Mitochondrial calcium uniporter Mcu controls excitotoxicity and is transcriptionally repressed by neuroprotective nuclear calcium signals

Abstract

The recent identification of the mitochondrial Ca²⁺ uniporter gene (Mcu/Ccdc109a) has enabled us to address its role, and that of mitochondrial Ca²⁺ uptake, in neuronal excitotoxicity. Here we show that exogenously expressed Mcu is mitochondrially localized and increases mitochondrial Ca²⁺ levels following NMDA receptor activation, leading to increased mitochondrial membrane depolarization and excitotoxic cell death. Knockdown of endogenous Mcu expression reduces NMDA-induced increases in mitochondrial Ca²⁺ , resulting in lower levels of mitochondrial depolarization and resistance to excitotoxicity. Mcu is subject to dynamic regulation as part of an activity-dependent adaptive mechanism that limits mitochondrial Ca²⁺ overload when cytoplasmic Ca²⁺ levels are high. Specifically, synaptic activity transcriptionally represses Mcu, via a mechanism involving the nuclear Ca²⁺ and CaM kinase-mediated induction of Npas4, resulting in the inhibition of NMDA receptor-induced mitochondrial Ca²⁺ uptake and preventing excitotoxic death. This establishes Mcu and the pathways regulating its expression as important determinants of excitotoxicity, which may represent therapeutic targets for excitotoxic disorders.

ARTICLE

Received 28 Aug 2012 | Accepted 20 May 2013 | Published 18 Jun 2013

Jing Qiu1,*, Yan-Wei Tan2,*, Anna M. Hagenston2, Marc-Andre Martel1, Niclas Kneisel2, Paul A. Skehel1, David J.A. Wyllie1, Hilmar Bading2 & Giles E. Hardingham1

The recent identication of the mitochondrial Ca2 uniporter gene (Mcu/Ccdc109a) has enabled us to address its role, and that of mitochondrial Ca2 uptake, in neuronal excitotoxicity. Here we show that exogenously expressed Mcu is mitochondrially localized and increases mitochondrial Ca2 levels following NMDA receptor activation, leading to increased mitochondrial membrane depolarization and excitotoxic cell death. Knockdown of endogenous Mcu expression reduces NMDA-induced increases in mitochondrial Ca2 ,

resulting in lower levels of mitochondrial depolarization and resistance to excitotoxicity. Mcu is subject to dynamic regulation as part of an activity-dependent adaptive mechanism that limits mitochondrial Ca2 overload when cytoplasmic Ca2 levels are high. Specically, synaptic activity transcriptionally represses Mcu, via a mechanism involving the nuclear Ca2

and CaM kinase-mediated induction of Npas4, resulting in the inhibition of NMDA receptor-induced mitochondrial Ca2 uptake and preventing excitotoxic death. This establishes

Mcu and the pathways regulating its expression as important determinants of excitotoxicity, which may represent therapeutic targets for excitotoxic disorders.

1 Centre for Integrative Physiology, University of Edinburgh, Edinburgh EH8 9XD, UK. 2 Department of Neurobiology, Interdisciplinary Center for Neurosciences (IZN), Im Neuenheimer Feld 364, D-69120 Heidelberg, Germany. *These authors contributed equally to this work. Correspondence and requests for materials should be addressed to H.B. (email: mailto:[email protected]

Web End [email protected] ) or to G.E.H. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 4:2034 | DOI: 10.1038/ncomms3034 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

& 2013 Macmillan Publishers Limited. All rights reserved.

DOI: 10.1038/ncomms3034 OPEN

Mitochondrial calcium uniporter Mcu controls excitotoxicity and is transcriptionally repressed by neuroprotective nuclear calcium signals

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3034

For B50 years, it has been known that mitochondria are able to take up Ca2 , achieved through the action of a membrane potential-driven carrier named the mitochon

drial calcium uniporter (Mcu)1,2. The neurotoxic potential of the excitatory neurotransmitter glutamate has been appreciated for a similarly long time3. Glutamate excitotoxicity was found to be due to excessive Ca2 inux through the NMDA subtype of glutamate receptor, and is implicated in promoting neuronal death and dysfunction in a variety of acute and chronic neurological disorders including stroke, traumatic brain injury and Huntingtons disease36.

Many important studies into the responses of mitochondria to NMDA receptor (NMDAR) activity suggest that mitochondrial Ca2 uptake by the uniporter has a role in excitotoxicity79.

Inappropriate levels of mitochondrial Ca2 uptake, in concert with nitric oxide production and activation of poly(ADP-ribose)

polymerase-1 (PARP-1), lead to loss of mitochondrial membrane potential, which in turn energetically compromises the neuron and may lead to ROS generation711. However, a denitive answer to the question of whether mitochondrial Ca2 uptake mediates excitotoxicity has been lacking because the molecular identity of the Mcu was not known. Early attempts to interfere with mitochondrial Ca2 uptake in neurons indirectly involved the use of protonophores in order to depolarize the mitochondria (the membrane potential is essential for uniporter activity)12. However, this intervention can dramatically impact the cells bioenergetics as well as potentially triggering changes to the plasma membrane potential7. Moreover, the protective effects of prior mitochondrial depolarization are controversial13. The use of a cocktail of mitochondrial toxins to dissipate the mitochondrial membrane potential, while preventing ATP depletion, has also been employed to indirectly prevent mitochondrial Ca2 uptake, with protective consequences14.

Pharmacological agents based on the hexavalent cation ruthenium red have also been utilized. Ruthenium red itself is able to selectively block the uniporter in isolated mitochondria, but has non-selective effects on certain ion channels in intact cells and is unable to cross the plasma membrane of many cell types15,16. The derivative Ru360 has been proposed to be more selective and cell-permeant (although there remain some doubts in these areas1517). Effects of Ru360 on glutamate-induced mitochondrial depolarization have been observed11, although investigations have focussed on early events, as it is unstable in aqueous solutions (it quickly becomes oxidized). Ru360 is of limited use for long-term experiments needed to assess the role of mitochondrial Ca2 uptake in excitotoxic cell death.

In two recent papers, the gene product encoding the uniporter channel (Mcu) was identied as the ubiquitously expressed gene previously known as Ccdc109a18,19, which acts in concert with regulatory proteins such as Micu1 and Mcur1 to mediate potential-driven mitochondrial Ca2 uptake20,21. This nding now allows selective approaches involving exogenous Mcu expression and knockdown to be employed to determine the role of mitochondrial Ca2 uptake in all aspects of cellular physiology and pathology.

Here we have manipulated Mcu expression in order to directly investigate the long-standing issue of a role for mitochondrial Ca2 uptake in excitotoxicity. Overexpression and knockdown of

Mcu reveals that it has an important role in mitochondrial Ca2 uptake following NMDAR activation, as well as in subsequent cell death. Furthermore, we nd that the Mcu gene is subject to dynamic regulation: it is transcriptionally repressed by neuro-protective nuclear Ca2 signals via a mechanism involving induction of the transcriptional regulator Npas4.

ResultsMcu expression promotes neuronal mitochondrial Ca2 uptake. Mcu is a ubiquitously expressed gene19 (although absent in yeast2) and we conrmed expression of Mcu in mouse cortical and hippocampal neurons: western analysis of whole-cell lysates using a previously validated anti-Mcu antibody18 revealed a band of expected size that was enriched in neurons over-expressing Mcu (Fig. 1a, Supplementary Fig. S1a). We employed immunouorescence and biochemical fractionation approaches to show that Mcu fused to the uorescent proteins eGFP or tDimer localized to neuronal mitochondria, consistent with its known subcellular distribution (Fig. 1b, Supplementary Fig. S1b and data not shown). Our overarching aim was to investigate the effect of manipulating Mcu expression on responses of forebrain neurons to NMDA treatment, focusing on mitochondrial and cytoplasmic Ca2 increases, mitochondrial depolarization, and cell death.

Throughout the study, we present data using two complementary approaches: (a) Liposome-mediated transfection of an Mcu expression vector or Mcu-directed siRNA in mouse cortical neurons, and (b) Recombinant adeno-associated virus (rAAV)-mediated transduction of expression vectors encoding Mcu (tagged with GFP or tDimer) or Mcu-directed shRNA in mouse hippocampal neurons. The reason for the liposome-mediated transfection approach was to achieve low efciency, enabling a direct comparison between transfected and juxtaposed untransfected cells, while the viral transduction technique was chosen for high efciency of infection (8090%), which facilitates ease of data analysis, particularly cell death. The two neuronal types used served to broaden the applicability of our ndings.

We rst studied the effect of Mcu overexpression on NMDAR-dependent increases in mitochondrial Ca2, loss of mitochondrial membrane potential and cell death. In all Ca2 imaging experiments we performed experiments with control and treated

conditions using sister cultures interleaved on the same day to control for any culture-to-culture variation in NMDA-induced Ca2 inux. Thus, the only control data shown in any particular gure are those data collected on the same day on the same culture as when the treated condition data (for example, Mcu overexpression) were collected. As an important additional control, we veried that Mcu overexpression did not have any effect on NMDA-evoked whole-cell currents (Fig. 1c).

