). The relative mRNA expression of MICU3 tends to be high in tissues where MICU2 expression is low (such as neural tissue and skeletal muscle), and MICU3 expression is low in most other tissues where MICU2 expression is high (93). This pattern suggests a reciprocal relationship between MICU2 and MICU3 expression within each cell type. Like MICU2, MICU3 heterodimerizes with MICU1 (96) and therefore competes with MICU2 for binding to MICU1 (FIGURE 4). MICU2 and MICU3 do not appear to heterodimerize with each other (96). Although MICU3 subcellular localization has not yet been experimentally demonstrated, binding to MICU1 would place MICU3 within the mitochondrial IMS.

MICU3 strongly potentiates histamine-induced _mCa²⁺ transients and increases _mCa²⁺ accumulation at low (400 nM) cytosolic Ca²⁺ concentration (96). These effects are dependent upon simultaneous expression of MICU1 (96), suggesting that, just like MICU2, MICU3 requires MICU1 to regulate mtCU function. MICU3 is less effective than MICU2 at increasing mtCU gatekeeping and minimizing basal _mCa²⁺ uptake when expressed in cells that also overexpress MCU and MICU1 (96). Therefore, Patron et al. (96) conclude that any gatekeeping functions of MICU3 are less potent than gatekeeping functions of MICU2. Indeed, they suggest that MICU3 disrupts normal mtCU gatekeeping function because it displaces MICU2 from MICU1, thereby causing loss of the more robust (less Ca²⁺ sensitive) MICU2-dependent gatekeeping (96) (FIGURE 4). Accordingly, increasing MICU3 expression in a cell that normally expresses MICU2 would replace MICU1/2 dimers with MICU1/3 dimers, resulting in the shift to greater channel activity at lower cytosolic Ca²⁺ concentrations, as observed by Patron et al. (FIGURE 4).

MICU3 silencing downregulates the amplitude of _mCa²⁺ transients in primary cortical neurons (96). Thus, a fundamental role of MICU3 is to augment Ca²⁺ uptake through the mtCU. In theory, this might occur via a mechanism similar to that described for MICU2, where Ca²⁺ binding to both MICU1 and its partner MICU2 (or MICU3) is required for MICU1’s cooperative activation of the channel (172, 173). The notion that MICU3 augments _mCa²⁺ uptake is challenged, though, by experiments in HeLa cells, where MICU3 overexpression prevents increased _mCa²⁺ uptake in MICU1-knockdown cells or MICU2-knockdown cells treated with 1 µM Ca²⁺ (232). This result suggests that MICU3 may attenuate _mCa²⁺ uptake rather than augmenting it. However, this study did not examine the consequences of MICU3 overexpression alone on a wild-type background, so it is difficult to dissect the specific contributions of MICU3 versus potential secondary effects due to stabilization of other mtCU subunits on _mCa²⁺ uptake.

One reported behavior unique to MICU3 is that its overexpression in wild-type cells shortens the time between an agonist-induced increase in cytosolic Ca²⁺ concentration and the initiation of subsequent _mCa²⁺ uptake (96). How this phenomenon arises is unclear, but one possibility is that it results from a lower binding affinity between the constituent subunits of MICU1/3 heterodimers versus MICU1/2 heterodimers. As introduced above, a lower binding affinity between the members of a MICU dimer makes it easier for each subunit to undergo the conformational changes needed to bind and respond to Ca²⁺. If the requisite Ca²⁺-induced conformational changes occur more readily in MICU1/3 heterodimers than in MICU1/2 heterodimers, it would shorten the time between a rise in cytosolic Ca²⁺ and the subsequent activation of _mCa²⁺ uptake in cells that have a greater relative population of MICU1/3 heterodimers than MICU1/2 heterodimers. Indeed, this concept agrees with the idea that MICU1/3 heterodimers have a greater Ca²⁺ affinity than MICU1/2 heterodimers and so do not shift the Ca²⁺ sensitivity of mtCU regulation as far to the right (relative to MICU1/1 homodimers) as MICU1/2 heterodimers.

3.3.3.3.1. MICU3 phylogenetic conservation.

MICU3 arose from an ancestral MICU1-like gene via gene duplication, similar to MICU2. MICU3 is not present in plants or protozoa but is conserved among vertebrate species (93). Drosophila contain MICU1 and a second MICU isoform, which Tufi et al. (195) identify as a paralog of MICU3. The presence of MICU3 but not MICU2 in flies suggests that the gene duplication that gave rise to MICU3 occurred prior in evolutionary history to a subsequent duplication that gave rise to MICU2. The evolution of MICU3 and MICU2 conferred species with an extra level of mtCU regulation—modulation of the function of the ancestral MICU1—and an enhanced ability to control mtCU activity in a tissue-specific manner based on the relative expression of the three MICU isoforms (96).

3.3.3.3.2. MICU3 structure.

MICU3 protein shares ∼25% sequence identity and 34% sequence similarity with MICU1, whereas MICU3 shares ∼47% sequence similarity with MICU2 (93, 96). MICU3 contains an NH₂-terminal mitochondrial targeting sequence and two Ca²⁺-binding EF-hands (93, 96). Mutation of MICU3’s EF-hands diminishes agonist-induced _mCa²⁺ uptake but does not completely abolish it, indicating that the EF-hands contribute to positive regulation of the mtCU but are not strictly required for it (96). A unique feature of MICU3, versus MICU1 and MICU2, is the inclusion of an extra 50-amino acid domain in the middle of the protein. This additional sequence makes MICU3 slightly larger than the other MICU isoforms, with a molecular mass of ∼55 kDa (96).

MICU3 dimerizes with MICU1 via a disulfide bond formed between MICU3 C515 and MICU1 C465. Both of these cysteine residues must be present for MICU1/3 dimer formation (96). Wu et al. (193) predict that E256, M445, and R460 of MICU3 also contribute to MICU1/3 heterodimerization, based on conservation with corresponding residues in MICU2 that promote MICU1/2 heterodimerization.

3.3.3.3.3. Interaction of MICU3 with the metazoan mtCU complex.

MICU3 binds to the mtCU via its heterodimerization with MICU1 (96). The structural features of MICU1’s uniporter interaction domain that allow it to bind to MCU are not conserved in MICU3 (185). MICU3, like MICU2, therefore strictly depends upon dimerization with MICU1 to bind to and regulate the uniporter. Accordingly, MICU3 does not directly contact MCU or directly block the uniporter pore. In the structural model for the mtCU as a dimer of uniplex channels discussed above in sects. 3.3.3.1 and 3.3.3.2, MICU3 is expected to replace MICU2 in tissues like the brain where MICU3 is highly expressed. This would result in each uniplex mtCU containing 4 MCU:3 or 4 EMRE:1 MICU1:1 MICU3 (184, 185).

Structural studies indicate some degree of functional redundancy between MICU3 and MICU2 in regulating mtCU activity. In Ca²⁺-free conditions, MICU3 can form back-to-back homodimers, just like MICU2 (232). The ability to form back-to-back dimers is likely a consequence of the structure of the MICU3 NH₂ domain, which is very similar in structure to the MICU2 NH₂ domain (232). Xing et al. (232) hypothesize that back-to-back MICU3 dimers act in the same way as back-to-back MICU2 dimers to create an inhibitory latch that positions MICU1 appropriately to keep the mtCU closed. However, this idea has not yet been experimentally tested and at first consideration conflicts with the current model of MICU3 increasing the Ca²⁺ affinity of MICU1-dependent mtCU regulation relative to MICU2. These conflicting views may be reconciled by a model in which back-to-back interactions between the MICU3 subunits of two MICU1/3 heterodimers occur and do contribute to overall channel gatekeeping, but are simply less stable and therefore less effective at inhibiting channel activity as cytosolic Ca²⁺ concentrations rise than back-to-back interactions between the MICU2 subunits of two MICU1/2 heterodimers. This interpretation is supported by the finding that MICU3 is not as effective as MICU2 at limiting basal _mCa²⁺ uptake in cells that overexpress MCU + MICU1 (96).

Ca²⁺ binding induces structural rearrangements within the first EF-hand of MICU3, but the second EF-hand does not change conformation upon binding Ca²⁺ (232). Thus, the overall three-dimensional structure of MICU3 does not appreciably differ between the Ca²⁺-free and Ca²⁺-bound states. MICU3 behavior resembles MICU1 in this respect (182) and stands in contrast to MICU2, which changes its overall conformation upon binding Ca²⁺ (232). Ca²⁺-bound MICU3 is capable of forming “face-to-face” dimers via its EF-hands, much like the face-to-face MICU1 dimers that form in the presence of Ca²⁺ (232). Thus, Xing et al. (232) suggest that Ca²⁺ binding to MICU1/2/3 triggers a shift in the distribution of different mtCU forms present within a cell. At low cytosolic Ca²⁺ concentration, MICU1/3 heterodimers form back-to-back dimers via the association of their constituent MICU3 subunits with the corresponding MICU3 subunit of a second MICU1/3 heterodimer. This organization of the MICU proteins generates the “inhibitory latch” that Xing et al. (232) propose helps keep the mtCU closed. MICU1 does not form back-to-back dimers itself (232), so it is less effective at channel inhibition on its own than when interacting with MICU2 or MICU3. At higher Ca²⁺ concentrations, a greater proportion of the individual MICU1/2 and MICU1/3 dimers adopt a face-to-face conformation that allows for mtCU activation as the inhibitory latch disassembles (232). Based on this proposed model, one can speculate that the apparent inhibitory function of MICU3 reported by Xing et al. (232) is not an intrinsic property of an individual MICU3 dimerizing with MICU1, which should lower the Ca²⁺ threshold for activation of the channel compared with a single MICU1/2 dimer. Rather, the inhibitory function of MICU3 may instead be an emergent property predicted to arise from the interaction between multiple MICU1/3 heterodimers acting within the same uniporter channel. Thus, when participating in back-to-back dimers, either MICU2 or MICU3 may contribute to channel gatekeeping or inhibition, albeit with different degrees of efficacy. Direct experimental evidence for or against this hypothesized function of MICU3 is needed to settle whether MICU3 indeed forms back-to-back dimers in situ and the effect they may have on mtCU function.

A relevant question for future studies is how the subunit-subunit binding affinity compares among MICU1 homodimers, MICU1/2 heterodimers, and MICU1/3 heterodimers. As noted above, a weaker binding affinity between the two constituent subunits of a MICU dimer is correlated with a higher Ca²⁺-binding affinity for the dimer, because it is easier for the individual subunits to undergo conformational changes related to Ca²⁺ binding (231). A higher Ca²⁺-binding affinity makes the MICU dimer more sensitive to increases in cytosolic Ca²⁺ concentration, and so reduces the gatekeeping threshold for the mtCU and lets the channel open at a lower cytosolic Ca²⁺ concentration. If MICU3 truly allows the mtCU to be activated at lower Ca²⁺ thresholds than when regulated by MICU2, then this function should be associated with a higher degree of subunit mobility (i.e., weaker subunit-subunit binding affinity) and a corresponding increase in the Ca²⁺-binding affinity of MICU1/3 heterodimers compared with MICU1/2 heterodimers. A MICU1-MICU3 subunit binding affinity that is weaker than the affinity between MICU1 and MICU2 subunits aligns well with the observation that MICU3-regulated mtCU gatekeeping is relieved at lower Ca²⁺ concentrations than MICU2-regulated gatekeeping (96) (FIGURE 4).

3.3.3.3.4. Special role of MICU3 in physiology.

Examination of MICU3’s particular role within neurons highlights how the distinct functional properties of the various mtCU regulatory subunits tune mtCU activity to match tissue-specific needs. The unique features of MICU3 and its interaction with MICU1 and the mtCU confer neural tissue with specific _mCa²⁺ handling properties that support its specialized physiology. Neurons express MICU1/3 and MICU1/2 heterodimers at similar levels (96), and this will result in a higher proportion of neuronal mtCU channels gated by MICU1/3 than in other cell types. Since MICU1/3 dimers have a lower Ca²⁺ activation threshold than MICU1/2 dimers, this makes neuronal mitochondria able to take up Ca²⁺ at lower cytosolic Ca²⁺ concentrations than mitochondria in other tissues. As a result, neuronal mitochondria are more sensitive to changes in cellular activity that elevates cytosolic Ca²⁺ than mitochondria in other cell types (235). The notion that, relative to MICU2, MICU3 increases the sensitivity of _mCa²⁺ uptake to small increases in cytosolic Ca²⁺ concentration agrees with the finding that MICU3 knockdown in primary cortical neurons decreases the amplitude of _mCa²⁺ transients caused by cytosolic Ca²⁺ oscillations (96).

The enhanced Ca²⁺ sensitivity conferred by MICU3 may be critical for presynaptic neurons to ramp up local mitochondrial metabolism when needed, such as during neurotransmission. After an action potential reaches the presynaptic terminal, the subsequent increase in cytosolic Ca²⁺ concentration is transmitted to the mitochondrial matrix, where Ca²⁺ stimulates ATP production needed to fuel the synaptic vesicle recycling that is required for sustained neurotransmitter release. The Ryan laboratory proposes that the MICU3-dependent reduction in the mtCU Ca²⁺ threshold allows for “feedforward regulation of ATP production” within the nerve terminals, such that the very stimulus (local elevation in cytosolic Ca²⁺) that triggers an energy-consuming process (neurotransmitter release and vesicle recycling) is the same signal that stimulates energy production needed to sustain this process (235). This is the same concept as the parallel stimulation of mitochondrial oxidative metabolism and ATP-consuming myofilament cross-bridge cycling during sympathetic stimulation of the heart, introduced above in sect. 3.3.1.

3.3.3.3.5. Genetic diseases associated with MICU3.

MICU3 has not yet been reported to cause genetic disease in humans. Also, very little is known about whether MICU3 is involved in the pathological mechanisms by which mutations in other genes cause disease. Given the importance of MICU3 in neurons, future studies should explore its potential involvement as a driver and/or modifier of neurological and neurodegenerative disease. Conditional knockout models are needed to begin to dissect the role of MICU3 in pathophysiology.

3.3.3.3.6. Animal models of MICU3.

Tufi et al. (195) recently described a Drosophila model with CRISPR/Cas9-induced genetic disruption of MICU3 (CG4662 in the fly) (TABLE 2). Homozygous MICU3-mutant flies are viable but have a 7% reduction in life span, in contrast to MICU1 disruption, which is lethal (195). MICU3-mutant flies exhibit a climbing defect, even at young ages, which the authors interpret as a neurological defect (195). However, MICU3 disruption has no significant effect on basal mitochondrial respiration in fly head tissue (195). This suggests that loss of MICU3 has little effect on basal _mCa²⁺ uptake and respiration, even though increased expression of MICU3 is reported to make _mCa²⁺ uptake in mouse neurons more sensitive to small increases in cytosolic Ca²⁺ concentration (235). These results imply that the most relevant effects of MICU3-dependent modulation of MICU1 function in neurons may occur at higher cytosolic Ca²⁺ concentrations, where MICU3 would be more important for enhancing mtCU activation than in gatekeeping. The interpretation is also consistent with the observation that although MICU1 overexpression prevents the deleterious effects of overexpression of MCU + EMRE in the Drosophila eye (i.e., _mCa²⁺ overload and cell death), overexpression of MICU3 with MCU + EMRE is not capable of preventing these effects (195). The finding that MICU3 + MCU overexpression gives a stronger deleterious eye phenotype than MICU1 + MCU overexpression likewise suggests that MICU3 is most relevant for enhancing rather than attenuating _mCa²⁺ uptake through the mtCU within neural tissue (195). Finally, Tufi et al. (195) report that genetic disruption of MICU1 is lethal to flies and that ubiquitous overexpression of MICU3 in MICU1-mutant flies is not able to rescue this lethality. This result reinforces the views that MICU1 and MICU3 are not functionally redundant and play unique roles within the cell, and that MICU3 depends on the presence of MICU1 to exert its effects on Ca²⁺ uptake through the mtCU.

A recent study from the Murphy laboratory notes that the ratio of MICU3:MCU protein in the heart is about threefold higher than in the liver, suggesting that MICU3 may be particularly relevant for cardiac physiology (236). Germline deletion of Micu3 in mice produces no effect on baseline cardiac function, but prevents left ventricular dilation and the corresponding decline in cardiac contractility upon chronic isoproterenol infusion. This protection is associated with increased inhibitory phosphorylation of PDH, suggesting that loss of MICU3 prevents _mCa²⁺ overload. Finally, loss of MICU3 preserves contractile function and reduces infarct size in hearts subjected to ex vivo I/R (236). Puente et al. (236) conclude that loss of MICU3 prevents pathological _mCa²⁺ overload in the heart. Conversely, then, the presence of MICU3 in the heart contributes to _mCa²⁺ overload during cellular Ca²⁺ stress, just as it appears to in the nervous system. Interestingly, Puente et al. note that constitutive disruption of Micu3 does not induce the same kind of compensatory adaptations in the heart as observed in earlier models of constitutive Mcu or Emre deletion (119, 164), but why this is so remains to be determined.

3.3.3.3.7. Regulation of MICU3.

The field has only recently begun to examine the function of MICU3 in _mCa²⁺ uptake, and little is known about how MICU3 expression and protein function are regulated. Recent work in the zebrafish retina suggests that MICU3 expression is responsive to changes in MCU expression. MCU expression is low in cone photoreceptors, presumably to limit _mCa²⁺ uptake and prevent _mCa²⁺ overload in this highly active, excitable cell type (237). Experimental overexpression of MCU in zebrafish cone photoreceptors increases _mCa²⁺ concentration and causes some mitochondria to swell and lose cristae (237). MICU3 transcripts are downregulated over time in retinas with MCU overexpression, perhaps as a compensatory mechanism that favors the presence of MICU1/2 heterodimers at the mtCU and thereby reduces the Ca²⁺ sensitivity of channel activation. This may be an adaptive response to limit pathogenic _mCa²⁺ uptake (237). Whether this transcriptional downregulation of MICU3 is a direct response to increased _mCa²⁺ content, other cell stress pathways secondary to _mCa²⁺ overload, altered mitochondrial energetics, or some other mechanism is yet to be determined.

3.3.3.4. mcub.

MCUB (Mitochondrial Calcium Uniporter Dominant-Negative Beta Subunit; gene name = MCUB or CCDC109B) was identified via its sequence similarity with MCU (90). MCUB protein shares ∼50% homology with MCU and like MCU contains two transmembrane domains (90). MCUB is located in the inner mitochondrial membrane, and MCUB immunofluorescence overlaps with that of MCU (89, 90). However, MCUB exhibits a distinct expression profile from MCU that likely reflects tissue-specific regulation of mtCU function. MCUB expression is particularly high in the lung, heart, and brain, which have relatively low MCU expression (MCU:MCUB mRNA ratio of ∼3:1 in heart and lung). In contrast, MCUB expression is low in tissues with high MCU expression, such as skeletal muscle (MCU:MCUB of ∼40:1) (90). This pattern suggests a reciprocal relationship between MCUB and MCU expression and is consistent with a model in which MCU and MCUB proteins compete for incorporation into the same uniporter complexes (FIGURE 2). MCUB expression is also enriched in hematopoietic and immune cells (90), suggesting that MCUB is highly relevant in the immune system. MCUB protein is detected only in mitochondria from the placenta and not in mitochondria from any of the 13 other mouse tissues examined in the MitoCarta database (156, 238).

MCUB acts as a dominant-negative inhibitor of mtCU function. MCUB protein alone is not capable of conducting Ca²⁺ when inserted into lipid bilayers (90). Furthermore, coexpression of MCUB along with MCU decreases the probability of observing MCU channel activity in lipid bilayers, consistent with an inhibitory effect of MCUB on uniporter function (90) (FIGURE 5).

FIGURE 5.

Consequences of MCUB incorporation on mitochondrial calcium uniporter channel function. Replacement of MCU with its dominant-negative paralog, MCUB, impairs Ca²⁺ uptake through the mtCU. At least 2 mechanisms are responsible for this phenomenon. Replacement of MCU with MCUB can disrupt binding between MICU1 and the mtCU, resulting in loss of MICU-dependent regulation of the channel. Here, cooperative activation of the mtCU is impaired, shifting the relationship between mtCU-dependent _mCa²⁺ uptake and cytosolic/IMS Ca²⁺ concentration to the right. However, the consequences of the loss of MICU-dependent gatekeeping function at low cytosolic/IMS Ca²⁺ concentration are mitigated by MCUB’s additional effect to disrupt Ca²⁺ flux through the channel. As a result, _mCa²⁺ uptake through the channel is reduced compared with the classical mtCU. Replacement of additional MCU subunits with MCUB further attenuates Ca²⁺ flux through the channel. Structural modeling suggests that this effect results from a widening of the channel pore in the presence of MCUB, which may disrupt Ca²⁺ coordination within the channel. See glossary for abbreviations.

In intact cells, MCUB overexpression suppresses histamine-induced _mCa²⁺ uptake. In contrast, MCUB silencing increases agonist-induced _mCa²⁺ transients (90). These studies suggest that control of MCUB expression is yet another mechanism that can regulate overall mtCU function.

3.3.3.4.1. MCUB phylogenetic conservation.

MCUB is conserved in vertebrates but is lacking in lower organisms that do contain an MCU homolog, including plants, kinetoplastids, Amoebozoa, nematodes, and arthropods (90, 157). Trypanosomatids are a peculiar example of organisms that contain not only MCUB but additional unique MCU paralogs as well. The parasites T. cruzi and T. brucei each have a total of four MCU homologs: MCU, MCUb, MCUc, and MCUd. A critical distinction between these trypanosome MCU paralogs and mammalian MCUB is that knockout of any of the four trypanosome genes (MCU and MCUb–d) diminishes _mCa²⁺ uptake, whereas their overexpression enhances _mCa²⁺ uptake (102–105). Functional studies further support the idea that trypanosome MCUb–d contribute to rather than inhibit _mCa²⁺ uptake. In T. cruzi, knockout of TcMCU has little phenotypic consequence, but knockout of TcMCUb, TcMCUc, or TcMCUd reduces the parasite’s growth, respiration, and infectivity (102, 105). These findings suggest that replication of the MCU gene in trypanosomatids has allowed for functional divergence of the various MCU isoforms, such that MCUb and the other variants (MCUc and MCUd) are most important for acute _mCa²⁺ uptake, whereas trypanosome MCU is relatively less important for _mCa²⁺ uptake than in higher species. T. brucei likewise contains MCUb–d, all of which promote rather than inhibit _mCa²⁺ uptake (104). T. brucei MCUb, MCUc, and MCUd all interact with MCU, and each of these four paralogs can immunoprecipitate all of the other three paralogs as well (104). Thus, T. brucei MCU, MCUb, MCUc, and MCUd can all exist within a given mtCU complex and are capable of interchanging with each other to form a functional complex.

Recent work provides insight into why mammalian MCUB functions as a dominant-negative inhibitor of the mtCU but trypanosomatid MCUB–D do not. All four T. cruzi MCU isoforms share the DIME (WDXXEPXTY) motif’s critical glutamic acid (E226 in TcMCU, corresponding to E264 in human MCU and E249 in human MCUB) (102). All four T. cruzi MCU isoforms also have at least one additional negatively charged residue between the linker domain and the DIME motif (102), which may provide sufficient negative charge near the pore of the uniporter to allow Ca²⁺ permeation. The region surrounding the DIME motif is largely conserved between human MCUB and T. cruzi MCUb, suggesting that differences in the ability of these two proteins to conduct Ca²⁺ may arise from structural difference elsewhere in the primary sequence of the proteins or possibly from species-specific differences in the interaction of MCUB with other mtCU components such as EMRE, which is not conserved in trypanosomes (89, 95).

3.3.3.4.2. MCUB structure.

Human MCUB protein is 336 residues long and contains two transmembrane domains that are connected by a short linker (90). MCUB has several critical amino acid substitutions in the pore region compared with MCU (90). A conserved arginine residue (R251 in mouse, R252 in human) in MCU’s first transmembrane domain is replaced by tryptophan in MCUB (W246 in mouse, W237 in human). Additionally, a conserved glutamic acid (E256 in mouse, E257 in human MCU) in MCU is replaced by a valine in MCUB (V251 in mouse, V242 in human MCUB), resulting in loss of a negative charge from the loop connecting the two transmembrane domains. As a result, the surface of MCUB is less negatively charged than the surface of MCU, which may interfere with Ca²⁺ permeation of the channel (90). Combined mutation of MCU R251W and E256V to make the protein resemble MCUB reduces histamine-induced _mCa²⁺ uptake, suggesting that these residues contribute to the differences in Ca²⁺ conduction between MCU and MCUB (90).

