The conserved aspartate ring of MCU mediates MICU1 binding and regulation in the mitochondrial calcium uniporter complex

  1. Charles B Phillips
  2. Chen-Wei Tsai
  3. Ming-Feng Tsai  Is a corresponding author
  1. Brandeis University, United States
  2. University of Colorado Anschutz Medical Campus, United States

Abstract

The mitochondrial calcium uniporter is a Ca2+ channel that regulates intracellular Ca2+ signaling, oxidative phosphorylation, and apoptosis. It contains the pore-forming MCU protein, which possesses a DIME sequence thought to form a Ca2+ selectivity filter, and also regulatory EMRE, MICU1, and MICU2 subunits. To properly carry out physiological functions, the uniporter must stay closed in resting conditions, becoming open only when stimulated by intracellular Ca2+ signals. This Ca2+-dependent activation, known to be mediated by MICU subunits, is not well understood. Here, we demonstrate that the DIME-aspartate mediates a Ca2+-modulated electrostatic interaction with MICU1, forming an MICU1 contact interface with a nearby Ser residue at the cytoplasmic entrance of the MCU pore. A mutagenesis screen of MICU1 identifies two highly-conserved Arg residues that might contact the DIME-Asp. Perturbing MCU-MICU1 interactions elicits unregulated, constitutive Ca2+ flux into mitochondria. These results indicate that MICU1 confers Ca2+-dependent gating of the uniporter by blocking/unblocking MCU.

https://doi.org/10.7554/eLife.41112.001

Introduction

The mitochondrial calcium uniporter is a multi-subunit Ca2+-activated Ca2+ channel complex located in the inner mitochondrial membrane (IMM). It catalyzes Ca2+ influx from the intermembrane space (IMS) into the mitochondrial matrix, where a large quantity of Ca2+ can be stored. Extensive studies have established that the uniporter regulates spatial and temporal dimensions of intracellular Ca2+ signals, as well as Ca2+-dependent mitochondrial processes, including oxidative phosphorylation and programmed cell death (Kamer and Mootha, 2015; Rizzuto et al., 2012).

The Ca2+-conducting function of mammalian uniporters are mediated by two subunits, MCU and EMRE, in the transmembrane (TM) region (Figure 1). The MCU protein possesses two TM helices and a highly-conserved ‘DIME’ signature sequence (Baughman et al., 2011; De Stefani et al., 2011). High-resolution structures (Baradaran et al., 2018; Fan et al., 2018; Nguyen et al., 2018; Yoo et al., 2018) show that MCU assembles into a tetrameric Ca2+ pore, with the DIME-Asp and -Glu forming two parallel side-chain carboxylate rings to constitute a Ca2+ selectivity filter at the pore’s IMS entrance (Figure 1). The single-pass EMRE protein binds to MCU via its TM helix (Tsai et al., 2016). This interaction is shown to be necessary for Ca2+ permeation (Tsai et al., 2016; Kovács-Bogdán et al., 2014; Sancak et al., 2013).

Molecular assembly of the mitochondrial Ca2+ uniporter.

The MCU protein assembles into a tetrameric Ca2+ pathway across the inner mitochondrial membrane (only two subunits are illustrated to reveal the Ca2+ pore). Conserved Asp and Glu residues in MCU’s DIME signature sequence form two parallel side-chain carboxylate rings at the IMS entrance of the pore to coordinate Ca2+. The EMRE protein binds to MCU and MICU1 via its TM helix and C-terminal tail, respectively. When an intracellular Ca2+ signal arrives at the IMS surface of the uniporter, Ca2+ binding to MICUs leads to activation of the uniporter to transport Ca2+ into the matrix.

https://doi.org/10.7554/eLife.41112.002

The uniporter is tightly regulated by intracellular Ca2+ signals. It stays quiescent in resting cellular conditions, and becomes activated only when IMS Ca2+ increases to low micromolar levels (Csordás et al., 2013; Mallilankaraman et al., 2012). This Ca2+-dependent gating is mediated by two EF-hand (a helix-loop-helix Ca2+-coordinating motif) containing subunits: MICU1 and MICU2 (the neuron-specific MICU3 is not discussed here) (Csordás et al., 2013; Mallilankaraman et al., 2012; Perocchi et al., 2010; Plovanich et al., 2013), which are tethered to the uniporter’s TM region via the C-terminal tail of EMRE (Tsai et al., 2016). Depletion of MICU1 eliminates Ca2+-regulation of the uniporter, causing the channel to constitutively load Ca2+ into the matrix (Tsai et al., 2016; Mallilankaraman et al., 2012; Plovanich et al., 2013; Tsai et al., 2017), a condition linked to debilitating neuromuscular disorders in humans (Logan et al., 2014). Currently, the mechanism by which MICUs control Ca2+ transport via MCU remains largely unknown.

Here, we demonstrate that MICU1 interacts with MCU’s DIME-Asp via a Ca2+-modulated electrostatic interaction. This is mediated by two closely-spaced Arg residues on the surface of MICU1. MICU2, which lacks these Args, does not bind MCU. Mutations that disrupt the MCU-MICU1 interaction severely perturbs Ca2+-regulation of the uniporter. These results led to a molecular mechanism in which MICUs open or close the uniporter in response to intracellular Ca2+ signals by physically blocking or unblocking the MCU pore.

Results

Evolutionarily conserved MCU-MICU1 interactions

Phylogenetic analyses (Sancak et al., 2013; Bick et al., 2012) have shown that uniporters in lower eukaryotes (e.g., plants and protists) contain only MCU and MICU1 subunits, raising a possibility that MICU1 might gate MCU via direct molecular contacts. If so, these interactions might be conserved in evolution to ensure proper regulation of the uniporter. To test this idea, we performed co-immunoprecipitation (CoIP) experiments to examine complex formation between human MICU1 and various MCU homologues in MCU/EMRE-KO HEK 293 cells (Tsai et al., 2016). The EMRE gene is deleted because EMRE can bind both MCU and MICU1 (Figure 1) (Tsai et al., 2016; Sancak et al., 2013), and would therefore complicate assessment of direct MCU-MICU1 contacts. Figure 2 shows that human MICU1 pulls down not only human MCU but also MCU homologues in D. melanogaster, C. elegans, D. discoideum, and A. thaliana, indicating that the MCU-MICU1 interaction is indeed evolutionarily conserved.

Conserved MCU-MICU1 interactions.

1D4-tagged MCU homologues from various species (HS: Homo sapiens, DM: Drosophila Melanogaster, CE: Caenorhabditis elegans, AT: Arabidopsis thaliana, and DD: Dictyostelium discoideum) were expressed in the presence or absence of FLAG-tagged WT human MICU1 in MCU/EMRE-KO cells. MICU1 was immobilized in FLAG-affinity resins to pull down MCU. Anti-FLAG and anti-1D4 antibodies were used to detect MICU1 and MCU, respectively. SDS-PAGE was performed under reducing conditions. WCL: whole cell lysate. IP: immunoprecipitation. Asterisk: non-specific Western signals. Hash: MCU homologues that contain untruncated mitochondrial-targeting sequences.

https://doi.org/10.7554/eLife.41112.003

The role of the DIME-Asp in Ca2+ transport and MICU1 binding

We reasoned that MICU1 might bind to the DIME-Asp, as MCU structures (Baradaran et al., 2018; Fan et al., 2018; Nguyen et al., 2018; Yoo et al., 2018) show that this Asp is the only fully-conserved residue with the side-chain exposed to the IMS, where MICU1 is localized. Accordingly, the DIME-Asp in human MCU was mutated to Ala (D261A), and the mutant was expressed in MCU-KO HEK 293 cells for analysis. Surprisingly, a standard mitochondrial Ca2+ uptake assay shows that D261A MCU is capable of importing Ca2+ (10 μM), with the rate of transport unaffected by adding 100 mM Na+, which has an ionic radius virtually identical to Ca2+ (Figure 3 and Figure 3—figure supplement 1). A quantitative 45Ca2+ flux experiment (Tsai et al., 2016) performed in 10 μM Ca2+ shows that D261A slows MCU’s Ca2+ transport by only 3.8-fold (Figure 3—figure supplement 2), an effect remarkably small considering the critical position of this residue in the pore (Baradaran et al., 2018; Fan et al., 2018; Nguyen et al., 2018; Yoo et al., 2018). In contrast, mutating the DIME-Glu (E264) to Ala, Asn, or Gln abolishes uniporter function (Figure 3 and Figure 3—figure supplements 2 and 3), as expected from its key role in coordinating Ca2+ in the selectivity filter (Baradaran et al., 2018; Fan et al., 2018; Nguyen et al., 2018; Yoo et al., 2018). To further pursue these observations, D261 was mutated to all other 18 amino-acids. Only D, E and A at this position support Ca2+ transport (Figure 3—figure supplement 3).

