Dysregulation of Mitochondrial Calcium Signaling and Superoxide Flashes Cause Mitochondrial Genomic DNA Damage in Huntington Disease*

Background: Oxidative damage has been implicated in the pathology of Huntington disease (HD). Results: Strikingly higher mitochondrial Ca2+ loading and superoxide generation cause significantly higher levels of mitochondrial DNA damage in HD cells. Conclusion: Excessive mitochondrial Ca2+ loading-dependent oxidant generation is a causative factor for HD. Significance: Our data reveal new links between dysregulated mitochondrial Ca2+ signaling and elevated oxidative DNA damage in HD. Huntington disease (HD) is an inherited, fatal neurodegenerative disorder characterized by the progressive loss of striatal medium spiny neurons. Indications of oxidative stress are apparent in brain tissues from both HD patients and HD mouse models; however, the origin of this oxidant stress remains a mystery. Here, we used a yeast artificial chromosome transgenic mouse model of HD (YAC128) to investigate the potential connections between dysregulation of cytosolic Ca2+ signaling and mitochondrial oxidative damage in HD cells. We found that YAC128 mouse embryonic fibroblasts exhibit a strikingly higher level of mitochondrial matrix Ca2+ loading and elevated superoxide generation compared with WT cells, indicating that both mitochondrial Ca2+ signaling and superoxide generation are dysregulated in HD cells. The excessive mitochondrial oxidant stress is critically dependent on mitochondrial Ca2+ loading in HD cells, because blocking mitochondrial Ca2+ uptake abolished elevated superoxide generation. Similar results were obtained using neurons from HD model mice and fibroblast cells from HD patients. More importantly, mitochondrial Ca2+ loading in HD cells caused a 2-fold higher level of mitochondrial genomic DNA (mtDNA) damage due to the excessive oxidant generation. This study provides strong evidence to support a new causal link between dysregulated mitochondrial Ca2+ signaling, elevated mitochondrial oxidant stress, and mtDNA damage in HD. Our results also indicate that reducing mitochondrial Ca2+ uptake could be a therapeutic strategy for HD.

Huntington disease (HD) is an inherited, fatal neurodegenerative disorder characterized by the progressive loss of striatal medium spiny neurons. Indications of oxidative stress are apparent in brain tissues from both HD patients and HD mouse models; however, the origin of this oxidant stress remains a mystery. Here, we used a yeast artificial chromosome transgenic mouse model of HD (YAC128) to investigate the potential connections between dysregulation of cytosolic Ca 2؉ signaling and mitochondrial oxidative damage in HD cells. We found that YAC128 mouse embryonic fibroblasts exhibit a strikingly higher level of mitochondrial matrix Ca 2؉ loading and elevated superoxide generation compared with WT cells, indicating that both mitochondrial Ca 2؉ signaling and superoxide generation are dysregulated in HD cells. The excessive mitochondrial oxidant stress is critically dependent on mitochondrial Ca 2؉ loading in HD cells, because blocking mitochondrial Ca 2؉ uptake abolished elevated superoxide generation. Similar results were obtained using neurons from HD model mice and fibroblast cells from HD patients. More importantly, mitochondrial Ca 2؉ loading in HD cells caused a 2-fold higher level of mitochondrial genomic DNA (mtDNA) damage due to the excessive oxidant generation. This study provides strong evidence to support a new causal link between dysregulated mitochondrial Ca 2؉ signaling, elevated mitochondrial oxidant stress, and mtDNA damage in HD. Our results also indicate that reducing mitochondrial Ca 2؉ uptake could be a therapeutic strategy for HD.
Huntington disease (HD) 2 is an inherited, progressive neurodegenerative disorder characterized by chorea, gradual but inexorable cognitive decline, and psychiatric disturbances (1,2). At the genetic level, the disease is caused by an abnormal expansion of the CAG repeat located in exon 1 of huntingtin (HTT), the gene encoding the huntingtin protein (3). A number of lines of evidence indicate that polyQ expansion in HTT (HTT exp ) leads to a "toxic gain of function" (4 -15). Recent results suggest that HTT exp toxicity may also involve disturbed glutamate-induced Ca 2ϩ signaling in MSNs (9,16). It has been reported that HTT exp facilitates activity of NMDA receptors (15,(17)(18)(19)(20)(21) and type 1 inositol 1,4,5-trisphosphate receptors (InsP 3 R1) (22,23), leading to the enhanced intracellular Ca 2ϩ signaling in HD neurons.
Oxidative damage may also contribute to neuronal loss in HD (24). Oxidative damage is clearly apparent in both HD patients and HD mouse models (25)(26)(27)(28)(29). Interestingly, oxidative damage in mitochondrial genomic DNA (mtDNA) preferentially occurs in the parietal region of the human HD brain, but not frontal cortex or cerebellum (26), and striatal MSN particularly accumulates more mtDNA mutations than any other brain cells (27), indicating the brain regional specificity of oxidative damage in HD. Deficiency of respiratory chain complexes has been suggested in enhanced reactive oxygen species generation in the HD yeast model (30), and complex II and III defects have been detected in HD brain (31)(32)(33), suggesting that defects of respiratory chain complexes may cause mitochondrial dysfunction and ROS overproduction. However, no significant deficiency of respiratory chain complexes has been dem-* This work was supported National Basic Research Program of China Grants onstrated in presymptomatic patients or in HD model mice expressing full-length HTT exp , suggesting that respiratory chain defects are a secondary feature in HD pathogenesis (34,35). Therefore, the mechanism by which HTT exp causes elevated oxidant stress in HD neurons remains elusive, and why striatal MSN are preferentially under oxidant stress is not known.