Computational modeling of predicted human MCU and MCUB protein reveals overall structural similarity between both monomeric and tetrameric MCU and MCUB, with the greatest structural divergence occurring at their NH₂ and COOH termini (239). In addition to the differences in the pore region noted above (90), several amino acid differences in the transmembrane domains of MCU and MCUB are predicted to affect their function. The first transmembrane domain of MCU contains several hydrophobic residues (L240, A244, F247, A251, and W255) that interact with other hydrophobic and neutral amino acids (F269, Y272, M276, and Y279) in the second transmembrane domain of the neighboring MCU subunit. Two of these residues, Y272 and Y279, are replaced by phenylalanines F257 and F264 in MCUB. What’s more, the hydrophobic alanine and phenylalanine residues present in MCU’s first transmembrane domain (A244 and F247) are replaced by neutral serine and glycine residues in MCUB (S229 and G332). In silico modeling predicts that these amino acid substitutions in MCUB together widen the opening of the uniporter pore (239), which may be sufficient to disrupt Ca²⁺ coordination and Ca²⁺ permeation through the channel (FIGURE 5). The current model therefore suggests that the primary mechanism by which MCUB inhibits uniporter function is by disrupting the three-dimensional structure and function of the channel pore (90). However, effects of MCUB to disrupt the negatively charged surface of the uniporter raise the possibility that MCUB also interferes with MICU1 binding to the pore, and therefore limits cooperative activation of the uniporter as well.

3.3.3.4.3. Properties and function of MCUB.

The current consensus is that MCUB is not capable of forming a Ca²⁺-permeable channel on its own (90, 240). Coexpression of MCU and MCUB at a 1-to-3 ratio, corresponding to tetrameric channels consisting of a single MCU subunit and three MCUB subunits, produces channels that are capable of conducting Ca²⁺. However, the incorporation of MCUB into the uniporter reduces the probability of recording channel activity from 89% in channels composed of MCU alone to 13% in channels composed of 1 MCU:3 MCUB (240). These data are consistent with MCUB exerting a net inhibitory effect on uniporter Ca²⁺ conductance. It is also interesting to note that MCUB alone does form a channel capable of conducting sodium (90). This observation suggests that the major functional difference between MCU and MCUB is in their ion selectivity, with both proteins capable of forming functional ion channels but only MCU being sufficient to transport Ca²⁺. A relevant question is whether EMRE may be required for MCUB to function as a Ca²⁺ channel, given that EMRE is required for the function of metazoan MCU (89, 157–159, 161, 162). Furthermore, does “functional” MCUB, when coupled to the appropriate binding partners (such as EMRE), conduct Ca²⁺ at all, albeit perhaps with a lower conductance than channels composed of MCU? This seems a distinct possibility, given that the addition of EMRE confers on MCU the ability to conduct Ca²⁺, whereas MCU alone is incapable of Ca²⁺ transport but, like MCUB, can conduct Na⁺ in the absence of EMRE (90, 180). One important caveat, though, is that these conclusions regarding the competence of MCUB and MCU for ion transport in the absence of EMRE (90, 180) were based on experiments conducted in lipid bilayers with reconstituted channels, which did not include all the relevant mtCU components and therefore may not fully or accurately recapitulate the in situ behavior of MCU or MCUB, as would be observed by using patch clamp of mitoplasts.

Intriguing data from genetic mouse models suggest a downregulation of EMRE protein upon MCUB overexpression (241). Thus, displacement of EMRE from the mtCU may be one way in which MCUB disrupts uniporter function. Alternatively, downregulation of EMRE with increased MCUB expression may indicate that EMRE is not required for the fundamental function(s) of MCUB. Using a variety of methodologies including reductionist systems to answer fundamental questions such as these will clarify the molecular mechanism(s) by which increased mitochondrial MCUB content diminishes acute _mCa²⁺ uptake. It will also help to determine whether MCUB may function as an alternative, MCU-independent _mCa²⁺ uptake pathway.

Recent research supports the view that, in cells that also express MCU, the net effect of MCUB is to disrupt mtCU-dependent Ca²⁺ transport. Deletion of MCUB in HeLa cells increases the amplitude of _mCa²⁺ transients while decreasing the amplitude of cytosolic Ca²⁺ transients, suggesting that when MCUB is expressed it limits the ability of mitochondria to buffer changes in cytosolic Ca²⁺ (242). Loss of MCUB from HeLa cells shifts the relationship of _mCa²⁺ uptake rate versus bath Ca²⁺ concentration to the left, with MCUB-deleted cells exhibiting an enhanced rate of _mCa²⁺ uptake at both high and low bath Ca²⁺ concentrations (242). This effect is particularly strong at bath Ca²⁺ concentrations above 3 µM and in some ways resembles the effects of MICU1 deletion to disrupt gatekeeping and cooperative activation of the uniporter (169). MCUB deletion likewise increases net mtCU current density in HeLa mitoplasts, consistent with an overall increase in uniporter function in the absence of MCUB (242). These findings led Lambert et al. (242) to conclude that loss of MCUB enhances net mtCU function and/or increases cooperative activation and Ca²⁺ uptake rate of the uniporter, especially in conditions of high cytosolic/IMS Ca²⁺. The physiological function of MCUB, then, may be to limit _mCa²⁺ uptake during periods of cytosolic Ca²⁺ stress. This hypothesis is supported by experiments performed in genetic animal models of MCUB deletion and overexpression.

The relative ratio of MCU:MCUB tracks closely with the net mtCU activity observed across different tissues. For example, skeletal muscle has an exceptionally high mtCU current density. Skeletal muscle also has a very high ratio of MCU:MCUB, with 4 times higher MCU expression and 3 times lower MCUB expression compared with the heart (90). The relative MCUB content within a tissue may therefore modify the mitochondria’s ability to take up Ca²⁺ in response to changes in cytosolic/IMS Ca²⁺ and so serve as a control point to regulate the mitochondria’s sensitivity to signals that are transduced by an increase in cytosolic Ca²⁺ concentration (90). This mode of regulation is analogous to the proposed mechanism in which the relative ratio of MICU1:MCU controls net mtCU activity in a tissue-specific manner (187).

3.3.3.4.4. Interaction of MCUB with the metazoan mtCU complex.

MCUB incorporates into the endogenous mtCU complex of HEK293T and HeLa cells and coimmunoprecipitates with MCU (89, 90). Huang et al. (104) propose that the interactions between MCU and MCUB are mediated by their transmembrane domains, just like the interactions between MCU subunits. MCUB is also capable of self-oligomerizing into complexes of ∼170 kDa in cells in which MCU is silenced, demonstrating that MCUB does not need MCU to form a tetrameric channel (90). The population of mtCUs present within a cell may therefore be a heterogeneous mix of mtCUs composed of MCU-MCUB heterotetrameric complexes, MCU homotetrameric complexes, and MCUB homotetrameric complexes (90). A shift in the relative distribution of these different types of uniporter channels is yet another way that a tissue may modulate its _mCa²⁺ handling.

Differences in the residues of the pore regions of MCU and MCUB and in their interactions with other mtCU subunits give clues as to how MCUB can inhibit Ca²⁺ uptake through the mtCU. Like MCU, MCUB binds to EMRE (242). However, EMRE and MCUB subunits appear to compete for inclusion in the mtCU, because MCUB deletion from HeLa cells or the mouse heart increases EMRE expression, and MCUB overexpression in cardiomyocytes decreases EMRE expression (241, 242). Since MCU depends on EMRE to conduct Ca²⁺, it is possible that MCUB’s specific displacement of EMRE from the mtCU is a key mechanism that disrupts the function of the remaining MCU subunits.

Overexpression of MCUB in the mouse heart also reduces the total molecular weight of the mtCU and reduces the total amount of MICU1 and MICU2 present in high-molecular mass (∼500–800 kDa) uniporter complexes (242). Immunoprecipitation experiments show that MCUB does not interact with MICU1 or MICU2 (242). These results indicate that when MCUB replaces MCU within a mtCU channel, it causes displacement of MICU1/2 from the channel as well because MCUB is not able to bind to the MICU dimer (FIGURE 5). Fitting with this view, deletion of MCUB from HeLa cells (allowing for greater MCU + MICU1/2 incorporation into the complex) increases MCU, EMRE, MICU1, and MICU2 protein expression and increases the overall molecular weight of the mtCU (242). At least two mechanisms may account for the difference in MCU’s and MCUB’s ability to bind the MICU1/2 dimer. First, the replacement of E257 (residue numbers in this paragraph refer to the human proteins) in MCU’s loop region with V242 in MCUB renders the IMS-facing surface of MCUB less negatively charged than the corresponding surface of MCU (90). Besides a possible direct effect to disrupt Ca²⁺ permeation through the channel, the loss of negative charge from the MCUB surface may also make MCUB less capable of binding MICU1 compared with MCU. A direct MCU-MICU1 interaction occurs between positively charged arginine residues (R440 and R442) on MICU1 and the negatively charged DIME motif residues D261 and E264 of MCU, and prevents Ca²⁺ permeation of the channel when cytosolic Ca²⁺ concentration is low (192). Given its proximity to the DIME motif, it is possible that the replacement of negatively charged E257 in MCU with neutral V242 in MCUB also affects the electrostatic interaction between MICU1 and the region around the channel pore. Second, EMRE helps to anchor MICU1 to the mtCU (158). MCUB’s displacement of EMRE, from the mtCU could therefore have an additional, indirect effect to dislodge the MICU1/2 dimer from the channel.

The overall consequence of increased MCUB incorporation on mtCU function, then, is to inhibit the channel by disrupting Ca²⁺ permeation directly and/or via displacement of EMRE, and by disrupting the channel gatekeeping and cooperative activation functions of the MICUs. Increased MCUB expression would result in a greater proportion of the mtCU population existing in an ungated, but still inhibited, state, resulting in less net _mCa²⁺ uptake over a range of cytosolic Ca²⁺ concentrations. Conversely, decreasing cellular MCUB content would allow more mtCUs to be gated by MICU dimers. The net effect here would be to increase the cooperative activation of the channel, particularly at higher Ca²⁺ concentrations. Basal _mCa²⁺ uptake at low (<3 µM) cytosolic Ca²⁺ does not appear to be affected as strongly by loss of MCUB (242). This phenomenon can be explained by the increased expression of MICU1/2 in observed MCUB^−/− cells (242). The addition of the MICU1/2 gatekeeping activity at each uniporter would functionally compensate for loss of MCUB to keep the channel inhibited at low Ca²⁺ concentrations. The relative ratio of MCUB:MCU in a tissue therefore may exert more stringent control over its net mtCU activity than the MICU1-to-MCU ratio, because MCUB content would affect both the channel pore and, via secondary displacement of MICU1/2, its gating properties.

3.3.3.4.5. Genetic diseases associated with MCUB.

No primary mutations in MCUB have yet been reported to cause human disease. However, changes in MCUB expression are noted as secondary features of a number of human pathologies including diabetes (243), cancer (244), atopic dermatitis (245), and muscular dystrophy (246, 247).

3.3.3.4.6. Animal models of MCUB.

Evidence from genetic mouse models implicates MCUB in tissues’ responses to pathological stress (TABLE 2). Our laboratory and others note that MCUB is not normally expressed in the mouse heart but is upregulated at both the mRNA and protein levels and incorporated into the mtCU after myocardial infarction or myocardial ischemia-reperfusion injury (241, 242). MCUB expression thus appears to be induced by pathological signals such as hypoxia. The first genetic mouse model of MCUB employed a flox-stop MCUB transgene crossed to cardiomyocyte-specific, tamoxifen-inducible αMHC-MCM to enable inducible overexpression of MCUB in the adult mouse heart (242). Acute ∼5-day cardiomyocyte MCUB overexpression largely replicates the phenotype of cardiac MCU deletion (119, 122, 126, 127) in that it diminishes the heart’s contractile responses to acute β-adrenergic stimulation (242). Acute MCUB overexpression also reduces the maximum oxygen consumption rate and reserve capacity of adult cardiomyocytes, although it does not affect basal cardiomyocyte respiration in vitro (242). Consistent with reduced _mCa²⁺ content, acute MCUB overexpression is associated with greater inhibitory phosphorylation of pyruvate dehydrogenase both at rest and after acute isoproterenol stimulation (242). When mice are stressed with 40 min of cardiac ischemia + 24 h of reperfusion, animals with acute MCUB overexpression have substantially increased mortality, with 12 of 13 MCUB overexpressor mice dying several minutes after the onset of ischemia. Lambert et al. (242) hypothesize that this increase in mortality is related to an inability of the MCUB overexpressor heart to quickly adapt to the energetic stress of the ischemia. Another prominent distinction between earlier models of cardiomyocyte MCU deletion and short-term cardiomyocyte MCUB overexpression is that acute MCUB overexpression causes substantial baseline contractile dysfunction (242).

In contrast, after ∼1 mo of cardiac MCUB overexpression, the predominant effects of increased cardiomyocyte MCUB content are protective. Acute _mCa²⁺ uptake is still inhibited in MCUB overexpressor cardiomyocytes, but there is no longer any detrimental effect of MCUB overexpression on baseline heart function or on survival during cardiac I/R (242). These improvements are associated with a normalization of PDH phosphorylation and cellular bioenergetics. Such gradual metabolic compensation also normalizes the MCUB overexpressor heart’s contractile responses to acute isoproterenol stimulation (242). These findings suggest that the metabolic deficits that impair β-adrenergic responsiveness in mice with 1-wk MCUB overexpression and that predispose them to death during I/R are largely overcome via undefined compensatory mechanisms that take effect over the course of several weeks. Since acute _mCa²⁺ uptake is still inhibited after 1 mo of MCUB overexpression, the compensatory response likely involves a shift toward _mCa²⁺-independent regulation of cellular energy production (242). Fitting with maintained inhibition of acute _mCa²⁺ uptake, 1-mo MCUB overexpression before I/R reduces myocardial infarct size. This indicates that MCUB overexpression is sufficient to protect the cardiomyocytes against _mCa²⁺ overload and necrosis when challenged with cellular Ca²⁺ stress during I/R (242).

Many of the observations in mice with chronic 1-mo MCUB overexpression were recently recapitulated by Huo et al. (241) using mice in which MCUB is overexpressed in the cardiomyocytes because of expression of an α-myosin heavy chain-driven tetracycline-transactivator transgene and a mouse MCUB cDNA under the control of tetracycline operator sequences. Here, chronic MCUB overexpression within the cardiomyocytes has no detrimental effect on baseline cardiac function but protects the heart from I/R injury and mPTP opening (241). These results agree that with sufficient time for metabolic compensation, the beneficial effects of MCUB overexpression to limit deleterious _mCa²⁺ uptake under stress outweigh any detrimental effects. Huo et al. (241) also generated a knockout first mouse line resulting in constitutive disruption of the Mcub gene (Mcub^−/−). Cardiomyocytes from MCUB^−/− mice do not exhibit any alteration in basal _mCa²⁺ uptake, oxygen consumption, or cardiac function, as expected from the observation that the unstressed heart expresses very little MCUB (241). However, constitutive MCUB deletion exacerbates myocardial ischemic injury, subsequent cardiac remodeling, and contractile dysfunction (241). These detrimental effects of MCUB deletion are associated with increased _mCa²⁺ uptake at 7 days post-I/R, consistent with loss of the MCUB upregulation that normally occurs as an adaptive response to ischemia (241). Huo et al. (241) additionally report that hindlimb remote ischemic preconditioning is sufficient to induce MCUB expression and diminish _mCa²⁺ uptake in the mouse heart. This suggests the existence of regulatory mechanisms that communicate the presence of stress in one part of the body in order to modulate _mCa²⁺ handing in other tissues and protect them from subsequent Ca²⁺ stress. Upregulation of MCUB appears to be just one component of this protective response, since remote ischemic preconditioning is still capable of eliciting a partial reduction in acute _mCa²⁺ uptake in Mcub^−/− cardiac mitochondria (241).

Considered together, these mouse models suggest that MCUB has a beneficial role in protecting tissues from or adapting them to pathologically high cytosolic Ca²⁺ levels that would otherwise cause _mCa²⁺ overload, permeability transition, and cell death. They also indicate that increased MCUB expression, and potentially any other alteration in that inhibits mtCU function, is sufficient to perturb mitochondrial energetics and cardiac contractility. Given sufficient time, though, the heart can adapt and rewire its metabolic regulation such that impaired _mCa²⁺ uptake no longer disrupts bioenergetics or cardiac pump function. The reasons for the discrepancies between findings in mice with constitutive MCU deletion, inducible but chronic MCU deletion, and acute versus chronic inducible MCUB overexpression likely relate to the time available for the activation of alternative cell death pathways and compensatory changes to cellular metabolism. This topic has recently been reviewed in depth elsewhere (121).

3.3.3.4.7. Regulation of MCUB.

Much less is known about the regulation of MCUB than about MCU. MCUB mRNA expression is responsive to changes in cellular Ca²⁺ handling in lymphocytes, and in silico analysis suggests that MCUB is transcriptionally controlled by NFAT (135). These data agree with the observation that ischemia, which increases cellular Ca²⁺ content via a shift to glycolysis and activation of plasma membrane Na⁺/H⁺ and Na⁺/Ca²⁺ exchange, upregulates MCUB (241, 242). Despite conserving the cysteine residue that is oxidized in MCU (C97 in MCU/C82 in MCUB), MCUB is not as readily oxidized as MCU (148). MCUB therefore may differ from MCU with regard to its regulation by cellular redox signaling.

Many questions remain about what controls MCUB versus MCU incorporation into the uniporter complex. Given that MCU protein has a relatively long half-life in vivo (126, 127), how is MCUB able to quickly incorporate into the mtCU upon its induction within ∼1 day of I/R? Does the relative incorporation of MCU and MCUB into the intact mtCU align directly with their relative protein expression? Or do active mechanisms regulate which of the available subunits is selected for incorporation into and retention within a uniporter complex? Answering such questions will provide useful insight into how the relative composition of the core mtCU can be modulated as a therapy for human disease.

3.3.3.5. mcur1.

MCUR1 (Mitochondrial Calcium Uniporter Regulator 1, gene name = MCUR1, previously annotated as CCDC90A) was identified in an RNAi screen for proteins that affect _mCa²⁺ uptake in HEK293T cells (248). MCUR1 is an integral membrane protein that localizes to the IMM. It is expressed ubiquitously and has particularly high expression in skeletal muscle, kidney, and the large intestine (248). MCUR1 is not sufficient to conduct Ca²⁺ when reconstituted in yeast expression systems, indicating that it does not form a channel on its own (159). Mallilankaraman et al. (248) suggest that MCUR1 physically interacts with MCU in COS7 cells, although this notion is contested by findings from Sancak et al. (89) that MCUR1 does not copurify with MCU isolated from HEK293T cells. Nevertheless, knockdown of MCUR1 in HEK293T and HeLa cells reduces acute _mCa²⁺ uptake (248, 249). MCUR1 overexpression enhances agonist-induced _mCa²⁺ uptake, and this effect depends upon the expression of MCU (248, 249). These findings support the view that MCUR1 regulates uniporter function and that both MCUR1 and MCU are required for full mtCU activity (248, 249).

3.3.3.5.1. MCUR1 phylogenetic conservation.

MCUR1 exhibits a unique pattern of conservation across species. MCUR1 is conserved among vertebrates but is not found in Drosophila or C. elegans (91, 250). MCUR1 homologs are present in yeast as well as species of fungi that do not express homologs of MCU (91, 251). These data are consistent with MCUR1 having cellular functions that are independent of the mtCU. For example, the yeast MCUR1 homologs Put6 (Fmp32) and Put7 (Ylr283w) localize to the IMM and form large heterooligomeric complexes that are involved in proline metabolism (91). Thus, a generalizable feature of MCUR1 across species is that it participates in large complexes of mitochondrial proteins.

3.3.3.5.2. MCUR1 structure.

Human MCUR1 consists of 359 amino acids and has a molecular mass of ∼40 kDa. The first 60 residues form an NH₂-terminal domain that is followed by two transmembrane helices and a short COOH-terminal domain. Both the NH₂ and COOH termini of MCUR1 extend into the IMS, whereas the bulk of the protein extends into the mitochondrial matrix (248). The conserved three-dimensional structure of MCUR1 and related proteins is described as a head-neck-stalk-anchor, where a coiled-coil stalk is anchored in a lipid membrane via a COOH-terminal transmembrane region and a β-layer neck connects the NH₂-terminal head to the other end of the stalk (251). The NH₂ terminus of MCUR1 contains a mitochondrial signal peptide that is followed by a disordered segment of 160 residues (251). In vitro, MCUR1 constructs lacking the first 160 residues and the COOH-terminal anchor form trimers, suggesting that the intact MCUR1 protein can oligomerize (251). The head domain has small patches of negative charge due to residues E176 and E177 (251). Immunoprecipitation of MCUR1 deletion constructs in HEK293T cells show that MCUR1 interacts with MCU and that MCUR1’s head domain is required for MCUR1-MCU binding (251). The MCUR1 coiled-coil stalk domain is also critical for the interaction with MCU (251). The head domain of MCUR1 is capable of binding Ca²⁺ (251). This raises the possibility that Ca²⁺ regulates MCUR1 binding interactions, with potential implications for MCUR1-dependent regulation of uniporter function.

3.3.3.5.3. Properties and function of MCUR1.

The function of MCUR1 remains more controversial than that of other mtCU components. MCUR1 is necessary for full MCU activity, and knockdown of MCUR1 disrupts oxidative phosphorylation and ATP production (248). These effects are very similar to the effects of MCU knockdown and support the hypothesis that MCUR1 promotes _mCa²⁺ uptake through MCU. Consistent with this idea, MCUR1 knockdown in HeLa cells decreases basal oxygen consumption and reduces resting _mCa²⁺ content (248).

Conflicting data indicate that the effects of MCUR1 on _mCa²⁺ uptake may be indirect and attributable to alteration of mitochondrial membrane potential (ΔΨ_m) rather than a direct influence on MCU. MCUR1 is required for appropriate assembly of cytochrome-c oxidase (COX; respiratory complex IV), with knockdown of MCUR1 in human fibroblasts causing an increase in the protein turnover of COXI and COXII (252). MCUR1 appears to facilitate the maturation of COXII or its incorporation into the cytochrome-c oxidase complex rather than residing in mature complex IV (252). This study agrees with Mallilankaraman et al. (248) that loss of MCUR1 impairs _mCa²⁺ uptake, but claims that impaired _mCa²⁺ uptake is a consequence of depleted membrane potential (the driving force for _mCa²⁺ uptake) secondary to respiratory impairment (252). Likewise, Paupe et al. (252) reason that the respiratory defect observed by Mallilankaraman et al. (248) is not a secondary effect of diminished _mCa²⁺ uptake but rather is a primary consequence of the failure to assemble COX and maintain ΔΨ_m in MCUR1-deficient cells. Fitting with the view that MCUR1 affects ΔΨ_m, which subsequently impacts _mCa²⁺ uptake (as opposed to disrupted _mCa²⁺ uptake compromising basal metabolism and ΔΨ_m), Paupe et al. (252) show that silencing of MCU alone is not sufficient to disrupt ΔΨ_m. On the other hand, Mallilankaraman et al. (248) report that knockdown of MCUR1 does not change ΔΨ_m. Mitoplast patch-clamp experiments in which membrane potential is controlled show that MCUR1 knockdown in HEK293 cells reduces mtCU Ca²⁺ currents, independent of changes in membrane potential (249). Thus, although MCUR1 may influence multiple protein complexes within mitochondria, it does appear to act as a direct regulator of the mtCU. MCUR1 directly promotes mtCU activity by acting as a scaffolding factor to help assemble the intact uniporter complex (FIGURE 2). MCUR1 binds to both of the core uniporter components, MCU and EMRE, and coexpression of MCU and MCUR1 increases the proportion of MCU in high-molecular weight protein fractions (253). Loss of MCUR1 decreases the higher-order oligomerization of MCU, suggesting that MCUR1 is necessary for proper formation of the intact, heterooligomeric uniporter complex (253).

