Bidirectional Ca2�-dependent control ofmitochondrial dynamics by the Miro GTPaseMasao Saotomea,1, Dzhamilja Safiulinab,c,1, Gyorgy Szabadkaib,d,1, Sudipto Dasa, Åsa Franssone, Pontus Aspenstrome,Rosario Rizzutob, and Gyorgy Hajnoczkya,2

aDepartment of Pathology, Anatomy and Cell Biology, Thomas Jefferson University, Philadelphia, PA 19107; bDepartment of Experimental and DiagnosticMedicine, Section of General Pathology, Interdisciplinary Center for the Study of Inflammation, University of Ferrara, I-44100 Ferrara, Italy; cDepartment ofPharmacology, Centre of Molecular and Clinical Medicine, University of Tartu, 51014 Tartu, Estonia; dDepartment of Physiology, University College London,London WC1E 6BT, United Kingdom; and eLudwig Institute for Cancer Research, Biomedical Center, Uppsala University, SE-751 24 Uppsala, Sweden

Edited by Clara Franzini-Armstrong, University of Pennsylvania School of Medicine, Philadelphia, PA, and approved October 22, 2008 (received for reviewSeptember 11, 2008)

Calcium oscillations suppress mitochondrial movements along themicrotubules to support on-demand distribution of mitochondria. Toactivate this mechanism, Ca2� targets a yet unidentified cytoplasmicfactor that does not seem to be a microtubular motor or a kinase/phosphatase. Here, we have studied the dependence of mitochon-drial dynamics on the Miro GTPases that reside in the mitochondriaand contain two EF-hand Ca2�-binding domains, in H9c2 cells andprimary neurons. At resting cytoplasmic [Ca2�] ([Ca2�]c), movementsof the mitochondria were enhanced by Miro overexpression irrespec-tive of the presence of the EF-hands. The Ca2�-induced arrest ofmitochondrial motility was also promoted by Miro overexpressionand was suppressed when either the Miro were depleted or theirEF-hand was mutated. Miro also enhanced the fusion state of themitochondria at resting [Ca2�]c but promoted mitochondrial fragmen-tation at high [Ca2�]c. These effects of Miro on mitochondrial mor-phology seem to involve Drp1 suppression and activation, respec-tively. In primary neurons, Miro also caused an increase in dendriticmitochondrial mass and enhanced mitochondrial calcium signaling.Thus, Miro proteins serve as a [Ca2�]c-sensitive switch and bifunc-tional regulator for both the motility and fusion-fission dynamics ofthe mitochondria.

M itochondria are dynamically distributed in the cell to optimizethe utilization of a limited amount of discrete organelles (1,

2). Cytoskeletal tracks and motor proteins have been identified formitochondrial transport (3–6) but the signaling mechanisms thatcontrol motility and positioning remain to be solved. TNF�-andNGF receptor-activated pathways and plasma membrane phospho-lipids have been implicated in the control of mitochondrial move-ments (7–10). Recently, it has been shown in several cell types thatphysiological rises of [Ca2�]c arrest mitochondrial motility (11–14),effectively creating a homeostatic feedback circuit that positionsthese organelles near Ca2� sources (12) enhancing Ca2� bufferingand ATP production where demand is high (15–18). Notably, in ratcortical neurons both the presence (11) and absence (19) ofsensitivity to Ca2� was described for mitochondrial motility. Al-though the calcium signal can stimulate the formation of multiplefactors that affect mitochondrial motility [e.g., adenine nucleotides(13, 20)], Ca2� by itself can control movement activity (12). Ca2�

does not seem to activate Ca2�/calmodulin-dependent kinases orthe Ca2�-dependent protein phosphatase and does not seem totarget directly the microtubular motors, dynein and kinesin toestablish control over mitochondrial motility (12). Thus, we haveproposed that a distinct Ca2� sensor molecule is required totranslate the Ca2� signal for the microtubular motor proteins (12).

A subfamily of the Ras GTPases (Miro 1 and 2 proteins) islocalized at the outer mitochondrial membrane (OMM) and hastwo potential Ca2� binding domains, so called EF-hands (21). Bothproteins consist of 618 amino acid residues and were found to be60% identical (21). Miro is present in yeast (Gem1p) (22), Dro-sophila (dMiro) (23), and mammalian cells as well (21). Mirointeracts with the kinesin-binding proteins, GRIF-1/Milton 2 andOIP106/Milton 1, suggesting that Miro forms a link between the

mitochondria and the trafficking apparatus of the microtubules (24,25). Both the GTPase domains and EF-hand motifs of Miro areexposed to the cytoplasm and are required for yeast Miro functionin mitochondrial morphology (22). Here, we have tested thehypothesis that Miro serves as a Ca2�-sensitive regulator of mito-chondrial motility and fusion-fission dynamics.