Figure 3 with 3 supplements see all
Functional analysis of MCU.

(A) A fluorescence-based mitochondrial Ca2+ uptake assay. MCU-KO HEK293 cells, transiently expressing WT MCU, were permeabilized with digitonin (dig) in the presence of an extracellular Ca2+ indicator Calcium Green-5N (CG5N). Adding 10 µM CaCl2 leads to an immediate increase of fluorescence, followed by a signal decline reflecting uniporter-mediated Ca2+ uptake. Ru360 (Ru) was added to inhibit the channel. In subsequent experiments, only traces obtained after applying Ca2+ (dashed box) are presented. (B) The activity of D261A or E264A mutants. These mutants were expressed in MCU-KO cells, with 100 mM NaCl added during Ca2+ uptake to test if the channel can select Ca2+ against Na+. The bar chart summarizes the initial rate of Ca2+ uptake, and the western blot compares expression levels of MCU constructs. Con: untransfected cells. *p<0.01.

https://doi.org/10.7554/eLife.41112.004

We then performed CoIP to test how wild-type (WT) MICU1 binding responds to MCU mutations at D261 and E264. Results show that MICU1 binds WT, D261E, and E264A MCU, but not D261A or D261Q (Figure 4). Although the D261Q mutant cannot transport Ca2+, it still assembles as oligomers (Figure 4—figure supplement 1), suggesting that the mutation does not compromise MCU’s structural integrity. These results demonstrate that the DIME-Asp mediates MCU interaction with MICU1, instead of contributing essentially to Ca2+ permeation.

Figure 4 with 2 supplements see all
The impact of D261 or E264 mutations on MICU1 binding.

FLAG-tagged WT MICU1 was used to pull down various MCU mutants co-expressed in MCU/EMRE-KO cells.

https://doi.org/10.7554/eLife.41112.008

It was observed that D261A loses sensitivity to a potent and specific uniporter inhibitor Ru360 (Matlib et al., 1998),20 (Figure 3B). This is consistent with the thought that D261 contributes to a Ru360 site in MCU (Arduino et al., 2017; Cao et al., 2017), and implies that MICU1 and Ru360 inhibitory sites overlap. A previous study shows that the S259A mutation diminishes Ru360 inhibition (Baughman et al., 2011), raising a possibility that S259 might also be involved in MICU1 binding. We confirm that S259A reduces Ru360 inhibition of the uniporter by 82 ± 3% (Figure 4—figure supplement 2), and show that this mutation indeed destabilizes the MCU-MICU1 complex (Figure 4—figure supplement 2), albeit to a lesser degree than D261A. It thus appears that MCU and MICU1 form a multi-residue contact surface containing S259 and D261 in MCU, with the latter playing a more critical role in mediating tight MCU-MICU1 interactions.

Electrostatic interactions between MCU and MICU1

As DIME-Asp appears as a fourfold ring of negative charges facing the IMS, it is tempting to picture MICU1 as a classic pore-blocker (Banerjee et al., 2013; Park and Miller, 1992) electrostatically stabilized on MCU’s ion entryway. This picture is strongly supported by the observation that the MCU-MICU1 interaction can be weakened or strengthened by raising or lowering ionic strength, respectively (Figure 5A). In contrast, neither dissociation of the MICU1-MICU2 dimer nor the 1D4-tag and anti-1D4 antibody epitope interaction is affected by varying ionic strength (Figure 5—figure supplement 1). To search MICU1 for electrostatic binding partners of the DIME-Asp, we launched an Ala mutagenesis screen targeting 18 conserved Arg or Lys residues in human MICU1 (Figure 5—figure supplement 2). Only R119 and R154, two residues closely spaced on the protein’s surface (Wang et al., 2014), were found to abolish MCU binding upon mutation to Ala (Figure 5B and Figure 5—figure supplement 3). These mutants, like WT MICU1, form heterodimers with MICU2 (Patron et al., 2014) (Figure 5—figure supplement 4), indicating proper protein folding. Moreover, R119K or R154K mutants remain associated with MCU, while Glu or Gln substitutions in these two positions strongly disrupt MCU binding (Figure 5—figure supplement 5). Neither of the two Arg residues is present in MICU2, and MICU2 is indeed unable to complex with MCU (Figure 5C). Taken together, the data suggest that R119 and R154 in MICU1 mediate electrostatic interactions with the DIME-Asp in MCU.

Figure 5 with 5 supplements see all
Electrostatic interactions between MCU and MICU1.

(A) Modulation of MCU-MICU1 complex stability by ionic strength. WT MCU and MICU1 were expressed in MCU/EMRE-KO cells, and CoIP experiments were performed in the presence of 50, 150, or 500 mM of NaCl. The IP signal of MCU was normalized to that of MICU1, with the ratio presented in the bar chart. (B) The effect of MICU1 Arg mutations on MCU binding. (C) A CoIP experiment testing if MCU and MICU2 form complexes. MICU2 was FLAG-tagged to precipitate WT MCU in MCU/EMRE-KO cells. *p<0.05.

https://doi.org/10.7554/eLife.41112.011

Functional roles of the MCU-MICU1 interaction

We have thus far utilized transiently expressed WT or mutant MICU1 to identify molecular determinants of the MCU-MICU1 interaction. However, as MICU1 exclusively forms a disulfide-connected heterodimer with MICU2 in mammalian cells (Patron et al., 2014; Petrungaro et al., 2015), it is necessary to exclude the possibility that dimerization with MICU2 could fundamentally alter how MICU1 contacts MCU. Accordingly, we employed MCU to pull down native MICUs. Results show that the D261A mutation disrupts MCU association with the physiological MICU1-2 heterodimer (Figure 6), indicating that the MICU2-bound form of MICU1 still interacts with MCU via the DIME-Asp.

Ca2+-dependent interaction between MCU and the MICU1-2 heterodimer.

1D4-tagged WT or D261A MCU was expressed in WT HEK cells. The cell lysate, after a portion was taken for whole-cell lysate (WCL) analysis, was split into two for CoIP under Ca2+-free (EG, 1 mM EGTA) or 10 µM Ca2+ conditions. MCU was used to pull down the native, disulfide-connected MICU1-2 heterodimer (Patron et al., 2014; Petrungaro et al., 2015), which has a molecular weight of ~90 kDa. SDS-PAGE was performed in non-reducing environments. MICU1 and MICU2 were detected using anti-MICU1 and -MICU2 antibodies, respectively. WCL signals of MICU1 and MICU2 are not as clean as in previous images (e.g., Figure 2) due to the low abundance of native MICUs and lower qualities of these polyclonal MICU1 and MICU2 antibodies.

https://doi.org/10.7554/eLife.41112.017

As binding of MICU1 to the DIME-Asp would likely block the uniporter’s pore, we hypothesize that MICU1 shuts the uniporter in resting Ca2+ (<1 µM) through this particular interaction. This hypothesis predicts that (1) raising Ca2+ to micromolar levels would disrupt MCU’s association with the MICU1-2 heterodimer, and that (2) perturbing the MCU-MICU1 interaction by mutating the DIME-Asp or R119/R154 would prevent MICU1 from shutting the uniporter. Indeed, CoIP experiments show that supplying 10 μM Ca2+ breaks the MCU-MICU1-MICU2 complex (Figure 6). The 45Ca2+ flux assay described above was subsequently used to quantify mitochondrial uptake under a low Ca2+ (0.5 μM) condition. In WT cells, little Ca2+ entry (1.6 ± 0.9 pmol/min/106 cells) into mitochondria was detected (Figure 7A). As expected, MICU1-KO induces robust Ca2+ influx (205 ± 11 pmol/min/106 cells), a phenotype partially reversed by expressing WT MICU1 (53 ± 4 pmol/min/106 cells, Figure 7A). We then introduced WT or D261A MCU into MCU-KO cells. In low Ca2+, WT MCU exhibits no activity (1.7 ± 0.5 pmol/min/106 cells) while D261A mediates a Ca2+ influx (34 ± 5 pmol/min/106 cells) 6.2-fold slower than that observed in MICU1-KO cells (Figure 7B). A few factors might underlie the rather small magnitude of the D261A-mediated Ca2+ uptake: (1) this mutant is 3.8-fold slower than WT MCU (Figure 3—figure supplement 2), (2) our transfection efficiency is ~80%, and (3) other residues (e.g., S259) are also involved in MICU1 binding. A S259A/D261A double mutant was constructed to further disrupt the MCU-MICU1 interface, but unfortunately its function could not be analyzed due to a low expression level (Figure 4—figure supplement 2). The finding that D261A catalyzes unregulated Ca2+ flux in submicromolar Ca2+ argues strongly that MICU1 must contact MCU to gate the uniporter. Lastly, we tested R119 or R154 mutants in MICU1-KO cells. All of these, except for R154Q, are less competent than WT MICU1 in restoring Ca2+ regulation of the uniporter (Figure 7C), a result confirming the critical role of the MCU-MICU1 interaction in Ca2+-activation of the uniporter.