A number of cellular functions are ascribed to MCUR1, although how each of these functions specifically relates to MCUR1’s effect on mtCU-dependent _mCa²⁺ uptake versus MCUR1’s effects on other protein complexes is still under investigation. MCUR1-dependent uniporter function is required for maintenance of mitochondrial bioenergetics. Without MCUR1, cells become energy starved and upregulate alternative metabolic pathways such as autophagy in order to survive (253). MCUR1 has also been implicated in mitochondrial permeability transition. Expression of MCUR1 in Drosophila cells (which normally lack MCUR1) renders the mPTP sensitive to _mCa²⁺ overload (250). Conversely, siRNA knockdown of MCUR1 in HeLa cells renders the mPTP less sensitive to permeability transition in response to an increase in _mCa²⁺ content (250). Chaudhuri et al. (250) report that MCUR1 interacts with the mPTP component cyclophilin D (CypD) and hypothesize that MCUR1 brings the mPTP into proximity with MCU, so that the mPTP is exposed to a local domain of high _mCa²⁺ concentration as Ca²⁺ flows through the mtCU into the mitochondria. However, these data conflict with a report that HeLa cells with stable MCUR1 overexpression take up more Ca²⁺ pulses into the mitochondria before permeability transition, which suggests that MCUR1 instead diminishes _mCa²⁺-sensitivity of the mPTP (248).

Since MCUR1 binds divalent cations and has a particularly high affinity for Ca²⁺ (10 times higher than its affinity for Mg²⁺) (251), another interpretation that may reconcile the existing data is that MCUR1 also functions as a Ca²⁺ sensor to mediate cellular responses—including susceptibility or resistance to mitochondrial permeability transition (PT)—upon changes in _mCa²⁺ content. Indeed, the head domain of MCUR1 is destabilized in the presence of Ca²⁺, and chronic exposure to elevated Ca²⁺ levels causes purified MCUR1 to form amyloid fibrils (251). The functional consequences of these Ca²⁺-dependent changes in MCUR1 have not yet been demonstrated experimentally, but Ca²⁺-dependent regulation of MCUR1 structure is consistent with a model in which Ca²⁺ binding to MCUR1 can act as a switch to regulate other proteins or cellular processes. In this model, the apparent effects of MCUR1 expression on susceptibility to Ca²⁺-induced mitochondrial PT would arise from a complex interaction between the separate effects of MCUR1’s ability to sense Ca²⁺, MCUR1’s promotion of mtCU activity, and MCUR1’s physical interaction with the mPTP modulator CypD, which may or may not be in competition with MCUR1’s interaction with the uniporter. These multiple functions of MCUR1 would allow it to contribute to both physiological and pathological responses to elevated _mCa²⁺.

In addition to its regulatory effects on the mtCU and possible responses to changes in _mCa²⁺, MCUR1 also has mtCU-independent functions that are specifically responsive to other intramitochondrial signals. For instance, protein expression of the yeast homologs of MCUR1, Put6 and Put7, is positively regulated by proline concentration (91). Deletion of Put6 and Put7 reveals that they are necessary for proline metabolism in yeast, serving as negative regulators of mitochondrial proline uptake, and are needed to maintain cellular redox balance (91). Further investigation into the multiple effects of MCUR1 will shed greater light on its involvement in such additional processes and how they are related to _mCa²⁺.

3.3.3.5.4. Interaction of MCUR1 with the metazoan mtCU complex.

MCUR1 interacts with the uniporter complex via direct binding to MCU as well as EMRE (248, 253) (FIGURE 2). The interaction between MCU and MCUR1 is mediated by the MCUR1 head domain and additionally depends on the coiled-coil stalk, which may position the MCUR1 head in the appropriate position to interact with MCU (251). The NH₂-terminal domain and coiled-coil domain of MCU are also critical for MCU-MCUR1 binding (253). The MCU-MCUR1 interaction occurs independent of Ca²⁺ concentration, although since Ca²⁺ destabilizes the MCUR1 head domain (251) changes in Ca²⁺ concentration near the uniporter may change the nature of the MCU-MCUR1 interaction. MCUR1 thus represents another potential control point at which the local Ca²⁺ microdomain may regulate mtCU function. MCUR1 interacts with EMRE (253) but, unlike MCU and EMRE, does not directly bind to MICU1 (248, 253). MICU1 coimmunoprecipitates MCU but not MCUR1, whereas MCU coimmunoprecipitates both MCUR1 and MICU1, and MCUR1 coimmunoprecipitates only MCU and not MICU1 (248). MCUR1 and MICU1 therefore do not appear to exist within the same uniporter complex, which suggests the presence of multiple forms of uniporter channels with distinct compositions, one containing MCU-MICU1 and another containing MCU-MCUR1 (248). A possible explanation for these data is that MCUR1 associates with the nascent core uniporter composed of MCU + EMRE to ensure appropriate assembly of the channel pore, but MCUR1 is displaced by MICU1 in the mature, fully gated uniporter (FIGURE 2).

Knockdown of MCUR1 results in specific upregulation of MCU but does not affect the expression of other mtCU components (248, 249). These data could be interpreted as MCUR1 and MCU being in competition for incorporation into mtCU channels, analogous to the competition between MCU and MCUB for inclusion in the mtCU. Alternatively, upregulation of MCU in MCUR1-knockdown cells could simply be a compensatory response to normalize the cell’s net _mCa²⁺ uptake capacity in the face of diminished _mCa²⁺ uptake through each MCUR1-deficient uniporter channel. MCUR1 was not included in the structural studies of the uniporter complex discussed above in section 3.3.3.1 (184, 185), so the typical stoichiometry of MCUR1 within the mtCU remains unknown.

3.3.3.5.5. Genetic diseases associated with MCUR1.

Primary mutations in MCUR1 have not yet been reported to cause human genetic disease, although MCUR1 is implicated in the pathogenesis of hepatocellular carcinoma (254).

3.3.3.5.6. Animal models of MCUR1.

C57BL/6 mice with deletion of Mcur1 from cardiomyocytes (Mcur1^fl/fl × αMHC-Cre) or from endothelial cells (Mcur1^fl/fl × VE-Cad-Cre) are viable and born at expected Mendelian ratios (253) (TABLE 2). However, cardiac Mcur1-knockout animals are smaller than expected and die within 3 wk of birth (253). Mcur1-null cardiomyocytes exhibit impaired _mCa²⁺ uptake and diminished mtCU currents, but have normal basal _mCa²⁺ content (253). In contrast, basal _mCa²⁺ content, _mCa²⁺ uptake, basal ATP levels, and agonist-induced increases in ATP concentration are all reduced in MCUR1-deficient endothelial cells. Proliferation and migration are attenuated in MCUR1-deficient endothelial cells, similar to observations in MCU-deficient endothelial cells (253). Mcur1 endothelial cell knockout mice have normal metabolic profiles and normal serum chemistry, but have increased heat dissipation that may be related to higher expression of uncoupling protein 2 and a greater degree of mitochondrial uncoupling within the MCUR1-deficient cells (253). Autophagy is increased in both Mcur1-KO endothelial cells and Mcur1-KO cardiomyocytes (253). This finding is consistent with a requirement for MCUR1 in normal mtCU-dependent _mCa²⁺ uptake that stimulates mitochondrial metabolism: without MCUR1, cells upregulate autophagic pathways as an alternative fuel source in order to survive the bioenergetic stress imposed by limiting mtCU function (253).

3.3.3.5.7. Regulation of MCUR1.

MCUR1 protein is subject to several posttranslational modifications. Binding of Ca²⁺ to the MCUR1 head region destabilizes its three-dimensional structure (251). A large portion of the MCUR1 expressed in HEK293T cells is proteolytically cleaved, resulting in loss of the first 140 residues from the NH₂ terminus of the protein (251). This cleavage removes most of the disordered region that precedes the MCUR1 head domain (251). Both full-length and cleaved MCUR are capable of binding MCU (251, 253), but it is unknown whether these forms differ in their ability to regulate uniporter function.

MCUR1 is also sensitive to metabolic stress. MCUR1 protein is downregulated in human pulmonary microvascular endothelial cells subjected to oxygen and glucose deprivation (255). MCUR1 downregulation is associated with increased autophagic flux in microvascular endothelial cells, in agreement with earlier findings by Tomar et al. (253) (255).The finding that MCUR1 is the only mtCU component significantly affected by oxygen and glucose deprivation in this cell type (255) suggests the presence of specific regulatory pathways that act through MCUR1 to impact mtCU function. However, it should be acknowledged that such regulatory pathways could preferentially affect MCUR1 protein that participates in other mitochondrial protein complexes besides the mtCU, such as respiratory complex IV. Any impact of MCUR1 downregulation on _mCa²⁺ handling in the context of oxygen and nutrient deprivation remains to be determined. Finally, experiments in mtNOD mice with polymorphisms in the mitochondrial COX3 gene indicate that MCUR1 protein expression is also responsive to changes in the Ca²⁺ microdomain surrounding the mitochondria. Aged mtNOD mice exhibit a hyperfused and elongated mitochondrial network, in contrast to a more fragmented mitochondrial network in aged control animals (256). Niemann et al. (256) observe that MCUR1 gene expression is diminished in aged mtNOD hepatocytes and propose that this change relates to changes in mitochondrial structure. In normal aged tissue, the fragmented mitochondrial network requires a high _mCa²⁺ uptake capacity (corresponding to higher MCUR1 expression), because some mitochondria will be located far from ER Ca²⁺ release sites and will not be exposed to as high of a cytosolic/IMS Ca²⁺ concentration as those mitochondria located closer to the ER (256). These fragmented mitochondria would require robust pathways for _mCa²⁺ uptake to ensure that they still import sufficient Ca²⁺ to support mitochondrial metabolism. However, this problem is not encountered in aged mtNOD cells, as the existence of hyperfused mitochondria means that each mitochondrion is more likely to have a portion of its structure in proximity to ER Ca²⁺ release sites. Each mitochondrion will consequently be exposed to a locally high Ca²⁺ microdomain upon ER Ca²⁺ release, which can trigger sufficient _mCa²⁺ uptake for cellular needs, independent of positive mtCU regulation by MCUR1 (256). mtNOD cells would thus be able to effectively couple changes in cytosolic Ca²⁺ concentration or ER Ca²⁺ release to mitochondrial responses, while requiring a lower level of MCUR1 expression than that needed in wild-type cells.

3.4. Structure and Stoichiometry of the Proteins Within the mtCU

Several features of the regulation of _mCa²⁺ uptake emerge only from consideration of the intact uniporter complex and are not readily apparent from studies of its individual protein constituents. The net mass of native, intact protein complexes that contain MCU varies from ∼450 kDa up to ∼800 kDa (99, 242, 253). The lower end of this range aligns closely with the 4 MCU:4 EMRE:1 MICU1:1 MICU2 channel stoichiometry proposed by Fan et al. (184) and is consistent with the incorporation of additional components such as MCUR1 into the complex (FIGURE 2). The higher end of the range is consistent with the formation of dimers of mtCU uniplex channels as observed in structural studies of the uniporter (184, 186). That the higher end of the range of reported mtCU weights is not double the molecular mass of the individual uniporter (∼450 kDa) perhaps reflects the loss of MICU1/2 from some dimerized channels (184) or the displacement of accessory subunits as the uniporter forms mature dimers. Unfortunately, cryo-EM studies of uniporter structure did not include accessory subunits like MCUR1 and MCUB, so the precise stoichiometry of their incorporation into mature mtCU uniplex channels and dimers remains unresolved.

How the appropriate uniporter stoichiometry is maintained requires further investigation. Several questions regarding this matter are particularly relevant to our understanding of uniporter regulation. For example, does the MICU1-to-MCU ratio at a given uniporter channel correspond directly to the relative gene expression of MICU1 and MCU, or do additional mechanisms control whether or not MICU1 will be selected for inclusion in a complex composed of MCU + EMRE? Are there additional physiological mechanisms besides replacement of MCU with MCUB that dictate whether or not the uniporter channel binds to and is gated by the MICU proteins?

What is clear is that alteration of the normal mtCU stoichiometry is a fundamental mechanism by which cells modulate _mCa²⁺ uptake. Transcriptional control of relative MCU, MICU1, and MICU2/3 content allows for tissue specificity of _mCa²⁺ handling and the sensitivity of the mitochondria to changes in cytosolic Ca²⁺ concentration (187). Uniporter current density varies widely across tissues and tends to be low in highly metabolically active tissues like the heart, possibly to prevent excess mitochondrial buffering of cytosolic Ca²⁺ signals or to prevent _mCa²⁺ overload in the face of repeated bouts of high cytosolic Ca²⁺ (117). Temporal changes in mtCU composition within a given tissue can also mediate changes in _mCa²⁺ handling over time as part of normal biological processes. For example, the placenta and myometrium exhibit distinct changes in mtCU component expression late in pregnancy. MCU, MICU1, MICU2, and EMRE mRNA and MCU and MICU1 protein expression increase in the placenta throughout gestation, along with an increase in the MICU1-to-MCU protein ratio (257). MICU2 protein expression in the placenta then decreases in the later stages of pregnancy. This response is consistent with a greater membrane depolarization observed in response to bolus Ca²⁺ addition (257), suggestive of increased _mCa²⁺ uptake. At the same time, protein expression of MCU and MCUB increases while MICU1 protein expression decreases in the myometrium through the late stages of pregnancy (257). How these specific changes may be important for uterine or placental function is still under investigation, but they highlight that coordinated changes in the expression of uniporter components occur in response to physiological stimuli.

Changes in uniporter component expression also allow the mitochondrion to modulate its Ca²⁺ uptake in order to withstand pathological cellular stress, as is seen with upregulation of MCUB in the heart in response to I/R injury (242). The plasticity of uniporter composition is also evident from studies in which experimental manipulation of a given subunit results in a reflex up- or downregulation of another homologous subunit to maintain overall uniporter structure. For example, knockout of MCUB increases MCU and EMRE protein expression (242), and MCUB overexpression downregulates EMRE (241). Likewise, overexpression of a uniporter subunit can increase the expression of its binding partners: MCU overexpression increases the expression of MICU1 and 2, and MICU1 overexpression results in increased MICU2 expression (93). These findings reflect the existence of homeostatic control mechanisms that maintain tissue-appropriate uniporter function and may minimize the abundance of inappropriately regulated channels.

Since certain subunits like MICU1 and MICU2/3 depend on the presence of others (i.e., MCU rather than MCUB) for uniporter complex binding and regulation, changes in the composition of the core uniporter channel (MCU vs. MCUB) also have secondary effects on the gating and regulation of each uniplex mtCU. Additionally, since back-to-back MICU2 dimers impact the function of dimerized mtCUs (232), changes in the relative MCU versus MCUB content and subsequent binding versus displacement of MICU1/2 from each individual uniplex channel may have further effects on the specific behavior of dimerized channels. For example, incorporation of increasing amounts of MCUB within the each uniplex channel would limit the amount of bound MICU2 available to participate in the back-to-back dimers proposed to act as the inhibitory latch within a set of dimerized uniplex channels.

3.5. Regulators of the mtCU

Several IMM proteins that are not considered part of the mtCU nevertheless have direct impacts on uniporter function.

3.5.1. SLC25A23.

The mitochondrial ATP-magnesium/phosphate carrier SLC25A23 (also called SCaMC-3) is required for Ca²⁺-dependent accumulation of adenine nucleotides and the subsequent increase in OXPHOS in liver mitochondria (258). This activity may depend on SLC25A23’s EF-hands, which lie in the IMS and sense an increase in IMS Ca²⁺ (258, 259). Interestingly, the Ca²⁺-induced increase in SLC25A23 activity is also associated with increased mitochondrial Ca²⁺ retention capacity (258). Amigo et al. (258) suggest that this effect on Ca²⁺ retention is attributable to SLC25A23’s transport of adenine nucleotides into the mitochondria, which may facilitate the formation of Ca²⁺-phosphate precipitates. SLC25A23 also has direct effects to promote _mCa²⁺ uptake. Knockdown of SLC25A23 reduces _mCa²⁺ uptake rate and cytosolic Ca²⁺ clearance in response to histamine stimulation (260). Likewise, mutation of the SLC25A23 EF-hands reduces _mCa²⁺ uptake (260). These observations suggest that SLC25A23 promotes mtCU activity in response to increased Ca²⁺ concentration in the cytosol or IMS. SLC25A23’s effect on _mCa²⁺ uptake may be mediated by a direct protein-protein interaction with the uniporter complex, as SLC25A23 binds both MCU and MICU1 when overexpressed in COS7 cells (260). Consistent with a role for SLC25A23 in promoting _mCa²⁺ uptake, knockdown of SLC25A23 reduces mitochondrial ROS production and protects against oxidant-induced cell death, both of which are phenomena related to increased _mCa²⁺ content (260).

3.5.2. Uncoupling proteins 2 and 3.

Uncoupling proteins 2 and 3 (UCP2 and UCP3) are implicated in the regulation of _mCa²⁺ uptake. Overexpression of UCP2 or UCP3 increases and knockdown of these proteins reduces agonist-induced _mCa²⁺ uptake (261). This activity of UCP2 and UCP3 is in contrast to UCP1, which is the major player in nonshivering thermogenesis but has no impact on _mCa²⁺ uptake (261). In UCP2^−/− liver mitochondria, all ruthenium red-sensitive _mCa²⁺ uptake is ablated, indicating that UCP2 is absolutely required for uniporter function in this tissue (261). Expression of UCP2 or UCP3 alone is insufficient to confer ruthenium red-sensitive _mCa²⁺ uptake in yeast, suggesting that UCP2/3 modulate mtCU function in mammalian cells rather than conducting Ca²⁺ themselves (261). Trenker et al. (261) demonstrate that mutation of the second intermembrane loop of either UCP2 or UCP3 has a dominant-negative effect to limit _mCa²⁺ uptake, and propose that wild-type UCP2 and UCP3 form homo- and heteromultimers that normally promote mtCU function. Uniporter channels containing MCU + EMRE mediate several distinct ruthenium red-sensitive Ca²⁺ currents that are observed in electrophysiological experiments, including the extra-large mitochondrial/mitoplast Ca²⁺ current (xl-MCC) and the intermediate mitochondrial/mitoplast Ca²⁺ current (i-MCC) (262–265). Overexpression of UCP2 increases and knockdown of UCP2 decreases the open probability of xl-MCC (265). Thus, even though UCP2 is not a core uniporter component, it stimulates the form of the uniporter responsible for conducting xl-MCC. The specific molecular composition of the distinct forms of mtCU that conduct xl-MCC and i-MCC, and how these forms may be differentially regulated by proteins besides UCP2, largely remains to be determined. However, one report suggests that UCP2/3 affect the Ca²⁺ sensitivity of MICU1, and thereby regulate the activity of MICU1-gated forms of the mtCU (222).

3.6. Additional Mitochondrial Ca²⁺ Influx Pathways Distinct From the mtCU

Apart from the mtCU, several additional channels and exchangers contribute to Ca²⁺ uptake across the inner mitochondrial membrane. These proteins are candidates for the pathways that allow for more gradual _mCa²⁺ uptake compared with the mtCU and may help normalize homeostatic _mCa²⁺ content as observed in some tissues with chronic genetic mtCU disruption (126, 127). They may also contribute to the eventual Ca²⁺ loading of MCU-deficient cardiac mitochondria upon prolonged dobutamine stimulation (127). Thus, these proteins may participate in pathological _mCa²⁺ loading, particularly in chronic stress or disease states. They therefore represent attractive, alternative targets to the mtCU for therapeutic strategies that aim to prevent _mCa²⁺ overload in pathologies characterized by sustained elevations in cytosolic Ca²⁺ levels.

3.6.1. Mitochondrial ryanodine receptor.

The Sheu laboratory identified a ryanodine receptor, referred to as the “mitochondrial ryanodine receptor (mRyR),” in the IMM of cardiac mitochondria (266, 267). Cardiac mitochondria bind both the RyR ligand ryanodine and ruthenium red in a competitive manner, and their net _mCa²⁺ uptake is sensitive to inhibition by ruthenium red, ryanodine, and the type 1 ryanodine receptor (RyR1) inhibitor dantrolene (266). Both ryanodine and dantrolene inhibit mitochondrial swelling upon exposure to high cytosolic Ca²⁺ concentration, suggesting that, like the mtCU, the mRyR mediates _mCa²⁺ uptake and contributes to pathogenic _mCa²⁺ overload (266) (FIGURE 1). _mCa²⁺ uptake through the mRyR is driven by the same thermodynamic forces as Ca²⁺ flux through the mtCU: the membrane potential across the IMM (ΔΨ_m, approximately −180 mV on the matrix side of the IMM relative to the IMS; favoring influx of positively charged Ca²⁺) and the concentration gradient of free Ca²⁺ ions [∼100 nM at rest in the matrix (40, 268, 269), up to micromolar concentrations in nearby cytosolic microdomains/the IMS during cellular Ca²⁺ release events (83, 270–274)]. Subsequent genetic studies confirmed the identify of cardiac mRyR as RyR1, the isoform that is typically found in the sarcoplasmic reticulum of skeletal muscle (275). Electrophysiological characterization of purified mitochondrial membranes fused with lipid bilayers further revealed that the Ca²⁺-dependent activation of mRyR more closely resembles that of RyR1 than that of the cardiac type 2 ryanodine receptor, RyR2 (276).

The function of a mRyR as an alternative route of _mCa²⁺ uptake may be important for certain cell types, particularly in excitable tissues. Since its initial characterization in the heart, mRyR activity has also been measured in neurons, and the RyR has been detected via immunofluorescence in neuronal mitochondria (277). Acute _mCa²⁺ uptake may only be partially inhibited in MCU-null neurons (278), consistent with a role for MCU-independent _mCa²⁺ uptake pathways such as the mRyR. Similar observations of MCU-independent, albeit gradual, _mCa²⁺ uptake are reported in MCU-null MEFs (126). Whether the mRyR may be relevant to _mCa²⁺ uptake in other tissues is less clear. The original studies investigating the role of MCU in _mCa²⁺ uptake in tissue culture cell lines relied on MCU knockdown, rather than complete genetic MCU knockout, and overexpression of nonfunctional MCU mutants (98, 99, 101). These studies reported reduced but not complete loss of acute _mCa²⁺ uptake, but the relative contribution of any residual MCU expressed in these cells, as opposed to alternative pathways such as the mRyR, to the residual _mCa²⁺ uptake is indiscernible. Relevant questions for future study are whether the extent of MCU-independent _mCa²⁺ uptake observed in different tissues is correlated with the relative expression and mitochondrial localization of mRyR and whether deletion of mRyR on MCU-null backgrounds is sufficient to ablate residual _mCa²⁺ uptake in cell types such as cardiomyocytes and neurons.

To date, studies of the function and physiological effects of the putative mRyR have mostly been carried out in cardiac mitochondria or cardiomyocytes by a single laboratory. Whereas _mCa²⁺ uptake following exposure of cardiac mitochondria to 10 µM Ca²⁺ is sensitive to both ryanodine and dantrolene, _mCa²⁺ uptake is less sensitive to these inhibitors at higher (50 µM) extramitochondrial Ca²⁺ concentration. This result suggests that mRyR may be most relevant for _mCa²⁺ uptake at lower cytosolic Ca²⁺ concentrations and less important at higher cytosolic Ca²⁺ concentrations, where its contribution to net _mCa²⁺ uptake may be masked by a greater relative contribution of _mCa²⁺ uptake through the mtCU (275). The observation that IP₃ receptor-dependent Ca²⁺ release from the ER/SR, but not global cytosolic Ca²⁺ elevation under adrenergic stimulation, elicits dantrolene-sensitive, mRyR-dependent _mCa²⁺ uptake suggests that mRyR is particularly responsive to changes in Ca²⁺ concentration in local microdomains adjacent to cellular Ca²⁺ release sites (279).

