Recent advances in the molecular mechanism of mitochondrial calcium uptake

The mitochondrial Ca²⁺ uniporter MCU

In 2011, two independent in silico screenings^22,23 identified the CCDC109a gene product as the long-sought pore-forming unit of the MCU channel mediating mitochondrial Ca²⁺ uptake in mammalian cells. Purified MCU protein reconstitution in lipid bilayers revealed a Ca²⁺ channel activity with the electrophysiological features of the hypothetical MCU²⁰. In addition, it was clearly shown that MCU downregulation in mammalian cells inhibits mitochondrial Ca²⁺ uptake while its overexpression increases mitochondrial Ca²⁺ accumulation, thus proving that it is bona fide the channel mediating Ca²⁺ uptake in mitochondria²³.

The sequence of MCU consists of two transmembrane α-helix domains spanning the inner mitochondrial membrane (IMM), the second of which contains the critical DIME motif responsible for the channel selectivity filter, linked by a short loop toward the mitochondrial intermembrane space (IMS), while both the C- and the N-termini face the mitochondrial matrix^22,24. A single MCU molecule per se is thus not able to form the channel but should organise in oligomers, and MCU was initially proposed to arrange itself into tetramers²⁵. This hypothesis was very recently confirmed by coherent high-resolution cryo-EM data from four independent groups on different MCU orthologs^24,26–28. These seminal studies finally unveiled the structure and arrangement of MCU protomers within the complex and the exact position of the channel selectivity filter. The 3D reconstruction of EM data unambiguously revealed that MCU forms tetramers with a nonobvious symmetry: the transmembrane domain displays a fourfold symmetry, while the N-terminal domain (NTD) on the matrix side shows a twofold symmetry axis. These findings confuted the pentameric MCU architecture that was previously proposed based on magnetic resonance and negative staining EM data from Caenorhabditis elegans MCU-ΔNTD protein, in which a significant portion of the NTD was removed to facilitate the analysis²⁹. In addition, the MCU DIME motif, which contains the two critical acidic residues directly involved in the cation coordination, has now been shown to reside at the beginning of––and thus integral to––the second transmembrane helix (TMH), and not in the loop connecting the two transmembrane helices, as previously suggested²⁹.

However, as Raffaello et al. showed, MCU is not the only pore-forming unit of the oligomer, but it associates with the protein encoded by the MCU paralog gene CCDC109b, which was thus named MCUb²⁵. The characterisation of MCUb demonstrated that it acts as a negative regulator of MCU activity both in lipid bilayer experiments and when overexpressed in mammalian cells²⁵ (see Figure 1). To note, in other organisms, such as Trypanosoma cruzi, the ortholog of MCUb acts as a Ca²⁺ conducting subunit and its overexpression enhances, rather than dampens, mitochondrial Ca²⁺ uptake³⁰. Interestingly, in another trypanosomatid, Trypanosoma brucei, two additional MCU isoforms were identified, MCUc and MCUd, which are also endowed with Ca²⁺ uniporter activity and are able to form heterotetrameric complexes with their homologues MCU and MCUb³¹. This highlights the existence of significant species-specific differences in the function and distribution of MCU orthologues, despite their sequence conservation, and this should be taken into account when studying different organisms. Moreover, the expression and relative proportion of MCU and MCUb vary significantly among different tissues in mammals; thus, each cell type may have different variants of the uniporter owing to the specific composition and stoichiometry of MCU subunits. This accounts for tissue-specific variations of the capacity of mitochondria to take up Ca²⁺ and perfectly fits with the electrophysiological recordings of MCU Ca²⁺ currents in mitochondria from different mammalian tissues³². In skeletal muscle, for example, the presence of a high MCU:MCUb ratio²⁵ matches the highest mitochondrial Ca²⁺ conductance recorded in this tissue³², whereas, in adult heart, the relatively elevated MCUb expression²⁵ results in a considerably low Ca²⁺ current in cardiomyocyte mitochondria³². In cardiac cells, in which ∼37% of the volume is occupied by mitochondria³³, this is crucial for preventing massive mitochondrial Ca²⁺ accumulation that would potentially cause undesired buffering of the [Ca²⁺]_i transients required for contraction, futile cycling of Ca²⁺ across the IMM, and, eventually, organelle Ca²⁺ overload and apoptosis. The control of the MCU:MCUb proportion in the mitochondrial Ca²⁺ channel pore domain is thus fundamental for the function and physiology of different tissues.