Rhodamine-based Ca2 indicators are a useful tool for measuring mitochondrial Ca222,23, as they partition preferentially into polarized mitochondria due to their positive charge, and responses of transfected and juxtaposed untransfected cells can be analyzed. Neurons loaded with Rhod-2 showed a punctate distribution of the indicator that closely resembled that obtained with mitotracker red23 and the pattern was abolished by the mitochondrial uncoupler FCCP (Supplementary Fig. S1c), conrming a predominant mitochondrial localization of the dye. Using Rhod-2 imaging, we found that Mcu over-expressing neurons (co-transfected with an eGFP marker for identication) exhibited a larger NMDA-induced increase in [Ca2 ]mit, compared with surrounding untransfected neurons (Supplementary Fig. S1d,e). In control experiments, neurons transfected with control plasmid (-globin) responded no differently than surrounding untransfected neurons (Supplementary Fig. S1d,e).

Although Rhod-2 imaging is well-suited for comparing the responses of adjacent transfected versus untransfected cells, its localization to mitochondria is not completely clear-cut: some cytoplasmic localization is unavoidable, potentially leading to an underestimation of the effects on mitochondrial Ca2 . We therefore complemented our Rhod-2 imaging studies with those involving a genetically encoded, mitochondrially targeted

2 NATURE COMMUNICATIONS | 4:2034 | DOI: 10.1038/ncomms3034 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2013 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3034 ARTICLE

ns

MitodsRed

GFPMcu

100

Merge

Relative I NMDA

Con

Mcu

51 kDa

39 kDa

-tubulin

Mcu

50

0 Con Mcu

Mito-dsRed

GCaMP2-mt

Merge

Control-transfected neurons

3

Mcu-transfected neurons

NMDA

Con Mcu

*

*

5

NMDA

5

GCaMP2-mt fluorescence

(FF min)/(F maxF)

GCaMP2-mt fluorescence

(FF min)/(F maxF)

GCaMP2-mt fluorescence

(FF min)/(F maxF)

4

4

2

1

3

3

2

1

0 0 50 100 150 200

Control

NMDA

2

1

Individual cells

Mean

Individual cells

Mean

0 0 50 100 150 200 Time (s)

0 Basal

NMDA

Time (s)

rAAVMcutDimer

800

Con Mcu

*

NMDA

1/n (% of Con)

500

Individual cells

Mean

1/n (% of Con)

1/n (% of Con)

700

500

Individual cells

Mean

600

4mtD3cpv FRET ratio

{%R fret/(R max%R fret)}

4mtD3cpv FRET ratio

{%R fret/(R max%R fret)}

4mtD3cpv FRET ratio

{%R fret/(R max%R fret)}

400

400

500

300

300

400

300

200

200

100

100

200

100

0

0

0

0

300Time (s) Time (s)

0 Basal NMDA

100

200

100

200

300

Figure 1 | Overexpression of Mcu promotes uptake of Ca2 into mitochondria following NMDA receptor activation. (a) Western blot of extracts from control neurons or neurons nucleofected with an Mcu-encoding plasmid. (b) Confocal image of a Mito-dsRed and GFP-Mcu co-expressing neuron. Scale bar 15 mm. (c) NMDA (150 mM)-evoked whole-cell currents measured in control- and Mcu-expressing neurons (co-expressing GFP for identication)

(means.e.m., n 7). (d) Conrmation that GCaMP2-mt is localized to mitochondria. Scale bar 15 mm. (eg) Mcu overexpression boosts mitochondrial

Ca2 uptake assayed using GCaMP2-mt. Neurons were transfected with vectors encoding GCaMP2-mt plus either Mcu or control (-globin). GCaMP2-mt uorescence was measured before and during exposure to NMDA (20 mM). (e,f) show example traces from single experiments involving control(e) or Mcu (f) over-expressing neurons. For each experiment, the mean value of all cells within the eld was calculated (line in bold shown) and the average pre- and post- stimulation level calculated (normalized to the maximal ionomycin-induced signal). 1G shows quantitation of experiments (means.e.m.). *Po0.05 (unpaired two-tailed t-test, Con: 45 cells, n 6; Mcu: 48 cells, n 6). (hj) rAAV-mediated expression of McutDimer boosts

mitochondrial Ca2 uptake assayed using 4mtD3cpv. Hippocampal neurons were infected with rAAVs containing vectors encoding 4mtD3cpvrAAV McutDimer. 4mtD3cpv FRET ratios were measured before and during NMDA exposure (10 mM) 200250 s following NMDA application. 1H and 1I show example traces from single experiments involving control (h) or rAAVMcutDimer (i) infected neurons. (j) shows quantitation (means.e.m.).

In all cases, [Ca2 ] levels obtained using 4mtD3cpv were normalized to the mean NMDA-induced [Ca2 ] in control-transfected neurons measured on that precise day of imaging. *Po0.05 (unpaired two-tailed t-test. Con: 204 cells from n 15 experiments; McutDimer: 100 cells from n 12 experiments).

NATURE COMMUNICATIONS | 4:2034 | DOI: 10.1038/ncomms3034 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

& 2013 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3034

Ca2 indicator, GCaMP2-mt24 (Fig. 1d). The GCaMP2-mt signal obtained in resting control neurons was B30% of the maximum signal obtained by ionomycin treatment and the indicator was found to have a Fmin (obtained with ionomycin in zero Ca2 medium) that was barely detectable over background (data not shown). Using this technique, we also found that Mcu overexpression strongly increased both basal [Ca2 ]mit and

NMDA-induced increases in [Ca2 ]mit, compared with control-transfected neurons treated in parallel experiments (Fig. 1eg).

We next studied the effect of Mcu overexpression on NMDA-induced mitochondrial Ca2 uptake using an rAAV-based approach. We co-infected neurons with an rAAV containing an expression cassette for Mcu (tagged with tDimer for identication) and an rAAV containing an expression cassette for the FRET-based mitochondrial Ca2 indicator 4mtD3cpv25. The effect of Mcu overexpression was compared with parallel experiments in which the neurons were infected with only rAAV4mtD3cpv. The rAAV-mediated infection resulted in high efciency (8090%) expression of McutDimer in mouse hippocampal neurons (data not shown). Compared with parallel experiments performed on control neurons Mcu tDimer overexpression led to enhanced elevation of [Ca2 ]mit, following NMDAR activation (Fig. 1hj). We have thus shown using independent methods of gene transfer and imaging, that Mcu overexpression in neurons leads to enhanced mitochondrial Ca2 uptake following activation of NMDARs.

Mcu promotes NMDAR-dependent mitochondrial depolarization. A reduction of mitochondrial membrane potential (Cm) is one of the earliest observable events in response to excitotoxic insults, and mitochondrial Ca2 uptake is thought to be a key factor in this process11,26,27. We therefore studied the effect of manipulating Mcu expression on NMDA-induced Cm loss, assayed using the indicator rhodamine-12310,28. We found that Mcu overexpressing neurons (co-transfected with an expression vector for mCherry for identication of the manipulated cells) exhibited a greater percentage change in rhodamine-123 signal following NMDA treatment, compared with surrounding untransfected neurons, indicative of a greater loss of Cm (Fig. 2b, see 2a for an example experiment). In control experiments, neurons transfected with control plasmid (expression vector for -globin) responded no differently than surrounding untransfected neurons (Fig. 2b). Thus, overexpression of Mcu exacerbates NMDA-induced loss of Cm, consistent with its effect on mitochondrial Ca2 uptake.

Mcu enhances vulnerability to NMDAR-dependent excitotoxicity. We next investigated the effect of manipulating Mcu expression on the vulnerability of neurons to excitotoxic insults. Pictures of cortical neurons transfected (Lipofectamine) with control or Mcu-encoding vectors plus an eGFP marker were taken before and 24 h after a 1-h exposure to an excitotoxic dose of NMDA, and their viability scored. We found that Mcu over-expression increased the level of basal toxicity in neurons (Fig. 3a). In addition, levels of neuronal death were higher in NMDA-treated neurons over-expressing Mcu compared with control-transfected neurons (Fig. 3a,b). Similar results were obtained using hippocampal neurons infected with rAAVMcu GFP: the expression of McuGFP led to increased basal cell death rates in neurons and elevated levels of NMDA-induced cell death (Fig. 3c,d). For both cortical and hippocampal neurons, we additionally analyzed the above data to calculate whether Mcu overexpression made neurons more vulnerable to NMDA treatment, taking into account the increased basal death (see Methods). This revealed that Mcu makes neurons signicantly

more vulnerable to 20 and 30 mM NMDA in cortical neurons and more vulnerable to all doses of NMDA in hippocampal neurons (Supplementary Fig. S2a,b).