Like the mtCU, mRyR contributes to Ca²⁺-dependent regulation of the TCA cycle and oxidative phosphorylation. Acute mRyR inhibition with ryanodine attenuates the increase in oxygen consumption induced by incubating isolated cardiac mitochondria or whole heart lysates with 10 µM bath Ca²⁺, although it does not affect basal oxygen consumption (275). The view of these channels’ roles in _mCa²⁺ uptake and metabolic regulation is more complicated, though, when mRyR versus mtCU function is studied within their native environment. Experiments comparing local IP₃ receptor-dependent Ca²⁺ release from the ER/SR to global cytosolic Ca²⁺ elevation with β-adrenergic stimulation in intact cardiomyocytes reveal differences in _mCa²⁺ uptake under these two conditions. IP₃ receptor activation by endothelin 1 stimulates ER/SR Ca²⁺ release, _mCa²⁺ uptake through the mRyR, and an increase in ATP production in both resting and electrically paced cardiomyocytes (279). In contrast, global cytosolic Ca²⁺ elevation with isoproterenol treatment stimulates _mCa²⁺ uptake through the mtCU, and this increases ATP production only in electrically paced and not quiescent cells (279). These results may reflect particular enrichment of mRyR versus the mtCU at distinct mitochondrial subdomains (i.e., at regions closer to or farther from ER Ca²⁺ release sites) or additional regulatory signals that differentially modulate the mtCU and mRyR in response to G_s- versus G_q-coupled receptor signaling. They also suggest that mRyR is relatively more important than the mtCU in maintaining basal mitochondrial ATP production, whereas the mtCU contributes more to enhancing mitochondrial ATP production when cellular workload and ATP consumption are increased. A role for mRyR in maintaining basal mitochondrial ATP production in unstressed cells could help to explain why the hearts of animals with adult cardiomyocyte-specific MCU deletion do not have any baseline defects in metabolism or cardiac contractility (126, 127, 279). In this model, mRyR would help to maintain homeostatic _mCa²⁺ uptake and mitochondrial metabolism, while the mtCU is specifically activated during periods of increased energetic demand. Mathematical modeling further suggests that the relative balance of cytosolic Ca²⁺ and Mg²⁺ concentrations may shift the balance between mtCU- and mRyR-dependent _mCa²⁺ uptake. It also reinforces the notion that mRyR-dependent uptake is favored under basal, physiological conditions (280).

Heart homogenates from neonatal RyR1-deficient mice exhibit elevated basal oxygen consumption, but their oxygen consumption is not increased further in response to an elevation in extramitochondrial Ca²⁺ (275) (TABLE 2). Overexpression of mRyR (RyR1) but not RyR2 or MCU in cardiac H9c2 myoblasts yields the opposite effect, enhancing _mCa²⁺ transients and augmenting mitochondrial ATP production triggered by an increase in cytosolic Ca²⁺ concentration (281). These observations suggest that mRyR has some physiological redundancy with the mtCU and, like the uniporter, contributes to Ca²⁺-dependent parallel activation of oxidative metabolism and myofilament cross-bridge cycling. They also highlight mRyR-dependent _mCa²⁺ uptake as a mechanism that may partially compensate for loss of uniporter function in order to maintain basal metabolism and _mCa²⁺ content in animal models with mtCU disruption, as well as to gradually increase mitochondrial Ca²⁺ concentration and oxygen consumption in response to prolonged elevations in cytosolic Ca²⁺ concentration, even in the absence of MCU. It will be informative for future studies to examine how disruption of mtCU function specifically affects the expression and activity of proteins such as mRyR/RyR1 that mediate alternative routes of _mCa²⁺ entry.

Two ion exchangers of the IMM that typically function in _mCa²⁺ efflux, NCLX and LETM1, exhibit reversible activity and so can mediate _mCa²⁺ uptake under appropriate cellular conditions.

3.6.2. NCLX in _mCa²⁺ uptake.

NCLX is the mitochondrial Na⁺/Ca²⁺ exchanger that is primarily responsible for Na⁺-linked _mCa²⁺ efflux (282). This forward-mode NCLX activity is discussed below in sect. 3.7. Mitochondrial Na⁺/Ca²⁺ exchange can occur in the reverse direction under nonrespiring conditions or perhaps pathological states where mitochondrial membrane potential is depolarized (283), leading to _mCa²⁺ uptake. Here, if cytosolic Ca²⁺ concentration becomes elevated relative to free matrix Ca²⁺ [∼100 nM at rest (40, 268, 269)], reverse-mode NCLX activity can allow the influx of 1 Ca²⁺ ion across the IMM in exchange for the efflux of 3 Na⁺ ions. Note that the resting concentration gradient for Na⁺ is relatively small [∼8 mM in cytosol and ∼6 mM in the mitochondrial matrix (21)] and so may offer minimal opposition to reverse-mode flux through NCLX compared with the driving force for this flux that could be provided by a large cytosolic-mitochondria Ca²⁺ gradient. Such reverse-mode flux through NCLX is electrogenic, resulting in the net loss of 1 positive charge from the matrix. Thus, under normal conditions where ΔΨ_m is maintained at approximately −180 mV, this potential tends to oppose reverse-mode NCLX activity.

Fitting with this model, when rat cardiomyocytes are subjected to metabolic inhibition mimicking hypoxia, reverse-mode Na⁺/Ca²⁺ exchange mediates influx of Ca²⁺ into the matrix (284). Reverse-mode mitochondrial Na⁺/Ca²⁺ exchange also occurs in metabolically inhibited kidney epithelial cells (285). The NCLX inhibitor CGP-37157 attenuates this reverse Na⁺/Ca²⁺ exchange, supporting the notion that it occurs via NCLX (286) Although forward-mode NCLX exchange would be favored under physiological conditions where ΔΨ_m is intact, reverse-mode NCLX exchange may have pathological consequences, such as contributing to _mCa²⁺ overload that occurs in conjunction with impaired mitochondrial metabolism during hypoxia and ischemia (287). However, a limitation of these studies of reverse-mode mitochondrial Na⁺/Ca²⁺ exchange is that they typically rely on the direct manipulation of ion gradients or the use of pharmacological inhibitors such as CGP-37157, which is not strictly specific to NCLX. Further validation of the purported role of NCLX in reverse-mode mitochondrial Na⁺/Ca²⁺ exchange via approaches that genetically target NCLX is therefore needed. Indeed, knockdown of NCLX is associated with a slight increase in _mCa²⁺ uptake rather than any appreciable decrease, arguing against NCLX operating in reverse mode (166).

3.6.3. LETM1 in _mCa²⁺ uptake.

LETM1 is a reversible IMM Ca²⁺/H⁺ antiporter (288). LETM1’s function in _mCa²⁺ efflux is discussed in detail below; here we focus on its role in _mCa²⁺ uptake. LETM1 mediates _mCa²⁺ uptake and mitochondrial H⁺ extrusion at cytosolic Ca²⁺ concentrations < 1 µM, where mtCU-dependent Ca²⁺ uptake is minimal (288, 289). At cytosolic Ca²⁺ concentrations > 1 µM, LETM1-independent _mCa²⁺ uptake predominates and is not coupled to mitochondrial H⁺ extrusion, consistent with _mCa²⁺ uptake through the mtCU (288). LETM1 has a specific role in _mCa²⁺ uptake following influx of Ca²⁺ across the plasma membrane rather than in response to Ca²⁺ release from intracellular stores (289, 290). This contrast with mtCU-mediated _mCa²⁺ uptake reflects LETM1’s high affinity for Ca²⁺, which makes _mCa²⁺ uptake through LETM1 less reliant on high-Ca²⁺ microdomains than _mCa²⁺ uptake through the mtCU (287). The existence of high- and low-affinity modes of _mCa²⁺ uptake enables mitochondria to differentiate slow, sustained increases in cytosolic Ca²⁺, which trigger _mCa²⁺ uptake through high-affinity pathways like LETM1 and mRyR/RyR1, from high-amplitude cytosolic Ca²⁺ oscillations, which trigger _mCa²⁺ uptake through lower-affinity pathways such as the mtCU (287).

Jiang et al. (288) propose an electrogenic 1 Ca²⁺:1 H⁺ stoichiometry for Ca²⁺/H⁺ exchange through LETM1. Consistent with this model, LETM1-dependent _mCa²⁺ uptake is driven by ΔΨ_m and favored by the Ca²⁺ concentration gradient [∼100 nM free resting matrix Ca²⁺, up to micromolar cytosolic Ca²⁺ (40, 83, 268–274)], but it is limited by the proton gradient [matrix pH ∼7.7–8.2; cytosolic/IMS pH ∼7-7.2 (21, 291–294)] across the IMM (288). The highly negative ΔΨ_m (approximately −180 mV) that exists under physiological conditions favors _mCa²⁺ uptake, although, assuming electrogenic uptake of Ca²⁺ through LETM1, equilibrium would be reached once matrix Ca²⁺ concentration reaches ∼10 µM (287). As a result, _mCa²⁺ uptake through an electrogenic LETM1 should not be able to cause pathological _mCa²⁺ overload (287). This suggests that whereas the mtCU contributes to both physiological _mCa²⁺ uptake and pathological _mCa²⁺ overload, LETM1 is most relevant in physiological _mCa²⁺ uptake.

3.7. Mitochondrial Ca²⁺ Efflux Pathways

Evidence for regulated, physiological efflux of Ca²⁺ from the mitochondria dates back to the 1970s and 1980s. Carafoli et al. (7) first reported that Na⁺ addition stimulates the release of Ca²⁺ from isolated heart mitochondria. Lithium also stimulates some _mCa²⁺ efflux, although other cations such as potassium, rubidium, cesium, and magnesium do not (7). Na⁺-induced _mCa²⁺ efflux persists in the presence of ruthenium red, so it does not occur through the mtCU and is not coupled to transport of H⁺ across the IMM (8). Crompton et al. (8) propose a mechanism for _mCa²⁺ cycling in which Ca²⁺ enters the matrix through a ruthenium red-sensitive influx pathway (the mtCU) and then is recycled via a Na⁺-dependent efflux pathway. Na⁺-coupled _mCa²⁺ efflux does not strictly require the mitochondria to be energized, although it occurs at a greater rate in energized mitochondria. Crompton et al. (8) therefore postulated the existence of a mitochondrial sodium/lithium-calcium antiporter and suggested that it operates with an electrogenic 3 Na⁺:1 Ca²⁺ stoichiometry. Bernardi and Azzone and colleagues (295–297) further suggested the presence of an electrogenic mitochondrial H⁺/Ca²⁺ exchanger that drives _mCa²⁺ efflux and is regulated by ΔΨ_m. Together, these mechanisms of regulated _mCa²⁺ efflux reflect the activities of at least two IMM cation exchangers: the Na⁺,Li⁺/Ca²⁺ exchanger, NCLX, and the H⁺/Ca²⁺ exchanger, LETM1 (FIGURE 1).

3.7.1. NCLX in _mCa²⁺ efflux.

NCLX (mitochondrial Na⁺/Ca²⁺,Li⁺ exchanger; gene name = SLC8B1, previously annotated as NCKX6 or NCLX) was first isolated as a 110-kDa mitochondrial protein that catalyzes Na⁺/Ca²⁺ exchange when reconstituted into proteoliposomes (298). Palty et al. (299) later identified the human gene encoding this particular exchanger and showed that the protein it encodes mediates K⁺-independent Na⁺/Ca²⁺ exchange. NCLX is unique in that it is the only member of the Na⁺/Ca²⁺ exchanger (NCX) family that also mediates Li⁺/Ca²⁺ exchange (299). NCLX was later detected specifically within the cristae of mitochondria (282). Its identity as the principal mitochondrial Na⁺/Ca²⁺ exchanger was confirmed by findings that NCLX overexpression enhances mitochondrial Na⁺-dependent Ca²⁺ efflux and that NCLX knockdown diminishes this activity (282). Moreover, Ca²⁺ transport through NCLX is activated by Li⁺ and is inhibited by the benzothiazepine CGP-37157, an accepted inhibitor of mitochondrial Na⁺/Ca²⁺ exchange (282). These results support NCLX as the primary mitochondrial Na⁺/Ca²⁺ exchanger. Fitting with such a fundamental function, NCLX is widely expressed across tissues, with particularly high expression in the pancreas, skeletal muscle, and stomach (298–300). In cardiomyocytes, the rate of _mCa²⁺ efflux is much slower than the rate of _mCa²⁺ uptake through the mtCU. _mCa²⁺ efflux is faster and roughly equal to _mCa²⁺ uptake rate in skeletal muscle, which has greater NCLX expression than cardiomyocytes. NCLX thus represents the rate-limiting component of _mCa²⁺ exchange (282), meaning that changes in NCLX expression and activity exert a major influence over _mCa²⁺ homeostasis.

3.7.1.1. nclx phylogenetic conservation.

Human and mouse NCLX share 83% amino acid identity (300). NCLX diverged early in the evolutionary history of mammalian sodium/calcium exchanger genes and forms its own subfamily, separate from the NCX1–3 and NCKX1–4 subfamilies (illustrated in Ref. 300). Phylogenetic analysis reveals the presence of cation/Ca²⁺ exchanger proteins related to NCLX in protozoa, fungi, plants, invertebrates, and mammals (301).

3.7.1.2. nclx structure.

Full-length NCLX protein is 584 residues long in humans and 585 residues long in the mouse, with a predicted mass of 64 kDa (300). NCLX contains 13 transmembrane regions, a NH₂-terminal sequence variably identified as a mitochondrial targeting sequence or signal peptide, and two consensus glycosylation sites on the extracellular loop between transmembrane regions M0 and M1 (300). NCLX contains a large regulatory loop region that contains two potential PKA phosphorylation sites and lies between transmembrane regions M5 and M6 (300). Whether NCLX is oriented so that this loop lies within the matrix or within the IMS has not yet been definitively demonstrated.

The structural features of NCLX resemble the structure of plasma membrane Na⁺/Ca²⁺ exchangers and plasma membrane Na⁺/Ca²⁺,K⁺ exchangers (301). NCLX also shares the two α-repeat domains (designated α1 and α2) that are typical of NCX proteins. These repeats are located in the clusters of hydrophobic transmembrane domains located on either side of the loop region, the first consisting of transmembrane regions M2 and M3 and the other consisting of transmembrane regions M9 and M10 (300). These α-repeats are hypothesized to form the ion binding sites required for ion exchange (301).

NCLX is detected as ∼70-kDa and ∼55-kDa proteins upon Western blotting, and Palty et al. (299) ascribe these bands to full-length NCLX and a shorter splice variant, s-NCLX. Others suggest that the 55-kDa form of NCLX represents the full-length variant, but that its protein is subject to protranslational cleavage that reduces its molecular mass (300). NCLX can also appear as a ∼100-kDa band, which likely represents dimers of the ∼50-kDa NCLX monomers noted above (282, 299). A technical challenge facing studies of NCLX structure and function, which is augmented by the apparent diversity of NCLX proteoforms, is that the research community at large suffers from a lack of specific, genetically validated, and reliably available antibodies for NCLX. Alternative methodologies for detecting, isolating, and manipulating NCLX protein are greatly needed. Care must also be taken in assessing the reliability of any reported literature that depends solely on the performance of NCLX-directed antibodies or pharmacological targeting of NCLX.

3.7.1.3. alternative splice/transcript variants of nclx.

A ∼60-kDa mouse NCLX variant results from the skipping of exons 13 and 14, which removes 149 nucleotides, causing a frame shift. The resultant protein is 552 amino acids long and has a unique COOH terminus. This short form exhibits K⁺-dependent Na⁺/Ca²⁺ exchange, whereas full-length NCLX can mediate Na⁺/Ca²⁺ exchange independent of K⁺ (300). The 58-kDa NCLX variant identified in humans results from the exclusion of exon 7. This results in an in-frame deletion and disruption of the 3rd and 4th transmembrane regions (M3 and M4), but has no impact on NCLX ion exchanger activity (299).

3.7.1.4. properties and function of nclx.

NCLX is proposed to operate with 3 Na⁺:1 Ca²⁺ stoichiometry and like other Na⁺/Ca²⁺ exchangers mediate electrogenic ion exchange (302–305). Physiological conditions [ΔΨ_m approximately −180 mV, free matrix and cytosolic Ca²⁺ ∼100 nM, matrix Na⁺ ∼6 mM, and cytosolic Na⁺ ∼8 mM (16, 21, 37, 40, 268, 269)] will favor forward-mode operation of NCLX, with _mCa²⁺ efflux and mitochondrial Na⁺ influx resulting in the net flow of 1 positive charge into the matrix. This ion exchange occurs via a ping-pong mechanism, in which Na⁺ and Ca²⁺ compete for binding to 4 potential ion-binding sites within the exchanger and are transported in separate steps. The ion binding sites are alternately exposed to opposite sides of the IMM, with the ion of highest local abundance outcompeting the other for access to the binding sites and subsequent transport across the IMM (306, 307). This model is consistent with the observed electrogenicity and voltage dependence of mitochondrial Na⁺/Ca²⁺ exchange (286). NCLX also exhibits sensitivity to pH and is inhibited at basic extramitochondrial pH, which increases the K_m of NCLX for Ca²⁺ (308). Although K⁺ is not required for NCLX activity (299), K⁺ can stimulate mitochondrial Na⁺/Ca²⁺ exchange because it reduces the exchanger’s K_m for Ca²⁺ (308).

Both α-repeats are not required within the same molecule of NCLX in order for it to catalyze ion exchange. Instead, NCLX may function as a dimer or trimer, in which the contribution of a single α-repeat from each subunit is sufficient for ion transport (309). This model agrees with the detection of ∼100-kDa dimers in Western blots for NCLX (282). That NCLX functions as an oligomer is further supported by the finding that expression of a nonfunctional mutant NCLX exerts a dominant-negative effect to inhibit the activity of single α-repeat forms of NCLX (309). These findings imply that short NCLX splice variants with disruption of either the α1 or α2 domain should still be capable of Na⁺/Ca²⁺ exchange when dimerized with an additional copy of NCLX.

NCLX catalyzes Na⁺/Ca²⁺ exchange and Li⁺/Ca²⁺ exchange at similar rates (299). NCLX S468 is highly conserved among NCX and NCKX exchangers and is critical for _mCa²⁺ transport by NCLX (282, 309). N149, P152, D153, and G176 within the NCLX α1 domain and N467, S468, G494, and N498 within the α2 domain are required for full _mCa²⁺ exchange activity (306). N149A, P152A, D153A, N467Q, S468T, and G494S mutations do not affect Li⁺/Ca²⁺ exchange but disrupt Na⁺/Ca²⁺ exchange, suggesting that these sites are critical specifically for Na⁺-mediated Ca²⁺ transport (306). In contrast, D471 is critical for Li⁺/Ca²⁺ exchange but is dispensable for Na⁺/Ca²⁺ exchange (306). Simultaneous disruption of four critical Ca²⁺-binding residues (N149A, D153A, N467A, D471A) is required to fully disrupt all Ca²⁺ exchange by NCLX (306). Nine out of 12 ion coordinating residues differ between the four ion binding sites of NCLX and archaeal NCX (307, 310). Replacement of these nine residues in NCX with the corresponding residues from NCLX (N159, G150, D153, A177, V181, N467, D471, G494, N498) allows the mutant NCX to catalyze Li⁺/Ca²⁺ exchange, suggesting that some of these sites confer the Li⁺-transporting activity of NCLX (307). Refaeli et al. (307) propose that this difference in Li⁺ binding in NCLX versus NCX is also related to the fact that all four of the residues that coordinate Ca²⁺ at the S_Ca ion binding site (N149, D153, D471, and N467 in NCLX) differ between these two proteins. This particular structural difference may induce structural rearrangements in the other three ion binding sites that ultimately affect the proteins’ ability to bind Na⁺ versus Li⁺ (307). Since intracellular Li⁺ concentration is low (0.6–0.8 mM) and far below NCLX’s K_d for Li⁺, whether NCLX actually transports Li⁺ under physiological conditions (306) and what functional consequences this Li⁺ transport may have for the cell are pertinent unanswered questions.

3.7.1.5. special roles of nclx in physiology.

NCLX’s role in _mCa²⁺ efflux allows it to shape cellular processes that depend on mitochondrial or cytosolic Ca²⁺ signals. _mCa²⁺ efflux through NCLX can limit _mCa²⁺ accumulation and so limit TCA cycle flux, NADH production, and the rate of oxidative phosphorylation (311). This means that fluctuations in cytosolic Na⁺ can indirectly impact metabolism and influence the mitochondria’s ability to buffer cytosolic Ca²⁺. This effect is apparent in tetanus-induced synaptic potentiation, where increased cytosolic Na⁺ drives _mCa²⁺ release, and this Ca²⁺ stimulates neurotransmitter release at neuromuscular synapses (312). In a similar fashion, NCLX contributes to cytosolic Ca²⁺ signals that cause pancreatic β-cells to secrete insulin after stimulation with glucose (313). NCLX-dependent _mCa²⁺ efflux is also essential for maintaining Ca²⁺ recycling from the mitochondria to the SR in HL-1 cardiomyocytes, and disruption of this process diminishes cardiomyocyte automaticity (314).

NCLX-dependent changes in _mCa²⁺ content affect mitochondrial ROS production, which itself affects cellular Ca²⁺ homoeostasis. During store-operated Ca²⁺ entry (SOCE), elevated cytosolic Ca²⁺ is transmitted to the mitochondria, causing a spike in ROS production. Without NCLX to extrude _mCa²⁺, excessive mitochondrial ROS causes oxidation and inactivation of Orai1. This prematurely terminates SOCE and limits the filling of intracellular Ca²⁺ stores (315). NCLX is likewise implicated in control of Ca²⁺-dependent signals relevant for diverse processes including nociceptive signaling (316) and B lymphocyte chemotaxis (317). Finally, recent work demonstrates that NCLX is necessary to limit _mCa²⁺ accumulation that would otherwise lead to permeability transition and cell death during processes involving increased cytosolic Ca²⁺ signaling, such as during the adrenergic stimulation of brown fat that triggers nonshivering thermogenesis (318).

3.7.1.6. genetic diseases associated with nclx.

No primary mutations in NCLX are yet reported to cause human disease, but altered NCLX function has been implicated in the pathogenesis of several neurological and neuromuscular disorders. Loss of frataxin in Friedreich’s ataxia causes a secondary increase in cytosolic Ca²⁺ concentration in neurons (319) and cardiomyocytes (320). Elevated Ca²⁺ activates calpains that degrade NCLX and thereby reduce _mCa²⁺ efflux activity (321, 322). As a result, mitochondria overload with Ca²⁺ and become susceptible to permeability transition, leading to cell death. Downregulation of NCLX is also implicated in Alzheimer’s disease and Parkinson’s disease (323, 324).

3.7.1.7. animal models of nclx.

The first NCLX loss-of-function model, developed by our laboratory, is a mouse in which exons 5–7 of Slc8b1 are floxed (TABLE 2). Tamoxifen-inducible deletion of Nclx in the adult cardiomyocytes of Slc8b1^fl/fl × αMHC-MerCreMer mice reduces _mCa²⁺ efflux and results in _mCa²⁺ overload, leading to excessive ROS production, necrotic cell death, and rapidly progressing heart failure (166). Eighty-seven percent of these Nclx conditional-knockout animals die within 2 wk of tamoxifen administration, reflecting the critical requirement for NCLX within the heart. Cardiac mitochondria from these animals exhibit increased permeability transition (PT) and swelling following a Ca²⁺ challenge, and already tend to be swollen at baseline, suggesting that they are already overloaded with Ca²⁺. EMRE protein expression decreases somewhat in Nclx-conditional knockout hearts, possibly as a compensatory response to reduce _mCa²⁺ uptake and limit _mCa²⁺ overload (166). Deletion of the mPTP component cyclophilin D rescues lethality, contractile function, cardiac hypertrophy, and mitochondrial structural defects in mice with acute cardiomyocyte Nclx deletion (166). This confirms that these defects result in large part from _mCa²⁺ overload, PT, and cardiomyocyte death. Despite the critical role of NCLX in maintaining cardiomyocyte _mCa²⁺ homeostasis and survival, Slc8b1^fl/fl mice crossed to the αMHC-Cre driver to enable constitutive deletion of Nclx in the cardiomyocytes are viable because of undefined compensatory responses (166). Cardiomyocytes from these animals have a profound reduction in _mCa²⁺ efflux rate but no change in cytosolic Ca²⁺ handling. _mCa²⁺ uptake is also somewhat reduced in these cells, even though the expression of mtCU components is not changed, suggesting that some aspect of their posttranslational regulation is altered in the absence of NCLX. Our laboratory also developed mice with cardiac-specific, doxycycline-controlled NCLX overexpression (TRE-NCLX × αMHC-tTA mice) that exhibit enhanced _mCa²⁺ efflux and increased resistance to PT (166). Cardiomyocyte NCLX overexpression in these animals reduces infarct size and cell death following cardiac ischemia-reperfusion injury, and protects contractile function, reduces cardiomyocyte superoxide production, and minimizes pathological cardiac remodeling following permanent myocardial infarction (166). Together, these models indicate that NCLX is the predominant route of _mCa²⁺ efflux within the heart and is required for normal mitochondrial function and cell survival. They also suggest that mitochondrial Ca²⁺ cycling occurs continuously within the heart, such that disruption of _mCa²⁺ efflux quickly results in pathological _mCa²⁺ overload.