After the molecular identification of the MCU pore components and auxiliary factors, the path for the genetic manipulation of mitochondrial Ca²⁺ uptake machinery in cells and animal models was finally opened. The first MCU^–/– mouse was generated by Finkel’s group in 2013. It shows a relatively mild phenotype, and, despite the expected abrogation of mitochondrial Ca²⁺ uptake, it develops normally and displays unaffected basal metabolism³⁴. However, the MCU^–/– mouse shows increased plasma lactate levels after starvation and impaired exercise performance accompanied by a reduction in the activity of pyruvate dehydrogenase in skeletal muscle³⁴. These findings, together with the fact that MCU^–/– mouse survival depends on the genetic background (MCU deletion is embryonically lethal in a pure C57/BL6 background³⁵), pointed to a more subtle and less obvious involvement of mitochondrial Ca²⁺ uptake in organ metabolism and organism development. This issue has been elegantly addressed by recent studies implementing tissue-specific modulation of MCU expression (by ablation/downregulation or overexpression) in adult tissues. For example, cardiac-specific tamoxifen-inducible MCU deletion in adult mice, differently from the germline genetic ablation of MCU, which did not protect from ischemic-reperfusion injury³⁴, clearly protects from ischemia-reperfusion heart damage, abrogates the contractile responsiveness to β-adrenergic stimulation responsible for the so-called “fight-or-flight” response, and reduces heart bioenergetics reserve capacity, even though it does not induce phenotypic abnormalities either in basal conditions or after cardiac overload and does not alter cardiomyocytes’ resting [Ca²⁺]_mt^36–39. These results suggest that MCU may be dispensable for cardiac homeostasis in basal conditions, while it appears to play a major role in cardiac metabolic flexibility during acute stress. Similarly to what has been reported for the heart tissue, germline deletion of MCU did not protect from hypoxic/ischemic brain injury⁴⁰. However, brains from conditional neuronal-specific inducible MCU-deleted mice as well as primary cortical neurons silenced for MCU show a significant reduction of the hypoxic/ischemic damage and decreased cell death without impairment of neuronal mitochondria metabolism⁴¹. Once more, these findings point to the existence of a strong drive for organs and tissues to respond to chronic MCU ablation by establishing adaptive metabolic shunts in order to cope with/bypass impaired mitochondrial Ca²⁺ signalling. Such adaptation to mitochondria dysfunction has also been described in the cardiac-specific Tfam^–/– mouse, the murine model for human cardiomyopathies with mtDNA depletion⁴², in which the alteration in mitochondrial metabolism due to oxidative phosphorylation (OXPHOS) deficiency induces a secondary upregulation of MCU protein and a concomitant downregulation of NCLX transcription⁴³. The resulting increased mitochondrial Ca²⁺ uptake and reduced Ca²⁺ efflux in Tfam^–/– cardiac mitochondria, although inducing an elevated and potentially damaging [Ca²⁺]_mt, appear somehow functional to ensure an efficient Ca²⁺-dependent mitochondrial respiration that otherwise would be compromised by the pathology⁴¹.

Overall, these findings lead us to the following consideration: experimental systems featuring long-term genetic manipulation of the mitochondrial Ca²⁺ uptake machinery should be evaluated with caution, and the potential induction of unpredicted phenotypes, as a consequence of adaptive/maladaptive responses, should be taken into account. This is particularly relevant when in vivo models are considered.