Enhanced cytosolic Ca 2ϩ signaling in striatal MSN has been linked to neuropathology in HD (18,22,23). Mitochondrion is not only a very efficient Ca 2ϩ buffer system but also constitutes the major ROS generator in eukaryotic cells, including neurons. We aim to look at the potential connection between mitochondrial Ca 2ϩ loading and ROS generation in HD cells/neurons. However, whether Ca 2ϩ loading into mitochondria matrix stimulates or decreases ROS generation is a topic of great debate in the field. From a thermodynamic point of view, it has been noted that Ca 2ϩ loading into mitochondria, which consumes mitochondrial membrane potential (⌬⌿ m ), should result in a partial uncoupling and thus would reduce electron flow from complex II and decrease ROS generation (36). However, a fundamentally different model has been proposed that mitochondrial matrix Ca 2ϩ loading-induced uncoupling leads to increased ROS generation by virtue of increased in electron flux (to restore and to maintain ⌬⌿ m ), and thus increased in the probability of electron slippage and superoxide formation (37,38). Both models were supported experimentally; however, the lack of specific ROS probes targeted to the mitochondria of living cells may be the main reason for this discrepancy. Nevertheless, Ca 2ϩ signaling system may regulate components of ROS homeostasis and thus influence the redox balance in cells (38,39), raising the possibility that excessive Ca 2ϩ loading into mitochondria creates mitochondrial oxidant stress.
To dissect the molecular events between disrupted cytosolic Ca 2ϩ signaling and mitochondrial oxidant stress, it is essential to visualize and quantify mitochondrial matrix ROS generation in situ and in real time. Recently, Wang et al. (40,41) developed a highly sensitive mitochondrially targeted superoxide indicator, circular permuted yellow fluorescent protein (cpYFP), which can sense bursts of superoxide generation (termed superoxide flashes) in single mitochondrion in live cells. Moreover, mitochondrial matrix Ca 2ϩ dynamics were also visualized by a mitochondria matrix-targeted Ca 2ϩ indicator, cameleon (42,43). We found that both mitochondrial Ca 2ϩ signaling and superoxide generation are dysregulated in embryonic fibroblasts and striatal neurons from a yeast artificial chromosome transgenic mouse model of HD. We then provided direct evidence to support a new causal link between dysregulated mitochondrial Ca 2ϩ signaling, elevated mitochondrial oxidant stress, and mtDNA damage accumulation in HD. Steady buildup of oxidative damage to mitochondrial DNA and proteins could eventually result in mitochondrial dysfunction and the defects of respiratory chain complexes found in postmortem studies of symptomatic HD patients (31)(32)(33). Because dysregulation of cytosolic Ca 2ϩ signaling occurs preferentially in striatal neurons of HD (18,21,22), excessive mitochondrial Ca 2ϩ loading and elevated mitochondrial oxidant stress are expected to affect striatal region selectively. Given that dysregulation of intracellular Ca 2ϩ signaling has been linked to many pathological conditions (16, 44 -48), our data also suggest that reducing mitochondrial Ca 2ϩ uptake could form part of a therapeutic approach to slow the progression of several neurodegenerative diseases, including HD.

EXPERIMENTAL PROCEDURES
MEF and Primary MSN Cultures-YAC128 HD transgenic mice were obtained from The Jackson Laboratory (Bar Harbor, ME). Generation and breeding of YAC128 HD transgenic mice (FVBN/NJ background strain) have been described previously (49). Heterozygous male YAC128 mice were crossed with wildtype (WT) female mice. MEFs were isolated from 12.5 to 13.5 postcoitum mouse embryos as described previously (50) and genotyped by PCR with primers specific for exons 44 and 45 of HTT. The MEF cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and immortalized through continuous passaging. Several immortalized MEF clones for each genotype were frozen for Ca 2ϩ imaging and superoxide flash imaging experiments. Primary MSN cultures were established as described previously (23,51).
MSN cultures that had been 13-14 days in vitro were transfected with pCDNA3.0 -2mt-cameleon using Lipofectamine 2000 (Invitrogen) and then imaged 2 days later after transfection. The MSNs were rinsed twice and maintained in artificial cerebrospinal fluid (140 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 2 mM CaCl 2 , and 10 mM HEPES, pH 7.3) at room temperature throughout the experiments. The stock solution of (RS)-3,5dihydroxyphenylglycine (DHPG) was dissolved in artificial cerebrospinal fluid, and the working concentration was 50 M.
Measurement of Superoxide Flashes in Mitochondria of Living Cells-We employed lentivirus-medicated gene transfer to express cpYFP in the mitochondrial matrix of WT and YAC128 MEFs. The superoxide flash measurement experiments (cpYFP imaging) were performed as described previously with modifications (40). Briefly, cells were rinsed twice, maintained in HBSS at room temperature, and then stimulated with BK. cpYFP was excited at 488 nm, and emission was collected at 500 -550 nm. Amplitude of superoxide flashes (⌬F/F 0 ) was expressed as (F max Ϫ F base )/(F base Ϫ F background ), where F max represented the peak superoxide flash fluorescence recorded, and F base represented the fluorescence recorded before the generation of superoxide flashes. In some experiments, MEFs were preincubated in 2-APB or Ru360 at room temperature for 30 min before stimulating with 5 M BK. 13-14 days in vitro MSN cultures were transfected with plasmid encoding cpYFP using Lipofectamine 2000 and then imaged the next day. MSNs were rinsed twice and then maintained in artificial cerebrospinal fluid at room temperature throughout the whole superoxide flashes experiment. The base-line measurements were acquired over 8 min, and the treatment with BK or DHPG also lasted for 8 min.