ResultsMiro Proteins Support Mitochondrial Motility Along the Microtubules.First we evaluated the effect of Miro1&2 (Miro) and Miro1&2-EF-hand mutants (24)(MiroEF) on basal mitochondrial movementin resting H9c2 cells. Cells were transiently transfected with Miroand MiroEF cDNAs and (i) the spatial relationship betweenmitochondria and microtubules as well as (ii) basal mitochondrialmotility was measured (12). Mitochondria were aligned with andmoved along the microtubules in mock-transfected cells as de-scribed before (12) and similar relation between mitochondria andmicrotubules was observed in Miro- and MiroEF-transfected cells(Fig. 1A). However, when [Ca2�]c was kept at resting level (�100nM), cells overexpressing Miro or MiroEF or a constitutively activeform of Miro1&2 (MiroV13) (24) displayed an increase in mito-chondrial movements (Fig. 1 B and C). Furthermore, Miro-depleted cells (Miro1 or Miro1&2 siRNA) showed a decrease inmitochondrial motility (Fig. 1D). Because Miro 1 and 2 showstructural homology including their GTPase and EF-hand domains(21), exerted similar effects on mitochondrial motility (Fig. 1B) anddisplay similar mitochondrial distribution in H9c2 cells (Fig. S1),indicating that these proteins may substitute each other, in manysubsequent experiments the expression of both Miro1 and 2 wasaltered. The overexpression and silencing of Miro1 and Miro2 wasconfirmed by both anti-Miro1 and anti-Miro2-specific antibodies inwestern blotting (Fig. S2) and in immunocytochemistry (data notshown). For the Myc tagged Miro constructs anti-Myc antibodieswere also used (SI 1 and 2). These results indicate that the Miroproteins facilitate the mitochondrial movements along the micro-tubules in low [Ca2�]c environment and this effect does not requirethe EF-hand Ca2�-binding site.

Author contributions: M.S., D.S., G.S., R.R., and G.H. designed research; M.S., D.S., G.S., S.D.,and A.F. performed research; A.F. and P.A. contributed new reagents/analytic tools; M.S.,D.S., G.S., S.D., and G.H. analyzed data; and M.S., D.S., G.S., P.A., R.R., and G.H. wrote thepaper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

1M.S., D.S., and G.S. contributed equally to this paper.

2To whom correspondence should be addressed at: Department of Pathology, Anatomyand Cell Biology, Suite 253 JAH, Thomas Jefferson University, Philadelphia, PA 19107.E-mail: gyorgy.hajnoczky@jefferson.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/0808953105/DCSupplemental.

© 2008 by The National Academy of Sciences of the USA

20728–20733 � PNAS � December 30, 2008 � vol. 105 � no. 52 www.pnas.org�cgi�doi�10.1073�pnas.0808953105

Dow

nloa

ded

by g

uest

on

July

10,

202

0

Miro Dependence of the [Ca2�]c-Induced Mitochondrial Motility Inhi-bition. Because we have previously shown that an increase in[Ca2�]c leads to inhibition of basal mitochondrial motility (12), nextwe tested the effect of stimulation with vasopressin (VP), a Ca2�

mobilizing hormone on mitochondrial movements in control, Miro-and MiroEF-overexpressing cells. Strikingly, whereas in controlscells the VP (100 nM)-induced Ca2� signal led to substantialinhibition of mitochondrial movements (by 68 � 4%), in MiroEFexpressing cells a significant part of this inhibition was lost (inhi-bition by 40 � 4% and 51 � 4% in Miro1EF and Miro2EFexpressing cells, respectively, Fig. 2A). Miro1EF-and Miro2EF-expression did not affect significantly either the resting [Ca2�]c orthe VP-induced [Ca2�]c spike, although it showed a tendency tosuppress the latter one (P � 0.06 and 0.05, respectively, Fig. 2A).The reversal of the inhibition of mitochondrial movements wasmore pronounced at lower agonist concentrations (e.g., 0.25 nM VPevoked a [Ca2�]c rise but failed to inhibit mitochondrial movementsin MiroEF expressing cells, Fig. 2B). Thus, next mitochondrialmotility inhibition was plotted as a function of [Ca2�]c measuredafter the application of a range of VP doses. The extent of inhibitionof mitochondrial movements showed a sigmoid function of [Ca2�]creached during VP stimulation (Fig. 2C), and 50% inhibition wasobserved at 380 nM [Ca2�]c in control cells. MiroEF overexpressioncaused a right shift of Ca2� sensitivity curve, indicating decreasedsensitivity (Fig. 2C), whereas in Miro-V13 (Fig. 2C) and in Miro-overexpressing cells (data not shown) showed increased sensitivity.Indeed, in the 300–400 nM [Ca2�]c range, the decrease in mito-chondrial motility was 45.6 � 3.6% in the control, 70.0 � 0.3% (P �0.01) in MiroV13-expressing, 54.9 � 3.1% (P � 0.05) in Miro-overexpressing and 34.5 � 3.6% (P � 0.05) in MiroEF-expressingcells (n � 10–13) (Fig. 2D). Moreover, similarly to the expressionof the MiroEF mutant, depletion of Miro by siRNA also attenuatedthe Ca2�-dependent mitochondrial motility inhibition in VP-stimulated cells (Fig. 3 A and B). The [Ca2�]c versus the motilityinhibition relationship in the different Miro expression conditionsindicated no change in cooperativity (Hill slopes were between3.4–4.6). These data provide the first evidence that the Miro maybe involved in the motility inhibition during the Ca2� signal.

