Rearrangement of MICU1 multimers for activation of MCU is solely controlled by cytosolic Ca2+

Mitochondrial Ca2+ uptake is a vital process that controls distinct cell and organelle functions. Mitochondrial calcium uptake 1 (MICU1) was identified as key regulator of the mitochondrial Ca2+ uniporter (MCU) that together with the essential MCU regulator (EMRE) forms the mitochondrial Ca2+ channel. However, mechanisms by which MICU1 controls MCU/EMRE activity to tune mitochondrial Ca2+ signals remain ambiguous. Here we established a live-cell FRET approach and demonstrate that elevations of cytosolic Ca2+ rearranges MICU1 multimers with an EC50 of 4.4 μM, resulting in activation of mitochondrial Ca2+ uptake. MICU1 rearrangement essentially requires the EF-hand motifs and strictly correlates with the shape of cytosolic Ca2+ rises. We further show that rearrangements of MICU1 multimers were independent of matrix Ca2+ concentration, mitochondrial membrane potential, and expression levels of MCU and EMRE. Our experiments provide novel details about how MCU/EMRE is regulated by MICU1 and an original approach to investigate MCU/EMRE activation in intact cells.

earlier than MCU, has two classical EF-hand Ca 2+ -binding domains and forms large complexes of app. 480 kDa 17,20 . The interaction of MICU1 with MCU is mediated by EMRE 16,21 . The two Ca 2+ -binding EF-hand domains of MICU1 can undergo large conformational changes upon binding of Ca 2+ 22,23 and face the intermembrane space 18 . MICU1 is a gatekeeper of MCU preventing an increase of the mitochondrial Ca 2+ concentration ([Ca 2+ ] mito ) at low resting Ca 2+ levels while it is thought to facilitate mitochondrial Ca 2+ uptake upon high Ca 2+ concentrations 18,24,25 . Recently, an elegant study determined the crystal structure of MICU1 indicating that in the absence of Ca 2+ MICU1 exists as a hexamer, while Ca 2+ binding to the two EF-hands results in a rearrangement to MICU1 dimers 23 . This study provided the first insights into the molecular mechanism by which MICU1 possibly controls MCU/EMRE activity and suggests that the hexameric MICU1 prevents MCU/EMRE Ca 2+ channel activity that gets released upon (re-)organization of MICU1 hexamers/multimers 7 . However, whether or not Ca 2+ -dependent rearrangements of MICU1 multimers indeed occur and regulate the activity of MCU in intact cells, and the impact of MCU and EMRE to the structural (re-)organization of MICU1 have not been investigated so far. Hence, the kinetics of the rearrangement of the MICU1 multimers upon cytosolic Ca 2+ elevation and whether or not the mitochondrial matrix Ca 2+ concentration or membrane potential (ψ mito ) have an impact on the structural reorganization of MICU1 remain elusive.
In this study we used a FRET-based live-cell imaging approach to dynamically monitor the kinetics of the structural (re-)organization of MICU1 and to explore its dependence from cytosolic and mitochondrial Ca 2+ signals as well as ψ mito . Simultaneous measurements of cytosolic Ca 2+ with either mitochondrial Ca 2+ signals or MICU1 rearrangement were performed to verify whether MICU1 multimers indeed retard MCU activation in intact cells. Finally, we tested whether the Ca 2+ -sensitive rearrangement of MICU1 is dependent of the expression level of MCU and EMRE. Our data provide new insights in the dynamics, regulation, and molecular effect of the structural (re-)organization of MICU1 that adds to our current understanding of the complex molecular mechanisms of MCU activation.