Normal (GM02185C) and HD patient (GM09197) primary human fibroblasts were from Coriell Cell Repositories. Human fibroblasts were grown in minimal essential medium supplemented with 15% fetal bovine serum at 37°C with 5% CO 2 . Lentivirus-mediated gene transfer was employed to express cpYFP in the mitochondrial matrix of human fibroblasts, and cpYFP imaging was performed as described above.
Assessment of Mitochondrial DNA Damage-The extent of mitochondrial DNA damage was evaluated by using a previously described qPCR assay with modifications (54,55). Briefly, WT and YAC128 MEFs, before and after repetitive BK stimulation (twice for 10 min, 2.5 M of BK) without or with Ru360 and SS31 pretreatment, were rinsed twice with cold PBS and digested with 0.25% trypsin. Total DNA was extracted using a TIANamp genomic DNA kit (TIANGEN) according to the manufacturer's protocol. DNA concentration and the A 260 / 280 ratio were measured using a SYNeRGY4 (Biotek) spectrophotometer. A 117-base pair (bp) mtDNA fragment was amplified by qPCR to quantify the relative mitochondrial genome DNA copy number in total DNA samples. We also amplified a 900-bp fragment of 18 S ribosome gene as a control to correct the 117-bp mtDNA copy number in qPCR. The qPCR cycling conditions were as follows: 75°C for 2 min; 94°C for 1 min; 94°C for 15 s; 62°C for 45 s, and 72°C for 30 s (40 cycles), and 72°C for 5 min. The data were analyzed using the ⌬⌬Ct method to reflect the relative mitochondrial genome copy number in 1.0 ng of total DNA. Standard PCRs were then carried out to further confirm the exact relative mitochondrial genome copy number using the 117-bp mitochondrial fragment primers and cycle numbers that were within the linear range of qPCR amplification. To assess mitochondrial DNA integrity, a 10.0-kilobase pair (kb) mitochondrial DNA fragment was then amplified by using same amount of mitochondrial DNA as templates. The PCR profile was as follows: 1 min at 94°C, followed by 30 cycles of 40 s at 94°C and 60°C for 15 min. A final extension at 72°C was performed for 10 min. The 10.0-kb PCR products were resolved on 1% agarose gels, whereas the 117-bp PCR products were resolved on 2% agarose gels. Ethidium bromide-stained products were imaged, and the density of the amplified bands was quantified using ImageJ software under Gel Analyzer Options. Amplification of the 10-kb mtDNA fragment was normalized to the amplification of the 117-bp mtDNA fragment, and the PCR signal from treated samples was compared with that from untreated control DNA that contained the same amount of mtDNA. DNA from control samples is defined as undamaged in this assay. Because the DNA polymerase amplifies only undamaged templates, the relative amplification ratio (A D /A O ) is inversely proportional to the extent of damage. A D ϭ amount of amplification from damaged DNA sample (BKstimulated); A O ϭ amount of amplification from undamaged DNA sample (no BK stimulation). The average lesion frequency per strand is calculated as Ϫln A D /A O as described previously (54,55). The results are expressed as relative amplification ratio (A D /A O ) and as lesion frequency per 10-kb per strand.
Drugs-BK and DHPG were purchased from Sigma. Ru360 and 2-APB were purchased from Merck. SS31 was a gift from Tao Ye (Laboratory of Chemical Genomics, Peking University Shenzhen Graduate School, China).
Statistics-Data were expressed as the means Ϯ S.E., and statistical significance of differences between different groups was assessed using the independent samples t test or analysis of variance for repeated measures at the 95% level (p Ͻ 0.05).
Study Approval-All animal experiments were reviewed and approved by the Institute of Zoology Institutional Animal Care and Use Committee and were conducted according to the committee's guidelines.

RESULTS
Disrupted Mitochondrial Ca 2ϩ Signaling in HD Cells-In our previous studies, we found that mutant huntingtin protein, HTT exp , specifically binds to and facilitates InsP3R1 activity (22), suggesting that excessive cytoplasmic Ca 2ϩ signaling may play an important role in HD pathogenesis (9,23). Because of the close connection between mitochondria and the endoplasmic reticulum (ER) (56,57), Ca 2ϩ may rush into the mitochondria and cause excessive Ca 2ϩ loading during Ca 2ϩ mobilization. To monitor the mitochondrial Ca 2ϩ concentration ([Ca 2ϩ ] m ) changes in HD cells, we took the advantage of a protein-based Ca 2ϩ indicator, cameleon. Cameleon was targeted into mitochondria, and the free Ca 2ϩ dynamics in the mitochondrial matrix were visualized using fluorescence resonance energy transfer (FRET) (42,43). The free Ca 2ϩ concentration in the mitochondria of WT and YAC128 MEF cells was continuously monitored by cameleon FRET imaging, and the data are presented as absolute Ca 2ϩ concentration in micromoles ( Fig.  1). On average, the basal mitochondrial Ca 2ϩ concentration before BK treatment in YAC128 MEFs was significantly higher than that in WT MEFs (6.23 Ϯ 0.19 M versus 5.48 Ϯ 0.23 M, p Ͻ 0.05) (Fig. 1, A-C). Treatment with 2.5 M BK caused significantly bigger mitochondrial Ca 2ϩ transients in YAC128 MEFs than in WT cells (22. To quantify the amount of Ca 2ϩ that loaded into mitochondria during BK stimulation, we calculated the area (M⅐s) below Ca 2ϩ transient traces. Significantly, more Ca 2ϩ was loaded into the mitochondria in YAC128 MEFs than in WT MEFs (2485 Ϯ 218 M⅐s versus 1155 Ϯ 209 M⅐s, p Ͻ 0.01) (Fig. 1D). Similar results were obtained using different WT and YAC128 MEF clones (data not show). In summary, mitochondria in YAC128 HD cells show significantly higher Ca 2ϩ loading compared with WT cells, even in resting cells, suggesting that mitochondrial Ca 2ϩ signaling in HD cells is dysregulated.