In line with this, the acute but transient manipulation of MCU expression in skeletal muscle provided additional intriguing evidence of the wide-ranging influence of mitochondrial Ca²⁺ signalling in the maintenance of organ and organism homeostasis. The decrease or enhancement of [Ca²⁺]_mt by AAV-mediated MCU silencing or overexpression, respectively, demonstrated that mitochondrial Ca²⁺ uptake regulates myofibre trophism through the modulation of the activity of the insulin growth factor-1/Akt pathway and the transcription of the peroxisome proliferator-activated receptor gamma coactivator 1 alpha 4 gene (PGC1α4). In particular, the overexpression of MCU in both adult and neonatal muscles induces significant fibre hypertrophy, which is not accompanied by alterations in mitochondrial membrane potential or other metabolic parameters (such as glycogen content and succinate dehydrogenase activity)⁴⁴. On the contrary, MCU downregulation leads to marked fibre atrophy as demonstrated by reduction of fibre size, inhibition of Akt activity, reduction of PGC1α4 transcripts, and impaired activation of the pyruvate dehydrogenase complex⁴⁴, in agreement with the data from the MCU^–/– mice³⁴ (see Figure 2).

Figure 2. The role of mitochondrial Ca²⁺ signalling in skeletal muscle trophism.

Schematic representation of the effects of manipulating mitochondrial Ca²⁺ uptake in skeletal muscle fibres. Increased mitochondrial Ca²⁺ accumulation by enhanced expression of mitochondrial Ca²⁺ uniporter (MCU) or mitochondrial Ca²⁺ uptake protein 1.1 (MICU1.1) leads to muscle hypertrophy by stimulating fibre growth and ATP production (left). Reduction of mitochondrial Ca²⁺ uptake by downregulation of either MCU or MICU1.1 causes fibre atrophy and impaired ATP production (right). [Ca²⁺]_mt, mitochondrial Ca²⁺ concentration.

Moreover, the recent description of a mouse model bearing constitutive skeletal muscle-specific deletion of MCU (skMCU^–/– mice) has been fundamental to uncover the metabolic rewiring occurring at the level of skeletal muscle cell but also, importantly, systemically⁴⁵. In line with previous data^34,44, the muscle of skMCU^–/– mice shows decreased performance and smaller fibre size compared to that of control mice⁴⁵. Moreover, it shows a shift toward fast fibre type and an enhanced preference for fatty acid oxidative metabolism⁴⁵. Interestingly, the ablation of MCU in skeletal muscle also impinges on other tissues, as shown by the increased catabolic response in both liver and adipose tissue of skMCU^–/– animals⁴⁵. This clearly points out how the mitochondrial metabolic adaptation in response to altered Ca²⁺ dynamics (i.e. MCU deletion) within the muscle tissue may have major systemic impact on different tissues and, importantly, on the modulation of global metabolism.

The MICU proteins

MICU1 was the first regulator of mitochondrial Ca²⁺ uptake to be identified²¹, even before the identification of MCU itself. It was initially described as an EF-hand Ca²⁺-binding protein associated with the IMM that stimulates MCU channel opening and thus acts as a positive modulator of mitochondrial Ca²⁺ uptake. Subsequent findings defined the crucial role of MICU1 in keeping the channel inactive under resting conditions, i.e. at low [Ca²⁺]_i^46,47, thus preventing massive Ca²⁺ entry that will otherwise cross the IMM down the steep electrochemical gradient and cause harmful mitochondrial Ca²⁺ overload. MICU1 was thus appropriately defined as the gatekeeper of the MCU channel. This concept has been recently confirmed by two independent in vivo MICU1^–/– mouse models. These MICU1^–/– animals, despite normal embryonic development and growth, show a variable degree of postnatal lethality (one out of six to one out of seven surviving pups in the case of CRISPR/Cas9 mutants⁴⁸ and 100% lethality in the case of Cre recombinase/LoxP deleted alleles⁴⁹) mostly due to breathing failure for the underdevelopment of the respiration coordination centres (cerebellum and Purkinje cells)⁴⁹. The few surviving knockout mice display muscle impairment and ataxia⁴⁸, a phenotype very similar to that described in children with a MICU1 mutation⁵⁰. The analysis of MICU1^–/– mitochondria revealed increased matrix [Ca²⁺]_mt, increased sensitivity to permeability transition pore (PTP) opening and cell death, and reduced production of ATP, similar to what was reported in MICU1-silenced hepatocytes and Hela cells^46,47. Overall, these data clearly highlight the relevance of MICU1 gatekeeping function in the control of mitochondrial Ca²⁺ homeostasis and metabolism. Nevertheless, an additional intriguing feature of MICU1 action has recently emerged, which consists of its ability to confer MCU selectivity for Ca²⁺ over Mn²⁺. Indeed, the DIME motif in the MCU pore subunit appears to be unable per se to discriminate between Ca²⁺ and Mn²⁺ in the absence of MICU1⁵¹. However, when present, MICU1 prevents Mn²⁺ entry in mitochondria, thus ensuring the stringent Ca²⁺ selectivity of MCU⁵¹. In this scenario, the proper stoichiometry of MICU1 is thus crucial to avoid mitochondrial Mn²⁺ uptake and to prevent Mn²⁺ toxicity. This concept may have relevant implications for human pathologies, especially for those diseases caused by MICU1 deficiency^50,52,53 but also for Parkinson’s disease, as already suggested⁵¹.