Mcu knockdown impairs Ca2 mit and is neuroprotective. We then performed loss-of-function experiments to investigate the effect of interfering with endogenous Mcu expression on NMDAR-dependent increases in mitochondrial Ca2, loss of mitochondrial membrane potential and cell death. Two independent approaches were taken: liposome-mediated transfection of Mcu-targeted pre-made siRNA, and rAAV-mediated expression of Mcu-directed small hairpin RNA (shRNA). The efcacy of the pre-made siRNA was investigated using a high-efciency transfection approach (nucleofection), which revealed a partial (50%) knockdown at protein and mRNA levels (Fig. 4a,b, Supplementary Fig. S1a) and no effect on NMDA-evoked whole-cell currents (Fig. 4c). Given the nucleofection efciency is never greater than 7080%, the 50% knockdown reects a 450%

knockdown in transfected cells. We found that neurons transfected with Mcu-directed siRNA exhibited a smaller NMDA-induced increase in [Ca2]mit, assayed either using the Rhod-2 indicator and comparing the results to surrounding untransfected neurons (Supplementary Fig. S3a,b), or using GCaMP2-mt and comparing the results to control-transfected neurons (Fig. 4d-f). Neurons transfected with Mcu-directed siRNA also exhibited a smaller loss of Cm than surrounding cells, while control-transfected neurons did not show differences compared with surrounding cells, (Fig. 4g,h). We then investigated whether Mcu knockdown inuenced NMDA-induced excitotoxicity, and found this indeed to be the case: levels of NMDA-induced death were lower than in control-transfected cells (Fig. 4i,j). Moreover, we conrmed that the mitochondrial membrane potential of surviving cells (24 h after the NMDA insult) was intact and no different from control cells (Supplementary Fig. S3c). We also

Mcu-transfected cell

Mean of surrounding cells

Individual surrounding cells

10 [afii9839]M 20 [afii9839]M NMDA

70

*

#

% Difference in NMDA-induced

rh-123 signal change

40

60

m loss: change in rh-123

fluorescence (% of FCCP value)

0 0

0 50 100 150 200

Time (s)

50

10 m

8

6

4

2

30

x

100 (xy)

y

20

y

10

Mcu

n=23

Con

n=16

Plasmid

Figure 2 | Overexpression of Mcu exacerbates NMDA-induced loss of mitochondrial membrane potential. (a,b) Example trace from a single experiment where Rhodamine-123 uorescence changes following NMDA application were studied in a cell overexpressing Mcu (plus an mCherry marker) and surrounding untransfected cells. Rhodamine-123 uorescence was normalized in all cases to the maximal signal obtained upon complete mitochondrial depolarization, achieved at the end of the experiment by exposure of neurons to the uncoupler FCCP. For each experiment, the difference between transfected and untransfected cells was calculated as shown in the gure, and combined with other repeats to give the quantitation shown in (b) (means.e.m.). # indicates a signicant difference between Mcu-expressing cells and surrounding untransfected cells (Po0.05, paired two-tailed t-test). * indicates a signicant difference between data obtained with Mcu-expressing cells and that with control-transfected cells (Po0.05, unpaired two-tailed t-test, n 23 Mcu,

n 16 Con).

4 NATURE COMMUNICATIONS | 4:2034 | DOI: 10.1038/ncomms3034 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2013 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3034 ARTICLE

Control

NMDA

NMDA

Mcu

*

Pre- Post-

% Cell death

% Cell death

80

60

40

Control ( globin)

*

*

20

Mcu

0

10NMDA ([afii9839]M)

NMDA ([afii9839]M)

30

0

20

No virus

rAAVGFP

*

rAAVMCUeGFP

*

80

60

40

20

0

rAAV MCUeGFP

*

No virus

rAAVGFP

*

*

0

5 30

GFP

Hoechst

10

20

Figure 3 | Overexpression of Mcu exacerbates excitotoxic cell death. (a,b) Cortical neurons were transiently transfected with an eGFP marker plus either control or Mcu-encoding plasmids. Twenty-four hours post-transfection, pictures of transfected cells were taken prior to a 1-h exposure to NMDA (where indicated). After a further 24 h, pictures were taken and cell death scored (means.e.m. shown). *Po0.05 (ANOVA, Bonferroni post hoc, 300500 cells analyzed from n 6 independent experiments). Panel (b) shows example pictures of cells from (a) transfected with the relevant plasmid

( eGFP) pre- and post-NMDA treatment (20 mM). White arrows indicate transfected neurons before NMDA treatment. Red/blue arrows in the

post-treatment panels indicate dead/live cells, respectively. Scale bar 15 mm (c) Hippocampal neurons were infected with the indicated rAAV at DIV 4,

and exposed to the indicated concentrations of NMDA at DIV 10, with cell death assessed after 20 h (means.e.m. shown). *Po0.05 (ANOVA, Bonferroni post hoc, n 3). (d) Example pictures from (c). Scale bar 15 mm. White arrows indicate infected neurons. Red/blue arrows in the Hoechst

panels indicate dead/live cells, respectively. In the rAAV-infected panels, only the infected (eGFP-expressing) neurons are scored as live/dead; in the non-infected control panel, all neurons are scored.

conrmed that the resting cytoplasmic Ca2 concentration of surviving cells was low and no different from that of control cells (Supplementary Fig. S3d). This showed that protection due to Mcu knockdown is (as expected) associated with the long-term preservation of mitochondrial capacity and Ca2 homeostasis.

Additionally, we found that the effect of Mcu-directed siRNA on neuronal survival was abolished by overexpression of a specially engineered siRNA-resistant form of Mcu, conrming the specicity of the effect (Supplementary Fig. S3e,f). Using the rAAV-mediated expression of Mcu-directed shRNAs, we achieved a 490% knockdown of Mcu (Fig. 5a,b, Supplementary Fig. S1aii), which when compared with control neurons (infected with an rAAV containing an expression cassette for a scrambled control shRNA) led to a reduction in NMDA-induced [Ca2 ]mit increases (Fig. 5c,d,e), a reduction in NMDA-induced mitochondrial membrane depolarization (Fig. 5f), and protection against NMDA-induced excitotoxicity (Fig. 5g,h).

Mcu is transcriptionally repressed by synaptic activity. It has long been known that episodes of synaptic activity can promote neuroprotection against excitotoxic insults23,2931. While multiple parallel mechanisms are likely to mediate this effect, the key role of Mcu-driven mitochondrial Ca2 uptake in excitotoxic cell death led us to investigate whether neuroprotective electrical activity inuences Mcu-mediated

effects. Burst activity was initiated in cortical neurons by treatment with the GABAA receptor antagonist bicuculline in the presence of the weak K channel blocker 4-aminopyridine (BiC/4-AP), as employed previously32,33. Consistent with previous studies, we found that BiC/4-AP pre-treatment, followed by wash-out, rendered neurons resistant to subsequently applied excitotoxic doses of NMDA (Fig. 6a). Strikingly, we also observed that the same stimulation resulted in the downregulation of Mcu expression at both protein and mRNA levels in cortical and hippocampal neurons (Fig. 6b,c).

This raises the exciting possibility that neuroprotective synaptic activity may actually inuence the degree of mitochondrial Ca2 uptake following subsequent NMDA treatment. Indeed, this was found to be the case: mitochondrial Ca2 uptake during NMDA treatment was substantially reduced in BiC/4-AP pre-treated cortical neurons (Fig. 6d) and BiC pre-treated hippocampal neurons (Fig. 6e). Importantly, BiC/4-AP pre-treatment of neurons did not reduce NMDAR whole-cell currents (Supplementary Fig. S4a) and moreover, Ca2 imaging using a non-mitochondrially targeted, cytoplasmic GCaMP2 revealed that basal and NMDA-induced cytoplasmic Ca2 levels were not reduced by BiC/4-AP pre-treatment (Supplementary

Fig. S4b). In fact, they were in both cases slightly larger, potentially a consequence of reduced mitochondrial Ca2 uptake. Thus, compared with control, neurons that had recently experienced strong synaptic activity showed a striking difference

NATURE COMMUNICATIONS | 4:2034 | DOI: 10.1038/ncomms3034 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

& 2013 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3034

*

siRNA

Con Mcu

Actin

Mcu

39 kDa

39 kDa

120

100

80

60

40

20

0 Con Mcu

siRNA

NS

Mcu mRNA

Relative I nmda

100

50

0 Con Mcu

siRNA

*

Mcu siRNA

NMDA

GCaMP2-mt fluorescence

(FF min)/(F maxF)

3

2

1

Control siRNA

NMDA

Con siRNA Mcu siRNA

GCaMP2-mt fluorescence

(FF min)/(F maxF)

5 4 3 2 1 0

GCaMP2-mt fluorescence

(FF min)/(F maxF)

5

4 3 2 1 0

Individual cells Mean

Individual cells Mean

100

0 200 300

Time (s)

100

0 200 300

Time (s)

0 Basal NMDA

30 [afii9839]M NMDA

10

8 6 4 2 0

0 50 100 150

Time (s)

m loss: change in rh rh-123

fluorescence (% of FCCP value)

Mcu siRNAMean of surrounding cells Individual surrounding cells

siRNA

Con

n=13

Mcu

n=11

y x

100 y

% Difference in NMDA-induced

rh-123 signal change

20 10

0 10 20 30

(xy)

200

#

*

100

80

60

40

20

*

Con siRNA

Mcu siRNA

NMDA

Pre- Post-

Control

siRNA

Mcu

siRNA

% Cell death

*

*

0 Con 10 20 30

NMDA ([afii9839]M)