Involvement of InsP 3 Rs and Mitochondrial Ca 2ϩ Uniporter (MCU) in the Disrupted Mitochondrial Ca 2ϩ
Signaling in HD Cells-InsP 3 R is involved in the abnormal Ca 2ϩ signaling in HD (16,22,23); therefore, we next tested if Ca 2ϩ release via InsP 3 R is important in the disrupted mitochondrial Ca 2ϩ signaling. 2-APB, a membrane-permeable blocker of InsP 3 Rs, can specifically block Ca 2ϩ release from InsP 3 Rs (58, 59). We compared BK-induced mitochondrial Ca 2ϩ signals in WT and YAC128 MEF cultures in the presence and absence of 2-APB. Preincubation with 20 M 2-APB significantly decreased the BK-induced mitochondrial Ca 2ϩ peak values in both WT and YAC128 MEFs (p Ͻ 0.01) (Fig. 2, A, B and D). Consistent with mitochondrial Ca 2ϩ peak value results, 2-APB was able to significantly reduce the peak area of BK-induced mitochondrial Ca 2ϩ transients in WT MEFs and YAC128 MEFs (p Ͻ 0.05, Fig.  2, B and E), indicating the important role of InsP3Rs in excessive mitochondrial Ca 2ϩ loading in HD cells. Moreover, in resting cells, 2-APB significantly reduced the basal mitochondrial Ca 2ϩ levels in YAC128 MEFs (p Ͻ 0.05) but not in WT MEFs (p Ͼ 0.05) (Fig. 2, B and D), suggesting that the higher basal mitochondrial Ca 2ϩ levels seen in YAC128 MEFs also involved InsP 3 Rs.
There are three types of InsP 3 R (InsP 3 R1/2/3), and which type of InsP 3 R is responsible for mitochondrial loading is not known. We performed InsP 3 R1/2/3 siRNA knockdown experiments to address further the role of InsP 3 R in elevated mitochondrial calcium loading. The efficiency and selectivity of siRNA knockdown experiments were confirmed by Western blot with antibodies against InsP 3 R1/2/3 (Fig. 2F). We found that knockdown of InsP 3 R1 but not InsP 3 R2/3 significantly reduced matrix Ca 2ϩ loading in HD cells (Fig. 2, G-L), indicating that HTT exp -induced elevated mitochondrial Ca 2ϩ loading in HD cells was mainly mediated through type 1 InsP 3 R. This is in agreement with our previous findings, which show that The MCU is located in the mitochondrial inner membrane and is considered the primary route of mitochondrial Ca 2ϩ uptake (60). Consistent with the involvement of MCU in our experiments, preincubation of MEFs with the membrane-permeable MCU blocker, Ru360 (61), significantly decreased the amplitudes of BK-induced mitochondrial Ca 2ϩ transients in both WT (p Ͻ 0.05) and YAC128 MEFs (p Ͻ 0.01) (Fig. 2, C and  D), and it also reduced the amount of Ca 2ϩ loaded into the mitochondria of both WT (p Ͻ 0.01) and YAC128 MEFs (p Ͻ 0.01) (Fig. 2, C and E). The resting mitochondrial Ca 2ϩ levels were also significantly reduced in YAC128 MEFs in the presence of Ru360 (p Ͻ 0.01). Based on these results, we concluded that excessive mitochondrial Ca 2ϩ loading in HD cells involved both InsP3Rs and MCU.
Excessive ROS Generation in YAC128 HD Cells-Compelling evidence suggests that oxidative damage contributes to neurodegeneration in HD. However, the origins of this oxidative stress remain unclear. ROS are primarily produced as by-prod-ucts of mitochondrial respiration due to electrons leaking from the electron transfer chain, and accumulating data indicate that mitochondrial matrix Ca 2ϩ is a positive effector of the tricarboxylic acid (TCA) cycle (62) and ROS generation (38,39). The quantitative estimation of the mitochondrial matrix redox state in situ in live cells was not possible until the development of the highly sensitive superoxide indicator, cpYFP (40,63). We therefore used cpYFP imaging to quantitatively compare the mitochondrial superoxide flashes between WT and YAC128 MEFs to determine the functional consequences of excessive mitochondrial Ca 2ϩ loading in YAC128 HD cells (Fig. 3A). In resting cells, the amplitudes of spontaneous superoxide flashes was sig- nificantly higher in YAC128 MEFs compared with WT MEFs (0.266 Ϯ 0.015 versus 0.143 Ϯ 0.008, p Ͻ 0.01) (Fig. 3, A-C). Stimulation with 2.5 M BK significantly increased the amplitudes of superoxide flashes both in WT MEFs (0.263 Ϯ 0.026 versus 0.143 Ϯ 0.008 without BK, p Ͻ 0.01) and YAC128 MEFs (0.416 Ϯ 0.022 versus 0.266 Ϯ 0.015 without BK, p Ͻ 0.01) (Fig.  3, B and C). Furthermore, BK stimulated significantly higher amplitude flashes in YAC128 MEFs compared with WT MEFs (0.416 Ϯ 0.022 versus 0.263 Ϯ 0.026, p Ͻ 0.01) (Fig. 3, B and C). Because superoxide flashes occur randomly in space and time, we also quantified the number of flashes that occurred per min per cell to indicate the flash frequency. In resting cells, spontaneous superoxide flashes occurred at similar frequencies (flashes/min/cell) in WT (Fig. 3D). More importantly, BK stimulated significantly higher flash frequencies in YAC128 MEFs compared with WT MEFs at 2-4 min (p Ͻ 0.01) after BK application (Fig. 3D). In summary, the amplitudes and frequencies of both spontaneous and BK-stimulated superoxide flashes are significantly higher in YAC128 MEFs compared with WT MEFs, indicating that mitochondria in HD cells produce elevated levels of oxidative stress.