Figure 7 with 1 supplement see all
The effect of D261 or R119/R154 mutations on the regulatory function of MICU1.

(A) Mitochondrial Ca2+ uptake in a low Ca2+ (0.5 µM) condition. Each data point represents a measurement of 45Ca2+ transported into mitochondria by the uniporter at a specific time point. These data points were fit with a linear function (red lines) to obtain the rate of Ca2+ transport. (B) The activity of WT or D261A MCU in 0.5 µM Ca2+. (C) A bar chart summarizing the rate of mitochondrial Ca2+ uptake. WT MICU1 or various R119/R154 mutants were expressed in MICU1-KO cells. Con: untransfected control. Paired t-test was performed between WT MICU1 and mutants. *p<0.05.

https://doi.org/10.7554/eLife.41112.018

Discussion

The mitochondrial Ca2+ uniporter plays a crucial physiological role of regulating cytoplasmic Ca2+ signals and controlling mitochondrial metabolic and apoptotic pathways. These processes require the uniporter to remain strictly quiescent in resting cellular conditions. Here, we propose a mechanism (Figure 8) in which MICU1 shuts the uniporter by binding to the DIME-Asp side-chain carboxylate ring to block the IMS entrance of the MCU pore. Upon arrival of intracellular Ca2+ signals, Ca2+ binding to MICU1 at its EF hands disrupts this interaction, thus leading to opening of this Ca2+-activated Ca2+ channel.

A model of Ca2+-dependent gating of the uniporter.

In resting cellular conditions, MICU1 shuts the uniporter by inserting Arg fingers into MCU’s Asp ring to occlude the pore. Ca2+ activates the channel by binding to MICUs to disrupt this MCU-MICU1 interaction. MICU2 forms a heterodimer with MICU1, but does not directly contact MCU. EMRE plays dual functional roles: it binds to MCU to enable Ca2+ permeation, and also interacts with MICU1 to maintain tight association of the MICU1-2 heterodimer with the uniporter during Ca2+ stimulation.

https://doi.org/10.7554/eLife.41112.020

The EMRE subunit, which binds both MCU and MICU1 (Tsai et al., 2016), plays an important role in this mechanism. It has been shown that the EMRE-MICU1 interaction is necessary to prevent MICU1 dissociation from the uniporter complex (Tsai et al., 2016). We can now understand this observation in light of new results here: When MCU and MICU1 separate due to Ca2+ elevation, EMRE’s tether to MICU1 would prevent this subunit from dissociating away. Thus, once the Ca2+ signal is over, MICU1 could rapidly bind to MCU to terminate Ca2+ influx (Figure 8).

In this model, MICU2 does not directly contact MCU to block the channel (Figure 8). This is consistent with previous work (Payne et al., 2017) (but c.f. other references, Plovanich et al., 2013; Kamer et al., 2017) showing that MICU2 is not required to gate the uniporter closed. A fundamental issue for the future would be to determine the function of MICU2 (Payne et al., 2017). MICU2 likely plays non-redundant roles, as MICUs are present exclusively in the form of MICU1-2 heterodimers in mammalian cells (Patron et al., 2014; Petrungaro et al., 2015), and as MICU2 depletion induces severe neuronal and cardiac pathologies (Bick et al., 2017; Shamseldin et al., 2017).

During the revision of this manuscript, Paillard et al. published an article (Paillard et al., 2018) showing that depletion of MICU1 sensitizes the uniporter to Ru360 inhibition. The interpretation was that MICU1 competes for the Ru360 inhibitory site, known to be formed by the DIME-Asp (Arduino et al., 2017; Cao et al., 2017). It follows that MICU1 must control the uniporter by interacting with the Asp ring. Our results similarly indicate that MICU1 and Ru360 sites in MCU likely overlap, as mutations in DIME-Asp and a nearby Ser (S259 in human MCU) perturb both Ru360 inhibition and MICU1 binding.

Paillard et al. further proposed that MICU1 uses a DIME-interacting domain (DID) that contains one Lys and two Args (K438, R440, and R443 in human MICU1) to bind MCU. However, it was also shown that with all these residues mutated to Ala, a portion of the MCU-MICU1 complex (~30% of that observed using WT MICU1) remains associated after tens of minutes of incubation in CoIP experiments. This result, which agrees with our finding that R440A or R443A MICU1 forms stable complexes with MCU (Figure 5—figure supplement 3), raises a possibility that the DID sequence might not play direct roles in mediating tight MCU-MICU1 interactions.

Our model instead posits that MICU1 uses two closely-spaced Arg in the N-terminal domain (Wang et al., 2014) (R119 and R154 in human MICU1) to bind the DIME-Asp (Figure 8). This picture is supported by the observations that, consistent with electrostatic interactions, the stability of the MCU-MICU1 complex can be modulated by varying the ionic strength, and that MICU2, which lacks these Args, is unable to bind MCU (the DID sequence is present in both MICU1 and MICU2). The crucial roles of these two Args in MCU binding are further highlighted by the fact that they are the only two basic residues that are conserved in MICU1 homologues in animals, plants, and protists (Figure 5—figure supplement 2), in which MCU and MICU1 co-evolve (Bick et al., 2012). However, we hasten to point out that, despite these observations, future biochemical and structural work is still required to determine the detailed chemistry that governs MICU1 interactions with the DIME-Asp.

It is known that the uniporter uses a classical multi-ion pore mechanism (Kirichok et al., 2004; Almers et al., 1984; Hess and Tsien, 1984) to select Ca2+ against >1000-fold more abundant cations such as Na+. In this mechanism, Ca2+ binding to a high-affinity site blocks permeation of other cations, while entry of a second Ca2+ knocks off the bound Ca2+ through electrostatic repulsion to enable high Ca2+ flux. New structures of MCU led to the hypothesis that the DIME-Glu forms the high-affinity site (S2) to coordinate a dehydrated Ca2+, while the DIME-Asp forms a second, low-affinity Ca2+ site (S1) (Baradaran et al., 2018; Fan et al., 2018; Nguyen et al., 2018; Yoo et al., 2018). We systematically mutated DIME-Asp (D261) in human MCU and found that most mutations abolish channel function, an outcome not unexpected considering the critical position of D261 in the pore. The fact that D261A exhibits a comparable activity as WT, however, raises a possibility that other Ca2+ sites might be present in proximity to S2 to mediate the electrostatic repulsion required for high Ca2+ throughput of the uniporter.

In conclusion, the current study provides a working model to understand how intracellular Ca2+ signals control the activity of the uniporter in the molecular level. Major challenges still lie ahead, including to understand MICU2’s physiological role, to determine how Ca2+ disrupts the MCU-MICU1 interaction, and to examine the individual roles of MICU1’s two EF hands in channel activation. New electrophysiological and structural tools (Baradaran et al., 2018; Fan et al., 2018; Nguyen et al., 2018; Yoo et al., 2018; Tsai and Tsai, 2018) will open exciting opportunities to address these in the future.