Figure 4 | siRNA-mediated knockdown of Mcu suppresses NMDA-induced increases in mitochondrial Ca2 uptake and protects neurons against excitotoxic cell death. (a,b) Efcacy of Mcu-directed siRNA assessed by nucleofecting Mcu or control siRNA into cortical neurons and assessing Mcu protein (a) or mRNA (b) expression (n 4). *Po0.05 (paired two-tailed t-test). Means.e.m. shown. (c) NMDA-evoked whole-cell currents measured in

neurons transfected with the indicated siRNAs (co-expressing eGFP for identication, n 5). Means.e.m. shown. (df) Mcu knockdown boosts

mitochondrial Ca2 uptake assayed using GCaMP2-mt. Methodology used as per Fig. 1eg except that NMDA was used at 30 mM. *Po0.05 (paired two-tailed t-test, Con siRNA: 33 cells analyzed from n 5 experiments; Mcu siRNA: 31 cells analyzed from n 5 experiments). Means.e.m. shown. (g,h) Mcu

knockdown reduces NMDA-induced loss of mitochondrial membrane potential. Means.e.m. shown. Methodology used as in Figure 2a,b, except that 30 mM NMDA was used. #Po0.05 (unpaired two-tailed t-test) indicates a signicant difference between Mcu siRNA-transfected cells and surrounding untransfected cells. *Po0.05 (unpaired two-tailed t-test) indicates a signicant difference between data obtained with Mcu siRNA-transfected cells and that obtained with control siRNA-transfected cells (n 13 Con siRNA, n 11 Mcu siRNA). (i,j) SiRNA-mediated Mcu knockdown protects neurons

against excitotoxic cell death. Methodology used as per Fig. 3a,b. *Po0.05 (paired two-tailed t-test, 200400 cells analyzed from n 4 independent

experiments). Means.e.m. shown. Panel (j) shows example pictures of cells from (i) transfected with the relevant siRNA ( eGFP) pre- and post-NMDA

treatment (30 mM). White arrows indicate transfected neurons before NMDA treatment. Red/blue arrows in the post-treatment panels indicate dead/live cells, respectively. Scale bar 15 mm.

in the relationship between cytoplasmic Ca2 increases and subsequent mitochondrial Ca2 uptake. Collectively these data support a model whereby prior ring activity suppresses toxic

mitochondrial Ca2 uptake following cytoplasmic Ca2 increases, at least in part via the repression of Mcu expression.

Of note, we found that forced expression of Mcu promotes

6 NATURE COMMUNICATIONS | 4:2034 | DOI: 10.1038/ncomms3034 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2013 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3034 ARTICLE

McumRNA (relative)

1.0

0.5

0.0

Con

AAVshRNA

Scr

Mcu-1

Mcu-2

Tubulin

Mcu

* *

Scr

Mcu-1

Mcu-2

AAVshRNA

55 kDa

34 kDa

rAAVshRNA (Scr)

rAAVshRNA (Mcu-1) NMDA NMDA

1/n (% of Scr)

1/n (% of Scr)

200 100

Scr

1/n (% of Scr)

*

Mcu-1

Mcu-2

80 60 40 20

0

200 Individual cells

Mean

4mtD3cpv FRET ratio

{%R fret/(R max%R fret)}

Individual cells

Mean

150 100

4mtD3cpv FRET ratio

{%R fret/(R max%R fret)}

150 100

50

0

4mtD3cpv FRET ratio

{%R fret/(R max%R fret)}

50

0

100 200 300 0 100 200 300 Basal

Time (s) Time (s)

0

60

NMDA

NMDA

NMDA-induced % change in

rh-123 fluorescence

Con

NMDA

rAAVshRNA

80

*

Con

Scr

Mcu-1

Mcu-2

40

20

% Cell death

60

*

mCherry

40 20

Hoechst

0 Scr Mcu-1

rAAVshRNA

0 Con

Scr

Mcu-1

Mcu-2

rAAVshRNA

Figure 5 | ShRNA-mediated knockdown of Mcu suppresses NMDA-induced increases in mitochondrial Ca2 uptake and protects neurons against excitotoxic cell death. (a,b) Efcacy of Mcu-directed shRNA assessed by qRT PCR for mRNA (a) or western blot for protein (b) expression. Hippocampal neurons were infected with rAAVs containing a vector encoding shRNA directed against Mcu (two sequences: Mcu-1 and Mcu-2) or scrambled sequence shRNA (Scr). Knockdown was assessed 6 days post-infection. *Po0.05 (ANOVA, Bonferroni posthoc, n 3). Means.e.m. shown. (ce) rAAV-mediated

expression of Mcu-directed shRNAs suppresses mitochondrial Ca2 uptake assayed using 4mtD3cpv. Hippocampal neurons were infected with rAAVs containing a vector encoding 4mtD3cpv, plus rAAVshRNA encoding Mcu-directed shRNA or scrambled sequence shRNA (Scr). 4mtD3cpv FRET ratios were measured before and during exposure to NMDA (10 mM) 200250 seconds following NMDA application. Panels (c) and (d) show example traces from single experiments involving neurons expressing Scr (c) or Mcu (d)-directed shRNA. Panel (e) shows quantitation of all experiments performed (means.e.m.). In all cases, [Ca2 ] levels obtained using 4mtD3cpv were normalized to the mean NMDA-induced [Ca2 ] in Scr-transfected neurons measured on that precise day of imaging. *Po0.05 (ANOVA, Bonferroni post hoc. Scr: 68 cells from n 9 experiments; Mcu-1: 130 cells

from n 10 experiments; Mcu-2: 85 cells from n 10 experiments). (f) Hippocampal neurons were infected with the indicated rAAV at DIV4, and at

DIV 10 subjected to Rhodamine-123 based mitochondrial membrane potential imaging. Percent changes in Rhodamine-123 uorescence upon NMDA (30 mM) application were measured. *Po0.05 (Scr: 211 cells from 13 experiments; Mcu-1: 257 cells from 13 experiments). (g) Hippocampal neurons were infected with the indicated rAAV at DIV4, and exposed to the indicated concentrations of NMDA at DIV 10, with cell death assessed at 24 h. *Po0.05 (paired two-tailed t-test, n 4). Means.e.m. shown. (h) Example pictures from (g). White arrows indicate infected neurons. Red/blue arrows

in the Hoechst panels indicate dead/live cells, respectively. In the rAAV-infected panels, only the infected (mCherry-expressing) neurons are scored as live/dead; in the non-infected control panel, all neurons are scored.

increased vulnerability to excitotoxicity even in synaptically active neurons (Fig. 6f). In other words, Mcu overexpression increases the level of neuronal death to a point that is that is beyond the protective effects of synaptic activity.

Mcu is repressed by nuclear Ca2 -dependent CaM kinases. Finally, we investigated the mechanism by which synaptic activity represses Mcu expression. The observed lowering of Mcu mRNA in response to synaptic activity was suggestive of either transcriptional repression or reduced mRNA stability (that is, elevated degradation rate). We studied mRNA stability using the standard method of inhibiting transcription with Actinomycin D and

harvesting RNA at multiple time points thereafter. By comparing the level of Mcu mRNA with a relatively stable mRNA (18S), one can gain a measure of mRNA degradation, and assess the inuence of prior synaptic activity on this stability. An episode of prior synaptic activity had no signicant difference in the decay kinetics of Mcu mRNA levels (Supplementary Fig. S5). This result strongly suggests that repression of Mcu mRNA levels is due to the repression of transcription.

We then investigated the mechanism of this transcriptional repression. We found that activity-dependent repression of Mcu was blocked by cycloheximide, indicative that de novo gene expression is required (that is, Mcu regulation is not immediate-early in nature, Fig. 7a). Moreover, repression of Mcu by synaptic

NATURE COMMUNICATIONS | 4:2034 | DOI: 10.1038/ncomms3034 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

& 2013 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3034

*

BiC/4-AP BiC/4-AP

Con 4 h 24 h Con

Mcu

Tubulin

80

3

Con

4 h 24 h

100 100

80 * *

*

Mcu

Actin mRNA

Protein

Bic

*

80 *

* *

*

mRNA

Protein

% Cell death

60

40

20

Mcu expression

60

40

20

0 Con 4 h 24 h

BiC/4-AP

Mcu expression

60

40

20

*

0 Con 20

NMDA ([afii9839]M)

30

0 Con 4 h 24 h

BiC/4-AP

*

Con

Bic

Con

Bic

No virus

AAVeGFP

AAVMCUeGFP

GCaMP2-mt fluorescence

(FF min)/(F maxF)

4mtD3cpv FRET ratio

{%R fret/(R max%R fret)}1/n (% of Con)

100

80 60 40 20

80

60

40

20

0

* *

% Cell death

2

1

*

*

*

*

+ +

NMDA

0 Basal

NMDA

0 Basal

Con BiC

NMDA

Figure 6 | Repression of Mcu-mediated mitochondrial Ca2 uptake by neuroprotective synaptic activity. (a) Cortical neurons were pre-treated overnight with bicuculline (50 mM) plus 4-aminopyridine (250 mM) (BiC/4-AP, BiC in graph legends), which was removed and replaced with medium containing the indicated concentrations of NMDA. After 1 h, NMDARs were blocked by the application of MK-801 and cell death levels were assessed 23 h later by xing and DAPI-staining the cells, and counting nuclear pyknosis. *Po0.05 (ANOVA, Bonferroni post hoc, 7001000 cells analyzed from n 3

independent experiments). Means.e.m. shown. (b,c) Mcu expression following BiC/4-AP treatment analyzed by qRTPCR (n 3) and western blot

(n 4) in both cortical (b) and hippocampal (c) neurons. Means.e.m. shown. (d) Synaptic activity suppresses NMDA-induced mitochondrial