Role of Mitochondrial Ca 2ϩ Overloading in Excessive ROS
Generation in YAC128 HD Cells-In the above mitochondrial Ca 2ϩ and cpYFP imaging experiments, we demonstrated excessive mitochondrial Ca 2ϩ loading (Figs. 1 and 2) and elevated ROS production (Fig. 3) in HD MEF cells. We next explored whether the elevated ROS production was dependent on mitochondrial Ca 2ϩ influx. Pretreatment with 20 M 2-APB or 10 M Ru360 significantly attenuated the amplitudes of both spontaneous and BK-induced superoxide flashes in YAC128 MEFs compared with untreated control cells (p Ͻ 0.01) (Fig. 4 A). For WT MEFs, however, 2-APB and Ru360 caused significant reduction in the amplitudes of only the BK-induced superoxide flashes (p Ͻ 0.01) (Fig. 4A). As for flash frequency, both spontaneous and BK-induced superoxide flashes occurred at significantly lower frequencies in YAC128 MEFs in the presence of 2-APB or Ru360 (p Ͻ 0.01) (Fig. 4, B and C).
Specific knockdown of InsP 3 R1 significantly reduced the amplitudes and flash frequencies of both spontaneous and BKinduced superoxide flashes in YAC128 MEFs compared with siRNA-control cells (p Ͻ 0.01) (Fig. 4, D and E), whereas knockdown of InsP 3 R2/3 failed to elicit any significant effect on superoxide flash generation (Fig. 4, F and G), indicating the important role of type 1 InsP 3 R in augmenting oxidant stress in HD. All these results together demonstrate that elevated oxidative stress is critically dependent on excessive mitochondrial Ca 2ϩ uptake in HD cells.
Disrupted Mitochondrial Ca 2ϩ Signaling in YAC128 MSNs-To determine whether mitochondrial Ca 2ϩ signaling is disturbed in HD neurons, we established MSN cultures from WT and YAC128 HD mice and compared the mitochondrial Ca 2ϩ signaling dynamics between WT MSNs and YAC128 MSNs. DHPG, a selective group I metabotropic glutamate receptor agonist that can trigger InsP 3 production in MSNs, was used to stimulate intracellular Ca 2ϩ mobilization in MSNs. Similar to the mitochondrial Ca 2ϩ results in MEFs, in resting MSNs, the basal mitochondrial Ca 2ϩ concentration before DHPG application in YAC128 MSNs was significantly higher than that in WT MSNs (p Ͻ 0.05) (Fig. 5, A-C). The difference in mitochondrial Ca 2ϩ concentration between WT and YAC128 MSNs was even more pronounced during intracellular Ca 2ϩ mobilization. Treatment with 50 M DHPG caused much larger mitochondrial Ca 2ϩ transients in YAC128 MSNs than in WT MSNs (p Ͻ 0.01) (Fig. 5, A-C). To quantify the amount of Ca 2ϩ that was loaded into mitochondria during DHPG stimulation, we also calculated the peak area (M⅐s) below each Ca 2ϩ transient trace. Significantly more Ca 2ϩ was loaded into mitochondria in YAC128 MSNs than in WT MSNs (547 Ϯ 207 M⅐s versus 135 Ϯ 56.7 (M⅐s, p Ͻ 0.01) (Fig. 5D). Therefore, mitochondria in YAC128 HD neurons show a significantly higher level of Ca 2ϩ loading compared with WT neurons, further indicating that mitochondrial Ca 2ϩ signaling is dysregulated in HD neurons.
Excessive ROS Generation in YAC128 HD Striatal Neurons-Excessive mitochondrial Ca 2ϩ uptake caused elevated oxidative stress in YAC128 HD MEFs (Figs. 3 and 4). Next, we determined whether this was also true in YAC128 HD striatal medium spiny neurons. In resting MSNs, the amplitude of spontaneous superoxide flashes was significantly higher in YAC128 MSNs compared with that in WT MSNs (p Ͻ 0.01) (Fig. 6, A-C). Stimulation with 50 M DHPG significantly increased the amplitudes of superoxide flashes both in WT MEFs (p Ͻ 0.01) and YAC128 MEFs (p Ͻ 0.01) compared with resting cells (Fig. 6, B and C). In addition, DHPG caused the amplitude of flashes to be significantly higher in YAC128 MSNs compared with WT MSNs (p Ͻ 0.01) (Fig. 6, B and C). Similarly to MEF cells, superoxide flashes in neurons also occur randomly. We monitored 5 min of neuronal flash events and cal-  FEBRUARY 1, 2013 • VOLUME 288 • NUMBER 5 culated the number of flashes occurring per neuron in 1 min to indicate the flash frequency. A striking difference in flash frequency between WT and YAC128 MSNs was already seen in resting cells (p Ͻ 0.01) (Fig. 6D). Stimulation with DHPG significantly increased the frequencies of superoxide flashes both in WT MSNs (p Ͻ 0.01) and YAC128 MSNs (p Ͻ 0.01) (Fig.  6D). Furthermore, DHPG induced significantly higher flash frequencies in YAC128 MSNs compared with WT MSNs (p Ͻ 0.01) (Fig. 6D). Therefore, the amplitudes and frequencies of both spontaneous and DHPG-stimulated superoxide flashes are significantly higher in YAC128 MSNs compared with WT MSNs, indicating that mitochondria in HD striatal neuronal cells produce elevated levels of oxidative stress.