Materials and methods

Key resources table
Reagent type
or resource
DesignationSource or referenceIdentifiersAdditional information
Cell lineHEK 293TATCCCat # CRL-3216
Cell lineMCU-KO HEK 293TPMID:27099988
Cell lineMCU/EMRE-KO
HEK 293T
PMID:27099988
Cell lineMICU1-KOPMID:28396416
Primary
Antibody
Mouse anti-FLAGSigma-AldrichCat # F1804Western 1:10000
Primary
Antibody
Mouse anti-V5ThermoFisherCat # R960-25Western 1:5000
Primary
Antibody
Mouse anti-β actinSanta CruzCat # 69879Western 1:500
Primary
Antibody
Rabbit anti-MICU1Sigma-AldrichCat # HPA037480Western 1:5000
Primary
Antibody
Rabbit anti-EFHA1
(MICU2)
AbcamCat # ab101465Western 1:10000
Primary
Antibody
Mouse anti-1D4PMID:6529569Western 50 ng/mL
Primary
Antibody
Mouse anti-C8PMID:8068416Western 50 ng/mL
Secondary
Antibody
IRDye 680RD
goat anti-rabbit
IgG
Li-CorCat # 925–68073Western 1:10000
Secondary
Antibody
IRDye 680RD
goat anti-mouse
IgG
Li-CorCat # 925–68072Western 1:15000
Chemical
compound
Ru360PMID:2036363
Chemical
compound
45CaCl2PerkinElmerCat # NEX01300
Commercial
kit
Lipofectamine 3000ThermoFisherCat # L3000015
Commercial
kit
Anti-FLAG M2 affinity gelSigma-AldrichCat # A2220
Commercial
kit
CNBr-activated
Sepharose 4B
GE HealthcareCat # 17043001
SoftwareIgor Pro 7WaveMetricsFigure production and
data fitting
SoftwareImageStudio 5Li-CorWestern-blot
quantification
SoftwareClustal OmegaPMID:21988835Sequence alignment
SoftwareExcel (office 365)Microsoftt-test

Reagents, cell culture, and molecular biology

Request a detailed protocol

Reagents were purchased at the highest grade available. Ru360 was synthesized in-house following a previously published protocol (Ying et al., 1991). Genes encoding uniporter subunits were cloned into a pcDNA 3.1 (+) expression vector. Site-directed mutagenesis was performed using a QuickChange kit (Agilent) and confirmed with sequencing. All MCU constructs used here contain a C-terminal 1D4 tag (TETSQVAPA) for Western detection. Similarly, MICU1 is tagged with a C-terminal FLAG (DYKDDDDK), and MICU2 with a C-terminal FLAG or V5 (GKPIPNPLLGLDST). Sequences of these have been reported in a previous manuscript (Tsai et al., 2016).

HEK 293 cells, obtained from ATCC and authenticated by short tandem repeat profiling, were cultured in Dulbecco’s modified Eagle’s medium (Gibco) supplemented with 10% FBS, and were incubated at 37°C with 5% CO2. Mycoplasma infection was routinely ruled out using an ATCC PCR detection kit (30–1012K). CRISPR knockout cell lines have been established in our previous work (Tsai et al., 2016; Tsai et al., 2017). Transient transfection was performed using Lipofectamine 3000 (ThermoFisher), following the manufacturer’s instructions. Cells were harvested for experiments 24–30 hr after transfection.

Co-immunoprecipitation (CoIP)

Request a detailed protocol

All CoIP experiments were performed at 4°C. Transfected cells in 2 wells of a 6-well plate were lysed in 0.5 mL solubilization buffer (SB, 100 mM NaCl, 20 mM Tris, 1 mM EGTA, 5 mM DDM, pH 7.5-HCl) supplemented with an EDTA-free protease inhibitor cocktail (cOmplete Ultra, Roche). The lysate was clarified by spinning down. 50 μL of the supernatant was removed, with total protein concentration determined using a BCA assay (Thermo-Fisher) and 10 μg of protein used for whole-cell lysate (WCL) analysis. Then, 25 μL of FLAG (Sigma-Aldrich, A2220)- or 1D4-conjugated beads (50% slurry) were added to the rest of the supernatant for a 30 min batch binding process. The beads were then collected on a spin column, washed with 2 mL of SB, and then eluted with 0.15 mL SDS loading buffer. 10–20 μL of the elute was used for SDS-PAGE, with 5% of 2-mercaptoethanol used to produce reducing conditions. The whole CoIP procedure was completed within 45 min after cell lysis (prolonged incubation of >2 hr could lead to complete dissociation of uniporter subcomplexes). 1D4-affinity gel was produced in house using 25 mg 1D4 antibody per 1 g of CNBr-activated Sepharose 4B resin (GE Healthcare).

To perform Western blot, proteins on SDS gels were transferred to low-fluoresce PVDF membranes (EMD-Millipore), which were then blocked in a TBS-based Odyssey blocking buffer (Li-Cor), and incubated with primary antibodies in TBST (TBS +0.075% Tween-20) at 4°C overnight. Then, after a 1 hr incubation with infrared fluorescent secondary antibodies in TBST at room temperature, signals were acquired using an Odyssey CLx imaging system (Li-Cor), and analyzed with an ImageStudio software (Li-Cor version 5.0). Unless specified, MCU and MICU1 were detected using α−1D4 and α-FLAG antibodies, respectively. See the key resources table for antibodies and dilutions. 1D4 and C8 antibodies were produced in house.

Mitochondrial Ca2+ flux assays

Request a detailed protocol

For the fluorescence-based assay, 2 × 107 HEK 293 cells were suspended in 10 mL of wash buffer (WB, 120 mM KCl, 25 mM HEPES, 2 mM KH2PO4, 1 mM MgCl2, 50 µM EGTA, pH 7.2-KOH), pelleted, and then resuspended in 2.5 mL of recording buffer (RB, 120 mM KCl, 25 mM HEPES, 2 mM KH2PO4, 5 mM succinate, 1 mM MgCl2, 5 µM thapsigargin pH 7.2-KOH). 2 mL of the cell suspension were placed in a stirred quartz cuvette in a Hitachi F-2500 spectrophotometer (ex: 506 nm, ex-slit: 2.5 nm, em: 532 nm, em-slit: 2.5 nm, sampling rate: 2 Hz). Reagents were added into the cell suspension in the following order: 0.5 µM calcium green 5N (Thermo-Fisher C3737), 30 µM digitonin (Sigma-Aldrich D141), 10 µM CaCl2, and 75 nM Ru360. Upon adding Ca2+, fluorescent signals would increase by 200 to 300 a.u. Without adding Ru360, the signal would eventually drop to a steady-state level roughly the same as that before Ca2+ addition. Quantification of data is done by linear fit to the fluorescent signal between 10 s and 15 s after adding Ca2+.

For the 45Ca2+ based assay, 1.2–2.4 * 106 viable cells were suspended in 1 mL WB, spun down, and then resuspended in 120 µL WB, supplemented with 5 µM thapsigargin (Sigma-Aldrich, T9033) and 30 µM digitonin. To initiate mitochondrial Ca2+ uptake, 100 µL cell suspension was transferred to 400 µL low-Ca2+ flux buffer (RB +0.69 mM EGTA, 0.5 mM CaCl2, 15 µM 45CaCl2, 30 µM digitonin, 5 µM thapsigargin, pH 7.2-KOH) or high-Ca2+ flux buffer (RB +20 µM 45CaCl2, 30 µM digitonin, 5 µM thapsigargin, pH 7.2-KOH). At desired time points, Ca2+ uptake was terminated by adding 100 µL of the sample to 5 mL ice-cold WB, and then filtered through 0.45 µM nitrocellulose membranes (Sigma-Aldrich WHA10402506) on a vacuum filtration manifold (EMD-Millipore model 1225). The membrane was washed immediately with 5 mL ice-cold WB, and later transferred into scintillation vials for counting. Nonspecific signals were measured using samples containing 75 nM Ru360 or using untransfected cells (for the Ru360-insensitive D261A mutant), and were subtracted to yield uniporter-specific Ca2+ transport. In a typical experiment, readings of 45Ca2+ in three time points were fit with a linear function to generate the rate of Ca2+ transport (e.g., Figure 7A). Rates obtained from at least three independent experiments were then averaged for data presentation (see Figure 7—figure supplement 1 for examples of the data analysis process). For experiments comparing WT and D261A, 1 µg WT DNA or 2.2 µg D261A DNA was used for transfection to ensure similar expression levels of these two constructs. Moreover, cells were harvested within 24 hr after transfection to avoid a molecular excess of overexpressed MCU over native MICU1. 45Ca2+ radioisotope was obtained from PerkinElmer, and has a specific activity of 12–15 mCi/mg.

Sequence analysis and statistics

Request a detailed protocol

Sequences of MICU1 homologues were collected using PSI-BLAST. Multiple sequence alignment was performed using the Clustal Omega online server (Sievers et al., 2011).

All experiments were repeated in at least three independent experiments, and the data were presented as mean ±standard error of the mean (SEM). Statistical analysis was performed using Student’s t-test, with significance defined as p<0.05.

Data availability

All data generated or analyzed during this study are included in the manuscript and supporting files. Source data for calcium-45 flux experiments are available via Dryad.

