Calcium Homeostasis and Mitochondrial Dysfunction in Striatal Neurons of Huntington Disease*

Dysfunctions of Ca2+ homeostasis and of mitochondria have been studied in immortalized striatal cells from a commonly used Huntington disease mouse model. Transcriptional changes in the components of the phosphatidylinositol cycle and in the receptors for myo-inositol trisphosphate-linked agonists have been found in the cells and in the striatum of the parent Huntington disease mouse. The overall result of the changes is to delay myo-inositol trisphosphate production and to decrease basal Ca2+ in mutant cells. When tested directly, mitochondria in mutant cells behave nearly normally, but are unable to handle large Ca2+ loads. This appears to be due to the increased Ca2+ sensitivity of the permeability transition pore, which dissipates the membrane potential, prompting the release of accumulated Ca2+. Harmful reactive oxygen species, which are produced by defective mitochondria and may in turn stress them, increase in mutant cells, particularly if the damage to mitochondria is artificially exacerbated, for instance with complex II inhibitors. Mitochondria in mutant cells are thus peculiarly vulnerable to stresses induced by Ca2+ and reactive oxygen species. The observed decrease of cell Ca2+ could be a compensatory attempt to prevent the Ca2+ stress that would irreversibly damage mitochondria and eventually lead to cell death.

Huntington disease (HD) 2 is a fatal disease characterized by chorea and psychiatric disturbance (1) caused by the expansion of CAG repeats in the first exon of the gene encoding huntingtin (Htt). In the normal gene, the repeats specify a stretch of up to 36 Gln in the N-terminal region of Htt. In the disease, the poly-Q tract is longer, and the mutant protein becomes harmful to cells. Htt is expressed ubiquitously in human tissues, but its mutation is particularly harmful to cortical and striatal medium size spiny neurons (MSNs) (2). The reasons for the specific damage to these neurons is an open problem in HD research, which reflects the incomplete knowledge on the function(s) of Htt. Htt interacts with several proteins in neurons, and plays roles in processes like axonal transport, regulation of transcription, exocytosis, calcium homeostasis, and prevention of apoptosis (3). It is easy to see that dysfunctions in any one of these processes, e.g. in Ca 2ϩ homeostasis, could be involved in the ethiology of HD (4). As for the molecular mechanisms of the harmful effects of the mutated protein, the idea is now gaining ground that the extended poly-Q tract, cleaved off from Htt, causes the transcriptional dysfunction of genes that are essential for neuronal survival. Htt, however, could also have non-transcriptional effects.
Mitochondrial defects may have a causative role in neurodegenerative diseases, and are considered important in HD ethiology (5). They have indeed been found in the brain of HD patients (6) and in the striatal and other cells of animal HD models (7)(8)(9). The defective mitochondrial component has been suggested to be complex II of the respiratory chain. The suggestion is supported by experimental findings, for example the demonstration that the specific inhibitor 3-nitropropionic acid (3-NPA) induces a degeneration of rat striatal neurons that mimics that seen in the disease (10,11). The alterations in Ca 2ϩ homeostasis commonly observed in neuronal damage (12,13) could also have a role in the HD ethiology. In neurons, the mobilization of Ca 2ϩ from the endoplasmic reticulum (ER) through channels modulated by InsP 3 produced in the phosphatidylinositol (PI) cycle is an important component of Ca 2ϩ homeostasis. In the striatal neurons studied here the cycle is only activated by two plasma membrane agonists: ATP and bradykinin (BK). ATP produces InsP 3 through fast acting ionotropic (P2X) receptors, and slower acting, G q protein-coupled P2Y receptors (14). The receptors for BK are all coupled to G q proteins (15). The alterations of Ca 2ϩ homeostasis, if resulting for instance in the increased release of Ca 2ϩ from the ER, could exacerbate the mitochondrial dysfunction, as the complex II defect would limit the ability of the respiratory chain to increase activity, presumably making mitochondria peculiarly unable to retain the Ca 2ϩ they have accumulated (16). However, the inability to retain Ca 2ϩ could also be due to the increased pro-pensity of the permeability transition pore to open: recent findings have shown that the membrane potential (⌬ m ) in mitochondria of cells expressing mutant huntingtin is peculiarly sensitive to Ca 2ϩ (17), and that poly-Q constructs de-energize isolated mitochondria exposed to Ca 2ϩ (17).
Reactive oxygen species (ROS) also damage mitochondria, i.e. they dissipate the ⌬ m , and are thus frequently mentioned in the pathogenesis of neurodegenerative diseases. They are produced by defective mitochondria, and in HD neurons they could be generated by the dysfunction of complex II. This is suggested by the finding that in normal neurons they increase following the inhibition of the complex by 3-NPA (18). An interesting development in the area of ROS has been the observation (19,20) that the striatal neurons are particularly sensitive to the mitochondrial damage caused by the lack of a co-activator of the transcription of the genes for ROS-scavenging enzymes (acronym PGC-1␣). PGC-1␣ regulates a number of cell processes, among them the response of mitochondria to oxidative stress. PGC-1␣ is down-regulated in the striatum of HD patients (21), in striatal neurons of HD knock-in mice, and even in the immortalized striatal cells used in this work. Mutant Htt has been found to associate with the promoter of the PGC-1␣ gene, decreasing its transcription. As a result, the transcription of ROS-scavenging enzymes is down-regulated, and the concentration of ROS in the neurons increases. These important observations were reinforced by the finding that striatal neurons from HD knock-in mice were resistant to 3-NPA when expressing exogenous PGC-1␣ (22).
To integrate the effects of Ca 2ϩ and ROS in the neuronal damage in HD, we have studied their dynamics in a commonly used HD model, i.e. in immortalized striatal neuron precursor cells from a HD knock-in mouse model (KI-Hdh Q111 ) (23), and in the striata of the parent model mice. The immortalization affects cells growth and differentiation, but important biochemical mechanisms of primary brain cells are faithfully conserved in the cell line, and have allowed the discovery of interesting paths affected by the Htt mutation (reviewed in Ref. 24). We have found that the expression of the enzymes of the PI cycle, and of the receptors of the two agonists that activate it in this cell model, was profoundly altered in the mutant cells and in the striatum of the parent mice. As a result, the basal Ca 2ϩ level in the cells decreased, and the production of InsP 3 that would mediate its liberation in response to agonist stimulation, was delayed. The handling of Ca 2ϩ by mitochondria was studied directly within the mutant cells, and found to be compromised, but only when the Ca 2ϩ loads presented to the mitochondria were large. The inability of mitochondria to absorb the Ca 2ϩ -induced stress was exacerbated if the damage to complex II, which could be possibly compromised in the cell line, was artificially augmented with inhibitors. The transcription of ROS-degrading enzymes was also examined: in three clones of the mutant cells, which were examined, it was found to be consistently up-regulated. That of PGC-1␣, instead, varied in the three mutant clones. The concentration of ROS in the mutant cells did not differ significantly from the controls, but increased markedly in both cell types if mitochondria were artificially damaged, for instance by exposing them to the complex II inhibitor 3-NPA. The increase was much more pronounced in mutant cells. Thus, mitochondria in mutant cells (and, as is plausible to assume, in HD neurons) appear to be peculiarly vulnerable to stress. The changes in Ca 2ϩ homeostasis in the cells studied here could reflect a compensatory attempt to limit, or at least to delay, mitochondrial damage.