To further clarify whether Miro played a role downstream to the

[Ca2�]c elevation, [Ca2�]c and motility were measured in cells thatwere incubated in a Ca2�-free buffer supplemented with EGTA,thapsigargin, an inhibitor of the sarco-endoplasmic reticulum Ca2�

pump and ionomycin, a Ca2� ionophore to ensure rapid equilibra-tion of the cytosol with the extracellular [Ca2�], and then varyingamounts of CaCl2 were added to set [Ca2�]c at different levels (Fig.3 C and D). The data confirmed that the [Ca2�]c vs. motilityrelationships were left shifted in MiroV13-expressing, while show-ing a right shift in MiroEF-expressing cells (Fig. 3C). Again, in the300–400 nM [Ca2�]c range, the decrease in mitochondrial motilitywas 56.7 � 3.2% in the control, 72.2 � 3.1% (P � 0.01) inMiroV13-expressing, and 42.8 � 2.5% (P � 0.05) in MiroEF-expressing cells (n � 29–54). Importantly, using this experimentalmodel we have also showed that both Miro1&2 N18 (a dominantnegative lack of function GTPase mutant of Miro) and �TM (amutant that lacks its transmembrane and mitochondrial targetingdomain) attenuated [Ca2�]c-induced motility inhibition [the inhi-bition was 49.4 � 2.7% in the control, whereas 33.9 � 3.9% (P �0.01) in N18-expressing and 33.0 � 3.5% (P � 0.01) in �TM-expressing cells (n � 43–51cells)]. Thus, according to the resultsobtained in cells that were either stimulated with a Ca2�-mobilizinghormone (Figs. 2 and 3 A and B) or were directly perfused withCa2� (Fig. 3 C and D) we concluded that (i) the Miro proteins areimportant for Ca2�-induced movement inhibition of mitochondriaand (ii) this effect requires both intact EF-hand and GTPasedomains. Our results do not exclude the possibility that other(possibly lower affinity) Ca2� sensors are involved in the process,because Miro knockdown caused only a shift in the Ca2� dose-response with only a small suppression of the maximal motilityinhibition. Alternatively, the residual Miro activity (e.g., the 25–30% remaining Miro expression in the silencing studies, or endog-enous Miro in MiroEF expressing cells) was sufficient to mediatethe motility inhibition when saturated by Ca2�.

Miro-Dependent Changes in Mitochondrial Morphology in H9c2 Cells.Previous studies have described Miro expression-dependentchanges in mitochondrial morphology (22, 24, 25), which may belinked to the altered motility. In addition, the present results on theCa2� sensitivity of Miro activity necessitated further studies of the

Fig. 1. Miro promotes mitochondrialmovements along microtubules at basal[Ca2�]c in H9c2 cells. (A) Confocal images ofTubulinGFP and mtDsRed taken in control(Left), Miro1&2 (Middle), and Miro1&2EF-expressing cells (Right). (B) Actual basal mi-tochondrial motility values (Left) in cellstransfected with mitoYFP alone ormitoYFP�Miro1EF or mitoYFP�Miro2EF. *,P � 0.01. (C) Summarized data of baselinemitochondrial motility (Mito-motility) incells transfected with mitoYFP (Control; n �20 cells), mitoYFP�Miro (n � 13),mitoYFP�MiroV13 (n � 16), andmitoYFP�MiroEF (n � 14). Before imaging,cells were pretreated with thapsigargin (2�M) in a Ca2�-free ECM for 7 min to elimi-nate both intracellular Ca2� mobilizationand Ca2� entry and in turn to stabilize[Ca2�]c under the basal level (�40 nM).Data are shown as % of control. (D) Sum-marized data of baseline Mito-motilities inMitoYFP expressing cells transfected witheither scrambled control (Scr; n � 10),Miro1-siRNA (Miro1; n � 11), or Miro1&2-siRNA (Miro1&2; n � 9). The mitochondrialmovements were quantitated in a CCDtime-series recorded at the resting [Ca2�]c

(�40 nM). Data present % of Scr.

Saotome et al. PNAS � December 30, 2008 � vol. 105 � no. 52 � 20729

CELL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

July

10,

202

0

Miro dependence of mitochondrial morphology determined by thebalance between fusion and fission. In H9c2 cells, Miro1-GFPcolocalized with the mitochondria and induced mitochondrialthread formation and condensation (Fig. S1), as previously de-scribed in other cell types (24). Myc-tagged wild type Miro1&2,Miro1&2-V13, and Miro1&2EF also showed similar mitochondrialdistribution (Fig. S1) and evoked mitochondrial thread formation

and condensation (Fig. 4A). Notably, immunostaining did not showa difference in Miro overexpression levels between cells showingthread formation and condensation, and the ratio of the cellsshowing thread formation and condensation did not change from24 h to 48 h overexpression (data not shown). On the other hand,dominant-negative Miro constructs (N18, �TM) and Miro knock-down (data not shown, n � 3) caused mitochondrial fragmentationand condensation (Fig. 4A). Silencing of Miro also evoked mito-chondrial fragmentation (data not shown, n � 3). Thus, we con-firmed that the connectivity state and thus the length of themitochondria is directly proportional to the availability of Miro atresting [Ca2�]c, depending on its intact GTPase domain but notrequiring the EF hands.