Our laboratory recently used these genetic models of NCLX to examine its contribution to the progression of Alzheimer’s disease (AD). NCLX^fl/fl mice were crossed to mice with neuronal-restricted Camk2a-Cre to delete NCLX from the forebrain and then crossed to the 3xTg-AD mouse model of AD (323). Loss of NCLX accelerates impairments in spatial working memory, contextual recall, and cued recall in 3xTg-AD mice. These defects are associated with increased amyloid plaque burden and tau hyperphosphorylation (323). These findings suggest that the reduced NCLX expression observed in AD patients contributes to AD progression. Rescue experiments crossed the 3xTg-AD line to mice with neuronal-specific, doxycycline-controlled NCLX overexpression (3xTg-AD × TRE-NCLX × Camk2a-tTA) to investigate the impact of increased neuronal NCLX activity on AD. Here, increased NCLX expression protects against age-related cognitive decline and reduces amyloid plaque burden and tau pathology in AD mice (323). NCLX overexpression also reduces the susceptibility to mitochondrial PT and limits superoxide formation and lipid peroxidation in 3xTg-AD brains (323). These animal studies and reports of altered NCLX expression or activity in neurodegeneration and heart failure suggest that NCLX could be a powerful therapeutic target for human diseases driven by _mCa²⁺ overload.

Recently, the Trebak laboratory studied a Slc8b1-knockout mouse model generated via CRISPR/Cas9 to produce constitutive, global disruption of NCLX (325). Interestingly, these animals are viable, pointing to compensatory mechanisms during development that minimize the deleterious effects of NCLX ablation, although it should be noted that loss of NCLX protein expression is not demonstrated in these animals. Pathak et al. (325) used this model to investigate the link between colorectal cancer and the downregulation of NCLX expression observed in human colorectal tumors. NCLX-knockout mice subjected to the colitis-induced colorectal cancer model develop fewer and smaller colorectal tumors than wild-type control mice. Likewise, NOD-SCID mice xenografted with NCLX-knockout human HCT116 colorectal cancer cells exhibit less tumor growth than animals receiving control HCT116 colorectal cancer cells (325). Surprisingly, though, mice receiving NCLX-knockout cells have increased metastasis and reduced overall survival, suggesting that NCLX has distinct roles in tumor cell proliferation versus tumor cell migration or invasion in vivo. Further in vitro work suggests that loss of NCLX in colorectal cancer cells depolarizes mitochondria, disrupts mitochondrial morphology, and increases ROS production. These changes are associated with prometastatic transcriptional reprogramming, chemoresistance, and stabilization of HIF-1α, which results in increased glycolysis that drives enhanced cellular migration (325). These findings suggest that targeting NCLX as a therapy for cancer is a complicated objective, as manipulating NCLX activity in this context could have both desirable and deleterious outcomes.

Null mutation of the C. elegans NCLX-type gene, ncx-9, causes defects in axon guidance within the GABAergic motor neuron circuit, leading Sharma et al. (326) to conclude that NCLX is critical for neural circuit patterning. Beyond its critical function in adult organisms, then, NCLX also appears to have important roles during development, even if it is not strictly required for viability. It will be interesting for future studies to determine whether any such developmental defects also occur within the brain or other tissues in mice with constitutive loss of NCLX. Continued research into NCLX will likely uncover more roles for this protein in normal physiology and additional consequences of altered NCLX function in organs other than the heart and brain.

3.7.1.8. regulation of nclx.

Na⁺/Ca²⁺ exchange through NCLX is electrogenic and therefore sensitive to the mitochondrial membrane potential (ΔΨ_m). Furthermore, NCLX protein is itself allosterically regulated by membrane potential, with small changes in ΔΨ_m that are insufficient to affect _mCa²⁺ uptake instead altering _mCa²⁺ efflux. Mild mitochondrial depolarization inhibits _mCa²⁺ efflux through NCLX (327). This effect is dependent on a cluster of positively charged arginine and lysine residues (R253, R255, R256, and K325) located near the interface between the regulatory loop of NCLX and the membrane. Kostic et al. (327) propose that these charged residues respond to changes in membrane potential by changing the interaction of NCLX’s regulatory loop with the rest of the exchanger. They suggest that such allosteric inhibition of NCLX allows the mitochondria to respond to metabolic stress (partial depolarization of ΔΨ_m) by increasing _mCa²⁺ content, which in turn would stimulate the TCA cycle and oxidative phosphorylation. As mitochondrial metabolism rises to meet cellular demand, membrane potential recovers, relieving the inhibition of NCLX and allowing _mCa²⁺ efflux to resume (327). Phosphorylation of NCLX S258, which lies near the positively charged residues involved in this mechanism [see Kostic et al. (327) for an illustration], overrides the ability of membrane depolarization to inhibit NCLX, possibly because it changes the net charge in this region of the protein. Alternatively, S258 phosphorylation may change the nature of the interactions between this residue and R253, R255, R256, and K325, leading to secondary conformational changes in the catalytic α-domains of NCLX that favor ion exchange (327).

Phosphorylation of NCLX S258 or phosphomimetic S258D mutation rescues impaired NCLX exchanger activity in PINK1-deficient neurons, a model for Parkinson’s disease (328). Pharmacological activation of PKA is also sufficient to stimulate NCLX activity in this context, and S258 of NCLX lies within a putative PKA phosphorylation site. Furthermore, purified NCLX is phosphorylated at S258 when incubated with the PKA catalytic subunit, suggesting that PKA is capable of directly phosphorylating NCLX at this site to stimulate its activity (328). Knockout or inhibition of the leucine-rich repeat kinase 2 (LRRK2) impairs _mCa²⁺ efflux through NCLX (324), although whether LRRK2 directly regulates NCLX activity is unclear. Phosphorylation of NCLX at the putative PKA-regulated site, S258, recovers NCLX activity in LRRK2-deficient cells (324), but whether PKA-dependent phosphorylation of NCLX represents a downstream step in LRKK2-dependent signaling or rather is an independent regulatory event remains to be determined.

How NCLX may be regulated by protein-protein interactions is another area of active investigation. The activity of NCX family proteins is regulated by other membrane and membrane-associated proteins such as CRMP2 and 14-3-3 proteins (329, 330). 14-3-3 proteins bind to the cytoplasmic loops of NCX proteins and inhibit their activity (330). Since NCLX contains an analogous regulatory loop, this may be a site at which binding partners can modulate NCLX function. The antiapoptotic protein Bcl-2 inhibits mitochondrial Na⁺/Ca²⁺ exchange and thereby enhances _mCa²⁺ content and oxidative phosphorylation (331). Bcl-2 thus is a putative candidate that may regulate NCLX activity via a direct physical interaction. As noted above, NCLX appears to be degraded by calpains in Friedreich’s ataxia (321, 322). Recent work suggests that calpains may also be responsible for downregulation of NCLX activity after traumatic brain injury (332). Thus, proteolytic cleavage of NCLX is yet another mechanism that can control net cellular NCLX activity.

3.7.1.9. pharmacological modulation of nclx.

A number of compounds including tetraphenylphosphonium (TPP⁺) and benzodiazepines inhibit mitochondrial Na⁺/Ca²⁺ exchange and may act directly on NCLX (333, 334). Several Ca²⁺ antagonists such as diltiazem, prenylamine, fendiline, nifedipine, and verapamil also inhibit mitochondrial Na⁺/Ca²⁺ exchange (335), although their lack of specificity toward NCLX limits their utility as NCLX modulators. Likewise, KB-R7942 shows partial inhibition of NCLX but also affects Na⁺/Ca²⁺ exchangers (299). The benzothiazepine CGP-37157 inhibits Na⁺/Ca²⁺ exchange in cardiac mitochondria without affecting the L-type calcium channel, plasma membrane NCX proteins, or the Na⁺-K⁺-ATPase (336). CGP-37157 directly inhibits NCLX (282) and so has become a widely accepted experimental tool to control NCLX activity. However, some reports indicate that CGP-37157 also inhibits other proteins involved in cellular Ca²⁺ handling including plasma membrane Ca²⁺ channels (337, 338), SERCA (339), LETM1 (288), and even, in contrast to other reports, the L-type calcium channel (340). Experiments using intact or permeabilized cells should consider potential effects of CGP-37157 on these other targets. CGP-37157 also causes Ca²⁺ leak through the RyR in both cardiac and skeletal muscle, further complicating interpretation of experiments using this drug (339). Possible effects of CGP-37157 on the RyR in cell types where the presence of a mitochondrial RyR is suspected (see above) should also be taken into account when using this drug. There is therefore a great need for the development of more specific pharmacological NCLX inhibitors. No pharmacological activators of NCLX are currently known. Given the attractiveness of enhancing NCLX activity as a strategy to treat human disease, the development of potent and specific NCLX activators is another high-priority goal for the _mCa²⁺ field.

3.7.2. LETM1 in _mCa²⁺ efflux.

LETM1 (Leucine Zipper and EF-Hand Containing Transmembrane Protein 1; gene name = LETM1 or SLC55A1) was initially identified as a gene that is deleted in the majority of patients with the developmental disorder Wolf–Hirschhorn syndrome (WHS) (341). The presence of two EF-hands in LETM1 hints at its potential role in intracellular Ca²⁺ signaling (341), and its localization to the mitochondria suggests that some neuromuscular features of WHS are related to mitochondrial defects (342). LETM1 is broadly expressed, with transcripts detected in the heart, brain, placenta, lung, liver, skeletal muscle, kidney, and pancreas (341). Within the mitochondria, LETM1 is specifically localized to the IMM (288, 343, 344). LETM1 is also detected in the endoplasmic reticulum (ER), and at least one report suggests that it may contribute to ER Ca²⁺ exchange (345).

Early studies proposed LETM1 as a mitochondrial K⁺/H⁺ exchanger (346). The Drosophila homolog of LETM1, CG4589, was identified specifically as a mitochondrial Ca²⁺/H⁺ exchanger in an RNA interference screen for genes affecting _mCa²⁺ transport (288). When permeabilized Drosophila cells incubated in Na⁺-free solution to limit mitochondrial Na⁺/Ca²⁺ and Na⁺/H⁺ exchange are exposed to low (<1 µM) Ca²⁺, mitochondria take up Ca²⁺ in exchange for extrusion of H⁺ from the matrix (288). This activity is inhibited by knockdown of LETM1, suggesting that LETM1 mediates _mCa²⁺ uptake under low-cytosolic Ca²⁺ conditions (288). Such _mCa²⁺ uptake/H⁺ extrusion through LETM1 is normally limited by the pH gradient across the IMM [pH ∼7.7–8.2 in the matrix; pH ∼7-7.2 in the cytosol/IMS (21, 291–294)], thus preventing free _mCa²⁺ concentration from reaching equilibrium under physiological conditions (288). Fitting with this model, reduction of cytosolic Ca²⁺ concentration or acidification of the cytosol reverses LETM1 transport, causing _mCa²⁺ extrusion coupled to mitochondrial H⁺ (_mH⁺) influx (288). Human cell lines overexpressing LETM1 exhibit augmented pH-driven _mCa²⁺ uptake and _mH⁺ extrusion. Conversely, LETM1 knockdown in HeLa cells ablates both early _mCa²⁺ uptake/_mH⁺ extrusion and subsequent _mCa²⁺ efflux/_mH⁺ uptake that are triggered by histamine stimulation (288). Loss of this reversible LETM1 activity, in particular the second phase of _mCa²⁺ efflux/_mH⁺ uptake, causes eventual _mCa²⁺ overload (288). The immediate effects of LETM1 knockdown on mitochondrial Ca²⁺/H⁺ exchange are independent of any impairment in electron transport chain function, as basal ΔΨ_m is slightly increased in LETM1-knockdown cells (288).

Further support for LETM1 as a Ca²⁺/H⁺ exchanger comes from experiments on purified His-tagged LETM1. LETM1-containing liposomes rapidly take up Ca²⁺, and this flux is blocked by ruthenium red and Ru360 and is partially inhibited by GCP-37157 (288). H⁺ efflux from liposomes is triggered by addition of Ca²⁺ to or alkalinization of the external buffer solution. Likewise, acidification of the external buffer solution drives Ca²⁺ efflux from the liposomes, whereas alkalinization of the external buffer stimulates Ca²⁺ uptake (288). Although LETM1 transports a number of cations, its affinity for Ca²⁺ is much greater than its affinity for other cations, including K⁺ (347). LETM1 thus has a fundamental function as a reversible Ca²⁺/H⁺ exchanger that will mediate either _mCa²⁺ uptake or _mCa²⁺ efflux depending on the given conditions within a cell.

3.7.2.1. letm1 phylogenetic conservation.

LETM1 is conserved throughout Eukarya, including in protozoa such as Toxoplasma gondii (342, 348). Evolutionary relationships among LETM1 homologs are illustrated by Austin and Nowikovsky (349). Mouse and human LETM1 proteins share 83.3% sequence identity at the amino acid level, and LETM1 homologs in species from yeast to C. elegans share ∼25–43% homology with human LETM1 (341). The central region of LETM1 containing the transmembrane domain is the most conserved region of LETM1 among such divergent species, whereas the rest of the protein varies more, indicating that LETM1 homologs may have distinct functions in different taxa (341). Of note, PKC and casein kinase phosphorylation sites located NH₂-terminal to the transmembrane domain are conserved, suggesting that LETM1 regulation is consistent across distant species (341). Several additional features are conserved among some species, including an α-helical domain predicted to form coiled coils, the leucine zipper domain, and a SAP-like region of unknown function that is located 138–145 residues after the transmembrane domain (342). A COOH-terminal, Ca²⁺-binding EF-hand domain is present in plants and animals but absent in yeast; and the leucine zipper is conserved only among animals (350).

Studies of LETM1 homologs in various species support conserved, Ca²⁺-independent functions of this protein related to its role in K⁺/H⁺ exchange. Knockout of the yeast LETM1 homolog, MDM38/YOL027C, causes mitochondrial swelling, defective mitochondrial K⁺/H⁺ exchange, increased mitochondrial K⁺ content, and a diminished ΔΨ_m (346, 351). Thus, a fundamental function of LETM1 is to extrude K⁺ from the matrix, thereby lowering matrix osmolarity in order to control mitochondrial volume (343, 346). This function of LETM1 in mitochondrial K⁺ handling and volume control is shared in T. gondii, T. brucei, C. elegans, Drosophila, and humans (343, 348, 352, 353). The potential conservation of LETM1’s role in _mCa²⁺ exchange has not been studied in as great detail as its role in K⁺ flux, although the finding that the Ca²⁺-sensing EF-hand of LETM1 is not as widely conserved indicates that LETM1’s role in _mCa²⁺ handling may not be as consistent across taxa. LETM1 and its homologs have been implicated in a number of processes, including mitochondrial translation and protein homeostasis via interactions with the mitochondrial ribosome (350, 354). Some reports indicate that these effects are specifically related to and downstream of LETM1’s role in mitochondrial K⁺ exchange (352, 355). In many cases, though, any potential contribution of LETM1-dependent _mCa²⁺ transport to these processes has not been thoroughly investigated.

3.7.2.2. letm1 structure.

Human LETM1 encodes an 83.4-kDa, 739-amino acid transmembrane protein (341). LETM1 is detected in high-molecular mass (∼550 kDa) protein complexes (343, 356). LETM1 can homooligomerize, and the ∼550-kDa molecular mass of LETM1-containing complexes corresponds to the oligomerization of up to six or seven LETM1 subunits (343). LETM1 also forms 140-kDa dimers and 210-kDa trimers, and blue native gel electrophoresis further reveals that LETM1 participates in at least three distinct high-molecular mass complexes, of 250 kDa, 500 kDa, and 650 kDa. Since these high-mass complexes persist regardless of LETM1 knockdown or overexpression, they likely contain other proteins in addition to LETM1 (356).

The protein structure of LETM1 begins with an NH₂-terminal mitochondrial targeting sequence within the first 167 residues (342), followed by a transmembrane region. The transmembrane region is followed by a region of α-helices that may form coiled coils, a leucine zipper motif, and two EF-hand motifs (341). The first EF-hand domain (residues 569–597 of human LETM1) is located within the leucine zipper, and divergence of its sequence from the consensus EF-hand divalent cation binding sequence suggests that it has lost the ability to bind Ca²⁺ (341). The second EF-hand (residues 667–695 of human LETM1) conforms to the EF-hand consensus sequence and so is predicted to bind Ca²⁺ (341). Expression of LETM1 either with the second EF-hand deleted or with point mutation of two critical residues within this EF-hand (D676A and D688K) impairs mitochondrial Ca²⁺ flux in HeLa cells (357). LETM1 also contains several predicted phosphorylation sites for protein kinase C, casein kinase, and tyrosine kinase, as well as a potential glycosylation site (341).

The transmembrane domain of LETM1 anchors it in the IMM. Protease protection assays indicate that the majority of the protein, which lies COOH-terminal to the transmembrane domain, extends into the matrix and only a small and/or protease-resistant region extends into the IMS (344, 356). Since a single-pass transmembrane protein should not be capable of ion exchange, homooligomerization of LETM1 or its heterooligomerization with other proteins may be required for LETM1 to form a functional K⁺/H⁺ or Ca²⁺/H⁺ exchanger (297). Recent studies labeling the matrix and IMS sides of the IMM via genetically targeted peroxidases challenge this structural model and suggest that both the NH₂ and COOH termini of LETM1 reside in the mitochondrial matrix (358). According to this model, LETM1 has two relevant transmembrane domains (residues 204–232 and 413–421), and only a small linker region between them is exposed in the IMS (358). Further investigation into the structure of LETM1 and whether or not oligomerization is in fact required for LETM1-mediated cation exchange is warranted.

3.7.2.3. properties and function of letm1.

This section focuses on the ion transport properties of LETM1 specifically as they pertain to Ca²⁺/H⁺ exchange. For a discussion of the properties of LETM1-mediated K⁺/H⁺ exchange, the reader is referred to recent reviews on this topic (297, 351, 359).

Several recent papers reinforce LETM1’s role as a Ca²⁺/H⁺ exchanger and indicate that, under normal conditions, it is particularly important for _mCa²⁺ efflux. LETM1 knockdown increases net _mCa²⁺ content (357). In HeLa cells, LETM1 knockdown increases _mCa²⁺ transients in response to histamine stimulation and overexpression of wild-type LETM1 reduces _mCa²⁺ transients (360). Similar effects are observed in pancreatic β-cells, where silencing of LETM1 causes mitochondria to accumulate more Ca²⁺ and take up less H⁺ in response to a rise in extramitochondrial Ca²⁺ (361). Mutation of E221 diminishes the ability of LETM1 to reduce _mCa²⁺ transients, suggesting that E221 is critical for its Ca²⁺ exchange activity (360). These studies collectively support a critical role for LETM1 in _mCa²⁺ efflux, although the relative contribution of LETM1 versus NCLX to _mCa²⁺ efflux is the matter of some debate. De Marchi et al. (362) find that LETM1 overexpression in HeLa cells has no effect on _mCa²⁺ efflux following agonist stimulation, whereas NCLX overexpression clearly increases _mCa²⁺ efflux. Heterozygous deletion of NCLX in chicken DT40 B lymphocytes or siRNA-mediated NCLX knockdown in mouse A20 B lymphocytes causes a marked reduction in, but not complete loss of, _mCa²⁺ efflux, and the residual efflux occurs even in the absence of Na⁺ (363). This suggests that additional, Na⁺-independent exchangers such as LETM1 contribute to _mCa²⁺ efflux in immune cells. An interesting question is whether either of these two exchange systems may be able to compensate for loss or dysfunction of the other in order to maintain _mCa²⁺ homeostasis. For example, does a compensatory increase in LETM1 function account for the viability of NCLX-constitutive knockout mice (325)? Much more work is clearly required to elucidate the relative importance of LETM1 and NCLX to _mCa²⁺ efflux, and to understand how the relative contributions of these proteins vary across distinct tissues.

How LETM1 mediates ion exchange while it may have only a single transmembrane domain is an area of active investigation. LETM1’s function as a Ca²⁺ exchanger is independent of the function of other proteins involved in _mCa²⁺ uptake, as knockdown of LETM1 does not affect MCU current or MCU, MCUR1, or MICU1 protein level (357). Likewise, in fibroblasts of patients with genetic disruption of LETM1, the loss of LETM1 protein has no effect on the expression of these mtCU components (357). LETM1 does not physically interact with NCLX, indicating that LETM1 has a direct role in _mCa²⁺ efflux rather than simply binding to NCLX to modulate its activity (357). Shao et al. report that reconstituted LETM1 forms hexamers with a pore of 10.5 Å at pH 8.0, which is predicted to allow ion flux (360). This pore becomes blocked by a “plungerlike” structure at pH 6.5, which presumably inhibits ion transport (360). LETM1 therefore oligomerizes to a channel that appears competent for both ion exchange and pH-dependent regulation, independent of any requirement for other proteins.

Initial studies proposed LETM1 to operate with a stoichiometry of 1 Ca²⁺:1 H⁺ and mediate electrogenic cation exchange (288). Such electrogenic exchange means that the transport of Ca²⁺ and H⁺ by LETM1 is influenced by membrane voltage (287) and so could respond to changes in ΔΨ_m. Other studies of purified LETM1 indicate that it instead mediates electroneutral cation exchange and operates with a stoichiometry of 1 Ca²⁺:2 H⁺ (347). However, the observation that cation exchange through LETM1 is indeed sensitive to changes in membrane voltage (288) favors the electrogenic 1 Ca²⁺:1 H⁺ model of LETM1 stoichiometry. Furthermore, the observation that LETM1 mediates _mCa²⁺ uptake at low cytosolic Ca²⁺ concentrations (<1 µM) (288, 289) is inconsistent with electroneutral Ca²⁺/H⁺ exchange, which would require cytosolic Ca²⁺ to rise to a >100-fold higher concentration than the matrix Ca²⁺ concentration in order to drive _mCa²⁺ uptake through LETM1 (297). So, assuming an electrogenic 1:1 stoichiometry of Ca²⁺/H⁺ exchange, LETM1 should mediate _mCa²⁺ efflux under conditions where mitochondrial Ca²⁺ is sufficiently high relative to cytosolic Ca²⁺ or where cytosolic pH is sufficiently low (acidic, high H⁺ concentration) relative to the matrix, and/or ΔΨ_m is sufficiently depolarized to reverse LETM1 directionality. Indeed, this prediction agrees with the observation that, in response to agonist-induced elevation of cytosolic Ca²⁺, LETM1 first contributes to _mCa²⁺ uptake but then subsequently contributes to _mCa²⁺ efflux as the cytosolic Ca²⁺ transient subsides (288). Furthermore, Na⁺-independent _mCa²⁺ efflux, likely via LETM1, is sensitive to total matrix Ca²⁺ content. In experiments conducted in the absence of Na⁺ so as to eliminate Ca²⁺ transport by NCLX, increased _mCa²⁺ load increases the rate of _mCa²⁺ efflux via Ca²⁺/H⁺ exchange (364). LETM1 thus may serve as a mechanism to protect mitochondria from _mCa²⁺ overload, because it favors _mCa²⁺ efflux as _mCa²⁺ concentration rises (364). Such a protective ability is apparently of limited efficacy in some tissues, though, as it is not sufficient to compensate for loss of Na⁺/Ca²⁺ exchange and prevent basal _mCa²⁺ overload when NCLX is deleted in the adult heart (166). A limitation of existing studies of LETM1 ion transport is that few have been conducted in the presence of physiological concentrations of K⁺ and Mg²⁺ (287, 297). Thus, understanding the truly “physiological” function of LETM1 as it operates in intact cells requires further investigation.

3.7.2.4. special roles of letm1 in physiology.