The following data sets were generated
    1. Phillips C
    2. Tsai C
    3. Tsai M
    (2019) Dryad Digital Repository
    Date from: The conserved aspartate ring of MCU mediates MICU1 binding and regulation in the mitochondrial calcium uniporter complex.
    https://doi.org/10.5061/dryad.1976kg3

References

    1. Matlib MA
    2. Zhou Z
    3. Knight S
    4. Ahmed S
    5. Choi KM
    6. Krause-Bauer J
    7. Phillips R
    8. Altschuld R
    9. Katsube Y
    10. Sperelakis N
    11. Bers DM
    (1998)
    Oxygen-bridged dinuclear ruthenium amine complex specifically inhibits Ca2+ uptake into mitochondria in vitro and in situ in single cardiac myocytes
    The Journal of Biological Chemistry 273:10223–10231.
    1. Ying WL
    2. Emerson J
    3. Clarke MJ
    4. Sanadi DR
    (1991)
    Inhibition of mitochondrial calcium ion transport by an oxo-bridged dinuclear ruthenium ammine complex
    Biochemistry 30:4949–4952.

Decision letter

  1. Baron Chanda
    Reviewing Editor; University of Wisconsin-Madison, United States
  2. John Kuriyan
    Senior Editor; University of California, Berkeley, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

[Editors’ note: the authors were asked to provide a plan for revisions before the editors issued a final decision. What follows is the editors’ letter requesting such plan.]

Thank you for sending your article entitled "MICU1 interacts with the conserved aspartate ring of MCU to mediate Ca2+ activation of the mitochondrial Ca2+ uniporter" for peer review at eLife. Your article is being evaluated by three peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by John Kuriyan as the Senior Editor.

Given the list of essential revisions, including new experiments, the editors and reviewers invite you to respond within the next two weeks with an action plan and timetable for the completion of the additional work. We plan to share your responses with the reviewers and then issue a binding recommendation.

Regarding the flux experiments you are correct. These flux experiments have been established earlier in the eLife paper of Tsai and Miller you cited above. However, the difference between the eLife paper you cited and the current submitted manuscript is that in the previous paper, the interpretations drawn from the assays are within the limits of the assay resolution. As you know, the fluorescent flux assay in question is very qualitative and basically reports whether there is or there is not import of Ca into mitochondria via the uniporter. If a mutation is made in any of the protein components required for function and Ca is no longer imported, the MCU is broken somehow, and the conclusion is drawn that the residue is important for either function or for interaction between the protein components. For example, in the previous paper, the fluorescent assay was used to determine which domains of EMRE are interacting with MCU to support Ca entry into mitochondria by expressing different deletion mutants of EMRE with MCU and concluding that the deletion of the domain that killed the transport was important in the interaction. Similarly, they used the assays to determine which side of the helix interacts with MCU by doing a Trp scanning mutagenesis and finding that the side where all mutations killed the transport, likely interacts with the MCU. These conclusions are, in my opinion, valid and within the resolution of the assay.

In contrast, in the current manuscript, as I previously stated, the authors draw very specific mechanistic interpretations based on this yes/no assay. For example:

1) Mutation of the D in the DIME signature sequence region of MCU to an A did not abolish Ca import but lead to loss of MICU1 interaction (as measured with co-IP). Furthermore, addition of 100 mM Na during the assay did not change the rate of Ca-import. The authors conclude that the D does not contribute to channel selectivity but mediates MICU1 interaction. The problem are the following: first, the assay does not address selectivity at all, but rather whether the mutation kills Ca transport. The authors should consider dropping any mention of the role of D in selectivity. Second, there should be a control experiment showing the effect of Na addition on WT MCU. Third, the finding that mutation of the D to any other amino acid (except E and A) abolished Ca-import is intriguing, and argues that the story is much more complex. Fourth, the authors should also present the coIPs with the D261E MCU mutant, which, according to the authors' hypothesis, is functional and expected to support the interaction with MICU1.

2) The authors claim that the MCU-MICU1 interaction is electrostatic as evidenced by weaker interactions as measured by lower ratios of MCU/MICU1 gel bands post coIPs in higher ionic strength buffer conditions. If the authors want to make this electrostatic claim, they should show both a positive and a negative control experiment. Otherwise, a direct binding interaction assay would be more appropriate in this case.

3) The authors screen for interacting partners for the Asp on MCU by mutating 18 conserved Arg and Lys residues on MICU1 to Ala and checking for interaction using coIPs. They find two Arg that disrupt MCU association and claim that these Rs on MICU1 mediate electrostatic interactions with the D on MCU. Perhaps these claims should be reduced to: the two Rs are important in mediating the interaction between MCU and MICU1, since there is no indication that the Rs interact with the D specifically. Furthermore, if the screen were not limited to positively charged residues, perhaps other residues would pop up as important to this interaction.

4) The authors conclude that the MICU1 interaction with MCU must work by MICU1 directly blocking the MCU pore, predicated on their conclusions that the Rs on MICU1 electrostatically interact with the Ds in MCU, which are located in the pore. Since there is no solid evidence for the Rs interacting with the D, the pore blocking interpretation is similarly weak and I suggest that it is removed.

5) In Figure 6-7, the authors test whether MICU1 interaction with the pore in resting Ca concentrations is necessary to keep the pore closed. They find by coIP that the MICU1-MCU interaction is disrupted in the presence of 10 μm Ca and that in MICU1 knockout cells, Ca fluxes inside mitochondria, while if MICU1 is added back, less Ca is uptaken. If the MICU1 R mutants are added back instead, except for one of them, they are less competent than WT in restoring Ca flux. It would have been useful for the authors to also show that, similar to the R mutants on MICU1, the MCU D to A mutant (the interaction partner of the Rs) is also less competent than WT to restore Ca flux.

In addition, this entire study is predicated on the finding that MICU1 and MCU specifically interact directly and not through EMRE. This is shown exclusively with co IPs, which are usually only preliminary indicators of specific binding. It would be helpful for the solidity of the argument and for all the experiments that follow, for this interaction to be shown with more direct binding assays.

Overall, for the reasons I outlined above, I believe that the authors overreach in their conclusions and interpretations. A few examples: Introduction last paragraph: "These results led to a molecular mechanism in which MICUs open or close the uniporter in response to intracellular Ca signals by physically blocking or unblocking the MCU pore."; first paragraph of Discussion: "Here, we establish a mechanism in which MICU1 shuts the uniporter by binding to the DIME-Asp side chain carboxylate ring to block the IMS entrance of the MCU pore..… Ca-binding to MICU1 at its EF hands disrupts this interaction, thus leading to opening of this Ca-activated Ca channel". Last two paragraphs of the Discussion: ".… a result that demonstrated unambiguously that S1 is not critical for Ca selectivity…..". "..the current study provides insight into the uniporter's ion selectivity mechanisms…". All these statements need to be rethought.

Reviewer #1:

Phillips et al. describe studies on the mitochondrial calcium uniporter (MCU) aimed at examining the mechanism of Ca selectivity, pore gating, and association with the MICU1 and MICU2 subunits. Using coimmunoprecipitation assays, they show that MICU1 and MCU interact directly even in the absence of the EMRE subunit, which was already known to mediate association of MICU1 to the complex. Next, they test the role of the acidic residues in the DIME signature sequence thought to comprise the selectivity filter of the uniporter. They demonstrate that no mutations of the glutamate are tolerated, consistent with the notion that it is central to forming the Ca selective pore. However, the aspartate is not required for selectivity or transport, but is required for gating the pore. Mutations of the DIME-asp also abrogate binding of MICU1 as shown by CoIP assays. Further, they show that MCU-MICU1 association is highly dependent on ionic strength, implicating electrostatic interactions, then identify 2 basic residues on MICU1 that are critical for interaction. These 2 basic residues are absent in MICU2, which the authors show does not directly bind MCU, but associates in vivo through covalent disulfide linkage to MICU1. This was further demonstrated by pulldown using the endogenous MICU1/MICU2 heterodimers and WT and mutant MCU. In the case of WT MCU, both MICU1 and MICU2 are pulled down, whereas in MCU D261A does not pull down either. Finally, they show that mutations to the basic residues of MICU1 similarly reduce the MCU gating ability as mutations to the DIME-asp, supporting the role in this interaction in gating the pore. Based on their experimental results, the authors propose a gating model for the MCU channel that is clear and illustrates their key findings well.

Overall, the study reveals several new features of MCU gating and selectivity such as the role of the DIME-asp in pore gating rather than selectivity, and the direct association of MICU1 and MCU mediated by electrostatic interactions with the DIME-asp. The claims reported by the authors are all supported by the experimental data. The writing is clear and concise but I have some concerns:

Concerns:

1) The authors should clarify how the rates are calculated when the time courses do not saturate. Is there a way to estimate the steady-state levels? One assumes that the errors are large if the time course is significantly slowed down. See D261E in Figure 3—figure supplement 1.