Mitochondrial DNA Damage in Huntington Disease
Disrupted Mitochondrial Ca 2ϩ Signaling and Excessive ROS Generation in HD Patient Fibroblasts-Mitochondrial Ca 2ϩ signaling dynamics and superoxide flashes were also visualized in human fibroblasts from healthy individuals and HD patients. Similar to YAC128 MSNs, HD patient fibroblasts also showed a significantly higher level of mitochondrial Ca 2ϩ loading compared with healthy fibroblasts (Fig. 7, A-C), indicating that mitochondrial Ca 2ϩ signaling is also disrupted in HD patient fibroblasts. Furthermore, the amplitudes and frequencies of both spontaneous and BK-stimulated superoxide flashes were significantly higher in HD patient fibroblasts compared with MSNs was significantly higher than that in WT MEFs (n ϭ 10) (**, p Ͻ 0.01). In response to DHPG, both WT (n ϭ 11) and YAC128 MSNs (n ϭ 7) showed significantly higher superoxide flash amplitudes compared with spontaneous flashes in resting MSNs (**, p Ͻ 0.01), and DHPG caused significantly higher superoxide flash amplitudes in YAC128 MSNs (n ϭ 7) compared with those in WT (n ϭ 11) MSNs (**, p Ͻ 0.01). D, in resting MSNs, spontaneous mitochondrial superoxide flashes occurred at a significantly higher frequency in YAC128 neurons than in WT neurons (**, p Ͻ 0.01). In response to DHPG, superoxide flashes in both WT (n ϭ 11) and YAC128 MSNs (n ϭ 7) occurred at a significantly higher frequency when compared with resting MSNs (**, p Ͻ 0.01), and DHPG induced a significantly higher flash frequency in YAC128 MSNs (n ϭ 7) compared with that in WT MSNs (n ϭ 11) (*, p Ͻ 0.05). Flashes occurring 5 min before and 5 min after DHPG application were averaged for superoxide flash frequency analysis. Similar results were obtained in three independent experiments. Data are shown as the mean Ϯ S.E. healthy fibroblasts (Fig. 8, A-D), demonstrating that mitochondria in HD patient fibroblasts produce elevated oxidative stress.
Mitochondrial Ca 2ϩ Loading Induces Elevated Mitochondrial Genomic DNA Damage in HD Cells-Age-dependent increases in mitochondrial DNA damage have been reported in the 3-nitropropionic acid-induced HD model and in R6/2 HD mice (64), although the underlying mechanism is not clear. Based on our results described above, excessive mitochondrial Ca 2ϩ loading-dependent ROS generation might be responsible. We employed a quantitative PCR technique to check whether mitochondrial Ca 2ϩ loading causes oxidative damage to mtDNA. The qPCR assay is based on the principle that DNA lesion blocks the progression of the DNA polymerase during PCR, resulting in decreased amplification of the fragment of interest. Thus, the amount of PCR amplification is inversely proportional to the extent of DNA damage on a given template. BK-stimulated WT and YAC128 MEFs show significant decreases in relative amplification, 50 and 69%, respectively, compared with control groups (Fig. 9, A and C), indicating damage to mtDNA. Pretreatment with MCU blocker, Ru360, or the mitochondrion-targeted antioxidant peptide, SS31, almost completely prevented the BK-induced decrease in relative amplifications (Fig. 9, A and C). The frequency of mtDNA lesions in BK-stimulated MEFs was 0.70 lesions/10 kb for WT and 1.29 lesions/10 kb for YAC128, indicating that BK induces an almost 2-fold higher level of mtDNA damage in YAC128 HD MEFs compared with WT MEFs during matrix Ca 2ϩ loading (Fig. 9D). Ru360 can significantly reduce the BK-stimulated mtDNA damage to 0.18 lesions/10 kb for WT and 0.29 lesions/10 kb for YAC128, and SS31 significantly reduces the BK-stimulated mtDNA damage to 0.18 lesions/10 kb for WT and 0.37 lesions/10 kb for YAC128, demonstrating that the BKstimulated increase in mtDNA damage is mitochondrial matrix Ca 2ϩ loading-dependent and matrix oxidant-dependent (Fig.  9D). Our mtDNA damage assay data clearly revealed that disturbed mitochondrial Ca 2ϩ signaling in HD cells causes excessive oxidant generation, which in turn leads to more serious mtDNA oxidative damage in HD cells. It should be noted that this repetitive BK-stimulated increase in mtDNA lesions may occur transiently, and most of the mtDNA damage could be repaired by the base excision repair pathway (65,66). However, a significantly higher level of mtDNA damage in HD cell is expected to accumulate mtDNA lesions much faster than WT cells.

DISCUSSION
The mechanisms causing elevated oxidative stress in HD neurons remain elusive. By using a highly sensitive mitochondrial matrix-targeted superoxide indicator, cpYFP, which can sense bursts of superoxide generation in situ in living cells (40), we show that, even in resting cells, mitochondria in HD cells exhibit strikingly higher mitochondrial Ca 2ϩ loading and oxidant generation, which in turn causes a significantly higher rate of mtDNA damage accumulation in HD cells. Steady buildup of oxidative damage to mitochondrial DNA could eventually result in mitochondrial dysfunction and be central to the etiology of HD.