LETM1 is implicated in a number of cellular processes, some of which have obvious associations with _mCa²⁺ handling. Knockdown of LETM1 impairs cellular bioenergetics and oxygen consumption and is associated with defects in complex IV (357), suggesting a critical role for LETM1 in oxidative phosphorylation. One possible mechanism linking LETM1-dependent changes in _mCa²⁺ homeostasis to disrupted bioenergetics is increased mitochondrial ROS production. Consistent with this notion, overexpression of antioxidant genes in LETM1-knockdown cells corrects cellular energetics (357). Other effects of LETM1 disruption, such as altered mitochondrial fusion, apoptosis, and mitophagy, may instead reflect LETM1’s functions in K⁺ homeostasis and mitochondrial volume regulation (355, 365). LETM1 and its homologs also have pleiotropic effects on mitochondrial DNA metabolism and pyruvate dehydrogenase activity (366); mitochondrial mRNA translation and respiratory chain biogenesis (367); and mitochondrial morphology (342) and cristae organization (368). LETM1 is also required for the maturation of preadipocytes and the differentiation of beige adipocytes (369). Upregulation of LETM1 causes upregulation of YAP and the stimulation of proliferative pathways (370), suggesting an additional role for LETM1 in cell proliferation. Which of these phenomena are specifically related to LETM1-dependent _mCa²⁺ exchange versus LETM1-dependent mitochondrial K⁺ homeostasis or binding and regulation of other mitochondrial proteins has not yet been fully elucidated. Any interpretation of the putative impact of LETM1’s Ca²⁺ exchanger activity on cellular and whole animal phenotypes should therefore take into consideration the possible contribution of these Ca²⁺-independent effects of LETM1 on cell function.

3.7.2.5. genetic diseases associated with letm1.

LETM1 is deleted in almost all patients who have Wolf–Hirschhorn syndrome (WHS), a developmental disorder caused by partial deletion of the short arm of chromosome 4 (341) (TABLE 1). WHS is characterized by impaired growth, developmental delay, microcephaly, and mental defects. It often presents with impaired muscle tone and seizures, and congenital heart defects are sometimes observed as well (341, 342). WHS patient fibroblasts have reduced LETM1 mRNA and protein content (356), suggesting that loss of LETM1 contributes to these phenotypes. Consistent with this idea, disruption of LETM1 is implicated in the seizure phenotype of WHS (371), and LETM1 haploinsufficiency disrupts normal glucose metabolism, reduces brain ATP levels, and increases the occurrence of seizures in mice (372). LETM1 haploinsufficiency also alters mitochondrial function in cells derived from WHS patients, which display diminished Ca²⁺ sensitivity of the mPTP, hyperpolarization of ΔΨ_m, and elevated mitochondrial superoxide production (373). WHS patient fibroblasts surprisingly do not exhibit the same mitochondrial fragmentation observed with acute LETM1 silencing, suggesting that some compensatory mechanisms exist to maintain mitochondrial network structure in the face of chronic LETM1 disruption, despite persistent disruption of mitochondrial ion homoeostasis (356).

Although LETM1 deletion is necessary for WHS, LETM1 deletion alone is not sufficient to cause the full spectrum of phenotypes that are associated with this disease (374). Rather, the deletion of additional genes adjacent to LETM1 contributes to WHS as well (374). Further complicating our understanding is the observation that duplication of genetic regions containing LETM1 can also cause several features of WHS such as microcephaly, seizures, and delayed development (375), as well as syndromes distinct from but related to WHS (376). These findings indicate that either increased or decreased LETM1 expression can be detrimental, and that the specific dosage of LETM1 is critical for maintaining normal cellular function. How each of the individual phenotypic features of WHS specifically relates to altered _mCa²⁺ homeostasis versus other consequences of altered LETM1 expression remains to be determined.

Altered LETM1 expression is observed in a number of conditions including temporal lobe epilepsy (377), obesity (378), and diabetes (379). LETM1 is also upregulated in various types of cancer, where it may inhibit mitochondrial function and secondarily enhance glycolysis and cell survival signaling (380, 381). Multiple studies have proposed LETM1 as a prognostic cancer biomarker, as LETM1 expression typically correlates with poorer outcomes (382–391). This has raised interest in targeting LETM1 as a cancer therapeutic (392–394), although several cases suggest that LETM1 instead is protective and inhibits tumor growth in liver and lung cancer (395, 396). How these potential detrimental versus protective effects of LETM1 may be related to its specific role in _mCa²⁺ exchange is an open question.

3.7.2.6. animal models of letm1.

Null mutation of the LETM1 homolog (F58G11.1 or letm-1) in C. elegans causes mitochondrial swelling (TABLE 2). C. elegans homozygous for this mutation arrest in the L3 larval stage and are small and infertile, although they can live longer than wild-type worms (343). Animals that are heterozygous for this mutation appear no different from wild type and, in contrast to human WHS patients, have no signs of seizures (343). Wild-type worms subjected to RNAi to knock down letm-1 after hatching mature more slowly and remain smaller than control worms upon reaching adulthood (343). These phenotypes are associated with swollen and highly disorganized mitochondria, consistent with accepted roles for LETM1 in mitochondrial volume regulation. Conversely, transgenic overexpression of LETM-1 in C. elegans results in the opposite effect, causing “crimping” of the mitochondrial matrix and in some cases detachment of the mitochondrial outer membrane from the IMM and matrix (343). These findings indicate that LETM1 is inversely related with mitochondrial matrix volume and support the notion that mitochondrial defects in LETM1-mutant animals have broad effects to limit organismal growth and development.

In higher animals, loss of LETM1 is lethal during development. In vivo RNAi to ubiquitously knock down LETM1 in Drosophila results in lethality in the early larval stage or the pupal stage (353). Specific knockdown of LETM1 in muscle cells reduces the size of larvae, reduces their physical behavior, and arrests the growth of most flies in the pupal stage (353). The few animals that develop to adulthood are smaller than normal, weak, and unable to fly (353). Knockdown of LETM1 in the fly eye results in smaller than normal eyes surrounded by scar tissue, with many ommatidia exhibiting swollen mitochondria (353). Finally, downregulation of LETM1 in the nervous system reduces the speed of locomotion and increases the time that flies remain immobile, suggestive of neuromuscular impairments (353). This locomotor defect is associated with a reduction in neurotransmitter release following nerve stimulation (353). McQuibban et al. (353) conclude that loss of LETM1 has direct effects on neurotransmission, with subsequent impairments of neuromuscular function. This particular study did not distinguish whether these effects are related to disrupted mitochondrial K⁺ transport, Ca²⁺ transport, or both.

In mice, homozygous deletion of Letm1 via gene trap is lethal by day 6.5 of embryogenesis (day E6.5) (372). About half of the mice with heterozygous Letm1 deletion die by day E13.5. Once past this critical developmental stage, surviving heterozygous embryos are largely normal, although their fibroblasts have reduced _mCa²⁺ uptake rates, low homeostatic _mCa²⁺ concentration, and lower matrix pH. Letm1 heterozygous mice that survive to birth do not exhibit craniofacial or midline developmental defects as seen in WHS patients and have normal body weight up to at least 6 wk of age (372). The mitochondrial morphology of adult Letm1^+/− mice is relatively normal, much like in human WHS patients (356), suggesting that over time the mechanisms regulating mitochondrial structure compensate for loss of LETM1. Interestingly, the brains of Letm1^+/− mice have an impaired ability to maintain ATP concentration and reduced PDH activity, suggesting that LETM1 is required for glucose metabolism and pyruvate oxidation (372). Letm1^+/− mice exhibit increased seizure activity when challenged with kainic acid, which increases intracellular Ca²⁺ concentration (372). This study did not specifically examine whether altered _mCa²⁺ handling specifically contributes to increased seizure susceptibility. However, observations in rats with lentivirus-mediated knockdown of LETM1 in the hippocampus and dentate gyrus suggest that disrupted mitochondrial K⁺ homeostasis is not the key factor causing neuronal dysfunction upon loss of LETM1. Knockdown of LETM1 in these brain regions results in mitochondrial swelling and reduces mitochondrial gene expression, and increases the susceptibility to and duration of seizures in experimental pilocarpine-induced epilepsy (377). Although the K⁺/H⁺ ionophore nigericin corrects mitochondrial swelling in other models of LETM1 deficiency (353, 355), nigericin treatment fails to reduce this seizure susceptibility in rats with LETM1 knockdown (377). This suggests that other consequences of LETM1 disruption, like altered _mCa²⁺ homeostasis, contribute to seizures in LETM1-deficient subjects.

3.7.2.7. regulation of letm1.

PTEN-induced kinase 1 (PINK1) phosphorylates LETM1 at T192 to stimulate LETM1-mediated _mCa²⁺ uptake and efflux (397). Phosphorylation of LETM1 at this site protects neurons against MPP⁺-induced permeability transition and cell death, indicating that PINK1-dependent activation of LETM1 protects cells against _mCa²⁺ overload (397). Several binding partners also regulate LETM1 function. The mitochondrial AAA-ATPase BCS1L binds directly to LETM1 via an interaction that depends on R155 of BCS1L but is independent of BCS1L ATP hydrolysis activity (344). LETM1 incorporates into high-molecular mass complexes of ∼300 kDa and ∼500 kDa, and BCS1L specifically regulates the incorporation of LETM1 into the larger ∼500-kDa complex (344). As LETM1 forms homooligomers in vitro (343), Tamai et al. (344) propose that BCS1L facilitates this oligomerization process. LETM1 and BCS1L also interact with the mitochondrial peptidase neurolysin, which like BCS1L is required for LETM1 complex formation (398). Thus, cleavage of LETM1 via neurolysin, or some other protease, may be a necessary step for its incorporation into intact high-mass complexes. The carboxyl-terminal modulator protein (CTMP), a tumor suppressor-like protein that negatively regulates PKB/Akt, also associates with LETM1 (365), although any direct effect of CTMP on LETM1 function remains to be determined. Finally, studies of LETM1 activity in liposomes of defined phospholipid composition indicate that cardiolipin is required for full LETM1 Ca²⁺ exchanger activity (399). The authors hypothesize that this effect may arise from the ability of cardiolipin to facilitate LETM1 oligomerization into hexameric, functional transporters (399).

3.7.2.8. pharmacological modulation of letm1.

LETM1 activity in liposomes is inhibited by CGP-37157, ruthenium red, and Ru360 (288). However, these effects are not consistently observed (347). Furthermore, these compounds all inhibit other proteins involved in mitochondrial Ca²⁺ flux, notably the mtCU, which is inhibited by ruthenium red and Ru360, and NCLX, which is inhibited by CGP-37157. The development of LETM1-specific activators and inhibitors would thus be of great experimental and perhaps clinical utility.

3.7.3. The mitochondrial permeability transition pore.

The mitochondrial permeability transition pore is a high-conductance channel spanning the IMM and OMM. It is best appreciated for its classical role in stress-induced mitochondrial depolarization, swelling, rupture, and necrotic cell death. This function of the mPTP in cell death and the regulation of the mPTP by _mCa²⁺ are reviewed extensively elsewhere (150, 400–402). The mPTP also has a physiological function in _mCa²⁺ efflux, in which transient, low-conductance opening of the mPTP, termed “flickering,” selectively releases Ca²⁺ and so helps to limit _mCa²⁺ accumulation (401, 403–406) (FIGURE 1). Indeed, such physiological “venting” of Ca²⁺ out of the mitochondria, while ΔΨ_m is still largely intact, would imply that such _mCa²⁺ efflux is driven by a free _mCa²⁺ concentration that is greater than the local cytosolic/IMS Ca²⁺ concentration. The structure of the mPTP is largely undefined, although numerous proteins are suggested to form or regulate the pore, including the adenine nucleotide translocator (ANT) (407), the F₀/F₁ ATP synthase (408), the voltage-dependent anion channel (409), cyclophilin D (CypD) (410, 411), and the mitochondrial phosphate carrier (PiC) (401, 402, 412).

The mPTP inhibitor cyclosporin attenuates physiological _mCa²⁺ efflux in healthy, isolated rat cardiomyocytes (413). What’s more, ablation of CypD, the only genetically confirmed mPTP component, results in diminished cardiomyocyte _mCa²⁺ efflux and increased _mCa²⁺ content (414) (TABLE 2). Protective, transient mPTP openings that prevent _mCa²⁺ overload are observed in neurons, and Ca²⁺ efflux via this mechanism is disrupted in the absence of CypD (415). CypD-null hearts exhibit increased TCA cycle flux and a shift in cardiac metabolism toward glucose rather than fatty acid utilization, all expected consequences of elevated _mCa²⁺. Despite protection from permeability transition, CypD-null mice have exaggerated cardiac dysfunction, cardiac hypertrophy, and maladaptive cardiac remodeling following chronic pressure overload or chronic exercise (414). These phenotypes are consistent with purported roles for _mCa²⁺ in cell growth and potentially deleterious mitochondrial ROS generation. These observations collectively reinforce the idea that the mPTP is a physiologically relevant route for _mCa²⁺ efflux that helps to prevent detrimental _mCa²⁺ overload.

4. BUFFERING OF MITOCHONDRIAL Ca²⁺

In addition to the proteins that mediate _mCa²⁺ exchange, the ability of the mitochondria to buffer Ca²⁺ can influence cellular Ca²⁺ homeostasis. During acute _mCa²⁺ uptake, free, ionic Ca²⁺ in the matrix increases from ∼0.1 to ∼5 µM. Beyond this point, additional influx of Ca²⁺ is buffered by inorganic phosphate (40, 268, 269). Inorganic phosphate exists as polyphosphate polymers within the matrix, and this conformation minimizes the precipitation of buffered Ca²⁺-phosphate complexes despite their high concentration (416). The Ca²⁺ buffering capacity of the mitochondria is exceptional, with the ratio of phosphate-bound Ca²⁺:free Ca²⁺ estimated to rise as high as 100,000:1 (268). The proportion of phosphate-bound to free Ca²⁺ is sensitive to matrix pH, with more acidic matrix conditions favoring dissociation of Ca²⁺-phosphate complexes and increased free matrix Ca²⁺ (268).

The high mitochondrial capacity for Ca²⁺ buffering can influence cytosolic Ca²⁺ dynamics, as the presence of inorganic phosphate in the matrix accelerates _mCa²⁺ uptake and thus accelerates the clearance of cytosolic Ca²⁺ spikes (417). It should be noted, though, that the extent of mitochondrial buffering of cytosolic Ca²⁺ signals varies across tissues and with the amplitude and frequency of the Ca²⁺ signal (FIGURE 6).

FIGURE 6.

Cell-type dependent effects of _mCa²⁺ uptake on cytosolic Ca²⁺ signals. Stimulation of secretory cells elicits cytosolic Ca²⁺ transients that are rapidly transmitted to the mitochondria, resulting in large _mCa²⁺ transients. Modulation of net _mCa²⁺ uptake affects the magnitude of both the cytosolic Ca²⁺ transient and the mitochondrial Ca²⁺ transient, and impacts physiologically relevant cellular functions. For example, increased net _mCa²⁺ uptake in adrenal chromaffin cells can limit the magnitude of the cytosolic Ca²⁺ transient and thereby limit the amount of cytosolic Ca²⁺ available to stimulate secretion of catecholamines such as norepinephrine (NE). In other cell types, however, mitochondrial Ca²⁺ signals are not coupled as directly to each cytosolic Ca²⁺ transient. For example, cardiomyocytes do not generally appear to have beat-to-beat oscillations in _mCa²⁺ content that mirror cytosolic Ca²⁺ transients. Rather, alterations in the amplitude and/or frequency of cytosolic Ca²⁺ transients drive integrated changes in _mCa²⁺ concentration. For instance, sympathetic stimulation of cardiomyocytes increases the amplitude and frequency of cytosolic Ca²⁺ transients, causing a progressive increase in _mCa²⁺ concentration over time. [Ca²⁺], Ca²⁺ concentration; [_cytoCa²⁺], cytosolic [Ca²⁺]; [_mitoCa²⁺], mitochondrial [Ca²⁺]; see glossary for other abbreviations.

For example, _mCa²⁺ buffering has very little effect on the high-frequency, high-amplitude cytosolic Ca²⁺ transients that stimulate cardiomyocyte contraction (418). In contrast, _mCa²⁺ uptake limits the increase in cytosolic Ca²⁺ concentration in presynaptic neurons that occurs during neurotransmission (84). The shaping of cytosolic Ca²⁺ signals via mitochondrial Ca²⁺ buffering regulates catecholamine secretion from chromaffin cells (419) and so has the potential to regulate physiologically relevant biological phenomena. Additional examples of how _mCa²⁺ buffering shapes cellular function and can spatially regulate cytosolic Ca²⁺ signaling are discussed below.

5. SPATIAL ARRANGEMENT OF MITOCHONDRIA AND THE MITOCHONDRIAL Ca²⁺-HANDLING MACHINERY

Efficient _mCa²⁺ uptake depends on the exposure of mitochondria to local concentrations of Ca²⁺ high enough to activate the Ca²⁺ uptake machinery. For acute uptake through the mtCU, this threshold is ∼400 nM Ca²⁺, roughly 4 times higher than the bulk resting cytosolic Ca²⁺ concentration (10–12, 16). The placement of mitochondria in close proximity to sites of intracellular Ca²⁺ release creates local microdomains where Ca²⁺ released from the store is concentrated in a small volume of cytosol. Ca²⁺ concentration in these microdomains can be 5- to 20-fold higher than Ca²⁺ concentration in the bulk cytosol and so is sufficient to activate low-affinity _mCa²⁺ uptake through the mtCU (83, 270–274). Thus, the spatial organization of the mitochondria within the cell and with respect to other organelles is a major factor controlling _mCa²⁺ flux and cellular Ca²⁺ homeostasis. Local Ca²⁺ release events can even reduce the motility of nearby mitochondria, thereby trapping them within the region of high Ca²⁺ concentration and further guaranteeing efficient Ca²⁺ transfer to the mitochondria (420). That mitochondrial localization is actively regulated by Ca²⁺ signals highlights just how critical the spatial organization of the mitochondria is for cellular Ca²⁺ handling.

5.1. Localization of Mitochondria With Respect to Cellular Domains and Other Organelles

5.1.1. Interactions with plasma membrane.

All mitochondria contact the ER, and a small portion are located in proximity to the plasma membrane. Even fewer of these peri-plasma membrane mitochondria actually contact the plasma membrane, and those that do seem to interact with the membrane via an ER stack (421). This population of mitochondria may be specifically involved in store-operated Ca²⁺ entry (421). As reviewed elsewhere, striated muscle contains distinct populations of subsarcolemmal and intermyofibrillar mitochondria with unique functional properties and features of Ca²⁺ handling. These distinctions likely relate to differences in proximity to the plasma membrane versus the SR and, respectively, exposure to Na⁺ and Ca²⁺ fluxes during the action potential versus exposure to large Ca²⁺ transients upon SR Ca²⁺ release (422, 423).

5.1.2. Interactions with the endoplasmic/sarcoplasmic reticulum.

Fluorescent proteins targeted specifically to the mitochondria or ER reveal the existence of numerous close contacts between these two compartments (273). The term “mitochondria-associated membranes (MAMs)” is used to describe regions of the ER that interact with the mitochondria. MAMs are critical sites of Ca²⁺ exchange between these organelles and generate local cytosolic microdomains where Ca²⁺ can accumulate to high enough concentrations to activate _mCa²⁺ uptake through the mtCU. These sites of ER-mitochondria contact also have essential roles in lipid biosynthesis, regulation of mitochondrial morphology, autophagy, and apoptosis, which have been reviewed extensively elsewhere (424–427).

MAMs are maintained by protein-based tethers that hold the OMM and ER membrane at a distance of ∼10–25 nm. Changes in the spacing between these two surfaces modulates the dynamics of ER-to-mitochondria Ca²⁺ transfer. Narrowing of the ER-mitochondrial distance is observed under conditions that trigger apoptosis, and the expression of shortened artificial tethers to reduce this distance to 5 nm increases _mCa²⁺ uptake and the susceptibility to permeability transition (428). These findings suggest that endogenous mechanisms can actively modify the structure of ER-mitochondrial contacts in order to perturb _mCa²⁺ homeostasis. Mitofusin 2 is suggested as one ER-mitochondria tether based on its enrichment at ER-mitochondria junctions. Consistent with this model, knockout of mitofusin 2 in mouse embryonic fibroblasts reduces the area of interorganelle contact and interferes with propagation of Ca²⁺ released from the ER into the mitochondrial matrix (429). ER/SR-mitochondrial Ca²⁺ transfer is disrupted enough in mitofusin 2-knockout cardiomyocytes that it compromises cellular bioenergetics during electrical pacing (430). de Brito et al. (429) propose that ER-localized mitofusin 2 forms a tether by interacting with mitofusin 1 and mitofusin 2 on the OMM. This view is challenged, though, by experiments showing that mitofusin 2 knockout or knockdown increases rather than decreases juxtaposition of the ER and mitochondria and sensitizes cells to _mCa²⁺ overload-induced cell death (431, 432). Filadi et al. (432) reason that mitofusin 2 antagonizes rather than promotes the apposition of the ER and mitochondria.

The search for alternative tethers is ongoing. Szabadkai et al. (60) report that the inositol-(1,4,5)-trisphosphate receptor (IP₃R) on the ER membrane and VDAC on the IMM physically interact via the chaperone glucose-regulated protein 75 (grp75). Thus, the very proteins that mediate Ca²⁺ flux out of the ER and across the OMM help anchor these organelles together. Furthermore, this physical interaction can permit VDAC-dependent transport of _mCa²⁺ across the OMM (60). Finally, the ER protein PDZD8 was recently described as another metazoan tethering protein. PDZD8 is orthologous to Mmm1, a component of the yeast ER-mitochondrial encounter structure (ERMES) (433). PDZD8 is required for the maintenance of ER-mitochondrial contacts, and PDZD8 knockout disrupts ER-to-mitochondria Ca²⁺ transfer (433). The tethering function of PDZD8 may have particular consequences for neuronal biology. Loss of PDZD8 in neurons and the subsequent reduction in ER-mitochondrial contacts results in loss of mitochondrial Ca²⁺ buffering and therefore allows greater cytosolic Ca²⁺ elevations in dendrites after synaptic stimulation and ER Ca²⁺ release (433). Thus, maintenance or disruption of the microdomain between the ER and mitochondria can have appreciable effects to shape cellular Ca²⁺ signaling.

5.1.3. Localization in neurons and effects at synaptic terminals.

Mitochondria are found at many presynaptic nerve terminals, where they can modulate local Ca²⁺ dynamics and ATP production (reviewed in Ref. 434). This population of mitochondria maintains neurotransmission during high-frequency stimulation because they provide the ATP needed to restore the readily releasable pool of synaptic vesicles (435, 436). Presynaptic mitochondria also buffer cytosolic Ca²⁺ and so limit the amount of cytosolic Ca²⁺ locally available to stimulate post-tetanic potentiation, a phenomenon in which synaptic transmission is enhanced for minutes after high-frequency stimulation (434).

5.1.4. Localization in exocrine gland acinar cells and segregation of Ca²⁺ signals.

A characteristic feature of the secretory cells found in exocrine glands, such as the acinar cells of the pancreas or the salivary glands, is that they are polarized, with distinct processes occurring at the apical versus basolateral domains (437). This includes compartmentalization of Ca²⁺ signaling at the apical domain, which allows the cell to separate Ca²⁺-dependent processes such as exocytosis from other processes such as transduction of extracellular signals that originate on the basolateral membrane (438). Mitochondria in pancreatic acinar cells help to segregate these cellular domains because they form a “belt” around the secretory granule area and buffer changes in cytosolic Ca²⁺. Consequently, Ca²⁺ signals that originate in the apical domain are restricted to that region, where they stimulate secretion, and do not propagate beyond the mitochondrial belt to the basolateral side of the cell (439–441).

5.2. Spatial Segregation of Ca²⁺ Influx and Efflux Machinery within the Mitochondria

The spatial arrangement of the proteins involved in Ca²⁺ transport across the OMM and IMM is another mechanism that ensures efficient _mCa²⁺ exchange. In cardiomyocytes, contact points between the IMM and OMM align with the junctions that link the OMM and the SR (442). VDAC is enriched in these regions of the OMM and so acts as a direct conduit for Ca²⁺ released from the SR to gain access to the IMS and the _mCa²⁺ uptake machinery of the IMM (442). This arrangement represents the most direct route for Ca²⁺ transfer from the SR to the mitochondrial matrix. In support of this model, within cardiac muscle MCU and EMRE are enriched at the regions of the IMM that contact the OMM (443). Interestingly, the proteins responsible for _mCa²⁺ uptake and _mCa²⁺ efflux are segregated to distinct regions of the IMM (FIGURE 7).