Ca2 uptake. Neurons were transfected with vectors encoding GCaMP2-mt and then treatedBiC/4-AP as in (a). Cells were then transferred to fresh medium and GCaMP2-mt uorescence imaged before and after NMDA treatment (30 mM). For each experiment, the mean value of all cells withinthe eld was calculated and the average pre- and post- stimulation level calculated (normalized to the maximal ionomycin-induced signal). *Po0.05 (unpaired two-tailed t-test. Con: 22 cells, n 4; BiC/4-AP: 43 cells, n 4). Means.e.m. shown. (e) Synaptic activity suppresses NMDA-induced

mitochondrial Ca2 uptake using 4mtD3cpv. Hippocampal neurons were infected with rAAVs containing a vector encoding 4mtD3cpv and treatedBiC overnight. 4mtD3cpv FRET ratios were measured before and during exposure to NMDA (10 mM) 200250 s following NMDA application. Shownis the quantitation of all experiments performed. In all cases, [Ca2 ] levels obtained using 4mtD3cpv were normalized to the mean NMDA-induced [Ca2 ] in control-transfected neurons measured on that precise day of imaging. *Po0.05 (paired two-tailed t-test. Con: 201 cells from n 16

experiments; BiC: 224 cells from n 16 experiments). Means.e.m. shown. (f) Neurons infected with the indicated AAVs were stimulated where indicated

with BiC for 16 h followed by exposure to fresh mediumNMDA. Cell death was assessed after a further 20 h. *Po0.05 (paired two-tailed t-test, n 3). Means.e.m. shown.

activity was blocked both by CaM kinase inhibition with KN-62 (Fig. 7b) and also by the nuclear-localized Ca2 /Calmodulin inhibitor CaMBP43335 (Fig. 7c). This suggested a mechanism of

Mcu repression that involved the nuclear Ca2 and CaM kinase-dependent induction of a gene product that in turn (directly or indirectly) represses Mcu expression.

Activity-dependent Mcu repression requires Npas4 induction. Our laboratories have previously described the neuroprotective effects of nuclear Ca2 signaling33 and also discovered a series of nuclear Ca2-induced neuroprotective genes, ve of which were found to inhibit NMDA-induced mitochondrial depolarization35.

We therefore screened these genes for an ability to repress Mcu expression and found that viral overexpression of Npas4 was sufcient to repress Mcu expression (Fig. 7d). Npas4 is a transcription factor that can promote negative gene regulation in neurons36, and it is an immediate-early gene: its induction by synaptic activity was not reduced by protein synthesis inhibition (cycloheximide treatment, Fig. 7e). Indeed, superinduction in the presence of cycloheximide was observed as is common with immediate-early genes. Moreover, Npas4 induction by synaptic

activity was strongly inhibited by both KN-62 and CaMBP4 (Fig. 7e,f), the same interventions that prevented Mcu repression (Fig. 7b,c). To directly test whether Npas4 induction was required for activity-dependent Mcu repression, we performed shRNA-mediated knockdown of Npas4 (Fig. 7g) and found that this inhibited activity-dependent repression of Mcu (Fig. 7h). Thus, activity-dependent Mcu repression involves the nuclear Ca2 and CaM kinase-dependent induction of Npas4, which in turn mediates the transcriptional repression of Mcu. These data establish Npas4 as a potentially central regulator of Mcu activity in neurons.

DiscussionIn this study, we used gain- and loss-of-function experiments to demonstrate the central role of the Mcu gene product in the control of mitochondrial Ca2 uptake in neurons following an excitotoxic insult. Moreover, we established that Mcu is an important mediator of death signal-induced loss of mitochondrial membrane potential and relays NMDA receptor stimulation to excitotoxic cell loss. Intriguingly, we also nd that Mcu expression and mitochondrial Ca2 uptake are subject to dynamic regulation by neuroprotective ring activity.

8 NATURE COMMUNICATIONS | 4:2034 | DOI: 10.1038/ncomms3034 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2013 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3034 ARTICLE

ns

*

* ns

* * ns

75

Mcu mRNA

100 Con

BiC

75

Mcu mRNA

100 Con Control

BiC 4 h

BiC 24 h

BiC

75

Mcu mRNA

100

50

50

50

25

25

25

0 Con CHX

0 Con KN-62

0 Con mCherry CaMBP4

600 Con

BiC

* *

*

*

550

Npas4 mRNA

100

Con

BiC

Mcu mRNA

100

75

500

75

50

Npas4 mRNA

100

50

25

50

25

0 Con

hrGFP

Npas4

0 Con KN-62 CHX

0 Con mCherry CaMBP4

*

* *

Con BiC (4 h) BiC (24 h)

Con

sh-Scr

sh-Npas4(1)

sh-Npas4(2)

BiC (4 h)

Con

sh-Scr

sh-Npas4(1)

sh-Npas4(2)

100

100

80

Con

Scr

sh(1)

sh(2)

Con

Npas4 mRNA

Scr

sh(2)

sh(1)

75

60

Npas4

Actin

Mcu mRNA

40

50

20

*

25

0

0

Con BiC (4 h) BiC (24 h)

Figure 7 | Activity-dependent repression of Mcu requires the nuclear Ca2 and CaM kinase-dependent induction of immediate-early gene Npas4. (a) qPCR analysis of Mcu expression in neurons treatedBiC/4-AP in the presence or absence of cycloheximide (CHX, 10 mg per ml), applied 1 h prior to BiC/4-AP *Po0.05 (paired two-tailed t-test, n 3). Means.e.m. shown. (b) qPCR analysis of Mcu expression in neurons treatedBiC/4-AP in the

presence or absence of a CaM kinase inhibitor (KN-62, 10 mM), applied 1 h prior to BiC/4-AP *Po0.05 (paired two-tailed t-test, n 3). Means.e.m.

shown. (c) Neurons were infected where indicated with rAAVs containing a vector encoding mCherry or CaMBP4, followed by BiC/4-AP treatment for the indicated times. *Po0.05 (ANOVA, Bonferroni post hoc, n 3). Means.e.m. shown. (d) Neurons were infected with rAAV containing a vector encoding

hrGFP control or Npas4 prior to RNA extraction and qPCR analysis for Mcu levels. *Po0.05 (paired two-tailed t-test, n 6). (e) Neurons were treated

with the indicated drugs prior to 4 h BiC/4-AP treatment prior to RNA extraction and qPCR analysis for Npas4 levels. *Po0.05 (paired two-tailed t-test, n 3). Means.e.m. shown. (f) Neurons treated as in (c) and Npas4 levels studied (BiC/4-AP treatment for 4 h). *Po0.05 (paired two-tailed t-test,

n 3). Means.e.m. shown. (g) Neurons were infected with rAAV-containing vectors encoding the indicated shRNAs targeting Npas4 (or control

scrambled shRNA), and the effect of this on BiC/4-AP induction of Npas4 mRNA (graph) or protein (inset) measured, *Po0.05 (ANOVA, Bonferroni post hoc, n 5). Means.e.m. shown. (h) As for 7 g, except that Mcu expression was measured. *Po0.05 (ANOVA, Bonferroni post hoc, n 5).

Means.e.m. shown.

Several key pro-death pathways are activated in parallel in response to an excitotoxic insult, including nitric oxide and ROS production, oxidative stress, JNK and PARP-1 activation, and CREB shut-off3,3740. Their relative importance may depend on the cell type, developmental stage, and severity of insult. Nevertheless, in most scenarios, it is likely that more than one pathway is important and that pathways can interact with each other. Future studies on Mcu-driven Ca2 uptake will tell us what role this process has in several key pro-death cascades.

For example, mitochondrial Ca2 uptake is implicated in the activation of PARP-1 via mitochondrial ROS generation, which combine with NO to trigger ONOO -mediated DNA damage, the trigger for PARP-1 activation11,41,42. Moreover,

mitochondrial Ca2 uptake and ROS production have also been implicated in JNK activation43. Another important issue surrounds the contribution of factors other than mitochondrial Ca2 uptake in mitochondrial membrane depolarization.

The coupling of the GluN2B C-terminal domain to downstream cell death pathways, including nitric oxide production, contributes to excitotoxicity3,4446, and excitotoxic mitochondrial depolarization via Ca2 uptake has been reported to involve nitric oxide10,47. Nitric oxide sensitizes mitochondria to

Ca2 -dependent depolarization10, raising the possibility that it affects mitochondrial energy metabolism, Mcu activity itself, or even efux via the mitochondrial Na /Ca2 exchanger. PARP-1 has also been shown to contribute to excitotoxic mitochondrial

NATURE COMMUNICATIONS | 4:2034 | DOI: 10.1038/ncomms3034 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

& 2013 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3034

depolarization, likely via depletion of NADH levels11. Despite the existence of multiple mechanisms leading to excitotoxicity, it is thus apparent that there is crosstalk between pathways and that mitochondrial dysfunction is a convergent hub for some and an upstream effector for others. In seeking to minimize neuronal death and dysfunction in excitotoxic disorders such as stroke, targeting multiple sites of control may be optimal. One can envisage that targeting upstream (big effect, short therapeutic window) as well as downstream (smaller effect, longer window) events could be more effective than any one strategy. Future studies will help to illuminate and clarify these issues.

It will also be of interest to determine whether and how mitochondrial Ca2 uptake contributes to the differential effects of synaptic versus extrasynaptic NMDAR activity28,38.

Extrasynaptic NMDAR activity is particularly well-coupled to mitochondrial depolarization28, and Ca2 inux triggered by bath activation of NMDARs is more strongly coupled to mitochondrial Ca2 uptake than inux through other routes27.