Additional MICU family members have been identified by genome sequence analysis soon after MICU1, namely MICU2 (EFHA1) and MICU3 (EFHA2)⁵⁴, which share 41% and 34% identity with MICU1, respectively, but display a broad and non-overlapping expression pattern compared to MICU1⁵⁵. In particular, MICU2 was shown to bind covalently to MICU1 through disulphide bridges, and the resulting dimer endows the genuine MCU gatekeeper activity⁵⁶. In this scenario, cytosolic Ca²⁺ elevation causes Ca²⁺ binding to the EF-hand domains of the MICU1–MICU2 dimer, allowing MICU1 to function as a positive regulator of MCU activity, thus efficiently promoting Ca²⁺ entry into the organelle⁵⁶ (see Figure 1).

As for MICU3, its evolutionary conservation and sequence similarity to MICU1 suggested that it may share a common biological function with MICU1⁵⁴. Along this line, its role as a positive regulator of mitochondrial Ca²⁺ uptake has been demonstrated recently⁵⁵. Indeed, the expression of MICU3 in Hela cells, which normally do not express it, showed that it interacts with MICU1, but not with MICU2, forming obligatory dimers⁵⁵ (see Figure 1). This interaction causes a significant increase of mitochondrial Ca²⁺ uptake, demonstrating the stimulatory action of MICU3 on MCU activity. In addition, while displaying a reduced gatekeeper activity, MICU3 has been shown to mediate a more rapid response of mitochondria to [Ca²⁺]_i changes, allowing a shorter delay between the [Ca²⁺]_i rise and the increase in mitochondria Ca²⁺ uptake compared to MICU1⁵⁵. MICU3 thus plays a pivotal role in determining the kinetics of mitochondrial Ca²⁺ signalling and consequently in the modulation of global [Ca²⁺]_i dynamics, which may be of utmost relevance for the decoding of Ca²⁺ signals in excitable cells, such as neurons, where indeed MICU3 shows the highest level of expression^54,55.

By the interaction of monomers with non-overlapping functions, the MICU proteins confer an important property of the mitochondrial Ca²⁺ uptake system: the sigmoidicity of the dose-response to Ca²⁺. Indeed, the activity of MCU, which is kept low at resting [Ca²⁺]_i (∼100 nM) owing to the MICU1–MICU2 gatekeeper action, as already discussed, becomes extremely efficient upon [Ca²⁺]_i elevation. Indeed, Ca²⁺ binding to the MICU1–MICU2 regulatory complex converts it from an inhibitory regulator (ensuring gatekeeping) at basal [Ca²⁺]_i to an activator, strongly enhancing the flux through the channel. The threshold for MCU activation is directly dependent on the Ca²⁺ affinity of MICUs’ EF hands (which range from 300 nM to 1.3 µM), since the cooperative Ca²⁺ binding to these domains induces a conformational change in the MICU1–MICU2 dimer, which acts as a molecular switch relieving MCU inhibition⁵⁷.

Despite the role for MICU2 in the regulation of MCU gatekeeping and activation being controversial still, an intriguing hypothesis is that it has evolved to spatially restrict the Ca²⁺ crosstalk between single inositol trisphosphate receptor (IP3R) and MCU channels⁵⁸. Indeed, the presence of MICU2 has been shown to decrease the Ca²⁺ affinity of the MICU1 gatekeeper, thus elevating the threshold for MCU opening⁵⁸. In the context of the intracellular environment, this increased threshold for mitochondrial Ca²⁺ uptake translates into the need for a shorter distance between the ER Ca²⁺-releasing channel (IP3R) and the mitochondrial Ca²⁺ uptake channel (MCU)⁵⁸.