FIGURE 7.

Spatial segregation of mitochondrial Ca²⁺ influx and efflux pathways in cardiomyocytes. In cardiac muscle, the mitochondrial calcium uniporter complex (mtCU) is found at regions of the inner mitochondrial membrane (IMM) near contacts between the outer mitochondrial membrane and Ca²⁺ release sites on the sarcoplasmic reticulum (SR). This arrangement enables Ca²⁺ released from the SR via the ryanodine receptor (RyR2) to be taken up efficiently into the mitochondrial matrix. NCLX, the principal route for mitochondrial Ca²⁺ efflux in cardiomyocytes, is instead distributed at regions of the IMM that are more distant from the sites of SR-mitochondria contact. This spatial separation of mitochondrial Ca²⁺ uptake and efflux is thought to minimize futile cycling of Ca²⁺ flux through the mtCU and NCLX that would otherwise depolarize mitochondrial membrane potential. See glossary for abbreviations.

Mitochondrial regions associated with the SR contain minimal NCLX and do not exhibit Na⁺-induced _mCa²⁺ efflux; instead, NCLX localizes to regions of the mitochondria that are more distant from the SR (444). This organization is proposed to minimize futile cycling of _mCa²⁺ uptake through the mtCU and immediate, subsequent _mCa²⁺ efflux via Na⁺/Ca²⁺ exchange. Such futile Ca²⁺ cycling could occur if matrix Ca²⁺ diffuses slowly after uptake and so persists at comparatively high concentrations in the vicinity of the mtCU. If this assumption of slow diffusion of matrix Ca²⁺ is valid and the mtCU and NCLX were located in close proximity, futile cycles of _mCa²⁺ uptake and efflux would ultimately run down ΔΨ_m because of the net charge influx through forward-mode operation of NCLX, without effectively increasing _mCa²⁺ content (444). Recent data indicating that discrete cristae within the mitochondria may maintain distinct ΔΨ_m and operate independently from each other reveal the possibility that distinct regions of the IMM can be electrically insulated from other regions (445). These findings fit with the notion that _mCa²⁺ uptake and stimulation of OXPHOS (thus contributing to ΔΨ_m at nearby cristae) could be spatially and functionally isolated from NCLX-dependent _mCa²⁺ efflux, which would otherwise dissipate ΔΨ_m and potentially compromise ATP synthesis. Thus, the spatial segregation of _mCa²⁺ uptake and extrusion is proposed to protect cellular bioenergetics (444). How precisely the distinct localization of the mtCU versus NCLX is maintained is an intriguing question with direct implications for cellular energy production.

6. PHYSIOLOGICAL CONSEQUENCES OF MITOCHONDRIAL Ca²⁺ FLUX

6.1. Influence on Metabolism

The best-appreciated physiological consequence of an acute increase in _mCa²⁺ concentration is the rapid stimulation of mitochondrial metabolism. _mCa²⁺ uptake is associated with an increase in respiration and ATP production (446), indicating that matrix Ca²⁺ stimulates oxidative phosphorylation (FIGURE 8).

FIGURE 8.

Physiological and pathological functions of acute _mCa²⁺ uptake. Physiological cellular Ca²⁺ signaling simulates signal transduction pathways in the cytosol and, if cytosolic Ca²⁺ concentration reaches a high enough threshold, can trigger _mCa²⁺ uptake through the mitochondrial calcium uniporter complex (mtCU). In some cell types, mitochondrial Ca²⁺ uptake can buffer further changes in cytosolic Ca²⁺ concentration and so affect Ca²⁺-dependent signaling throughout the cell. Increasing matrix Ca²⁺ concentration stimulates the activity of pyruvate dehydrogenase (PDH), isocitrate dehydrogenase (IDH), and α-ketoglutarate dehydrogenase (αKGDH) to increase TCA cycle flux and delivery of electrons to the electron transport chain to stimulate ATP production. Matrix Ca²⁺ may also stimulate ATP synthase directly, although this concept is controversial. Rapid _mCa²⁺ uptake through the mtCU allows cells to couple Ca²⁺-dependent, energy-consuming processes in the cytosol, such as muscle contraction, with a coordinated increase in mitochondrial ATP production in order to supply the cell’s increased energetic demand. A pathological increase in cytosolic Ca²⁺ concentration, as occurs during ischemia-reperfusion, can drive excessive Ca²⁺ uptake through the mtCU, leading to activation of the mitochondrial permeability transition pore, collapse of mitochondrial membrane potential, influx of water and swelling of the matrix, and rupture of the outer mitochondrial membrane (OMM), culminating in cell death. See glossary for abbreviations.

Ca²⁺ exerts this effect via several distinct mechanisms: 1) Activation of pyruvate dehydrogenase through the stimulation of pyruvate dehydrogenase phosphatase, which removes inhibitory phosphorylation at S232, S293, and S300 of the PDH E1α regulatory subunit (447–449). Increased PDH activity increases pyruvate oxidation to generate acetyl-CoA, which enters the TCA cycle. This reaction also generates NADH, which can deliver electrons to the electron transfer chain (450). 2) Stimulation of the TCA cycle enzymes isocitrate dehydrogenase (IDH) and α-ketoglutarate dehydrogenase (αKGDH) (451–454). Ca²⁺ lowers the K_m of these enzymes for their substrates and thereby enhances TCA cycle flux. Increased TCA cycle activity increases the production of reducing equivalents (NADH and FADH₂) that deliver electrons to the ETC (311). 3) Stimulation of respiratory complexes I, III, and IV (455) to increase ETC activity. Whether the activity of complex V (ATP synthase) is also directly increased in response to elevated _mCa²⁺ is debated (210, 456, 457). On an acute timescale, these effects of increased _mCa²⁺ allow cells such as cardiomyocytes, neurons, and pancreatic β-cells to rapidly increase ATP production needed for energy-consuming processes like muscle contraction or vesicle trafficking for neurotransmission and insulin secretion (126, 127, 235, 290, 361, 458–461). _mCa²⁺-dependent effects on mitochondrial metabolism can also persist and allow for a sustained increase in mitochondrial ATP production, thus supporting continued cellular energy consumption even after the initial rise in _mCa²⁺ concentration subsides (462).

Since TCA cycle function is intricately linked to the metabolism of biosynthetic building blocks such as amino acids, nucleotides, and lipids, as well as signaling molecules such as ROS (reviewed in Refs. 463–465), Ca²⁺-dependent changes in mitochondrial metabolism can have wide-reaching effects on cellular function and growth. Consistent with this notion, ablation of _mCa²⁺ uptake disrupts mitochondrial bioenergetics, alters pyruvate and proline metabolism, and impairs growth in T. brucei (103).

6.2. Buffering of Cytosolic Ca²⁺ and Modulation of Processes Regulated by It

Depending on the cell type, _mCa²⁺ uptake can buffer changes in cytosolic Ca²⁺ concentration and thereby influence cellular functions and signaling cascades that respond to cytosolic Ca²⁺. For example, uptake of Ca²⁺ by mitochondria helps to maintain cellular Ca²⁺ signaling that is required for T-cell activation. Stimulation of T cells by antigens causes IP₃-dependent Ca²⁺ release from the ER. Depletion of ER Ca²⁺ stores activates the store-operated Ca²⁺ release-activated Ca²⁺ current (I_CRAC) to allow entry of Ca²⁺ across the plasma membrane. This replenishes the ER Ca²⁺ stores and allows for a sustained elevation in cytosolic Ca²⁺ that is required for T-cell activation and associated transcriptional responses (17, 466–468). If mitochondria are depolarized, diminishing the driving force for _mCa²⁺ uptake, I_CRAC becomes inactivated, which is likely due to a local elevation in cytosolic Ca²⁺ concentration in the absence of _mCa²⁺ buffering (469). Inactivation of I_CRAC impairs store-operated Ca²⁺ entry and consequently attenuates the broader cytosolic Ca²⁺ response required for the activation of the transcription factor NFAT (469). Thus, _mCa²⁺ uptake is required to maintain sustained cytosolic Ca²⁺ signaling that ultimately drives T cell-mediated immune responses.

Another recent example is the control of AMPK signaling in hepatocytes. _mCa²⁺ uptake through the mtCU minimizes cytosolic Ca²⁺ concentration, but, when _mCa²⁺ uptake is disrupted, elevated cytosolic Ca²⁺ activates protein phosphatase-4, leading to dephosphorylation and inhibition of AMPK. Reduced AMPK signaling then impairs hepatocyte lipid metabolism (470). Thus, by altering cytosolic Ca²⁺, _mCa²⁺ exchange may exert secondary effects on diverse, seemingly unrelated cellular processes.

6.3. Role in Cell Death

Excessive accumulation of Ca²⁺ in the mitochondrial matrix causes sustained activation of the mitochondrial permeability transition pore (mPTP), a process known as permeability transition (PT) (FIGURE 8). Opening of the mPTP results in collapse of the mitochondrial membrane potential, release of Ca²⁺ from the matrix, mitochondrial swelling, and rupture of the outer mitochondrial membrane. This leads to cell death via either apoptosis or necrosis, depending on the cellular ATP supply (reviewed in Refs. 150, 400–402). PT thus contributes to much of the tissue injury that occurs in pathologies marked by excessive _mCa²⁺ uptake, such as ischemia-reperfusion and myocardial infarction. Although _mCa²⁺ is a clear trigger for PT and mitochondrial swelling (471), whether _mCa²⁺ directly activates the mPTP is less certain. Increased _mCa²⁺ stimulates ROS production, which is itself a stimulus for PT (401). Increased ROS production can occur as a direct result of increased TCA cycle flux and electron transport stimulated by elevated _mCa²⁺ (472, 473). Another idea proposes that, with continued _mCa²⁺ uptake, excessive _mCa²⁺ loading causes precipitation of Ca²⁺-phosphate complexes within the matrix, which interfere with the ETC by inhibiting complex I, disrupting electron transfer between NADH and complex I, and altering the structure of cristae where the ETC is located, ultimately increasing ROS production (474). Increased _mCa²⁺ stimulates nitric oxide production, which inhibits complexes I and IV, thus blocking electron flow through the ETC and favoring ROS production (401). Increased ROS production due to increased _mCa²⁺ interferes with cytochrome c’s association with cardiolipin and the IMM, further disrupting the ETC and driving even more ROS production (475). Increased _mCa²⁺ also interacts with cardiolipin to dissociate complex II and generate a subcomplex that makes excessive ROS (476). Thus, excessive _mCa²⁺ accumulation drives ROS production via multiple mechanisms that can contribute to PT. In turn, increased mitochondrial ROS production drives further _mCa²⁺ accumulation by promoting leak of Ca²⁺ from intracellular stores such as the ER (477), setting up a vicious cycle that triggers PT via the combined effects of _mCa²⁺ overload and oxidative stress. Breaking this Ca²⁺-ROS cycle via the administration of antioxidants is sufficient to prevent PT and cell death in the context of I/R (478). Likewise, VDAC1 knockdown, MCU knockdown, or Ru360 administration to minimize _mCa²⁺ uptake prevents PT and apoptosis of cells subjected to oxidative stress (59, 479, 480). These findings underscore the relevance of the interplay between _mCa²⁺ and ROS in permeability transition and cell death.

6.4. Role in Autophagy and Mitophagy

Autophagy is a highly conserved process whereby cells envelope organelles and other cytosolic components in a double-membraned vesicle, which then fuses with lysosomes enabling degradation of the contents (481). This catabolic process provides cells with energy when nutrients are scarce. Mitophagy is a related process allowing cells to selectively degrade damaged mitochondria and thus functions in mitochondrial quality control (482, 483). Several studies suggest that these processes are influenced by _mCa²⁺. In fibroblast lines from patients with distinct complex I mutations, those with mutations associated with reduced _mCa²⁺ uptake have enhanced autophagic flux and increased mitophagy (484). This is due to increased AMPK activity, which stimulates the autophagic machinery. The same relationship between MCU downregulation and increased autophagy is seen in a dog model of electrical pacing-induced heart failure and cardiac resynchronization therapy (485). Autophagy and, specifically, mitophagy are noted to increase upon MCU inhibition and to provide cardioprotection in a mouse model of pressure overload-induced heart failure (486). A link between MCU-dependent _mCa²⁺ uptake and autophagy is also observed in neuronal cell lines treated with MPP⁺, an in vitro model for Parkinson’s disease that involves autophagic cell death. Here, pharmacological MCU activation or MCU overexpression reduces MPP⁺-induced autophagic death, whereas downregulation of MCU activates AMPK and enhances autophagy (487). These studies did not distinguish whether AMPK activity increases as a result of metabolic stress due to impaired Ca²⁺-dependent stimulation of mitochondrial metabolism or rather as a result of increased cytosolic Ca²⁺ singling, which could activate AMPK through upstream kinases such as CaMKK. However, knockout or pharmacological inhibition of IP₃ receptors to limit ER-to-mitochondria Ca²⁺ transfer triggers energetic stress, AMPK activation, and autophagy (488). This suggests that energetic starvation related to impaired _mCa²⁺ uptake contributes to autophagy, independent of gross changes in cytosolic Ca²⁺ signaling.

_mCa²⁺ uptake is also specifically involved in mitophagy. A key trigger for mitophagy is mitochondrial depolarization associated with permeability transition (489–491), which can occur as a consequence of _mCa²⁺ overload. MacVicar et al. (492) further suggest that ER-to-mitochondria Ca²⁺ transfer “primes” mitochondria for mitophagy, because knockdown of MCU prevents mitophagy that otherwise follows pharmacological mitochondrial depolarization. Thus, changes in _mCa²⁺ handling influence autophagy and mitophagy indirectly via effects on cellular energetics, and directly by affecting the susceptibility to PT as well as the mitochondria’s response to depolarization.

6.5. Influence on Mitochondrial Signaling to Other Cellular Compartments

An emerging concept is that mitochondria actively signal to other cellular compartments (FIGURE 9).

FIGURE 9.

Mitochondrial Ca²⁺ handling as a master regulator of cellular signaling pathways. Clockwise from top left: _mCa²⁺ buffering and shaping of cytosolic Ca²⁺ signaling: Net _mCa²⁺ flux influences the amount of cytosolic Ca²⁺ available to participate in cytosolic signaling pathways. For instance, changes in net _mCa²⁺ uptake can influence NFAT signaling and affect gene expression by influencing amount of cytosolic Ca²⁺ available to activate the Ca²⁺-sensitive phosphatase calcineurin. Retrograde signaling to the cytosol and nucleus: Prolyl hydroxylases (PHDs) hydroxylate the transcription factor hypoxia-inducible factor (HIF-1α) and target it for degradation. Increased net _mCa²⁺ accumulation triggers the mitochondrial production of reactive oxygen species (ROS), which inhibit PHDs, thereby favoring HIF-1α accumulation and the transcription of its target genes. Mitochondrial Ca²⁺ in plasma membrane repair: Lesion of the plasma membrane allows extracellular Ca²⁺ to flow into the cell, elevating cytosolic Ca²⁺ concentration. Subsequent _mCa²⁺ uptake drives increased mitochondrial production of ROS. ROS activates the GTPase RhoA, which then promotes F-actin polymerization to facilitate plasma membrane repair. Effective resealing of the plasma membrane completes this feedback loop by preventing further influx of extracellular Ca²⁺. Mitochondrial control of epigenetics and cellular differentiation: Profibrotic stimulation of fibroblasts upregulates the mtCU gatekeeper MICU1, thereby attenuating _mCa²⁺ uptake. Altered _mCa²⁺ handling remodels fibroblast metabolism and increases glutaminolysis to increase the cellular supply of α-ketoglutarate (αKG). αKG serves as a cofactor for histone demethylases and so modulates gene expression that ultimately causes differentiation of the fibroblasts into myofibroblasts. See glossary for abbreviations.

A classic example of retrograde signaling between the mitochondria and nucleus is that mitochondrial dysfunction triggers depolarization of ΔΨ_m, which favors release of _mCa²⁺ to the cytosol. Increased cytosolic Ca²⁺ stimulates the phosphatase calcineurin and the kinases PKC, JNK, CamKIV, and MAPK, leading to activation of the transcription factors NFAT, NF-κB, ATF2, CEBP/δ, CREB, Egr-1, and CHOP (25). This allows changes in mitochondrial function to alter nuclear gene transcription. Another example of retrograde signaling that can be regulated by _mCa²⁺ is mitochondrial ROS production, which activates the transcription factor hypoxia-inducible factor (HIF-1) (493) and can affect cell differentiation and proliferation (494). TCA cycle intermediates such as acetyl-CoA, succinate, fumarate, and α-ketoglutarate can function as signaling molecules by affecting the activity of proteins such as prolyl hydroxylases and Jumonji domain-containing histone demethylase (reviewed in Refs. 495–497). Notably, acetyl-CoA and succinyl-CoA also act as substrates for the posttranslational modification of lysine residues (498–500); a familiar example is histone acetylation. Mitochondrial metabolism influences levels of S-adenosyl methionine, which is required for DNA methylation (501, 502). By altering the availability of such compounds, changes in mitochondrial metabolism can thus affect epigenetic marks and patterns of gene expression. That altered _mCa²⁺ homeostasis is sufficient to influence this kind of signaling is supported by the observation that MICU1 upregulation and diminished _mCa²⁺ uptake drives metabolic reprogramming to control epigenetic signals critical for myofibroblast differentiation (165, 211).

Several additional examples illustrate how changes in _mCa²⁺ handling affect signaling to other parts of the cell. In lung carcinoma cells, depletion of mitochondrial DNA leads to loss of mitochondrial membrane potential, diminishing the driving force for _mCa²⁺ uptake and so elevating cytosolic Ca²⁺. This in turn increases the expression of genes involved in tumor invasion and increases invasive behavior, indicating that altered _mCa²⁺ handling and retrograde signaling contribute to cancer metastasis (503). In plants, the upregulation of genes involved in adaptive responses to salt stress is blocked by MCU inhibition, suggesting that increased _mCa²⁺ is a key step in this signaling pathway (504). Together these studies reveal that both increased and diminished _mCa²⁺ uptake affect retrograde signaling to regulate gene expression. Finally, in C. elegans _mCa²⁺ uptake through MCU stimulates superoxide production near skin wound sites; subsequent redox modulation of RHO-1 GTPase then facilitates actin-dependent wound closure (505). A similar phenomenon occurs in mammals to allow resealing of the sarcolemma in damaged skeletal muscle cells, although in this instance _mCa²⁺-driven ROS production instead activates RhoA GTPase to promote F-actin accumulation near the site of the membrane lesion and facilitate membrane resealing (208). _mCa²⁺ uptake following sarcolemmal injury is impaired in MICU1-deficient muscle cells, leading to impaired membrane repair (207). Thus, defective _mCa²⁺-dependent signaling and plasma membrane sealing likely contribute to ongoing muscle damage and exaggerated muscle weakness in patients with MICU1 mutations (199).

7. MITOCHONDRIAL Ca²⁺ FLUX IN DISEASE

As noted above, several human genetic diseases are caused by primary disruption of _mCa²⁺-handling proteins, such as myopathy (MICU1), a neurodevelopmental disorder (MICU2), and Wolf–Hirschhorn syndrome (LETM1) (TABLE 1). Secondary alterations in the expression of nearly all the proteins involved in _mCa²⁺ exchange are reported in the context of at least one acquired human disease. Here, we discuss evidence for perturbations in _mCa²⁺ homeostasis as a driving factor behind these pathologies.

7.1. Cardiac Ischemia-Reperfusion Injury and Heart Failure

Acute mitochondrial Ca²⁺ overload resulting in permeability transition is a major contributor to cardiomyocyte death that results from myocardial infarction and I/R injury (reviewed in Refs. 400, 401) (FIGURE 10).

FIGURE 10.

Either too much or too little _mCa²⁺ can contribute to disease. _mCa²⁺ has pleiotropic roles in diseases such as cancer, neurodegeneration, and heart failure that relate to its fundamental functions in supporting mitochondrial bioenergetics, influencing mitochondrial signaling, and cell death. For example, acute _mCa²⁺ overload during cardiac ischemia-reperfusion injury causes excess reactive oxygen species (ROS) production and elicits mitochondrial permeability transition, which compromises mitochondrial ATP production and triggers cardiomyocyte death. These immediate effects of _mCa²⁺ overload impair the heart’s mechanical pump function and over time cause the heart to fail. In end-stage heart failure, however, _mCa²⁺ content may be reduced below normal levels because of excessive mitochondrial Na⁺/Ca²⁺ exchange. Here, diminished _mCa²⁺-dependent stimulation of TCA cycle flux can disrupt mitochondrial redox balance, favor ROS production, and impair mitochondrial ATP production. The diminished ATP supply and excessive oxidative stress drive further impairment of the heart’s contractile function. See glossary for abbreviations.

Research in mouse models shows that acute deletion of MCU from adult cardiomyocytes protects the heart from I/R injury and cell death (126, 127). In a similar fashion, overexpression of NCLX to enhance cardiomyocyte _mCa²⁺ efflux, or upregulation of MCUB to attenuate _mCa²⁺ uptake, can also protect against I/R injury and, in the case of NCLX upregulation, is sufficient to preserve contractile function even after permanent myocardial infarction (166, 242). The heart has some endogenous capacity to upregulate such adaptive proteins in order to minimize I/R injury, as both NCLX and MCUB gene expression are increased after ischemic insult to the heart (166, 242). Interestingly, MICU1 transcripts are also elevated in end-stage ischemic heart failure (166). In the short term (up to 24 h of reperfusion), MICU1’s gating of the mtCU has a protective role in I/R, as knockdown of MICU1 exaggerates I/R-induced _mCa²⁺ overload, increases infarct size and apoptosis, and worsens contractile function (209). However, whether MICU1 upregulation in end-stage failure is an ongoing mechanism to minimize _mCa²⁺ overload, or rather is an adaptation to facilitate cooperative activation of _mCa²⁺ uptake in support of cardiac energetics, is less clear.

To this point, beyond its initial role in triggering cardiomyocyte death during I/R, excessive _mCa²⁺ uptake through the mtCU has additional consequences that lead to further deterioration of contractile function over time after I/R injury. Increased mitochondrial ROS production after myocardial infarction oxidizes critical residues on the cardiac RyR, leading to SR Ca²⁺ leak that further exacerbates _mCa²⁺ loading and ROS production (506). This interorganelle cross talk can set up a cycle of positive feedback favoring continued SR Ca²⁺ leak, mitochondrial dysfunction, and further ROS production that contributes to the progressive decline in contractile function in the weeks following infarction (506). Thus, therapeutic strategies that limit excessive, pathological _mCa²⁺ uptake should be beneficial both during I/R and during the course of cardiac decompensation that occurs afterward.

Current understanding of the role of excessive _mCa²⁺ uptake in chronic, nonischemic heart failure that develops in response to hemodynamic stresses such as hypertension or aortic stenosis is more complicated. During I/R, cardiomyocytes are subjected to acute elevations in cytosolic Ca²⁺ and subsequent _mCa²⁺ overload. In contrast, in conditions of chronic hemodynamic stress, cardiomyocytes are subjected to a sustained increase in cytosolic Ca²⁺ signals that may be of lower magnitude than the massive cytosolic Ca²⁺ stress experienced in I/R. Thus, blockade of acute _mCa²⁺ uptake through the mtCU may be insufficient to prevent an increase in _mCa²⁺ content in this context, because there may be ongoing, increased _mCa²⁺ entry via alternative Ca²⁺ transporters that can be activated at lower Ca²⁺ thresholds than the mtCU. This topic has recently been reviewed elsewhere (121).