2) Figure 7A. The authors should consider adding more data points so that they are not using just three data points for regression.

Reviewer #2:

In the study entitled "MICU1 interacts with the conserved aspartate ring of MCU to mediate Ca2+ activation of the mitochondrial Ca2+ uniporter", Phillips et al. seek to understand the mechanism by which the gate-keeping proteins MICU1 and MICU2 confer their regulatory effect on the pore forming subunit MCU of the mitochondrial calcium uniporter. In the absence of structural information on the entire MCU complex, it remains unclear how these MICU proteins regulate the ion conducting pore subunit. Using a combination of pull-down assays and fluorescent calcium uptake assays, the authors identified two conserved arginines (R119 and R154 of the human ortholog) in MICU1 that interact with the highly conserved aspartate residue of the selectivity filter of MCU. This direct interaction allows MICU1 to inhibit the Ca2+ uptake function of MCU at low cellular calcium concentrations.

Overall, this study was well done and it addresses a fundamentally important question about how MCU is regulated. Additionally, the body of literature points to MICU1 as an important regulatory component of the uniporter since loss of MICU1 leads to mitochondrial calcium overload from cells to animal models to humans. Thus, the manuscript being considered is highly relevant and appropriate for publication in eLife.

Reviewer #3:

The manuscript by Phillips et al. reports on the role of the DIME signature sequence of the mitochondrial uniporter Ca channel, which is to both determine Ca selectivity (the E) and to mediate binding of the accessory subunit MICU1 in the absence of Ca (the D), and blocking the MCU pore. Thus, in the absence of Ca, the authors hypothesize that MICU1 binds to and sterically blocks the pore of the Ca channel, while in the presence of Ca, Ca binding to the EF hand of MICU1 leads to its dissociation from the channel, unblocking the pore and consequently allowing Ca flux.

Major issue:

The topic is interesting, and understanding the mechanism of functioning of this important channel complex is of high impact especially in the view of the recent structures of this channel. However, the detailed mechanistic model that the authors propose, and which I outlined above, is only lightly supported by the experimental data presented in this paper. The experiments are all indirect. They all are either co-immunoprecipitations or fluorescent assays performed on permeabilized HEK cells transiently transfected with WT/mutants of either the pore-forming subunit (MCU) or the interacting proteins (MICU1/2). These experiments appear well-executed but they fall short of demonstrating essentially any of the authors' claims on the MCU mechanism. The obvious components of the mechanism not demonstrated here are: MICU1 as MCU pore blocker, MICU1 unblocking the pore upon Ca-binding, the location of Ca binding for this mechanism, the involvement of the D in the DIME sequence in the interaction between the two proteins, how does Ca break the Asp-Arg mediated interaction between the two proteins, etc.

Other comments:

The Introduction is too short and lacks presentation of the selectivity of the channel (which is actually brought up in the results), as well as even a brief discussion of the existing structure(s). Despite the precision of the conclusions drawn regarding the mechanism of Ca-induced activation of MCU, the authors do not discuss this mechanism from the perspective of the structures, which is I believe necessary.

The conclusion that the Asp in the DIME sequence does not contribute to the channel's selectivity filter is premature as it is based on only one mutant that for some reason still supports Ca uptake and is not affected by Na addition in these particular assays. However, there is no control showing how Na affects these fluorescent assays in the WT or in any of the other mutants, and furthermore, the assay is still indirect.

In the D261A MCU mutant, what keeps the channel closed in normal conditions, if D is crucial to MICU1 binding and if it's this pore block that keeps the channel closed? Is this mutant constitutively active?

I am assuming that the radioactive Ca flux assay was used instead of the fluorescent one for Figure 6 because of the low Ca involved. However, this must be spelled out in the manuscript, because it looked like the fluorescent assay was quite sensitive. Furthermore, more information is needed for both the fluorescent assay and the radioactive assay (the section in the Materials and methods is not detailed enough). For instance, for the radioactive assays, I don't get a sense of what is plotted in Figure 6. Is this only one experiment, since there are no error bars? What's the signal to noise here? What do the signals look like before and after application of the Ru360? Etc…

To summarize, I believe that although the topic is interesting, timely, and of high impact, and the experiments are well-done, the detailed signature-sequence aspartate-mediated protein-protein interaction mechanism of Ca activation of the MCU is an overreach and only indirectly supported by the experimental data presented.

https://doi.org/10.7554/eLife.41112.025

Author response

[Editors’ notes: the authors’ response after being formally invited to submit a revised submission follows.]

Reviewer #1:

Overall, the study reveals several new features of MCU gating and selectivity such as the role of the DIME-asp in pore gating rather than selectivity, and the direct association of MICU1 and MCU mediated by electrostatic interactions with the DIME-asp. The claims reported by the authors are all supported by the experimental data. The writing is clear and concise but I have some concerns:

Concerns:

1) The authors should clarify how the rates are calculated when the time courses do not saturate. Is there a way to estimate the steady-state levels? One assumes that the errors are large if the time course is significantly slowed down. See D261E in Figure 3—figure supplement 1.

This is a very legitimate concern, and we apologize that this issue has not been made clear in the original manuscript. Basically, after adding 10 µM Ca2+ (box in Figure 3A), free Ca2+ in the extra-mitochondrial solution would eventually drop back to a steady-state level that is roughly the same as that before Ca2+ addition. This is because mitochondria have high Ca2+ buffering capacity to sequester added Ca2+. A main problem in this assay is that the amplitude of fluorescence-signal increase upon adding 10 µM Ca2+ can vary from 200 to 300 a.u., presumably due to variations in extra-mitochondrial Ca2+ buffering capacity. Moreover, it is difficult to control protein-expression to exactly the same level for all constructs. Thus, the initial rate, presented as a.u./s, is qualitative in nature, and we strictly used this to address yes/no questions. We have now revised the Materials and methods section to highlight these issues. The s.e.m. for D261E is indeed smaller than WT, but in addition to the time course, there could be other factors, such as the consistency of protein expression, that contribute to errors.

2) Figure 7A. The authors should consider adding more data points so that they are not using just three data points for regression.

In Figure 7, we used a quantitative 45Ca2+ flux assay to determine the rate of mitochondrial Ca2+ uptake. In a typical experiment, we obtain readings of 45Ca2+ at 3 different time points, and then fit the data with a linear function to obtain the rate of Ca2+ uptake (as shown in Figure 7A). Rates from at least 3 independent experiments were than averaged for reporting. As this method is highly sensitive, we found that 3 time points are sufficient for reliable quantification of the initiate rate. We now supplement data from several individual experiments in Figure 7—figure supplement 1 to give the readers a better sense about the variation in these experiments, and have also revised Materials and methods to make these issues clear.

Reviewer #2:

In the study entitled "MICU1 interacts with the conserved aspartate ring of MCU to mediate Ca2+ activation of the mitochondrial Ca2+ uniporter", Phillips et al. seek to understand the mechanism by which the gate-keeping proteins MICU1 and MICU2 confer their regulatory effect on the pore forming subunit MCU of the mitochondrial calcium uniporter. In the absence of structural information on the entire MCU complex, it remains unclear how these MICU proteins regulate the ion conducting pore subunit. Using a combination of pull-down assays and fluorescent calcium uptake assays, the authors identified two conserved arginines (R119 and R154 of the human ortholog) in MICU1 that interact with the highly conserved aspartate residue of the selectivity filter of MCU. This direct interaction allows MICU1 to inhibit the Ca2+ uptake function of MCU at low cellular calcium concentrations.

Overall, this study was well done and it addresses a fundamentally important question about how MCU is regulated. Additionally, the body of literature points to MICU1 as an important regulatory component of the uniporter since loss of MICU1 leads to mitochondrial calcium overload from cells to animal models to humans. Thus, the manuscript being considered is highly relevant and appropriate for publication in eLife.

Reviewer #3:

We fully appreciate the reviewer’s suggestion to rely more on quantitative assays, such as FRET or ITC, to probe the interaction between MCU and MICU1. We do eventually hope to obtain binding parameters using these assays, but this would require purification of high-quality MCU proteins from higher eukaryotes, a challenging task that has not been achieved in the field. (MCU structures were determined recently, but these are homologues in fungi, which have no MICU1. Some of these fungal MCUs also show no function. An NMR structure published in 2016 used C. elegans MCU, but the protein was extracted from inclusion bodies in Fos-Choline- 14, a harsh detergent rarely used in membrane-protein biochemistry).