It is tempting to speculate that physical proximity of extrasynaptic NMDARs to mitochondria may contribute to this. Given the relatively low afnity and huge Ca2 -carrying capacity of the uniporter1 (half-saturation reached at 20 mM Ca2 ),

Ca2 uptake may be substantially enhanced if mitochondria are positioned extremely near the site of Ca2 entry where local

Ca2 levels are high. Synaptic NMDARs may be somewhat shielded from close proximity to mitochondria by the post-synaptic density, and more signicantly by the fact that dendritic spines rarely contain mitochondria48. It will also be important to determine whether other mitochondrial Ca2 transporters such as Letm1 or uncoupling proteins contribute in any way in neurons49,50, and which Mcu regulatory proteins (for example, Micu1/2 and Mcur1) are important. We occasionally observed biphasic kinetics in the increase of [Ca2]mit (for example, see

Fig. 1i), potentially indicating functionally different uptake mechanisms.

Our ndings suggest that pharmacological modulation of Mcu offers a realistic therapeutic target for excitotoxic injury, as ischemia-induced excitotoxicity is likely to develop over a relatively long timescale. Interference with early events in the excitotoxic cascade can yield protective outcomes, such as those employing NMDAR antagonists, and PSD-95 and JNK inhibitors3,37,43,46,51. Now that the molecular machinery of the Mcu has been identied18,19, and its role in excitotoxicity established (this study), screens for small molecules that inhibit Mcu activity may lead to novel neuroprotective compounds. Of course, the appropriateness of the Mcu as a therapeutic target needs to be assessed in the light of all its physiological role(s). Evidence supports the notion that mitochondrial Ca2 controls mitochondrial energy metabolism, including in neurons52,53. The impact of interfering with Mcu activity on these processes awaits further investigation, although we note from the current study that Mcu knockdown has no adverse effects on basal viability.

Our observations that Mcu is subject to transcriptional repression by neuroprotective Ca2 signals may reect an important facet of Mcus function in the cell. As Mcu-mediated mitochondrial Ca2 overload contributes to mitochondrial dysfunction and neuronal death in response to excitotoxic insults, repression of Mcu expression by Ca2 signals may represent a potentially important feedback mechanism acting to prevent mitochondrial Ca2 overload, (potentially in concert with other mechanisms54). It is tempting to speculate that this mechanism enables a cell to tune its Mcu activity to reect cytoplasmic Ca2 levels, and that Mcu transcriptional repression is an important factor in Ca2 -dependent tolerance/preconditioning mechanisms.

Protection against excitotoxic insults by Ca2-dependent transcriptional changes has centered mainly on the Ca2 responsive transcription factors involved. Induction of CREB-mediated transcription and inhibition of FOXO-mediated gene expression have been shown to contribute to protection against excitotoxic insults by neuroprotective Ca2 signals29,31. Our observation that repression of Mcu expression underlies an important part of the anti-excitotoxic effects of sub-toxic Ca2 signals demonstrates that a central effector of excitotoxicity is under direct control by neuroprotective pathways. Furthermore, we have demonstrated that this repression is a delayed response to the immediate-early induction of Npas4, although whether Npas4 is acting directly on the Mcu promoter in an inhibitory capacity or acting indirectly (for example, by inducing the expression of a repressor) awaits further investigation. Given that the MCU gene is also subject to Ca2 -mediated transcriptional repression in human ESC-derived neurons (Qiu and Hardingham, unpublished observations), a search for phylogenetically conserved elements within the Mcu promoter may yield the sequences responsible for mediating Ca2 sensitivity.

To conclude, we have established Mcu both as an effector of excitotoxic cell death and a target of neuroprotective signals. Exploiting these facts may lead to new interventions aimed at ameliorating neuronal dysfunction and death in disorders associated with aberrant NMDAR activity.

Methods

Neuronal culture. Hippocampal and cortical neurons from newborn C57Bl/6 mice (sex not determined) were cultured and maintained as described previously33,35. Neurons were cultured in Neurobasal media (Invitrogen, Gaithersburg, MD, USA) containing 1% rat serum and B27 (Invitrogen), and penicillin/streptomycin. Experiments were performed after a culturing period of 1013 days during which cultured neurons develop a rich network of processes, express functional NMDA-type and AMPA/kainate-type glutamate receptors, and form synaptic contacts.

Transfection and nucleofection. Neurons were transfected in trophic transfection medium55 with plasmids (2 mg ml 1 total) and/or siRNA (100 nM) using

Lipofectamine 2000. Nucleofection was performed using the Amaxa rat Neuron Nucleofector Kit (Lonza). See Supplementary Methods for further details and siRNA target sequences.

Western blotting and Subcellular fractionation. Gel electrophoresis and western blotting were performed using the Xcell Surelock system (Invitrogen) using precast gradient gels (420%) as described45. The following antibodies were used: Ccdc109a (Sigma, 1:500), -actin (Abcam, 1:2,000), a-Tubulin (Sigma, 1:400,000), anti PDH-E1-alpha (Abcam, 1:1,500); anti C-V-alpha (Abcam, 1:1,500); anti Erk 1/2 (NEB, 1:5,000). For visualization of western blots, HRP-based secondary antibodies were used followed by chemiluminescent detection on Kodak X-Omat lm. Western blots were analyzed by digitally scanning the blots, followed by densitometric analysis (ImageJ). For gure preparation of western blots, linear adjustment of brightness/contrast was applied (Photoshop) equally across the image. See Supplementary Methods for details of subcellular fractionation.

Plasmids and virus generation. The vector containing the mouse CaMKIIa promoter used to construct and package rAAV has been described previously56. For knockdown using shRNA, we used a rAAV vector containing the U6 promoter driving shRNA expression and that also contained a CaMKII promoter driving mCherry expression (to identify infected neurons). See Supplementary Methods for details of cloning and sequences used to generate rAAV vectors encoding Mcu, 4mtD3cpv and shRNA targeting Mcu and Npas4.

Neurons were infected with 1011 rAAV particles per ml at DIV 4. Infection efciencies were determined at DIV 10 by analyzing the uorescence of eGFP or mCherry; they ranged from 80 to 90% of the viable neurons. rAAV expression vectors for Npas4, CaMBP4 and mCherry have been described previously35,56. Mcu-GFP was a gift from Vamsi Mootha18, pCAGGS-GCaMP2 was a gift from Karel Svoboda57; GCaMP2-mt was a gift from Xianhua Wang24, Mt-cameleonpcDNA3 (containing 4mtD3cpv cDNA) was a gift from Roger Tsien25.

Electrophysiological recording and analysis. NMDA-evoked whole-cell steady-state currents (normalized to cell capacitance) were measured 48 h after

10 NATURE COMMUNICATIONS | 4:2034 | DOI: 10.1038/ncomms3034 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2013 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3034 ARTICLE

transfection as described45. NMDA (150 mM) was applied for 30 s, repeated twice for each cell. Data were ltered at 1 kHz and digitized at 5 kHz for subsequent off-line analysis.

Mitochondrial Ca2 imaging. In all Ca2 imaging experiments, we performed experiments with control and treated conditions using sister cultures interleaved on the same day to control for any culture-to-culture variation in NMDA-induced Ca2 inux. Thus, the only control data shown in any particular gure are those data collected on the same day on the same culture as when the treated condition

data (for example, Mcu overexpression) were collected. Several imaging approaches were used: (a) Rhod-2 imaging: this was performed as described23 using a Leica AF6000 LX imaging system, with a DFC350 FX digital camera. Neurons loaded with Rhod-2 showed a punctate distribution of the indicator that very closely resembled that obtained with mitotracker red (data not shown) and the pattern was abolished by the mitochondrial uncoupler FCCP (Supplementary Fig. S1c), conrming the mitochondrial localization of the dye. Transfected cells were identied by co-transfecting eGFP expression plasmid. The NMDA-induced increase in Rhod-2 signal (DF/F, ex 5466 nm; em 60020 nm) of each transfected cell was compared to that observed in surrounding untransfected cells in a 100 mm radius and subsequently expressed as a percentage of that observed in those untransfected cells. (b) GCaMP2-mt imaging: neurons were transfected with GCaMP2-mt-encoding vector, the uorescence signal of which was detected using a standard GFP lter set (ex 48020; em 52715). NMDA-induced changes in mitochondrial Ca2 were expressed as (F Fmin)/(Fmax F) according to the

equation [Ca2 ] Kd (F Fmin)/(Fmax F). Fmax was obtained when cells were

treated with the cell-permeable Ca2 ionophore ionomycin which both inserts into the plasma membrane and passes into the cell, inserting into mitochondrial membranes58, leading to saturation of the indicator when in regular medium(2 mM Ca2). Fmin was obtained under the same conditions except in zero Ca2 medium. The linear relationship between [Ca2 ] and (F Fmin)/(Fmax F) was

conrmed by calibrating the indicator as expressed in neurons, exposing them to ionomycin in the presence of sequentially different solutions of precise [Ca2], obtained by mixing K2EGTA and CaEGTA solutions (Calcium Calibration Buffer Kit, Invitrogen) at different ratios (data not shown). (c) 4mtD3cpv imaging was performed as described25. Fluorescence images were acquired at 2 Hz using a cooled CCD camera (iXon, Andor) through a 20 water-immersion objective

(XLMPlanFluor, Olympus) on an upright microscope (BX51W1, Olympus). Fluorescence excitation (CFP 43012; YFP 50010) was provided by a xenon arc lamp in combination with an excitation lter wheel (cell^R, Olympus). CFP (47012) and YFP (53515) emission wavelengths were separated and ltered using a DualView beam splitter (MAG Biosystems). Data were collected using proprietary software (cell^R, Olympus), and analyzed using ImageJ and IgorPro (WaveMetrics). When appropriate, only cells co-expressing both a Ca2 indicator and an mCherry or tDimer tag were chosen for analysis. Ca2 concentration changes were quantied using the crosstalk- and bleaching-corrected FRET ratio25.