Mitochondria are not only central to energy metabolism but are also a very efficient Ca 2ϩ buffer system. Mitochondria are spatially and functionally organized in a network of dynamically interconnected organelles and are often in close contact with the ER and intracellular Ca 2ϩ -releasing channels (57). Once Ca 2ϩ is released from the ER into the cytoplasm, mitochondria take up a portion of the released Ca 2ϩ primarily via the MCU at the expense of mitochondrial membrane potential (⌬⌿ m ) (67). Recently, a number of studies have indicated that toxicity of HTT exp may involve disturbed glutamate-induced Ca 2ϩ signaling in HD medium spiny neurons (15)(16)(17)(18)(19)(20)(21)(22)68). InsP 3 R1 is an C, in resting cells, the amplitude of spontaneous mitochondrial superoxide flashes in HD patient fibroblasts (n ϭ 18) was significantly higher than that in healthy fibroblasts (n ϭ 17) (**, p Ͻ 0.01). In response to BK, both healthy (n ϭ 15) and HD patient fibroblasts (n ϭ 16) showed significantly higher amplitudes of superoxide flashes when compared with spontaneous flashes in resting MEFs (**, p Ͻ 0.01), and BK caused significantly higher amplitudes of superoxide flashes in HD patient fibroblasts (n ϭ 16) than in healthy (n ϭ 15) fibroblasts (**, p Ͻ 0.01). D, in resting cells, spontaneous mitochondrial superoxide flashes occurred at a similar frequency in healthy and HD patient fibroblasts (p Ͼ 0.05). In response to BK, superoxide flashes in both healthy and HD patient fibroblasts occurred at a significantly higher frequency at 2 min after BK treatment when compared with resting cells (*, p Ͻ 0.05; **, p Ͻ 0.01). Data are shown as the mean Ϯ S.E. intracellular Ca 2ϩ -releasing channel located on the ER, and HTT exp binds directly and specifically to the InsP 3 R1 C-terminal region and facilitates the sensitivity of InsP 3 R1 to activation by InsP 3 (22). It is conceivable that mitochondrial Ca 2ϩ loading is excessive in HD striatal neurons due to the potentiation of Ca 2ϩ releasing activity of InsP 3 R1. Indeed, we found that the amount of mitochondrial Ca 2ϩ loading in HD cells was 2-4fold higher than in WT MEFs (Fig. 1) or WT MSNs (Fig. 5) during intracellular Ca 2ϩ mobilization. Our evidence also demonstrated that Ca 2ϩ release via InsP 3 R1 (Fig. 2) and the mitochondrial Ca 2ϩ uniporter (Fig. 2) are all important for excessive Ca 2ϩ loading into mitochondria. Our data indicate that disturbed cytosolic Ca 2ϩ mobilization can directly cause disrupted mitochondrial Ca 2ϩ signaling in HD neurons.
Altered mitochondrial Ca 2ϩ handling in HD cells has been previously implicated in HTT exp toxicity (69). Mitochondria isolated from HD patient lymphoblasts and from transgenic HD mouse brain have a lower membrane potential and are more sensitive to Ca 2ϩ -mediated mPTP opening (70), and mitochondria in MSNs from HD mice have enhanced loss of ⌬⌿ m and enhanced mPTP opening in response to acute glutamate or NMDA stimulation compared with wild-type control mice (23,51,71). In this study, by monitoring mitochondrial matrix Ca 2ϩ dynamics in situ and in real time, we provide direct evidence of a strikingly higher level of mitochondrial Ca 2ϩ uptake during Ca 2ϩ mobilization in HD neurons. Furthermore, different from our previous basal cytosolic Ca 2ϩ level measurement results, we here observed significantly higher basal mitochondrial Ca 2ϩ levels in HD cells (including MEFs, MSNs, and human fibroblasts) compared with those in WT/healthy cells even without stimulation. This suggested that InsP3R is more active at the resting state in HD cells than in WT cells and that this extra released Ca 2ϩ was largely loaded into mitochondria and elevated the basal mitochondrial Ca 2ϩ levels in HD cells. This is in agreement with a recent finding that HD cells have higher store-operated Ca 2ϩ entry activity than WT cells (72), implying a constant level of ER depletion (in other words, Ca 2ϩ release from ER) in HD cells. Besides the direct effect of HTT exp on mitochondrial Ca 2ϩ retention capacity (70), our observations of excessive mitochondrial matrix Ca 2ϩ loading provide D, comparison of mtDNA lesion frequencies from WT and YAC128 MEFs. The mean lesion frequency per 10 kb per strand was calculated using the Poisson equation as described under "Experimental Procedures." In response to the twice for 10 min BK treatment, both WT (n ϭ 6) and YAC128 MEFs (n ϭ 6) showed significantly higher mtDNA lesion frequencies compared with control MEFs (p Ͻ 0.01), and BK caused significantly higher mtDNA lesion frequencies in YAC128 MEFs than in WT MEFs (**, p Ͻ 0.01). Pretreatment with MCU blocker, Ru360, and mitochondrion-targeted antioxidant peptide, SS31, significantly reduced BK-induced elevation of mtDNA lesion frequencies in both WT and YAC128 MEFs (**, p Ͻ 0.01). Similar results were obtained in three independent experiments. Data are shown as mean Ϯ S.E.
another explanation for the mitochondrial Ca 2ϩ defect observed in HD cells, and higher matrix Ca 2ϩ loading could result in higher probability of mPTP open to release matrix Ca 2ϩ into cytosol in HD cells.