Results
Ca 2+ elevations yield rearrangement of MICU1 multimers in intact cells. In order to dynamically monitor whether and, if so, how intracellular Ca 2+ -mobilization by an inositol 1,4,5-trisphosphate-(IP 3 -)generating agonist affects the arrangement of MICU1 in intact cells, Förster energy transfer (FRET) imaging was applied in cells co-expressing MICU1 fused to either cyan fluorescent protein (MICU1-CFP) or yellow fluorescent protein (MICU1-YFP). Because MICU1 has been shown to assemble in hexamers in the absence of Ca 2+ and rearranges to dimers upon Ca 2+ binding 23 , we assumed that under resting conditions the expressed MICU1-CFP and MICU1-YFP chimeras exist as hexamers, thus, facilitating FRET from CFPs to YFPs (Fig. 1A). However, elevation of Ca 2+ and its subsequent binding to MICU1 should yield disaggregation of MICU1 hexamers and reduce the inter-MICU1 FRET signal. Indeed, cell treatment with the IP 3 -generating agonist histamine 11 considerably reduced the inter-MICU1 FRET ratio (Fig. 1B,C), while the agonist had no effect on fluorescence of cells expressing MICU1-YFP alone ( Supplementary Fig. 1). Upon removal of the agonist the signal was restored (Fig. 1B,C), indicating the reassembly of MICU1 to higher multimers upon the decline of Ca 2+ levels. In the nominal absence of extracellular Ca 2+ , the histamine-triggered decrease of inter-MICU1 FRET was more transient and slowly developed upon subsequent addition of extracellular Ca 2+ ( Supplementary Fig. 2), thus, highlighting that inter-MICU1 FRET strictly follows cellular Ca 2+ signals under these conditions ( Supplementary Fig. 3). In contrast to the fast intracellular Ca 2+ mobilization in response to histamine, slow Ca 2+ mobilization from the endoplasmic reticulum (ER) by the sarcoplasmic/endoplasmic reticulum Ca 2+ ATPase (SERCA) inhibitor 2,5-di-tert-butylhydroquinone (BHQ) 26 only slowly and weakly reduced the inter-MICU1 FRET signal ( Supplementary Fig. 4). Under this condition mitochondrial Ca 2+ uptake in response to cell treatment with BHQ was largely increased in cells treated with the 3´UTR siRNA against MICU1 ( Supplementary Fig. 5), despite an almost unaffected cytosolic Ca 2+ elevation ( Supplementary Fig. 6). These findings indicate that the slow and weak ER depletion with BHQ yields only in insufficient MICU1-rearrangement, while the inhibitory function of MICU1 multimers on MCU remains under these conditions. However, expression of FP tagged MICU1 completely restored the inhibition of mitochondrial Ca 2+ uptake at low Ca 2+ ( Supplementary Fig. 5). In contrast, using a MICU1 variant mutated in both canonical EF hands 17,23 was neither able to rescue this siRNA mediated effect of MICU1 ( Supplementary Fig. 7) nor showed any MICU1 FRET rearrangement upon stimulation with histamine ( Supplementary Fig. 8).
In order to determine the affinity of the MICU1 multimers for Ca 2+ to evoke their rearrangement in situ, the Ca 2+ ionophore ionomycin was used to clamp different cytosolic Ca 2+ concentrations (Fig. 1D,E). These experiments revealed a half maximal effective Ca 2+ concentration (EC 50 ) to trigger rearrangement of the MICU1 multimers of 4.4 (3.7-5.2) μ M in HeLa cells (Fig. 1E). Our findings are in line with studies demonstrating the existence of Ca 2+ micro-domains of up to 16 μ M in hot spots between the ER and mitochondria in response to an IP 3 -generating agonist 27 . Such Ca 2+ hot spots would efficiently destruct MICU1 multimers and, hence, activate mitochondrial Ca 2+ uniport. Our results provide a first demonstration that MICU1 oligomerization is reversibly controlled by high and low [Ca 2+ ] cyto in intact cells.