Mitochondrial Ca2 responses to NMDA were expressed as (%Rfret/ (DRmax %Rfret))1/n according to the equation [Ca2 ] K0d (%Rfret/

(DRmax %Rfret))1/n25, where %Rfret represents the percent of the maximum

response obtained in SGG medium containing 10 mM ionomycin and 10 mM Ca2 . The values for DRmax and n (105.3% and 0.74, respectively) were derived from published calibrations25. The linear relationship between [Ca2] and (%Rfret/

(DRmax %Rfret))1/n was conrmed in neurons exposed to ionomycin in the

presence of sequentially different solutions of precise [Ca2 ], obtained by mixing K2EGTA and CaEGTA solutions (Calcium Calibration Buffer Kit, Invitrogen) at different ratios (data not shown). Control values for mitochondrial [Ca2 ]

obtained using 4mtD3cpv were subject to some culture-to-culture variation, which is likely to be due to the fact that it is operating closer to its maximum than GCaMP2-mt is. As a result of this, we have normalized all 4mtD3cpv [Ca2 ] data to the average NMDA-induced [Ca2 ] obtained in control cells on that day of recording. This allows us to directly assess the effect of Mcu expression, Mcu knockdown, or BiC pre-treatment on NMDA-induced mitochondrial [Ca2 ], expressed as a percentage of the NMDA-induced mitochondrial [Ca2 ] level obtained in that exact day of recording.

Mitochondrial membrane potential imaging. Mitochondrial membrane potential was analyzed as described10,28 using Rh123 (Molecular Probes). Briey, neurons were loaded with Rh123 (10 mg ml 1 or 26 mM) in SGG medium for 10 min followed by extensive washing with SGG. Rh123 partitions into polarized mitochondria where it self-quenches at the concentration used. When mitochondria depolarize, Rh123 leaks out of the mitochondria into the cytoplasm where it dequenches and uoresces strongly. Maximum Rh123 signal (ex 48020; em 52715) was obtained by completely eliminating the mitochondrial potential by exposing the neurons to the mitochondrial uncoupler FCCP (5 mM; Sigma). For analyzing the effect of transfecting Mcu expression vectors or Mcu-directed siRNA, transfected cells were identied by co-transfecting mCherry expression plasmid. The NMDA-induced increase in Rh123 signal of each transfected cell (as a % of the FCCP-induced level for that cell) was compared with that observed in surrounding untransfected cells in a 100-mm radius and subsequently expressed as a percentage of that observed in those untransfected cells. The NMDA-induced increase in

Rh123 signal in Mcu-1 shRNA-infected hippocampal neurons was compared with that observed in Scr shRNA-infected cultures analyzed in parallel.

Studying transfected neurons after excitotoxic insult. Excitotoxicity experiments performed on Lipofectamine-transfected neurons were performed as described45,59 (for full details see Supplementary Methods). The induction and analysis of NMDA-induced neuronal cell death of rAAV-infected neurons were performed as described60, with slight changes. For full details, see Supplementary Methods. Also see Supplementary Methods for details of methodology used to calculate whether Mcu overexpression specically rendered neurons more vulnerable to NMDA exposure, taking into account the increased basal death observed.

Quantitative reverse transcriptase PCR. To determine Mcu silencing by the pre-made Mcu-directed siRNA, and study activity-dependent Mcu regulation in cortical neurons, QRTPCR was performed using an Mx3000P QPCR System as described45. For further details including primer sequences, see Supplementary Methods. To determine Mcu silencing by Mcu-directed rAAVshRNA, and study activity-dependent Mcu regulation in hippocampal neurons, QRTPCR was performed using real-time TaqMan technology with a sequence detection system model 7300 Real Time PCR System (Applied Biosystems, Foster City, CA, USA). For further details including primer sequences see Supplementary Methods.

Statistical analysis. Statistical testing involved a two-tailed paired or unpaired Students t-test, as appropriate. For studies employing multiple testing, we used a one-way ANOVA followed by Bonferroni post hoc test. All data were presented as means.e.m.

References

1. Kirichok, Y., Krapivinsky, G. & Clapham, D. E. The mitochondrial calcium uniporter is a highly selective ion channel. Nature 427, 360364 (2004).

2. Carafoli, E. The interplay of mitochondria with calcium: an historical appraisal. Cell Calcium 52, 18 (2012).

3. Lau, A. & Tymianski, M. Glutamate receptors, neurotoxicity and neurodegeneration. Pugers Arch. 460, 525542 (2010).

4. Lipton, S. A. Paradigm shift in neuroprotection by NMDA receptor blockade: memantine and beyond. Nat. Rev. Drug Discov. 5, 160170 (2006).

5. Fan, M. M. & Raymond, L. A. N-methyl-D-aspartate (NMDA) receptor function and excitotoxicity in Huntingtons disease. Prog. Neurobiol. 81, 272293 (2007).

6. Bano, D. & Nicotera, P. Ca2 signals and neuronal death in brain ischemia.

Stroke 38(2 Suppl): 674676 (2007).7. Nicholls, D. G. Mitochondrial calcium function and dysfunction in the central nervous system. Biochim. Biophys. Acta 1787, 14161424 (2009).

8. Pivovarova, N. B. & Andrews, S. B. Calcium-dependent mitochondrial function and dysfunction in neurons. FEBS J. 277, 36223636 (2010).

9. Duchen, M. R. Mitochondria, calcium-dependent neuronal death and neurodegenerative disease. Pugers Arch. 464, 111121 (2012).10. Keelan, J., Vergun, O. & Duchen, M. R. Excitotoxic mitochondrial depolarisation requires both calcium and nitric oxide in rat hippocampal neurons. J. Physiol. 520(Pt 3): 797813 (1999).

11. Abramov, A. Y. & Duchen, M. R. Mechanisms underlying the loss of mitochondrial membrane potential in glutamate excitotoxicity. Biochim. Biophys. Acta 1777, 953964 (2008).

12. Stout, A. K., Raphael, H. M., Kanterewicz, B. I., Klann, E. & Reynolds, I. J. Glutamate-induced neuron death requires mitochondrial calcium uptake. Nat. Neurosci. 1, 366373 (1998).

13. Johnson-Cadwell, L. I., Jekabsons, M. B., Wang, A., Polster, B. M. & Nicholls,D. G. Mild Uncoupling does not decrease mitochondrial superoxide levels in cultured cerebellar granule neurons but decreases spare respiratory capacity and increases toxicity to glutamate and oxidative stress. J. Neurochem. 101, 16191631 (2007).14. Budd, S. L. & Nicholls, D. G. Mitochondrial calcium regulation, and acute glutamate excitotoxicity in cultured cerebellar granule cells. J. Neurochem. 67, 22822291 (1996).

15. Hajnoczky, G. et al. Mitochondrial calcium signalling and cell death: approaches for assessing the role of mitochondrial Ca2 uptake in apoptosis.

Cell Calcium 40, 553560 (2006).16. Robert, V. et al. Beat-to-beat oscillations of mitochondrial [Ca2 ] in cardiac

cells. EMBO J. 20, 49985007 (2001).17. David, G. & Barrett, E. F. Stimulation-evoked increases in cytosolic [Ca(2 )]

in mouse motor nerve terminals are limited by mitochondrial uptake and are temperature-dependent. J. Neurosci. 20, 72907296 (2000).18. Baughman, J. M., Perocchi, F., Girgis, H. S., Plovanich, M., Belcher-Timme, C.A. & Sancak, Y. et al. Integrative genomics identies MCU as an essential component of the mitochondrial calcium uniporter. Nature 476, 341345 (2011).

NATURE COMMUNICATIONS | 4:2034 | DOI: 10.1038/ncomms3034 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 11

& 2013 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3034

19. De Stefani, D., Raffaello, A., Teardo, E., Szabo, I. & Rizzuto, R. A forty-kilodalton protein of the inner membrane is the mitochondrial calcium uniporter. Nature 476, 336340 (2011).

20. Perocchi, F. et al. MICU1 encodes a mitochondrial EF hand protein required for Ca(2 ) uptake. Nature 467, 291296 (2010).

21. Mallilankaraman, K. et al. MCUR1 is an essential component of mitochondrial Ca2 uptake that regulates cellular metabolism. Nat. Cell Biol. 14, 13361343

(2012).22. Kovacs, R., Kardos, J., Heinemann, U. & Kann, O. Mitochondrial calcium ion and membrane potential transients follow the pattern of epileptiform discharges in hippocampal slice cultures. J. Neurosci. 25, 42604269 (2005).

23. Soriano, F. X., Papadia, S., Hofmann, F., Hardingham, N. R., Bading, H. & Hardingham, G. E. Preconditioning doses of NMDA promote neuroprotection by enhancing neuronal excitability. J. Neurosci. 26, 45094518 (2006).

24. Chen, M., Wang, Y., Hou, T., Zhang, H., Qu, A. & Wang, X. Differential mitochondrial calcium responses in different cell types detected with a mitochondrial calcium uorescent indicator, mito-GCaMP2. Acta Biochim. Biophys. Sin. (Shanghai) 43, 822830 (2011).