Disruption of the dynein motor complex in HeLa cells has beenshown to cause mitochondrial condensation around the nucleus andmitochondrial elongation that was attributed to Drp1 inhibition(26), prompting us to determine whether the Miro effects onmitochondrial structure were due to increased fusion or decreasedfission activity. First, we showed that overexpression of a dominantnegative form of Drp1 (Drp1K38A) prevented the mitochondrialfragmentation induced by Miro1&2-N18 or �TM in H9c2 cells

Fig. 2. MiroEF suppresses and MiroV13 promotes the VP-induced Mito-motilityinhibition. (A) Actual mitochondrial motility values (Left) and [Ca2�]c (Right)during stimulation with VP (100 nM) in cells transfected with mitoYFP alone ormitoYFP�Miro1EF or mitoYFP�Miro2EF. For VP the lowest motility and highest[Ca2�]c were calculated (n � 27–33). *, P � 0.01. For motility the baseline valueswere shown in Fig. 1B. (B) Simultaneous measurements of Mito-motility and[Ca2�]c in H9c2 cells expressing mitoYFP (Upper) and coexpressing mitoYFP withMiroEF (Lower). Left images show mitoYFP fluorescence (grayscale); Middle andRight images show the sites of mitochondrial movement calculated by subtrac-tion of sequential images (�F: change in fluorescence between the two timepoints; red for positive changes, green for negative changes) before and afterapplication of 0.25 nM VP in each condition. Graphs show both [Ca2�]c (Upper)andmotility inhibitionincellsexpressingmitoYFP(Control;black)orcoexpressingmitoYFP and MiroEF (red). (C) Dose-response relationships between Mito-motility and [Ca2�]c. Vector (black, n � 58), MiroEF (pink, n � 24) and MiroV13(cyan, n � 27) expressing cells. Mito-motility decrease during VP stimulation wasnormalized to baseline motility in each cell (% inhibition) and is plotted againstthe corresponding [Ca2�]c elevations. The IC50 and Hillslope values were in eachcondition: Control; 380 � 10 nM and �4.6, V13; 300 � 20 nM and �4.4, EF; 450 �20 nM and �4.5. (D) Mito-motility inhibition at the range of [Ca2�]c � 300–400nM were calculated in Control (n � 13), V13 (n � 10), and EF (n � 11).

Fig. 3. Effect of Miro-silencing on the VP-induced and V13, EF overexpressionon Ca2�-induced mitochondrial motility inhibition (A) Dose-response relation-ship between [Ca2�]c and motility inhibition in Miro1&2 siRNA (blue, n � 51) andscrambled control (Scr; black, n � 41) transfected cells. The IC50 and Hillslopevalues were 360 � 10 nM and �4.2 in Scr and 430 � 20 nM and �3.6 in Miro1&2siRNA expressing cells. (B) Summarized data of mitochondrial motility inhibitionin scrambled control (n � 17), Miro1siRNA (n � 7), and Miro1&2siRNA transfectedcells (n � 19) in the range of [Ca2�]c � 200–300 nM. *, P � 0.01 vs. Scr. (C)Dose-response relationship between [Ca2�]c and motility in MiroEF (pink, n � 26)and MiroV13 (cyan, n � 31) expressing cells. For the quantitative estimation of[Ca2�]c-dependent mitochondrial motility inhibition, [Ca2�]c and motility weremeasured in cells that were incubated in a Ca2�-free buffer supplemented withEGTA (2 mM), thapsigargin (2 �M), an inhibitor of the sarco-endoplasmic retic-ulum Ca2� pump, and ionomycin (10 �M) to ensure rapid equilibration of thecytosol with the extracellular [Ca2�], and then varying amounts of CaCl2 wereadded. (D) Summarized data of mitochondrial motility inhibitions in Control (n �54), V13 (n � 36), and EF expressing cells (n � 29) at the range of [Ca2�]c �300–400 nM.

20730 � www.pnas.org�cgi�doi�10.1073�pnas.0808953105 Saotome et al.

Dow

nloa

ded

by g

uest

on

July

10,

202

0

(Fig. S3), demonstrating that an intact fission machinery is requiredfor the effect. Then we directly assessed the fusion/fission activityin Miro and MiroEF overexpressing cells, by cotransfecting themwith a matrix-targeted photoactivated GFP (mtPA-GFP) construct(Fig. 4B) that allowed quantification of the number and duration offusion events in a subset of mitochondria following its photoacti-vation. Whereas in control cells PA-GFP fluorescence remainedconfined to the area of the photoactivation, it rapidly filled largecomplex networks in Miro- and MiroEF-overexpressing cells (Fig.4B), confirming the increased continuity of the matrix space in theelongated mitochondria. However, the increased connectivity of

mitochondria in those cells was not due to the increased number offusion events, but rather to an increased duration of the transientfusion events [time elapsed between PA-GFP transfer and apparentseparation of the organelles for control, Miro and MiroEF were:49 � 5, 75 � 8, and 96 � 13 s, respectively; n � 64 � 120]. Becausethe duration of fusion events is determined by the occurrence ofsubsequent fission, our results indicated a decrease in the incidenceof mitochondrial fission. Thus, Miro appears to exert its effect onmitochondrial morphology at least partially by suppressing Drp1-mediated mitochondrial fission, although recruitment of fusionpromoting cytoplasmic proteins may also be affected. Finally,our results also showed that long mitochondria show less move-ment activity than the short ones in H9c2 cells (X. Liu and G.H.,unpublished work), indicating that the Miro effect on mitochon-drial fusion-fission cannot account for the Miro effect onmotility.