Mitochondria constitute the major ROS generator in eukaryotic cells, including neurons. ROS are primarily produced as by-products of mitochondrial respiration due to electrons leaking from the electron transfer chain. It has long been of great interest to quantitatively monitor mitochondrial matrix ROS production in situ in live cells. By using the recently developed cpYFP, a highly sensitive superoxide indicator (40,41), we quantitatively estimated mitochondrial superoxide generation in situ in HD MEFs and MSNs. HD cells exhibit significantly higher superoxide flash amplitudes and frequencies compared with WT cells (Figs. 3, 6, and 8), suggesting that mitochondria in HD cells are producing excessive ROS. More importantly, the elevated ROS generation in HD cells is mitochondrial matrix Ca 2ϩ -dependent, because higher basal mitochondrial Ca 2ϩ levels and greater mitochondrial Ca 2ϩ loading in HD cells promoted ROS generation, whereas blockage of mitochondrial Ca 2ϩ loading diminished the elevated superoxide flashes seen in HD cells (Figs. 3 and 4), indicating the causal link between increased mitochondrial matrix Ca 2ϩ loading, elevated superoxide generation, and mtDNA damage in HD cells/neurons. Striatal neurons are most affected in HD. The striatal neurons particularly accumulate more mitochondrial genomic DNA (mtDNA) mutations than any other brain cells (27), suggesting that there is a brain regional specificity of oxidative damage in the HD patient. It has been reported that, due to the preferential expression of metabotropic glutamate receptor 1/5 (coupling to phospholipase C-inositol 1,4,5-trisphosphate pathway that can trigger InsP 3 R open) and NR2B NMDA receptor in striatal MSN, Htt exp preferentially augments intracellular Ca 2ϩ signaling in striatal MSN (18,21,22). Moreover, it has been reported that InsP 3 R-mediated calcium signaling can be further enhanced by the dopamine D 1 -class receptor that is primarily expressed in striatal MSN (73). Therefore, excessive mitochondrial Ca 2ϩ loading and elevated mitochondrial oxidant stress are expected to occur selectively in the striatal region, which may explain why striatal neurons accumulate more mtDNA mutations than other brain cells.
Exactly how mitochondrial matrix Ca 2ϩ loading stimulates ROS production remains an open question. Our results show that an increase in mitochondrial Ca 2ϩ uptake was closely correlated with higher mitochondrial superoxide flash generation (Figs. 3, 4, 6, and 8). This is in agreement with recent findings that Ca 2ϩ influx through L-type Ca 2ϩ channels during pacemaking increases mitochondrial Ca 2ϩ loading and oxidant stress in substantia nigra pars compacta dopaminergic neurons (74). One possibility is that matrix Ca 2ϩ loading induces ROS production simply by stimulating the tricarboxylic acid cycle and increasing electron flux (38). Another possibility is that matrix Ca 2ϩ loading induces transient openings, or flickers, of the mPTP, and a brief opening of the mPTP has been reported to stimulate ROS generation (40,75). The frequency of brief mPTP opening is primarily determined by the matrix Ca 2ϩ concentration (76). Therefore, higher matrix Ca 2ϩ loading may induce a higher frequency of brief mPTP opening and higher superoxide flash generation in HD mitochondria.
As an ROS marker, superoxide flashes are bursts of superoxide generation in single mitochondria and represent the oxidative stress in mitochondria (40,63). ROS play an important role in physiological cellular functions, such as contraction, secretion, and metabolism, but excessive ROS signals are detrimental to nucleic acids, lipids, proteins, and carbohydrates. There is a delicate balance in mitochondria between ROS generation and ROS removal to maintain a physiological oxidative homeostasis. Both overproduction of ROS and/or a deficit in the ROS defense system could disrupt this balance and cause oxidative stress. This study revealed that, due to the excess mitochondrial Ca 2ϩ loading, mitochondria in HD neurons are producing significantly higher levels of ROS. Recently, Li et al. (77) showed that impairment of cysteine uptake led to a decline of antioxidant synthesis in HD neurons expressing full-length HTT exp protein. Therefore, both overproduction of ROS and deficiency in ROS defense systems may account for the elevated oxidative stress in HD neurons.
Mitochondria are not only the major ROS producer but are also the main target of ROS. Overproduction of ROS in HD neurons causes oxidative damage to mitochondrial nucleic acids, proteins, and lipids, which in turn impair the function of mitochondria. Naked mitochondrial genomic DNA (part of which encodes subunits of complex I, III, IV, and V) and the iron-sulfur cluster-containing proteins, including the succinate dehydrogenase B subunit of complex II and the Rieske protein of complex III, are particularly vulnerable to oxidative stress (30,78,79). Postmortem studies of symptomatic HD patients revealed a drastic deficiency of complexes II and III and a lesser deficiency of complex IV in the caudate or putamen, with relatively normal levels in the frontal cortex or cerebellum (31,32). Studies with HD cell models expressing Htt exon-1 containing elongated CAG repeats showed decreased complex II enzymatic activity and deficiency of complex III (33,80,81). These observations suggest that impairment of mitochondrial function is important in the progression of the disease. However, no significant deficiency of respiratory chain complexes has been demonstrated in presymptomatic patients or in HD model mice expressing full-length HTT exp , suggesting that respiratory chain defects are a secondary feature in HD pathogenesis (34,35). Based on our results showing excessive mitochondrial ROS generation and greater mtDNA damage in HD cells/neurons (Figs. 3, 4, 6, 8, and 9), it is plausible that excess Ca 2ϩ loadingdependent mitochondrial oxidant stress could be a primary contributor to the respiratory chain complex defects. The appearance of mitochondrial dysfunction in advanced HD may be the result of a gradual accumulation of oxidative damage to mitochondrial proteins, DNA, and lipids due to excessive ROS production. Indeed, in this study, for the first time, we show that repetitive BK application led to increased damage of mitochondrial genomic DNA in HD cells (Fig. 9). This observation strongly supports the role of elevated ROS levels in damaging mitochondrial genomic DNA in HD. Because the mitochondrial oxidant stress in HD cells/neurons was attenuated by exposure to an InsP 3 R blocker, an MCU blocker (Fig. 4), and a mitochondrial antioxidant (Fig. 9), our data also suggest that a