In line with this, in the last few years, the stoichiometry of MCU components has emerged as a crucial feature of the modulation of mitochondrial Ca²⁺ uptake rate. The relative proportion of the MICU proteins, in particular, has been demonstrated as the basis of the tissue-specific differences in cellular Ca²⁺ dynamics⁵⁹. Indeed, the different ratio of MICU1:MCU and MICU1:MICU2 proteins in the liver, the heart, and skeletal muscle has been correlated with the distinct mitochondrial Ca²⁺ handling of these three tissues⁵⁹. The level of MICU1 expression, which is much lower in cardiac tissue (while MCU and MICU2 are present at similar levels) compared to liver, has been suggested to determine the lower maximal Ca²⁺ uptake capacity and less-steep Ca²⁺ dependence that characterise cardiac mitochondria compared to liver mitochondria, as electrophysiological data³² and intracellular Ca²⁺ measurements⁵⁹ revealed. The low amount of MICU1 in cardiac mitochondria is instrumental to reduce mitochondrial Ca²⁺ uptake upon cytosolic Ca²⁺ elevation, allowing the decoding of the repetitive cytosolic Ca²⁺ spikes of the beating heart into a graded increase of matrix Ca²⁺. This integration of frequency fluctuations is crucial to prevent mitochondrial Ca²⁺ overload, which will be otherwise detrimental to the cardiac myocytes. By contrast, the relatively high MICU1 level and its cooperative activation of MCU in hepatocytes ensure the punctual and massive mitochondrial Ca²⁺ increase at every single cytosolic Ca²⁺ spike⁵⁹, which is functional to liver oxidative metabolism.

An additional degree of complexity came with the discovery of an alternative MICU1 isoform, extremely conserved and expressed at the highest level in skeletal muscle, which originates from the alternative splicing of the MICU1 mRNA, the MICU1.1 splice variant⁶⁰ (see Figure 1). This alternative isoform was shown to dimerise with MICU2, showing gatekeeper activity similar to that of MICU1, but it binds Ca²⁺ one order of magnitude more efficiently than MICU1 and activates MCU at lower [Ca²⁺]_i than MICU1⁶⁰. Thus, in skeletal muscle, the MICU1.1–MICU2 dimer ensures a much higher net Ca²⁺ entry and elevated ATP production than the MICU1–MICU2 dimer⁶⁰. These results highlight a novel mechanism of the molecular plasticity of the MCU Ca²⁺ uptake machinery that allows skeletal muscle mitochondria to adapt to the intense metabolic challenge that characterises this tissue (see Figure 2).

The essential mitochondrial Ca²⁺ uniporter regulator EMRE

To add further complexity to the study of MCU channel regulation, other molecules have been shown to interact with MCU in recent years. Indeed, the quantitative mass spectrometry analysis of the MCU-interacting proteins revealed the presence of a critical component of the MCU complex, the essential MCU regulator (EMRE) (see Figure 1), a 10 kDa single-pass transmembrane protein that was proposed to be necessary for MCU function by bridging MCU interaction with MICU1 in mammalian cells⁶¹. Interestingly, EMRE appears to be required for MCU channel activity in metazoans only, since it is not found in plants, fungi, or protozoa, such as Dictyostelium discoideum, although they express MCU proteins with functional channel activity. While its exact role in the MCU complex is still uncertain, EMRE was proposed to mediate MCU sensitivity to matrix [Ca²⁺]_mt, suggesting that MCU activity may be modulated by Ca²⁺ sensors facing both the IMS (MICUs) and the matrix (EMRE)⁶². However, this hypothesis was subsequently questioned by other groups, whose data supported an opposite topology for the EMRE protein, with the short N-terminus exposed to the matrix while its acidic C-terminus faces the IMS^63,64. In this conformation, EMRE action would be that of supporting MCU Ca²⁺ transport activity by interacting with the first TMH of MCU within the IMM and to interact also with MICU1 at the IMS via its C-terminus poly-aspartate tail, thus ensuring MICU1 binding to the channel and gatekeeping activity⁶³.