Miro Modulates Mitochondrial Morphology in Resting Neurons. Ourstudies provided evidence that Miro proteins are implicated inincreasing mitochondrial connectivity and in Ca2�-induced mito-chondrial arrest in cultured cell lines, but these models did not allowus to study whether the protein and its EF-hand domain is alsoinvolved in distribution of mitochondria in distinct cellular domains.Thus, in further studies we used primary cortical neurons, wheredistinct signaling domains exist in the cell body, axon and dendritesand the mitochondrial distribution among these compartments isprecisely controlled. First, primary cortical neurons were trans-fected with different Miro constructs together with mitochondriallytargeted DsRed2 (mtDsRed) and GFP, and neuronal processes andcell bodies were analyzed for mitochondrial length and occupancy(Fig. 5 A and B). Mitochondria in neuronal processes were mostlytubular in shape and highly variable in size consistent with previous

Fig. 4. Mitochondrial morphology and fusion-fission activity in H9c2 cellsoverexpressing Miro constructs. (A) Mitochondrial patterns were classified (nor-mal, thread, fragmented, and condensed) (24). H9c2 cells were transfected mi-toYFP (Control), or cotransfected mitoYFP with Miro DNAs (WT; Miro1&2, V13;Miro1&2-V13, EF; Miro1&2-EF, �TM; Miro1&2-�TM, N18; Miro1&2-N18). TheMyc-tagged Miro proteins were visualized by using a Myc-specific monoclonalantibody. For each experiment at least 300 cells were scored. (B) Visualization ofthe mitochondrial connectivity and fusion-fission activity in real time. Labeling ofthree subsets of mitochondria by photoactivation of PA-GFP in an H9c2 cellexpressingmtPA-GFP,mtDsRedandMiro(lowerrowof images),MiroEF(datanotshown) or pcDNA (Control, upper row of images). Irreversible photoactivationwas achieved at 0 s as described in Materials and Methods. The rapid spreading(40-s images) of the mtPA-GFP fluorescence (green) indicates the matrix connec-tivity in the Miro and Control mitochondria. Images obtained at 100 s and 300 sillustrate several fusion events in Miro expressing cells and several fission eventsin the control cells.

Fig. 5. Mitochondrial distribution and length in Miro overexpressing neurons.(A–E) mtDsRed-labeled mitochondria (grayscale) in neuronal processes of control(A) and Miro1-over-expressing (B) neurons. Wide-field fluorescence micrographshows the best focus slice of z-stack. Panel C shows the different frequencydistribution of mitochondrial lengths in the Miro1 group, compared with thecontrol. *, P � 0.001 vs. control, **, P � 0.001 vs. control and P � 0.05 vs. Miro1(2) wild-type group (Dunn’s multiple comparison test).

Saotome et al. PNAS � December 30, 2008 � vol. 105 � no. 52 � 20731

CELL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

July

10,

202

0

reports (11, 27). To analyze mitochondrial distribution in processespyramidal neurons were chosen and their dendrite regions �10–100 �m away from the soma where mitochondria are mostlyseparated from each other. Compared with controls, Miro1-overexpressing neurons showed a right-shifted pattern in frequencydistribution of mitochondrial length (Fig. 5C). Indeed, the meanlength of mitochondria in both Miro1 and Miro2-overexpressingneurons was �20% higher than in the corresponding controlgroups. Importantly, Miro1EF and Miro2EF overexpressioncaused even larger increase in mitochondrial length in the processes(Fig. 5 D and E), indicating that dendritic Ca2� levels under basalneuronal activity exert an inhibitory effect on Miro-dependentmitochondrial fusion. Moreover, the mitochondrial index [ratio oftotal mitochondrial length in the dendrite to dendritic length in agiven dendritic segment (27)] was also significantly increased inMiro1 (2)-overexpressing group (Fig. S4 A and E), and showed atendency to increase in MiroEF cells (Fig. S4 B and F). Because thenumber of mitochondria per 10 �m of dendrite was unchanged(data not shown), the increase in mitochondrial index reflected thepresence of elongated mitochondria in the dendrites. The length ofindividual mitochondria in neuronal cell bodies could not bequantified because of the complexity of mitochondrial network inthis region, but the total area occupied by mitochondria in the cellbody was unchanged in Miro-overexpressing neurons (data notshown).

Repetitive Depolarization-Induced Changes in Mitochondrial LengthAre Dependent on Miro’s Ca2�-Binding Activity in Neurons. Ourfindings indicated that Miro increases mitochondrial mass in thedendrites of resting neurons, which process is inhibited by thepresence of EF-hand domains. To further evaluate the role of Ca2�

occupancy of Miro EF-hands in regulation of mitochondrial dy-namics we used a model of repeated depolarizations, which isassociated with transient [Ca2�]c increase, previously shown toevoke changes in neuronal plasticity (28) and mitochondrial redis-tribution (27). Mitochondria in neuronal processes were examined1 h after four repetitive depolarizations with 90 mM KCl andcompared with a parallel set of nonstimulated neurons. In controlcultures no changes in mitochondrial morphology, index, or numberwere observed (Fig. S4A). In contrast, in Miro1-overexpressingneurons, mitochondria became shorter, dendritic mitochondrialindex slightly decreased (Fig. S4 B and C) and the number ofmitochondria was decreased (1.33 � 0.36 mitochondria per 10 �mof dendrite in control, 1.26 � 0.26 in Miro1-overexpressing versus1.06 � 0.29 in Miro1 after stimulations, P � 0.05, n � 20), indicatingmitochondrial redistribution after repetitive stimulation. A similartendency was observed in the experiments with Miro1V13 (Fig. S4C and D) and with Miro2 (Fig. S4 E and F). Most importantly,EF-hand mutations repressed these effects, and over-expression ofMiro1EF and Miro2EF even caused a small but significant increasein mitochondrial length and number in the processes after pre-stimulation (Fig. S4). Thus, we concluded that overexpression ofMiro, bearing intact Ca2� binding EF-hand domains, mediatesCa2�-dependent fragmentation of dendritic mitochondria, whichleads to their elimination or repositioning into the cell body.Abolition of Ca2� binding by EF-hand mutations reversed theeffect, probably by unmasking an opposing, independent Ca2�

mediated process. Importantly, endogenous Miro levels underphysiological Ca2� signaling events were not sufficient to supportCa2�-induced fragmentation and redistribution.

Effect of Miro Proteins on Mitochondrial Ca2� Uptake in Neurons.Finally, to test the consequences on physiological Ca2� signals ofMiro-dependent mitochondrial redistribution, evidence was soughtwhether Miro affects mitochondrial Ca2� uptake. We cotransfectedneurons with mitochondrially targeted aequorin (29) and Miro1,and induced plasma membrane depolarization by perfusing neu-rons with 40 mM KCl containing Krebs-Ringer buffer. The peak

[Ca2�]m response was significantly increased in Miro1-overexpress-ing cells (143.6 � 11.2 �M in control versus 245 � 8.9 �M inMiro1-overexpressing cells, P � 0.001, Fig. 6A). The increasein mitochondrial Ca2� uptake was not a consequence of changes in[Ca2�]c signaling because the peak of [Ca2�]c response was un-changed in Miro1 overexpressing neurons as measured in cytosolicaequorin expressing cells (Fig. 6B). Similar results were obtained byoverexpressing either Miro2 or Miro1/V13 (Fig. 6C). WhenMiroEF was overexpressed in neuronal cultures, similarly to wild-type Miro overexpression, MiroEF increased [Ca2�]m upon plasma-membrane depolarization (Fig. 6C). Thus, Miro mediated redistri-bution and fusion of mitochondria increased their ability toaccumulate Ca2� independently of the Ca2� binding activity of theprotein.

DiscussionThis work revealed that Miro proteins mediate multiple effects of[Ca2�]c signals on mitochondrial motility, fusion-fission dynamicsand function (summarized in Fig. 6D). Miro, in resting nonpolarizedcells, facilitates mitochondrial movements presumably by optimiz-ing the anchorage of the kinesin/dynein motor complexes to themitochondrial surface through GRIF-1/Milton 2 and OIP106/Milton 1 (24, 25). In neurons, this mechanism may be involved indirectional transport of mitochondria into dendrites. However,exposure of the Miro’s EF-hand to a [Ca2�]c rise relays a stop signalto the motors, and redirects dendritic transport of mitochondria inneurons. Importantly, this regulation takes place in the physiolog-ical range of global [Ca2�]c. Miro seems to be competent to conferthe Ca2� effect only if its GTPase domain is intact. Down-regulation of Miro resulted in a rightward shift in the Ca2�

dose-response for mitochondrial motility inhibition, whereas up-regulation exerted an opposite effect, indicating that a directrelationship exists between the availability of Miro and Ca2�

sensitivity. Thus, Miro is a Ca2�-sensing element of the molecularcomplex controlling mitochondrial motility. Miro may collaborate

Fig. 6. Effect of Miro on [Ca2�]c and [Ca2�]m signaling in neurons. (A and B)Miro1 over-expression increases 40 mM KCl-induced [Ca2�]m uptake. Corticalcultures were cotransfected with aequorin targeted to mitochondria (A) orcytosol (B) and Miro1WT and Ca2� uptake was measured luminometrically asdescribed in Materials and Methods. The [Ca2�]c peak was 1.97 � 0.16 �M incontrol versus 1.87 � 0.08 �M in Miro1-overexpressing neurons (n � 8, P � 0.579).(C) Changes in mitochondrial Ca2� uptake induced by overexpression of Miro1and Miro2 mutants. Data presented as % of control, from 3–6 independentexperiment (n � 15), **, P � 0.001 vs. control group. (D) Scheme illustrating thebidirectional Ca2�-dependent control of mitochondrial motility and fusion-fission dynamics by Miro proteins.

20732 � www.pnas.org�cgi�doi�10.1073�pnas.0808953105 Saotome et al.

Dow

nloa

ded

by g

uest

on

July

10,

202

0

with another Ca2� sensor that can mediate the high [Ca2�]c-induced maximal motility inhibition in Miro-depleted cells. Immu-nocytochemistry data showed fairly homogeneous distribution ofMiro proteins in the mitochondria, indicating that Miro can medi-ate the arrest of the mitochondrial movements wherever [Ca2�]crises. Because the presence of Miro is required for both mitochon-drial motility at low [Ca2�]c and for mitochondrial arrest at sites ofa [Ca2�]c elevation, it emerges as a key molecule for the homeo-static control of the mitochondrial distribution that allows strategicpositioning of mitochondria where for example, ATP supply orCa2� buffering is needed. This represents a novel mechanism forremodeling of the Ca2� signaling system (30).