Simultaneous imaging of cytosolic and mitochondrial Ca 2+ signals in individual single cells using a red-shifted mitochondria targeted cameleon (mtD1GO-Cam) in combination with the near ultra-violet excitable cytosolic Ca 2+ sensor fura-2 was performed in order to compare the kinetics of cytosolic and mitochondrial Ca 2+ signals. This approach revealed a lag time (Δ T) of 1.50 ± 0.08 s between the cytosolic Ca 2+ rise and its transition into the mitochondrial matrix upon intracellular Ca 2+ mobilization by the IP 3 -generating agonist histamine ( Fig. 2A). The molecular mechanism responsible for this temporal shift between rises of [Ca 2+ ] cyto and [Ca 2+ ] mito is so far unknown. To investigate whether or not the Ca 2+ -dependent de-oligomerization of MICU1 multimers might retard mitochondrial Ca 2+ signals, identical experiments were performed in cells transfected with siRNA specifically against MICU1. In these MICU1-diminuted cells the coupling between cytosolic and mitochondrial Ca 2+ signals was greatly improved and the mitochondrial Ca 2+ signal followed almost instantly the cytosolic Ca 2+ elevation upon histamine stimulation (Δ T = 0.93 ± 0.05 s) (Fig. 2B,C).
In order to visualize how the temporal pattern of MICU1 de-oligomerization refers to cytosolic Ca 2+ signals, we imaged dynamic changes of MICU1 FRET simultaneously with cytosolic Ca 2+ in response to histamine (Fig. 2D). This approach revealed that MICU1 de-oligomerization is only slightly delayed from the histamine-induced increase of [Ca 2+ ] cyto and, thus, clearly precedes mitochondrial Ca 2+ uptake (Fig. 2E). Notably, app. 90% of the maximal histamine-induced increase of [Ca 2+ ] cyto trigger ≥ 50% of MICU1 de-oligomerization that appears essential to initiate considerable uptake of Ca 2+ by mitochondria  Fig. 2D,E). These data dissect the transfer of cytosolic Ca 2+ into mitochondria into (at least) four sequential steps: (i) cytosolic Ca 2+ elevation and transfer into the intermembrane space, (ii) binding of Ca 2+ to MICU1 oligomers, (iii) MICU1 de-oligomerization, and (IV) activation of the MCU.

Rearrangement of MICU1 multimers occurs independently of the mitochondrial membrane potential and matrix Ca 2+ elevation.
In order to test if the mitochondrial membrane potential (Ψ mito ) impacts rearrangement of MICU1 multimers, cells were treated with oligomycin and the uncoupling agent carbonyl cyanide-4-(trifluoromethoxy) phenylhydrazone (FCCP) to efficiently depolarize mitochondria 12 , thus, only very small mitochondrial Ca 2+ signals in response to histamine were observed (Fig. 3A) approving the loss of Ψ mito under these conditions. However, depolarization of mitochondria did neither affect basal MICU1 FRET nor the histamine-triggered reduction of the inter-MICU1 FRET signal (Fig. 3B). These data indicate that the rearrangement of MICU1 multimers upon intracellular Ca 2+ mobilization with an IP 3 -generating agonist occurs independently of Ψ mito and the mitochondrial matrix Ca 2+ elevation.

MICU1 multimers rearrange irrespective of the expression level of MCU and EMRE. MICU1
is known to interact with EMRE 16 that represents the second pore forming protein of the mitochondrial Ca 2+ complex beside MCU [28][29][30] , which is regulated by MICU1 19,21,25 . To investigate the importance of EMRE and MCU for the Ca 2+ -triggered rearrangement of MICU1 multimers, expression of either EMRE or MCU were diminished by respective verified siRNAs (Supplementary Fig. 9). Diminution of the expression of either MCU or EMRE strongly reduced mitochondrial Ca 2+ uptake ( Supplementary  Fig. 10) while no effect on cytosolic Ca 2+ signaling was observed ( Supplementary Fig. 11). However, the rearrangement of the MICU1 multimers upon stimulation with histamine was neither affected by knock-down of MCU (Fig. 4A,B) nor EMRE (Fig. 4C). In line with these findings the EC 50 of Ca 2+ to reduce the inter-MICU1 FRET signal remained unaffected in cells diminished of either MCU or EMRE (Fig. 4D).