25. Palmer, A. E. & Tsien, R. Y. Measuring calcium signaling using genetically targetable uorescent indicators. Nat. Protoc. 1, 10571065 (2006).

26. White, R. J. & Reynolds, I. J. Mitochondrial depolarization in glutamate-stimulated neurons: an early signal specic to excitotoxin exposure. J. Neurosci. 16, 56885697 (1996).

27. Peng, T. I. & Greenamyre, J. T. Privileged access to mitochondria of calcium inux through N-methyl-D-aspartate receptors. Mol. Pharmacol. 53, 974980 (1998).

28. Hardingham, G. E., Fukunaga, Y. & Bading, H. Extrasynaptic NMDARs oppose synaptic NMDARs by triggering CREB shut-off and cell death pathways. Nat. Neurosci. 5, 405414 (2002).

29. Lee, B., Butcher, G. Q., Hoyt, K. R., Impey, S. & Obrietan, K. Activity-dependent neuroprotection and cAMP response element-binding protein (CREB): kinase coupling, stimulus intensity, and temporal regulation of CREB phosphorylation at serine 133. J. Neurosci. 25, 11371148 (2005).

30. Tauskela, J. S. et al. Elevated synaptic activity preconditions neurons against an in vitro model of ischemia. J. Biol. Chem. 283, 3466734676 (2008).

31. Dick, O. & Bading, H. Synaptic activity and nuclear calcium signaling protects hippocampal neurons from death signal-associated nuclear translocation of FoxO3a induced by extrasynaptic NMDA receptors. J. Biol. Chem. 285, 1935419361 (2010).

32. Hardingham, G. E., Arnold, F. J. & Bading, H. Nuclear calcium signaling controls CREB-mediated gene expression triggered by synaptic activity. Nat. Neurosci. 4, 261267 (2001).

33. Papadia, S., Stevenson, P., Hardingham, N. R., Bading, H. & Hardingham, G. E. Nuclear Ca2 and the cAMP response element-binding protein family

mediate a late phase of activity-dependent neuroprotection. J. Neurosci. 25, 42794287 (2005).34. Wang, J., Campos, B., Jamieson, G. A., Kaetzel, M. A. & Dedman, J. R. Functional elimination of calmodulin within the nucleus by targeted expression of an inhibitor peptide. J. Biol. Chem. 270, 3024530248 (1995).

35. Zhang, S. J. et al. Nuclear calcium signaling controls expression of a large gene pool: identication of a gene program for acquired neuroprotection induced by synaptic activity. PLoS Genet. 5, e1000604 (2009).

36. Lin, Y. et al. Activity-dependent regulation of inhibitory synapse development by Npas4. Nature 455, 11981204 (2008).

37. Borsello, T. et al. A peptide inhibitor of c-Jun N-terminal kinase protects against excitotoxicity and cerebral ischemia. Nat. Med. 9, 11801186 (2003).

38. Hardingham, G. E. & Bading, H. Synaptic versus extrasynaptic NMDA receptor signalling: implications for neurodegenerative disorders. Nat. Rev. Neurosci. 11, 682696 (2010).

39. Andrabi, S. A. et al. Iduna protects the brain from glutamate excitotoxicity and stroke by interfering with poly(ADP-ribose) polymer-induced cell death. Nat. Med. 17, 692699 (2011).

40. Rodriguez-Rodriguez, P., Fernandez, E., Almeida, A. & Bolanos, J. P. Excitotoxic stimulus stabilizes PFKFB3 causing pentose-phosphate pathway to glycolysis switch and neurodegeneration. Cell Death Differ. 19, 15821589 (2012).

41. Yu, S. W. et al. Mediation of poly(ADP-ribose) polymerase-1-dependent cell death by apoptosis-inducing factor. Science 297, 259263 (2002).42. Duan, Y., Gross, R. A. & Sheu, S. S. Ca2 -dependent generation of

mitochondrial reactive oxygen species serves as a signal for poly(ADP-ribose) polymerase-1 activation during glutamate excitotoxicity. J Physiol 585(Pt 3): 741758 (2007).43. Soriano, F. X. et al. Specic targeting of pro-death NMDA receptor signals with differing reliance on the NR2B PDZ ligand. J. Neurosci. 28, 1069610710 (2008).

44. Dawson, T. M., Zhang, J., Dawson, V. L. & Snyder, S. H. Nitric oxide: cellular regulation and neuronal injury. Prog. Brain Res. 103, 365369 (1994).

45. Martel, M. A. et al. The subtype of GluN2 C-terminal domain determines the response to excitotoxic insults. Neuron 74, 543556 (2012).

46. Cook, D. J., Teves, L. & Tymianski, M. Treatment of stroke with a PSD-95 inhibitor in the gyrencephalic primate brain. Nature 483, 213217 (2012).

47. Almeida, A., Bolanos, J. P. & Medina, J. M. Nitric oxide mediates glutamate-induced mitochondrial depolarization in rat cortical neurons. Brain Res. 816, 580586 (1999).

48. Li, Z., Okamoto, K., Hayashi, Y. & Sheng, M. The importance of dendritic mitochondria in the morphogenesis and plasticity of spines and synapses. Cell 119, 873887 (2004).

49. Trenker, M., Malli, R., Fertschai, I., Levak-Frank, S. & Graier, W. F. Uncoupling proteins 2 and 3 are fundamental for mitochondrial Ca2 uniport. Nat. Cell

Biol. 9, 445452 (2007).50. Jiang, D., Zhao, L. & Clapham, D. E. Genome-wide RNAi screen identies Letm1 as a mitochondrial Ca2 /H antiporter. Science 326, 144147 (2009).

51. Chen, H. S., Wang, Y. F., Rayudu, P. V., Edgecomb, P., Neill, J. C. & Segal, M.M. et al. Neuroprotective concentrations of the N-methyl-D-aspartate open-channel blocker memantine are effective without cytoplasmic vacuolation following post-ischemic administration and do not block maze learning or long-term potentiation. Neuroscience 86, 11211132 (1998).52. Glancy, B. & Balaban, R. S. Role of mitochondrial Ca2 in the regulation of

cellular energetics. Biochemistry 51, 29592973 (2012).53. Duchen, M. R. Ca(2 )-dependent changes in the mitochondrial energetics in

single dissociated mouse sensory neurons. Biochem. J. 283(Pt 1): 4150 (1992).54. Pivovarova, N. B., Stanika, R. I., Watts, C. A., Brantner, C. A., Smith, C. L. & Andrews, S. B. Reduced calcium-dependent mitochondrial damage underlies the reduced vulnerability of excitotoxicity-tolerant hippocampal neurons.J. Neurochem. 104, 16861699 (2008).55. Bading, H., Ginty, D. D. & Greenberg, M. E. Regulation of gene expression in hippocampal neurons by distinct calcium signaling pathways. Science 260, 181186 (1993).

56. Lau, D. & Bading, H. Synaptic activity-mediated suppression of p53 and induction of nuclear calcium-regulated neuroprotective genes promote survival through inhibition of mitochondrial permeability transition. J. Neurosci. 29, 44204429 (2009).

57. Mao, T., OConnor, D. H., Scheuss, V., Nakai, J. & Svoboda, K. Characterization and subcellular targeting of GCaMP-type genetically-encoded calcium indicators. PLoS One 3, e1796 (2008).

58. Nicholls, D. G. Simultaneous monitoring of ionophore- and inhibitor-mediated plasma and mitochondrial membrane potential changes in cultured neurons.J. Biol. Chem. 281, 1486414874 (2006).59. Puddifoot, C. et al. PGC-1alpha negatively regulates extrasynaptic NMDAR activity and excitotoxicity. J. Neurosci. 32, 69957000 (2012).

60. Zhang, S. J. et al. A signaling cascade of nuclear calcium-CREB-ATF3 activated by synaptic NMDA receptors denes a gene repression module that protects against extrasynaptic NMDA receptor-induced neuronal cell death and ischemic brain damage. J. Neurosci. 31, 49784990 (2011).

Acknowledgements

We are grateful to Vamsi Mootha and Xianhua Wang for providing plasmids. This work is supported by a MRC Senior Non-Clinical Research Fellowship (G.E.H.), an ERC Advanced grant (H.B.), the BBSRC, the Wellcome Trust and the German Ministry of Education and Research (BMBF, 01GQ1003A).

Author contributions

G.E.H and H.B. conceived the project and designed the research. G.E.H., H.B., P.A.S. and D.J.A.W. supervised the research. J.Q., Y-W.T., A.M.H., M.M., N.K. and G.E.H. performed the experiments. G.E.H. and H.B. wrote the manuscript.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Qiu, J. et al. Mitochondrial calcium uniporter Mcu controls excitotoxicity and is transcriptionally repressed by neuroprotective nuclear calcium signals. Nat. Commun. 4:2034 doi: 10.1038/ncomms3034 (2013).

This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. To view a copy of this licence visit http://creativecommons.org/licenses/by/3.0/

Web End =http:// http://creativecommons.org/licenses/by/3.0/

Web End =creativecommons.org/licenses/by/3.0/ .

12 NATURE COMMUNICATIONS | 4:2034 | DOI: 10.1038/ncomms3034 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2013 Macmillan Publishers Limited. All rights reserved.