At resting low [Ca2�]c levels, when the EF-hands are presumablynot occupied by Ca2� Miro also facilitates the formation ofelongated mitochondria. Real time analysis of mitochondrial fu-sion/fission and the use of a dominant negative Drp1 constructprovided evidence that (i) the Miro-mediated elongation of mito-chondria involved suppression of Drp1-mediated fission and (ii) theshortening after Miro depletion is the result of Drp1-mediatedfission, respectively. Notably, Miro-overexpressing cells lack mito-chondrial necklaces that are present when a dominant negativeDrp1 was expressed, suggesting that the Miro effect is not restrictedto Drp1 inhibition. Collectively, the effects on mitochondrial mo-tility and morphology show a unique capacity of Miro to maintainthe mobility of the organelles while their size is increasing. How-ever, extreme mitochondrial elongation may be accompanied bycondensation because elongated organelles might present oversizedcargo for the transport machinery (31). In contrast to resting conditions,repetitive or prolonged high [Ca2�]c signals trigger Miro-dependentshortening, mediated by the EF-hand domains, as shown in neuronsexpressing supraphysiological levels of the Miro proteins.

Finally, mitochondrial function can be modulated through thefusion state of the mitochondria and the activity of fusion-fissionproteins (32, 33). Recently, docking of mitochondria by syntaphilinin the axons has been shown to affect local calcium signaling (34).

The present study also uncovered that Miro supports distribution ofmitochondria to the dendrites close to the Ca2� entry sites andpromotes the calcium signal propagation to the mitochondria. Thelatter effect does not seem to result from an increase in Ca2� entryor mobilization but may be due to the mitochondrial distributionclose to the Ca2� source or to facilitation of the mitochondrial Ca2�

uniport. Increased Ca2� transfer by the mitochondria provides ameans to enhance the mitochondrial Ca2� buffering and to stim-ulate the Ca2�-dependent reactions in mitochondrial energy pro-duction. Thus, Miro proteins both mediate the effects and regulatecalcium signaling to coordinate the mitochondrial dynamics andcontribution to cell function.

Materials and MethodsDetailed methods are described in SI Methods. Constructs of Miro and antibodieshave been described previously (21, 24, 29). H9c2 cells were cultured and werecotransfected with Miro DNAs and mtYFP employing electroporation or withsiRNAs with GeneSilencer (Genlantis). Primary cultures of cortical neurons wereprepared from the cortices of neonatal Wistar rats, were grown in Neurobasal Amedium, were transfected using Lipofectamine 2000TM with mtDsRed, GFP oraequorin and imaging and aequorin measurements were performed 6–10 dayslater (29). Simultaneous measurements of [Ca2�]c with mitochondrial motility inH9c2 cells were carried out by using a fluorescence or a confocal imaging setupas described before (12). For PA-GFP photoactivation three 25-�m2 areas werechosen per cell and illuminated with maximum power 442-nm excitation. PA-GFPfluorescence was then monitored using 488-nm excitation. The change in mito-chondrial motility was evaluated as described previously (12). The length ofprocesses and mitochondria in the neuronal cultures was measured manuallywith the aid of MetaMorph software.

ACKNOWLEDGMENTS. This work was supported by grants from the EstonianScience Foundation Grant 7175 and European Community Contract MTKD-CT-2004-517176 (to D.S.), Swedish Cancer Society and the Swedish Research Council(to P.A.), the Italian Association for Cancer Research, Telethon, local funds fromthe University of Ferrara, the Italian University Ministry, the EU (fondi strutturaliObiettivo 2), the PRRIITT program of the Emilia Romagna Region, and the ItalianSpace Agency and by National Institutes of Health Grants 1P01AG025532–01A1(to R.R.) and DK51526 and GM59419 (to G.H.).

1. Rice SE, Gelfand VI (2006) Paradigm lost: Milton connects kinesin heavy chain to miroon mitochondria. J Cell Biol 173:459–461.

2. Reynolds IJ, Rintoul GL (2004) Mitochondrial stop and go: Signals that regulateorganelle movement. Sci STKE 2004:PE46.

3. Hollenbeck PJ, Saxton WM (2005) The axonal transport of mitochondria. J Cell Sci118:5411–5419.

4. Vale RD (2003) The molecular motor toolbox for intracellular transport. Cell 112:467–480.5. Frederick RL, Shaw JM (2007) Moving mitochondria: Establishing distribution of an

essential organelle. Traffic 8:1668–1675.6. Anesti V, Scorrano L (2006) The relationship between mitochondrial shape and func-

tion and the cytoskeleton. Biochim Biophys Acta 1757:692–699.7. De Vos K, et al. (2000) Tumor necrosis factor induces hyperphosphorylation of kinesin

light chain and inhibits kinesin-mediated transport of mitochondria. J Cell Biol149:1207–1214.

8. Chada SR, Hollenbeck PJ (2004) Nerve growth factor signaling regulates motility anddocking of axonal mitochondria. Curr Biol 14:1272–1276.

9. De Vos KJ, Sable J, Miller KE, Sheetz MP (2003) Expression of phosphatidylinositol (4,5)bisphosphate-specific pleckstrin homology domains alters direction but not the levelof axonal transport of mitochondria. Mol Biol Cell 14:3636–3649.