Discussion
The present study describes the dynamics of MICU1 (re-)organization in response to cytosolic Ca 2+ elevations in intact cells. Using the FRET technology we could correlate the Ca 2+ induced rearrangement of MICU1 multimers with the activation of mitochondrial Ca 2+ uniport and examined the impact of mitochondrial Ca 2+ , Ψ mito and the expression levels of MCU and EMRE for MICU1 (re-)organization. Our data highlight that an elevation of cytosolic free Ca 2+ rearranges MICU1 multimers to smaller complexes in intact cells. These findings are consistent with a recent report showing that recombinant Ca 2+ -free MICU1 exists as hexamer and rearranges in the presence of Ca 2+ to dimers 7,23 . Hence, our data that point to the existence of large MICU1 complexes that suppresses mitochondrial Ca 2+ uptake at low cytosolic Ca 2+ is in line with previous reports about the inactivity of MCU under resting conditions 8,18 .
Our approach of live-cell monitoring the inter-MICU1 FRET allows the visualization of MICU1 (re-) organization within the intact cell in life-time. Notably, calculation of the changes in MICU1 FRET probability in the transition from hexamers to dimers revealed a theoretically achievable reduction of FRET by app. 52% ( Supplementary Information, Supplementary Fig. 12). However, the actual measured changes in FRET upon MICU1 rearrangement triggered by histamine was app. 7%. The discrepancy between the theoretical FRET changes upon the rearrangement of hexamers to independent dimers and the actual measured one might be due to basically two reasons: first, the high expression of FP tagged MICU1 proteins might result in lower oligomerization states, and second, upon histamine-induced intracellular Ca 2+ release, MICU1 de-oligomerization most likely occurs predominantly within the junctions between the ER and mitochondria where high Ca 2+ gradients are developed 27 , thus, only a small portion of MICU1 hexamers actually undergoes rearrangement under this conditions. This assumption is further supported by the app. 15% FRET reduction achieved in the assessment of the concentration response curve of MICU1 FRET dynamic to Ca 2+ .
Our approach that allowed correlation of the kinetics of MICU1 (re-)organization with cellular Ca 2+ dynamics data revealed that the rearrangement of MICU1 multimers by cytosolic Ca 2+ strictly correlates with the cytosolic Ca 2+ concentration and is rapidly reversible. The in situ calibration indicates that the rearrangement of MICU1 senses Ca 2+ changes in the range between the low basal Ca 2+ levels of 100 nM up to 100 μ M. With an EC 50 of around 4 μ M, the Ca 2+ sensitivity of MICU1 rearrangement lays exactly in the range of Ca 2+ hot spots (3.78 and 16.4 μ M) that have been measured at the outer mitochondrial membrane between the junction of mitochondria and the ER 27 . In line with this report, the IP 3 -mediated intracellular Ca 2+ mobilization almost instantly yielded MICU1 rearrangement, while Ca 2+ entry via the store-operated Ca 2+ entry pathway 31,32 triggered only a slow and moderate re-organization of MICU1 multimers. These differences between the MICU1 rearrangement upon intracellular Ca 2+ release and entering Ca 2+ are consistent with previous reports that described a rather slow on-kinetics of entering Ca 2+ at the mitochondrial surface without the formation of Ca 2+ hot spots and slow mitochondrial Ca 2+ accumulation 14,27,33 .