CoIP indeed has its limitation, but it also has unique advantages: the uniporter complex is properly assembled by cellular machineries in mitochondria, and the function of the complex can be assessed in native environments using Ca2+ flux assays. Importantly, the ability to substitute native proteins with point mutants using CRISPR/Cas9 now enables the detection of highly- specific protein-protein interactions. Our goal is to take full advantage of these strengths to address important questions in uniporter mechanisms. Below we provide a point-to-point response to explain why our conclusion regarding Ca2+ activation of the uniporter is within the resolution limit of our assays. We separate the reviewer’s comments into 3 parts, about issues related to the Ca2+ flux assay, CoIP, and writing.

Ca2+ flux assay:

In contrast, in the current manuscript, as I previously stated, the authors draw very specific mechanistic interpretations based on this yes/no assay. For example:

Mutation of the D in the DIME signature sequence region of MCU to an A did not abolish Ca import but lead to loss of MICU1 interaction (as measured with co-IP). Furthermore, addition of 100 mM Na during the assay did not change the rate of Ca-import. The authors conclude that the D does not contribute to channel selectivity but mediates MICU1 interaction. The problem are the following: first, the assay does not address selectivity at all, but rather whether the mutation kills Ca transport. The authors should consider dropping any mention of the role of D in selectivity. Second, there should be a control experiment showing the effect of Na addition on WT MCU. Third, the finding that mutation of the D to any other amino acid (except E and A) abolished Ca-import is intriguing, and argues that the story is much more complex. Fourth, the authors should also present the coIPs with the D261E MCU mutant, which, according to the authors' hypothesis, is functional and expected to support the interaction with MICU1.

The reviewer is concerned that the fluorescence-based Ca2+ flux assay is not quantitative, and is suitable mostly for yes/no questions. Indeed, as in our response to reviewer #1, we strictly limit the use this assay for addressing yes/no questions. For instance, we show that D261A is Ca2+-transport competent, and that the transport is unaffected by adding 100 mM Na+ or Ru360. These are all valid conclusions well within the resolution of the assay.

As for the selectivity, the Clapham lab has shown that MCU employs a classical multi-ion pore mechanism in which Na+ can rapidly permeate the channel in the absence of Ca2+ and adding Ca2+ blocks the Na+ flux and leads to Ca2+ permeation. Therefore, if D261A significantly reduces MCU’s Ca2+ selectivity, adding 100 mM Na+ should strongly suppress the Ca2+ (only 10 µM) flux. That being said, a rigorous test of selectivity indeed requires electrophysiological experiments not performed in this work. We therefore decide to change the wording “selectivity” to “permeation” to be more accurate. As requested by the reviewer, we have performed experiments adding Na+ to WT MCU, and showed that Na+ has no effect on Ca2+ transport (Figure 3—figure supplement 1).

It’s true that several D261 mutations are non-functional, but this is not surprising as D261 sits in a critical position in the pore, right above the high-affinity Ca2+ site formed by E264. Mutations of D261 could in many non-specific ways perturb the chemistry required for Ca2+ transport. Finally, the original manuscript did provide the D261E CoIP data—it binds to MICU1 (please see Figure 4).

The conclusion that the Asp in the DIME sequence does not contribute to the channel's selectivity filter is premature as it is based on only one mutant that for some reason still supports Ca uptake and is not affected by Na addition in these particular assays. However, there is no control showing how Na affects these fluorescent assays in the WT or in any of the other mutants, and furthermore, the assay is still indirect.

We now change the sentence in the Results to “these results demonstrate that the DIME-Asp mediates MCU interaction with MICU1, instead of contributing essentially to MCU’s Ca2+ permeation.” The reviewer is concerned that this argument is based on a single D261A mutation. However, this is a very powerful positive result. Enzymatic reactions are known to require very specific chemistry. If the D261 side-chain is necessary for high-throughput Ca2+ permeation, it is extremely unlikely that other protein components can somehow compensate for a drastic change of the D261 side-chain to produce Ca2+ transport. That’s why we conclude that D261 does not contribute essentially to Ca2+ transport. It’s true that several other D261 mutants are non-functional, but again, such negative results are not particularly surprising, considering the critical position of the D261 residue in the pore. The assay is not quantitative, but it is direct and highly specific—it is a standard assay widely used to directly detect the uniporter’s transport activity. Our conclusion is derived well within the ability of the assay.

In Figure 6-7, the authors test whether MICU1 interaction with the pore in resting Ca concentrations is necessary to keep the pore closed. They find by coIP that the MICU1-MCU interaction is disrupted in the presence of 10 μm Ca and that in MICU1 knockout cells, Ca fluxes inside mitochondria, while if MICU1 is added back, less Ca is uptaken. If the MICU1 R mutants are added back instead, except for one of them, they are less competent than WT in restoring Ca flux. It would have been useful for the authors to also show that, similar to the R mutants on MICU1, the MCU D to A mutant (the interaction partner of the Rs) is also less competent than WT to restore Ca flux.

We have now performed 45Ca2+ flux experiments using MCU-KO cells transfected with WT or D261A MCU. Figure 7B shows that in submicromolar Ca2+, D261A produces mitochondrial Ca2+ uptake, while WT is completely inactive. This provides strong support for our model that MICU1 must bind to MCU to shut the uniporter in resting cellular conditions. While carrying out these experiments, we noticed that the rate of D261A-mediated Ca2+ “leak” is 6.2-fold slower than that observed using MICU1-KO cells. We investigated, and identified a few factors that might underlie the small magnitude of the leak, including a slower turnover rate of D261A than WT MCU, and other residues (e.g., S259) being involved in MICU1 binding. These are now described in the revised Result section.

CoIP related issues:

This entire study is predicated on the finding that MICU1 and MCU specifically interact directly and not through EMRE. This is shown exclusively with co IPs, which are usually only preliminary indicators of specific binding. It would be helpful for the solidity of the argument and for all the experiments that follow, for this interaction to be shown with more direct binding assays.

We share the reviewer’s concerns regarding the limitation of CoIP. Therefore, we endeavored to obtain multiple lines of evidence before drawing conclusions. For the MCU-MICU1 interaction, we show that it is highly-conserved so that human MICU1 can pull down MCU in lower eukaryotes (i.e. plants and protists), whose uniporters contain only MCU and MICU1 (no EMRE).

Moreover, this MCU-MICU1 interaction can be manipulated by point mutations: D261A at the cytoplasmic surface of MCU disrupts the complex, while E264A deeper in the pore does not affect MICU1 binding. This is analogous to the classical approach of utilizing point-directed mutagenesis to identify specific protein-protein interactions in binding assays.

The authors claim that the MCU-MICU1 interaction is electrostatic as evidenced by weaker interactions as measured by lower ratios of MCU/MICU1 gel bands post coIPs in higher ionic strength buffer conditions. If the authors want to make this electrostatic claim, they should show both a positive and a negative control experiment. Otherwise, a direct binding interaction assay would be more appropriate in this case.

The finding that the stability of the MCU-MICU1 complex can be manipulated by varying the ionic strength is diagnostic of electrostatic interactions (Figure 5A). This together with the observation that D261A and D261Q, but not D261E, abolishes MICU1 binding (Figure 4) provide strong evidence that the MCU-MICU1 interaction is electrostatic. The reviewer has a good point that controls should be provided. We now present data showing that the MICU1-2 interaction and the epitope interaction between 1D4-tagged MCU and the anti-1D4 antibody are unaffected by the ionic strength (Figure 5—figure supplement 1).

The authors screen for interacting partners for the Asp on MCU by mutating 18 conserved Arg and Lys residues on MICU1 to Ala and checking for interaction using coIPs. They find two Arg that disrupt MCU association and claim that these Rs on MICU1 mediate electrostatic interactions with the D on MCU. Perhaps these claims should be reduced to: the two Rs are important in mediating the interaction between MCU and MICU1, since there is no indication that the Rs interact with the D specifically. Furthermore, if the screen were not limited to positively charged residues, perhaps other residues would pop up as important to this interaction.

We completely agree with the reviewer that our data do not demonstrate directly that these two Rs interact with D261. Instead, the results indicate that these residues are crucial for the MCU- MICU1 interaction. We have been careful to make this clear, and have further revised the manuscript to emphasize this point.

The authors conclude that the MICU1 interaction with MCU must work by MICU1 directly blocking the MCU pore, predicated on their conclusions that the Rs on MICU1 electrostatically interact with the Ds in MCU, which are located in the pore. Since there is no solid evidence for the Rs interacting with the D, the pore blocking interpretation is similarly weak and I suggest that it is removed.