10. Minin AA, Kulik AV, Gyoeva FK, Li Y, Goshima G, Gelfand VI (2006) Regulation ofmitochondria distribution by RhoA and formins. J Cell Sci 119:650–670.

11. Rintoul GL, Filiano AJ, Brocard JB, Kress GJ, Reynolds IJ (2003) Glutamate decreases mitochon-drial size and movement in primary forebrain neurons. J Neurosci 23:7881–7888.

12. Yi M, Weaver D, Hajnoczky G (2004) Control of mitochondrial motility and distributionby the calcium signal: A homeostatic circuit. J Cell Biol 167:661–672.

13. Brough D, Schell MJ, Irvine RF (2005) Agonist-induced regulation of mitochondrial andendoplasmic reticulum motility. Biochem J 392:291–297.

14. Quintana A, Schwarz EC, Schwindling C, Lipp P, Kaestner L, Hoth M (2006) Sustainedactivity of CRAC channels requires translocation of mitochondria to the plasma mem-brane. J Biol Chem 281:40302–40309.

15. Jouaville LS, Ichas F, Holmuhamedov EL, Camacho P, Lechleiter JD (1995) Synchroni-zation of calcium waves by mitochondrial substrates in Xenopus laevis oocytes. Nature377:438–441.

16. Hoth M, Button DC, Lewis RS (2000) Mitochondrial control of calcium-channel gating:A mechanism for sustained signaling and transcriptional activation in T lymphocytes.Proc Natl Acad Sci USA 97:10607–10612.

17. Park MK, Ashby MC, Erdemli G, Petersen OH, Tepikin AV (2001) Perinuclear, perigranu-lar and sub-plasmalemmal mitochondria have distinct functions in the regulation ofcellular calcium transport. EMBO J 20:1863–1874.

18. Malli R, et al. (2003) Sustained Ca2� transfer across mitochondria is Essential formitochondrial Ca2� buffering, sore-operated Ca2� entry, and Ca2� store refilling.J Biol Chem 278:44769–44779.

19. Beltran-Parrazal L, Lopez-Valdez H, Brennan KC, Diaz-Munoz M, Vellis JD, Charles AC(2006) Mitochondrial transport in processes of cortical neurons is independent ofintracellular calcium. Am J Physiol Cell Physiol 291:C1193–1197.

20. Mironov SL (2006) Spontaneous and evoked neuronal activities regulate movements ofsingle neuronal mitochondria. Synapse 59:403–411.

21. Fransson A, Ruusala A, Aspenstrom P (2003) Atypical Rho GTPases have roles inmitochondrial homeostasis and apoptosis. J Biol Chem 278:6495–6502.

22. Frederick RL, McCaffery JM, Cunningham KW, Okamoto K, Shaw JM (2004) Yeast MiroGTPase, Gem1p, regulates mitochondrial morphology via a novel pathway. J Cell Biol167:87–98.

23. Guo X, et al. (2005) The GTPase dMiro is required for axonal transport of mitochondriato Drosophila synapses. Neuron 47:379–393.

24. Fransson S, Ruusala A, Aspenstrom P (2006) The atypical Rho GTPases Miro-1 andMiro-2 have essential roles in mitochondrial trafficking. Biochem Biophys Res Commun344:500–510.

25. Glater EE, Megeath LJ, Stowers RS, Schwarz TL (2006) Axonal transport of mitochondriarequires milton to recruit kinesin heavy chain and is light chain independent. J Cell Biol173:545–557.

26. Varadi A, Cirulli V, Rutter GA (2004) Mitochondrial localization as a determinant ofcapacitative Ca2� entry in HeLa cells. Cell Calcium 36:499–508.

27. Li Z, Okamoto K, Hayashi Y, Sheng M (2004) The importance of dendritic mitochondriain the morphogenesis and plasticity of spines and synapses. Cell 119:873–887.

28. Wu GY, Deisseroth K, Tsien RW (2001) Spaced stimuli stabilize MAPK pathway activa-tion and its effects on dendritic morphology. Nat Neurosci 4:151–158.

29. Chiesa A, et al. (2001) Recombinant aequorin and green fluorescent protein as valuabletools in the study of cell signalling. Biochem J 355:1–12.

30. Berridge MJ, Bootman MD, Roderick HL (2003) Calcium signalling: Dynamics, ho-meostasis and remodelling. Nat Rev Mol Cell Biol 4:517–529.

31. Campello S, Lacalle RA, Bettella M, Manes S, Scorrano L, Viola A (2006) Orchestrationof lymphocyte chemotaxis by mitochondrial dynamics. J Exp Med 203:2879–2886.

32. Youle RJ, Karbowski M (2005) Mitochondrial fission in apoptosis. Nat Rev Mol Cell Biol6:657–663.

33. Szabadkai G, et al. (2006) Mitochondrial dynamics and Ca2� signaling. BiochimBiophys Acta 1763:442–449.

34. Kang JS, et al. (2008) Docking of axonal mitochondria by syntaphilin controls theirmobility and affects short-term facilitation. Cell 132:137–148.

Saotome et al. PNAS � December 30, 2008 � vol. 105 � no. 52 � 20733

CELL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

July

10,

202

0