Recently, the control of EMRE protein levels has emerged as a crucial aspect of MCU complex regulation. Indeed, a correct amount of EMRE is fundamental to guarantee the exact stoichiometry of the different MCU components. Alteration of EMRE protein levels by ablation of the AAA-proteases responsible for its degradation⁶⁵, or by expression of proteolytic-resistant EMRE constructs⁶², leads to uncontrolled MCU channel opening causing mitochondrial Ca²⁺ leakage and mitochondrial Ca²⁺ overload, resulting in neuronal cell death^65,66. The existence of a tight control of EMRE protein levels has also been demonstrated in vivo in the MICU1^–/– mice, where EMRE is reduced in a genetic condition characterised by increased [Ca²⁺]_mt basal levels⁴⁸. These findings clearly underline a negative regulatory mechanism that keeps EMRE expression in check in order to cope with changes in MCU activity⁴⁸.

The mitochondrial Ca²⁺ uniporter regulator 1 MCUR1

Another protein that has been associated with the MCU complex^67,68, despite the original proteomic analysis of the MCU interactors failed to recover it⁶¹, is the MCU regulator 1 (MCUR1). MCUR1 is a coiled-coil-containing protein encoded by the CCDC90A gene initially identified by a RNAi screen searching for a mitochondrial membrane protein involved in MCU homeostasis⁶⁷, whose precise function in the regulation of MCU activity and mitochondrial metabolism is still debated. Indeed, MCUR1 has been shown to be required for MCU-dependent Ca²⁺ uptake, since its knockdown dampens mitochondrial Ca²⁺ entry while its overexpression enhances it⁶⁷. However, a direct action of MCUR1 on MCU activity was questioned by the finding that its silencing leads to a significant loss of mitochondrial membrane potential (ΔΨ_m), which per se can account for the decreased mitochondrial Ca²⁺ uptake⁶⁹. In this context, MCUR1 has been suggested to act primarily as a cytochrome c oxidase (COX) assembly factor⁶⁹, and this concept would also be supported by the fact that a MCUR1 orthologue is present in budding yeast⁷⁰, which lack MCU, and is required for yeast COX activity and survival in non-fermentable medium⁶⁹. Nevertheless, the patch clamp data of Ca²⁺ currents from voltage-clamped mitoplasts point to an effective role for the MCUR1 protein on MCU channel conductance, independently of ΔΨ_m⁷¹. In line with this, experiments in Drosophila cells, which are resistant to Ca²⁺-induced mitochondrial permeability transition (MTP)⁷² and in which no MCUR1 homologs are found, show increased sensitivity to Ca²⁺-dependent PTP opening when heterologous human MCUR1 is expressed, which is not accompanied by alterations in the rate of mitochondrial Ca²⁺ uptake⁷². Conversely, MCUR1 knockdown in mammalian cells renders them resistant to Ca²⁺ overload, providing protection from cell death⁷². This led the authors to conclude that MCUR1 regulates the Ca²⁺ threshold for the MPT and may act as a molecular bridge connecting the MCU channel to the MTP complex. Whether this hypothesis holds true still has to be tested.

More recently, the genetic deletion of MCUR1 in endothelial and cardiac tissues in vivo by mouse conditional knockout models⁶⁸ points to an involvement of MCUR1 in the assembly and function of the MCU channel. In particular, MCUR1 binding to MCU and EMRE appears to be necessary for MCU oligomerisation and stability, since MCUR1^–/– cells will display a significantly decreased amount of MCU-containing high-molecular-weight complexes and reduced rate of mitochondrial Ca²⁺ uptake⁶⁸. This, as expected, impinges on cellular energetics, as it is reflected by the dramatic reduction of ATP levels and consequent activation of the pro-survival autophagy pathway in cells depleted of MCUR1⁷². Whether the pleiotropic roles for MCUR1 in the regulation of mitochondrial Ca²⁺ uptake, OXPHOS efficiency, and the apparent discordant phenotypes of MCUR1-depleted cells depend on the intrinsic transcriptional and metabolic differences of different cell types or on the species-specific features of the mitochondrial Ca²⁺ machinery is still an open question, and further investigation is needed to clarify it.