We disagree with this comment. An electrostatic interaction of MICU1 with D261 at the entrance of the MCU pore, whether it’s directly mediated by R119/R154 or not, would prevent Ca2+ entry into the pore, as in the classical example of charybdotoxin block of K channels. This is further confirmed by the quantitative functional analysis showing that mutations that disrupt MCU- MICU1 interaction compromise MICU1’s ability to close MCU. These data argue strongly that MICU1 shuts MCU in resting cellular conditions by blocking the pore.

These experiments appear well-executed but they fall short of demonstrating essentially any of the authors' claims on the MCU mechanism. The obvious components of the mechanism not demonstrated here are: MICU1 as MCU pore blocker, MICU1 unblocking the pore upon Ca- binding, the location of Ca binding for this mechanism, the involvement of the D in the DIME sequence in the interaction between the two proteins, how does Ca break the Asp-Arg mediated interaction between the two proteins, etc.

We hope that we have explained clearly why our results provide strong evidence that MICU1 closes MCU by electrostatically interact with the DIME-Asp to block the pore. We are unable to answer all questions in a single paper. As already stated in the last paragraph of our Discussion, future work is needed to test the role of MICU1’s individual EF hand in this mechanism, and to understand how Ca2+ breaks the MCU-MICU1 complex. The later most likely requires an atomic structure of the MCU-MICU1 subcomplex.

Comments on manuscript writing:

The Introduction is too short and lacks presentation of the selectivity of the channel (which is actually brought up in the results), as well as even a brief discussion of the existing structure(s). Despite the precision of the conclusions drawn regarding the mechanism of Ca-induced activation of MCU, the authors do not discuss this mechanism from the perspective of the structures, which is I believe necessary.

As the selectivity of the channel is not the main issue in this manuscript, we feel that discussing this in the Introduction would distract the readers from the main question of how MICU1 gates MCU. We did describe the classical multi-ion pore mechanism underlying Ca2+-channel selectivity in the Discussion. The Ca2+-activation mechanism of MCU is mediated by the MICU1 protein. Currently, there is no MCU-MICU1 subcomplex structure. Structures of MCU homologues from fungal species reveal MCU’s oligomer state and a possible Ca2+-selectivity mechanism, but provide very little insights into the mechanism by which MICU1 regulates MCU. (These fungal species also do not have MICU1). There are potential issues with the MICU1 structure as described in our response to reviewer #2. Thus, we feel that it’s premature to discuss these structures in the Introduction.

I am assuming that the radioactive Ca flux assay was used instead of the fluorescent one for Figure 6 because of the low Ca involved. However, this must be spelled out in the manuscript, because it looked like the fluorescent assay was quite sensitive. Furthermore, more information is needed for both the fluorescent assay and the radioactive assay (the section in the Materials and methods is not detailed enough). For instance, for the radioactive assays, I don't get a sense of what is plotted in Figure 6. Is this only one experiment, since there are no error bars? What's the signal to noise here? What do the signals look like before and after application of the Ru360? Etc

Please see our response to reviewer #1’s major concerns. We have extensively revised Materials and methods to provide more detailed information, and have also included the data before and after adding Ru360 in Figure 7—figure supplement 1.

Overall, for the reasons I outlined above, I believe that the authors overreach in their conclusions and interpretations. A few examples: Introduction last paragraph: "These results led to a molecular mechanism in which MICUs open or close the uniporter in response to intracellular Ca signals by physically blocking or unblocking the MCU pore."; first paragraph of Discussion: "Here, we establish a mechanism in which MICU1 shuts the uniporter by binding to the DIME-Asp side chain carboxylate ring to block the IMS entrance of the MCU pore..… Ca-binding to MICU1 at its EF hands disrupts this interaction, thus leading to opening of this Ca-activated Ca channel". Last two paragraphs of the Discussion: ".… a result that demonstrated unambiguously that S1 is not critical for Ca selectivity…..". "..the current study provides insight into the uniporter's ion selectivity mechanisms…". All these statements need to be rethought.

We hope that we have explained clearly that our conclusions are based on strong experimental supports. The word “selectivity” has been changed to “permeation.”

https://doi.org/10.7554/eLife.41112.026

Article and author information

Author details

  1. Charles B Phillips

    Department of Biochemistry, Brandeis University, Waltham, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Validation, Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
  2. Chen-Wei Tsai

    1. Department of Biochemistry, Brandeis University, Waltham, United States
    2. Department of Physiology and Biophysics, University of Colorado Anschutz Medical Campus, Aurora, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Validation, Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
  3. Ming-Feng Tsai

    1. Department of Biochemistry, Brandeis University, Waltham, United States
    2. Department of Physiology and Biophysics, University of Colorado Anschutz Medical Campus, Aurora, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    ming-feng.tsai@ucdenver.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4277-1885

Funding

National Institute of General Medical Sciences (R01-GM129345)

  • Chen-Wei Tsai
  • Ming-Feng Tsai

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication. The funders pay for the authors' salary and other research expenses.

Acknowledgements

We thank Carole Williams for technical assistance in molecular biology, and Dr. Christopher Miller for critical reading of this manuscript as well as providing unconditional support during the development of this project. We thank Dr. Vamsi Mootha for kindly providing an independent strain of MICU1-KO cells for us to verify results in Figure 7. CWT and MFT are partly supported by the NIH grant R01-GM129345. The authors declare no conflict of interests.

Senior Editor

  1. John Kuriyan, University of California, Berkeley, United States

Reviewing Editor

  1. Baron Chanda, University of Wisconsin-Madison, United States

Publication history

  1. Received: August 14, 2018
  2. Accepted: January 7, 2019
  3. Accepted Manuscript published: January 14, 2019 (version 1)
  4. Accepted Manuscript updated: January 15, 2019 (version 2)
  5. Version of Record published: January 25, 2019 (version 3)

Copyright

© 2019, Phillips et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,140
    Page views
  • 355
    Downloads
  • 41
    Citations

Article citation count generated by polling the highest count across the following sources: Scopus, Crossref, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Charles B Phillips
  2. Chen-Wei Tsai
  3. Ming-Feng Tsai
(2019)
The conserved aspartate ring of MCU mediates MICU1 binding and regulation in the mitochondrial calcium uniporter complex
eLife 8:e41112.
https://doi.org/10.7554/eLife.41112

Further reading

    1. Biochemistry and Chemical Biology
    2. Chromosomes and Gene Expression
    Kanishk Jain, Matthew R Marunde ... Brian D Strahl
    Short Report Updated

    In nucleosomes, histone N-terminal tails exist in dynamic equilibrium between free/accessible and collapsed/DNA-bound states. The latter state is expected to impact histone N-termini availability to the epigenetic machinery. Notably, H3 tail acetylation (e.g. K9ac, K14ac, K18ac) is linked to increased H3K4me3 engagement by the BPTF PHD finger, but it is unknown if this mechanism has a broader extension. Here, we show that H3 tail acetylation promotes nucleosomal accessibility to other H3K4 methyl readers, and importantly, extends to H3K4 writers, notably methyltransferase MLL1. This regulation is not observed on peptide substrates yet occurs on the cis H3 tail, as determined with fully-defined heterotypic nucleosomes. In vivo, H3 tail acetylation is directly and dynamically coupled with cis H3K4 methylation levels. Together, these observations reveal an acetylation ‘chromatin switch’ on the H3 tail that modulates read-write accessibility in nucleosomes and resolves the long-standing question of why H3K4me3 levels are coupled with H3 acetylation.

    1. Biochemistry and Chemical Biology
    2. Medicine
    Dmitry Ter-Ovanesyan, Tal Gilboa ... David R Walt
    Research Advance

    Extracellular vesicles (EVs) are released by all cells into biofluids such as plasma. The separation of EVs from highly abundant free proteins and similarly sized lipoproteins remains technically challenging. We developed a digital ELISA assay based on Single Molecule Array (Simoa) technology for ApoB-100, the protein component of several lipoproteins. Combining this ApoB-100 assay with previously developed Simoa assays for albumin and three tetraspanin proteins found on EVs (Ter-Ovanesyan, Norman et al., 2021), we were able to measure the separation of EVs from both lipoproteins and free proteins. We used these five assays to compare EV separation from lipoproteins using size exclusion chromatography with resins containing different pore sizes. We also developed improved methods for EV isolation based on combining several types of chromatography resins in the same column. We present a simple approach to quantitatively measure the main impurities of EV isolation in plasma and apply this approach to develop novel methods for enriching EVs from human plasma. These methods will enable applications where high-purity EVs are required to both understand EV biology and profile EVs for biomarker discovery.