Finally, whereas excessive _mCa²⁺ loading drives cardiac dysfunction in the early stages of ischemic heart failure, an interesting hypothesis is that the end-stage failing heart is instead characterized by insufficient _mCa²⁺ content that is inadequate to meet bioenergetic needs and maintain mitochondrial redox balance (FIGURE 10). In the failing heart, cytosolic Na⁺ concentration is increased by ∼3–6 mM over its normal concentration of ∼7–8 mM (21, 507). This increases the driving force for Na⁺ entry into the mitochondria (matrix Na⁺ ∼6 mM) and so stimulates Na⁺/Ca²⁺ exchange through NCLX, which limits the accumulation of _mCa²⁺ (21, 508, 509). This effect is sufficient to impair TCA cycle NADH production, particularly under conditions of increased cardiac workload (508, 509). By limiting NADH production, enhanced _mCa²⁺ extrusion may have secondary effects to limit ETC flux and mitochondrial ATP production, further limiting contractility of the failing heart. In agreement with this concept, compensatory changes in _mCa²⁺-handling proteins are observed as an adaptive response in a mouse model of mitochondrial cardiomyopathy. Mice with cardiac-specific ablation of the mitochondrial transcription factor TFAM develop dilated cardiomyopathy. In these animals, increased MCU expression enhances _mCa²⁺ uptake and decreased NCLX expression minimizes _mCa²⁺ efflux in order to increase the sensitivity of mitochondrial metabolism to increases in cytosolic Ca²⁺ concentration, ultimately helping to maintain cardiac bioenergetics (510). The recent report that increasing cardiac MCU expression in a guinea pig model of severe heart failure protects contractile function strengthens this notion (511, 512). Together, these findings indicate that promoting _mCa²⁺ accumulation, rather than inhibiting it, may be an appropriate therapeutic goal in a heart that is already failing.

7.2. Stroke

I/R injury constitutes a fundamental mechanism that damages the brain upon recanalization of an occluded artery following ischemic stroke (513). Just like in cardiac I/R injury, excessive uptake of Ca²⁺ through the mtCU is a critical factor in the ensuing damage to brain tissue. In rats subjected to middle cerebral artery occlusion and reperfusion, I/R causes progressive inhibition of the ETC, impairment of ATP production, dissipation of ΔΨ_m, and increased ROS production, leading to neuronal damage and apoptosis (514). Pharmacological MCU inhibition prior to ischemia attenuates these changes and reduces brain infarct volume, whereas pharmacological MCU activation exacerbates brain injury (514). Conditional knockout of MCU in adult neurons similarly protects mice from hypoxic/ischemic brain injury and minimizes neuron loss and mitochondrial damage. Neuronal MCU ablation also attenuates sensorimotor deficits associated with hypoxic/ischemic injury (515), suggesting that this therapeutic strategy improves functional outcomes at the level of the intact organism. These findings together provide therapeutic rationale for inhibiting _mCa²⁺ uptake in stroke. Inhibition of _mCa²⁺ uptake also attenuates brain damage following acute traumatic brain injury, which like stroke is characterized by excessive ROS production, impaired mitochondrial energetics, and apoptosis (516).

7.3. Neurodegenerative Disease

Altered mitochondrial function leading to impaired oxidative phosphorylation, oxidative stress, and apoptosis are common features of neurodegenerative diseases such as Alzheimer’s disease (AD), Parkinson’s disease (PD), amyotrophic lateral sclerosis (ALS), and Huntington’s disease (HD) (517, 518). _mCa²⁺ overload is a common feature of neurodegenerative diseases and may impair cellular energetics by inducing oxidative stress and by triggering mitochondrial permeability transition, leading to collapse of ΔΨ_m that compromises ATP production (519, 520). In some cases, though, diminished _mCa²⁺ accumulation is implicated in neurodegeneration.

7.3.1. Alzheimer’s disease.

Alzheimer’s disease (AD) is a form of dementia characterized by neuronal dysfunction and cell death, and has hallmark pathological features including the deposition of amyloid-β plaques and neurofibrillary tangles comprised of tau protein (521). Increased connection of the ER and mitochondria is noted in cell models of AD, raising the possibility of increased communication between these organelles (522). Fitting with this view, overexpression of presenilin 2 mutants linked to familial AD increases Ca²⁺ transfer from the ER to the mitochondria (523). Several reports demonstrate that this phenomenon results from effects of mutant presenilin to increase ER Ca²⁺ release through the IP₃R, thereby enhancing _mCa²⁺ uptake that in turn increases the likelihood of _mCa²⁺ overload and permeability transition (524–527). That disruption of a single Itpr1 allele and subsequent reduction of IP₃R1 expression by ∼50% protects against AD pathology and ameliorates memory impairments in the 3xTg mouse model of AD (528) supports a causative role for elevated cytosolic Ca²⁺ signaling in this disease. Increased phosphorylation and altered oxidative modification of RyR2 in AD neurons similarly favor excessive ER Ca²⁺ leak, and such altered RyR2 activity has been implicated in cognitive dysfunction (529). Amyloid-β oligomers can also stimulate ER Ca²⁺ release, leading to _mCa²⁺ overload and neuronal death (530, 531). These findings are consistent with the observation that senile plaques impair neuronal Ca²⁺ homeostasis in the APP mouse model of AD (532) and that the presence of neurofibrillary tangles is associated with elevated _iCa²⁺ concentration in neurons (533, 534). These and numerous other observations beyond the scope of this discussion have collectively led to the formulation of a “Ca²⁺ hypothesis of AD,” which posits that dysregulated cellular Ca²⁺ handling drives AD pathogenesis (535). The reader is directed to several comprehensive reviews (535–537) for an in-depth examination of data in support of this hypothesis.

Altered _mCa²⁺ homeostasis resulting directly from either aberrant cytosolic Ca²⁺ signaling or altered expression or function of _mCa²⁺ handling proteins is emerging as a central mechanism behind neuronal dysfunction in AD (reviewed in Ref. 538). For example, the presence of misfolded or mutated tau protein is sufficient to inhibit NCLX activity in neurons and astrocytes, and thereby increases their susceptibility to Ca²⁺-induced death (539). Our laboratory’s findings of reduced NCLX expression in AD brains, and the ability for NCLX overexpression to rescue numerous AD phenotypes both in vitro and in vivo, further support the notion that increased _mCa²⁺ load drives AD progression (323). However, one recent report suggests that, in aged neurons, amyloid-β oligomers still increase ER-mitochondria contacts but surprisingly inhibit ER-to mitochondria Ca²⁺ transfer. Impaired _mCa²⁺ uptake has direct consequences to limit TCA cycle flux and thus can alter mitochondrial redox balance, contributing to excessive ROS accumulation and apoptosis (540). This work complicates our understanding of the role of _mCa²⁺ in AD and suggests that treatments aimed at limiting versus increasing _mCa²⁺ content may have to be matched to the particular stage of the disease.

7.3.2. Parkinson’s disease.

Parkinson’s disease (PD) is caused by death of dopaminergic neurons in the substantia nigra pars compacta and is characterized by motor symptoms including bradykinesia, tremor, muscle rigidity, and gait impairments (541). In models of PD caused by mutant α-synuclein, excessive _mCa²⁺ uptake causes mitochondrial ROS production, collapse of ΔΨ_m, and impaired ATP production (542, 543). Loss of function mutations in PTEN-induced kinase 1 (PINK1) also cause PD and are associated with _mCa²⁺ overload. PINK1 deficiency impairs NCLX activity and elicits _mCa²⁺ overload in dopaminergic neurons (328, 544). PKA-dependent phosphorylation of NCLX is able to restore _mCa²⁺ extrusion in these cells, suggesting a therapeutic avenue to prevent _mCa²⁺ overload in this setting (328). Likewise, genetic disruption or pharmacological inactivation of MCU improves mitochondrial bioenergetics and prevents the death of dopaminergic neurons in pink1-mutant zebrafish (545). The E3 ubiquitin ligase Parkin promotes the degradation of MICU1 protein, and Parkin mutations cause familial PD. Thus, Matteucci et al. (219) propose that altered proteostasis of MICU1 contributes to altered _mCa²⁺ handing in PD. Finally, expression of mutant forms of leucine-rich repeat kinase 2 (LRKK2) associated with PD increase the transcription of MCU and MICU2 and increase _mCa²⁺ uptake in mouse cortical neurons and PD patient fibroblasts, making them susceptible to injury and neurite and dendrite shortening (118). Impaired NCLX activity also contributes to _mCa²⁺ overload in LRKK2-deficient neurons (324). _mCa²⁺ overload and associated deleterious effects are attenuated by pharmacological MCU inhibition, knockdown of MCU, expression of the constitutively active phosphomimetic S258D NCLX mutant, or pharmacological activation of PKA to promote stimulatory NCLX phosphorylation (118, 324). These results support a causative role for both increased _mCa²⁺ uptake and diminished _mCa²⁺ efflux in PD-associated neurodegeneration.

7.3.3. Amyotrophic lateral sclerosis.

Amyotrophic lateral sclerosis (ALS) is a progressive neurodegenerative disorder characterized by muscle weakness, muscle atrophy, and eventual paralysis due to the degeneration of motor neurons in the brain, brain stem, and spinal cord (546). Impaired neuronal _mCa²⁺ uptake is noted in cell and mouse models of ALS and is associated with impaired complex I activity, depolarized ΔΨ_m, and impaired ATP synthesis (547–549). Likewise, induced pluripotent stem cell-derived motor neurons from ALS patients with C9ORF72 or TARDBP mutations exhibit impaired _mCa²⁺ buffering. As a result, cytosolic Ca²⁺ concentration becomes abnormally elevated during glutamate stimulation, causing excitotoxity (550). These studies suggest that impaired _mCa²⁺ uptake contributes to motor neuron death in ALS. In agreement with this notion, MCU is downregulated in motor neurons from presymptomatic and symptomatic ALS model mice (551). However, motor neurons that still remain at the end stage of ALS show upregulation of MCU, MICU1, and LETM1, suggesting that neurons with the highest capacity for _mCa²⁺ uptake or overall _mCa²⁺ buffering may be the most resistant to death (552). This implies that upregulation of _mCa²⁺ levels could be protective in ALS. Fitting with this view, inhibition of mitochondrial Na⁺/Ca²⁺ exchange with the NCLX inhibitor GCP-37157 prevents kainate-induced excitotoxicity in a mouse motor neuron model of ALS (553). Conflicting reports, though, indicate instead that, in embryonic motor neurons from ALS mice, MCU is upregulated and pharmacological suppression of MCU is protective against excitotoxicity (551). Thus, like in AD, properties of _mCa²⁺ exchange may change throughout the progression of ALS. Therapeutic goals centering on control of _mCa²⁺ homeostasis may therefore need to be tailored to distinct stages of the disease.

7.3.4. Huntington’s disease.

Huntington’s disease (HD) results from the expansion of a poly-glutamine tract in the huntingtin gene (517). Acute _mCa²⁺ uptake is impaired in HD (554); however, the susceptibility to Ca²⁺-induced depolarization and permeability transition is increased in HD mitochondria (555, 556). These changes in Ca²⁺ handling precede the development of pathology and behavioral abnormalities in HD mutant mice (554), but whether they indeed contribute to HD progression remains to be tested.

7.4. Diabetes and Obesity

Mitochondrial Ca²⁺ exchange has direct roles to regulate pancreatic insulin secretion by supporting mitochondrial ATP production and shaping cytosolic Ca²⁺ signals required for this process (290, 313, 361, 458, 460, 461). Therefore, alterations in _mCa²⁺ uptake and efflux have the potential to influence whole body glucose homeostasis. Mutations in _mCa²⁺ handling genes have not been causatively linked to either type 1 or type 2 diabetes mellitus. However, some reports suggest that diabetes and obesity are associated with changes in _mCa²⁺ handling proteins that can contribute to the characteristic sequelae of these conditions.

Chronically increased circulating glucose in uncontrolled diabetes impairs pancreatic β-cell insulin secretion (557), further disrupting glucose homeostasis. Chronic exposure of human pancreatic islets to high glucose in vitro upregulates EMRE, downregulates MICU3, and upregulates NCLX (558), indicating that impaired insulin secretion under chronic glucose stress may result from remodeling of _mCa²⁺ exchange. Likewise, changes in _mCa²⁺ handling proteins are observed in multiple tissues that show dysfunction in diabetes. MCU protein expression and _mCa²⁺ content are reduced in cardiomyocytes exposed to high glucose (559). Similarly, MCU and EMRE expression are decreased and MCUB expression is increased in the hearts of mice with streptozotocin-induced type 1 diabetes, resulting in impaired _mCa²⁺ uptake, compromised oxidative metabolism, and impaired cardiomyocyte contractility (560). Viral overexpression of MCU corrects these defects. At the same time, impaired _mCa²⁺ uptake due to downregulation of MICU1 is implicated in contractile dysfunction, cardiac hypertrophy, and fibrosis in the db/db model of type 2 diabetes (215). This study indicates that diminished _mCa²⁺ content causes myocardial apoptosis because it limits NAD(P)H production, thereby limiting mitochondrial antioxidant potential and increasing mitochondrial ROS signaling. These findings implicate impaired _mCa²⁺ uptake in diabetic cardiomyopathy. Downregulation of MICU1 is also reported in a mouse model of high-fat diet-induced obesity, but in this instance it instead leads to increased _mCa²⁺ accumulation and associated pathological cardiac remodeling (216).The discrepancies between these studies of diabetes versus obesity on the effect of MICU1 downregulation on _mCa²⁺ content likely reflect the biphasic nature of MICU1’s regulation of the mtCU.

Finally, hyperglycemia upregulates MCU and MCUR1 and increases _mCa²⁺ content in endothelial cells. These changes contribute to increased ROS production and apoptosis, suggesting that increased _mCa²⁺ uptake plays a role in endothelial dysfunction in diabetes (561). _mCa²⁺ uptake increases in the liver of rats with streptozotocin-induced type 1 diabetes, although in this case enhanced _mCa²⁺ results from downregulation of MCUB (243). Collectively, these studies support the notion that modulation of mtCU composition in response to metabolic dysregulation plays a role in pathological changes that commonly present alongside diabetes and obesity.

7.5. Cancer

Cancer is characterized by metabolic reprogramming including a shift to aerobic glycolysis (the “Warburg effect”) and associated phenotypic changes that allow unrestricted cell proliferation, resistance to cell death, and the spread of cancerous cells throughout the body. Given the central role of mitochondria in providing molecular building blocks for anabolism, modulating cellular Ca²⁺ homeostasis and cell death, and coordinating changes in nuclear gene transcription, increasing research is directed at understanding the role of mitochondrial biology in cancer (reviewed in Refs. 562, 563). Alterations in Ca²⁺ flux can affect all of these mitochondrial functions, and, indeed, growing evidence indicates that the mitochondrial Ca²⁺-handling machinery and _mCa²⁺ homeostasis are altered in diverse forms of cancer.

MCU expression is increased in some breast cancers and positively correlates with tumor size and infiltration (564, 565). This may result from downregulation of miR-340, which normally inhibits MCU expression (141). Upregulation of MCU promotes metabolic adaptations that favor prosurvival signaling (564) and metastasis (141). For example, increased _mCa²⁺ uptake through MCU increases mitochondrial ROS production, leading to induction of HIF-1α (565). Induction of this same pathway occurs in colorectal cancer as a result of NCLX downregulation and _mCa²⁺ overload (325). RIPK1 is overexpressed in colon cancer, binds to MCU, and stimulates _mCa²⁺ uptake to promote cancer growth and oncogenesis (566). Thus, procancer signaling also promotes _mCa²⁺ uptake, independent of changes in MCU expression. Finally, silencing of MCU in colon cancer due to increased expression of miR-25 is associated with increased resistance to apoptosis (139). This effect on MCU still favors tumor survival, albeit by a distinct mechanism that prevents lethal _mCa²⁺ overload.

In pancreatic cancer, downregulation of histidine triad nucleotide-binding protein (HINT2), a protein that sensitizes cells to apoptosis, is correlated with poor prognosis. Restoration of HINT2 expression increases expression of EMRE and downregulates the mtCU gatekeepers MICU1 and MICU2, and resensitizes cells to apoptosis by favoring _mCa²⁺ overload (567). These data suggest that modulation of regulatory mtCU components to limit lethal _mCa²⁺ accumulation is one mechanism that cancers cells employ to avoid apoptosis. Fitting with this view, EMRE expression is decreased in pancreatic adenocarcinoma, and expression of EMRE is positively correlated with better prognosis in this type of cancer. This protective effect of EMRE is related to increased cell death signaling, which likely results from a higher propensity for _mCa²⁺ overload (568). Increased MICU1 expression in ovarian cancer, resulting from reduced expression of miR-195, can limit _mCa²⁺ uptake and so prevent apoptosis to enable tumor cell survival. Increased MICU1 expression in this setting also stimulates aerobic glycolysis and associated metabolic reprogramming that drives cell growth, migration, and invasion (214, 569). Likewise, increased expression of the ribosomal protein S3 in melanoma and enhancer of zeste homolog 2 in squamous cell carcinoma can drive MICU1 upregulation, and so increase resistance to cell death (570, 571). In contrast, other studies indicate that increased expression of mtCU regulators that promote rather than oppose _mCa²⁺ uptake favor cancer progression. Upregulation of MCUR1 in hepatocellular carcinoma promotes _mCa²⁺ uptake and drives mitochondrial ROS signaling that leads to degradation of p53 and disruption of apoptotic pathways (572), and to Notch signaling that stimulates epithelial-mesenchymal transition and metastasis (254).

These diverse pro- and anticancer effects of the expression of _mCa²⁺-handling genes reflect the wide range of consequences of _mCa²⁺ uptake, from the more physiological range of _mCa²⁺ concentrations that regulate metabolism and signaling up to toxic levels of _mCa²⁺ that elicit permeability transition and cell death (FIGURE 8). The conclusion that altering _mCa²⁺ flux may therefore promote or inhibit cancer cell growth and survival, depending on the context, highlights the challenges in targeting _mCa²⁺ exchange as a generalizable therapeutic approach for cancer.

8. CONCLUSIONS AND FUTURE DIRECTIONS

8.1. The Case for Modulating Mitochondrial Ca²⁺ Flux as a Therapy for Disease

Mitochondrial Ca²⁺ exchange has emerged as a fundamental cellular process that, when perturbed, can initiate disease or help to drive disease progression. Since 2010, the genetic identification of many of the proteins that mediate _mCa²⁺ uptake and _mCa²⁺ efflux has greatly advanced our understanding of how mitochondrial Ca²⁺ exchange operates on a molecular level. It has also enabled the generation of genetic animal models that have been used to understand the cellular and molecular mechanisms by which an imbalance between _mCa²⁺ uptake and efflux impairs cellular, tissue, and whole organism function. Increasing evidence suggests that _mCa²⁺ homeostasis is indeed disrupted both in response to acute pathological stress such as ischemia-reperfusion and as a secondary consequence of chronic acquired diseases such as neurodegeneration and cancer. The next challenge is to understand 1) whether therapeutic manipulation of _mCa²⁺ exchange is sufficient to alter the course of these diseases and 2) whether the appropriate goal in each case should be to increase or decrease net _mCa²⁺ accumulation. Research into cardiac I/R injury in genetic animal models with knockout or overexpression of _mCa²⁺ exchangers offers powerful proof-of-concept evidence that modifying _mCa²⁺ flux is indeed an effective strategy to protect the heart from ischemic damage. Recent studies in mouse models of Alzheimer’s disease likewise argue for the utility of this approach in slowing the progression of neurodegeneration. More examples of how modifying _mCa²⁺ exchange in vivo improves or exacerbates disease are sure to follow. Future research should focus on the development of therapies that can modulate _mCa²⁺ exchange specifically in those tissues affected by a given disease. More research is also needed to understand how _mCa²⁺ handling may change and influence pathology via distinct mechanisms throughout different stages of chronic disease. This would provide needed insight into how therapeutic strategies for a given patient may need to be altered over time. Fulfilling these two goals will bring us closer to translating our fundamental understanding of _mCa²⁺ exchange into clinically useful approaches for the treatment of human disease.

8.2. Unanswered Questions and Future Directions

Several major unanswered questions regarding the basic molecular biology of _mCa²⁺ exchange remain and are directly relevant to the development of therapeutic approaches targeting _mCa²⁺ flux. First, have we identified all of the relevant players in _mCa²⁺ exchange? Our incomplete understanding of the molecular composition of the mPTP is one example of this problem, with relevance for both physiological _mCa²⁺ efflux and cellular responses to _mCa²⁺ overload. Likewise, do any additional proteins besides those discussed above contribute to MCU-independent _mCa²⁺ uptake? Second, do we fully understand the function of the _mCa²⁺-handling proteins we have already identified? For example, might MCUB, when paired with the appropriate regulatory subunits, have some capacity to conduct Ca²⁺? And how do we appropriately distinguish the _mCa²⁺-dependent versus _mCa²⁺-independent effects of a protein like MICU1, which has additional cellular functions independent of the mtCU? Third, how are the composition and stoichiometry of the mitochondrial calcium uniporter complex maintained? Is mtCU protein composition a direct reflection of gene expression of the various mtCU components, or do active mechanisms select particular subunits such as MCUB for incorporation into or exclusion from the complex? Fourth, how is the molecular function of each protein involved in _mCa²⁺ exchange regulated? This is a particularly relevant translational question for NCLX, since increasing NCLX activity is protective in several mouse models of human disease but no reliable pharmacological activators for NCLX yet exist. And finally, what are the compensatory mechanisms that eventually enable tissues to metabolically adapt to the loss of mtCU function? How might we exploit these pathways therapeutically to prevent deleterious _mCa²⁺ overload, while at the same time preserving mitochondrial bioenergetics? Resolving these fundamental questions regarding _mCa²⁺ exchange will help us to attain the therapeutic objectives outlined above.

GLOSSARY

AA

Amino acid

AD

Alzheimer’s disease

ALS

Amyotrophic lateral sclerosis

Ca²⁺

Calcium

COX

Cytochrome-c oxidase

CypD

Cyclophilin D

EMRE

Essential MCU regulator

ER

Endoplasmic reticulum

ETC

Electron transport chain

H⁺

Proton

HD

Huntington’s disease

_iCa²⁺

Intracellular calcium

IDH

Isocitrate dehydrogenase

IMM

Inner mitochondrial membrane

IMS

Intermembrane space

IP₃

Inositol-(1,4,5)-trisphosphate

IP₃R

Inositol-(1,4,5)-trisphosphate receptor

I/R

Ischemia-reperfusion

K⁺

Potassium

KO

Knockout

LETM1

Leucine zipper and EF-hand containing transmembrane protein 1

Li⁺

Lithium

_mCa²⁺

Mitochondrial Ca²⁺

MCU

Mitochondrial calcium uniporter

MCUB

Mitochondrial calcium uniporter dominant-negative beta subunit

MCUR1

Mitochondrial calcium uniporter regulator 1

MEF

Mouse embryonic fibroblast

Mg²⁺

Magnesium

_mH⁺

Mitochondrial H⁺

MICU1

Mitochondrial calcium uptake protein 1

MICU2

Mitochondrial calcium uptake protein 2

MICU3

Mitochondrial calcium uptake protein 3

Mn²⁺

Manganese

mPTP

Mitochondrial permeability transition pore

mRyR

Mitochondrial ryanodine receptor

mtCU:

Mitochondrial calcium uniporter channel

Na⁺

Sodium

NCLX

Mitochondrial Na⁺/Ca²⁺,Li⁺ exchanger

OCR

Oxygen consumption rate

OMM

Outer mitochondrial membrane

OXPHOS

Oxidative phosphorylation

PD

Parkinson’s disease

PDH

Pyruvate dehydrogenase

PT

Permeability transition

RaM

Rapid mode of Ca²⁺ uptake

ROS

Reactive oxygen species

RyR

Ryanodine receptor

SERCA

Sarco(endo)plasmic reticulum Ca²⁺-ATPase

SLC25A23

Mitochondrial ATP-magnesium/phosphate carrier

SR

Sarcoplasmic reticulum

TCA

Tricarboxylic acid

VDAC

Voltage-dependent anion channel

WHS

Wolf–Hirschhorn syndrome

WT

Wild type

αKGDH

α-Ketoglutarate dehydrogenase

ΔΨm

Mitochondrial membrane potential

GRANTS

This work was supported by National Institutes of Health (T32HL091804 to J.F.G.; F32HL151146 to J.F.G.; P01HL147841, R01HL142271, R01HL136954, P01HL134608, and R01NS121379 to J.W.E.) and the American Heart Association (20EIA35320226 to J.W.E.).

DISCLOSURES

J.W.E. is a paid consultant for Mitobridge and Janssen. J.F.G. has no conflicts of interest, financial or otherwise, to disclose.

AUTHOR CONTRIBUTIONS

J.F.G. prepared figures; J.F.G. and J.W.E. drafted manuscript; J.F.G. and J.W.E. edited and revised manuscript; J.F.G. and J.W.E. approved final version of manuscript.