Similar to entering Ca 2+ , the inhibition of SERCA by BHQ yielded only slow and small rearrangement of MICU1 multimers and, thus, explains the marginal effect of SERCA inhibition on mitochondrial Ca 2+ uptake. Since the diminution of MICU1 expression substantially gains mitochondrial Ca 2+ uptake in response to intracellular Ca 2+ release by SERCA inhibition, the prominent role of MICU1 as a negative regulator of mitochondrial Ca 2+ uniport is clearly demonstrated. This assumption is further supported by our experiments in which cytosolic and mitochondrial free Ca 2+ were simultaneously measured and the lag time of app. 1.5 s between cytosolic Ca 2+ elevation and the mitochondrial Ca 2+ uptake was strongly reduced in cells with diminished expression of MICU1. Hence, the correlation between cytosolic Ca 2+ elevation, the rearrangement of MICU1 multimers and mitochondrial Ca 2+ signals, revealed that MICU1 reorganization temporally occurs in between the Ca 2+ rise within the two compartments. Notably, significant mitochondrial Ca 2+ uptake appears to occur at app. 50% rearrangement of MICU1 multimers. These data demonstrate that for activation of the MCU/EMRE complex to achieve mitochondrial Ca 2+ influx, app. 50% of the MICU1 multimers have to be rearranged. Although these findings might be due to the overexpression of MICU1 in our model, the MICU1-dependent lack of mitochondrial Ca 2+ uptake in response to SERCA inhibition supports this assumption.
Mitochondrial Ca 2+ uptake strongly depends on Ψ mito that establishes the driving force for Ca 2+ uniport into the organelle. However, whether or not Ψ mito also impacts on MICU1 organization has not been investigated so far. Our data with completely depolarized mitochondria revealed an unchanged MICU1 dynamics despite a greatly reduced mitochondrial Ca 2+ uptake. These data demonstrate that neither Ψ mito nor matrix Ca 2+ are involved in the rearrangement of MICU1 complexes and confirms cytosolic Ca 2+ as possible the sole regulator of MICU1 (re-)organization 23 .
Considering the interaction of MICU1 with EMRE and MCU 7,16,18,19,23,25,29 , the importance of these two pore-forming proteins 16,28,29 for the structural organization of MICU1 was evaluated. Although the siRNA-mediated diminution of either of these proteins resulted in strongly reduced mitochondrial Ca 2+ uptake, the Ca 2+ -triggered rearrangement of MICU1 multimers, their arrangement upon the reduction of cytosolic Ca 2+ to basal levels, and the sensitivity to cytosolic Ca 2+ remained unaffected by the reduction of MCU or EMRE. These data demonstrate that the structural (re-)organization of MICU1 upon elevation of cytosolic free Ca 2+ does not involve MCU or EMRE indicating that the Ca 2+ -induced rearrangement of MICU1 multimers is a robust process that does most likely not require other interaction partners.
Despite the topology of MICU1 is still under debate 18,21,25,30 , our data that, first, MICU1 FRET rearrangement follows cytosolic but not matrix mitochondrial Ca 2+ , second, MICU1 knockdown results in a faster mitochondrial Ca 2+ uptake, third, rearrangement of MICU1 FRET precedes mitochondrial Ca 2+ uptake, forth, MICU1 FRET is independent from Ψ mito , and fifth, neither knockdown of MCU nor of EMRE influences MICU1 FRET rearrangement indicate that MICU1 anchors with its N-terminus in the IMM while the core protein is oriented towards intermembrane space and not to the matrix.
Our results provide new mechanistic insights in the regulation of mitochondrial Ca 2+ uptake. For the first time, the kinetics and adjustments of one of the most important molecular gatekeeper of mitochondrial Ca 2+ uptake was visualized in intact cells. Our data revealed cytosolic Ca 2+ as the most prominent regulator of the structural organization of MICU1, which in the form of a homo-multimere potently inhibits the MCU/EMRE mitochondrial Ca 2+ channel complex. Moreover, neither Ψ mito nor matrix Ca 2+ , nor MCU or EMRE were found to affect the Ca 2+ -controlled (dis)assembly of MICU1 multimers. Finally, our data provide important details for a better understanding of the molecular regulation of an intricate mitochondrial Ca 2+ uptake machinery.