Ca 2؉ Measurements
Aequorin-STHdh Q7 and STHdh Q111 cells were transfected with plasmid DNAs encoding cytosolic (cytAEQ), mitochondrial (mtAEQ), or endoplasmic reticulum (erAEQ) aequorin using TransFectin Lipid Reagents (Bio-Rad Laboratories, Hercules, CA). Ca 2ϩ measurements were performed as described elsewhere (25). All measurements were carried out at 37°C. The light signal was collected and stored in an IBM-compatible computer for further analysis. The luminescence data were converted off-line into [Ca 2ϩ ] values, using a previously described computer algorithm (25).
Fura-2-For Fura-2 measurements cells were plated on 24-mm coverslips in 6-well plates at a density of 400,000 cells per well. 24-h later, cells were loaded with 5 M Fura-2 AM (Invitrogen, San Giuliano Milanese, Italy) in KRB (Krebs-Ringer modified buffer: 125 mM NaCl, 5 mM KCl, 1 mM Na 3 PO 4 , 1 mM MgSO 4 , 5.5 mM glucose, 20 mM HEPES, pH 7.4) supplemented with 1 mM Ca 2ϩ (KRB-Ca) for 30 min at room temperature. After washing with KRB-Ca cells were left for additional 30 min at room temperature for de-esterification of Fura-2. The coverslip was then placed on the stage of a Zeiss Axiovert 100 epifluorescence microscope, equipped with a 16-bit digital CCD videocamera (Micromax, Princeton Instruments, Trenton, NJ). Samples were alternatively illuminated at 340 and 380 nm, and the emitted light (filtered with an interference filter centered at 510 nm) was collected by the camera. Images were acquired using the MetaFluor software (Universal Imaging, West Chester, PA). The ratio values (1 ratio image/sec) were calculated off-line, after background subtraction from each single image. To quantify the differences in the peaks of the responses the data were normalized using the formula F i (F 0 -F i ) commonly used in experiments of this type. The data on basal Ca 2ϩ levels are expressed as 340/380 ratio values.

InsP 3 Quantification
For InsP 3 quantification, cells were stimulated with ATP (100 M) and/or BK (100 nM). The reaction was arrested at the indicated time-points with one-third volume of 1.05% PCA (Sigma). 1-h later extracts were collected and subjected to an AlphaScreen InsP 3 assay according to the manufacturer's protocol (AlphaScreen-GST and AlphaScreen InsP 3 supplement, Perkin Elmer, Wellesley, MA).

RNA Preparation and Reverse Transcription
STHdh Q7 and STHdh Q111 cells were plated in 100-mm dishes and grown to 80 -90% confluency. They were washed three times with cold phosphate-buffered saline and collected with 1 ml of TRIzol reagent (Invitrogen). Total RNA was extracted according to the manufacturer's protocol. Striata from KI-Hdh Q111 and WT-Hdh Q7 mice were dissected, homogenized in 1 ml of TRIzol reagent, and total RNA was isolated according to the manufacturer's instructions. 1 g of total RNA was reverse-transcribed using Moloney murine leukemia virus reverse transcriptase (Invitrogen) according to the manufacturer's protocol.

Primers Design and Q-RT-PCR
Specific oligonucleotide primers were designed using Prim-er3 software (see supplemental Table S1). Q-RT-PCR was performed on a Rotor-Gene 3000 platform (Corbet Research, Sydney, Australia). An amount of cDNA corresponding to 1-10 ng of total RNA was amplified in 25 l of a mixture containing 12.5 l of Platinum SYBR Green qPCR SuperMix-UGD (Invitrogen), 2 l of primers mixture (2.5 M each). The PCR cycling parameters were: 94°C for 7 min, 45 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 15 s. The relative amount of amplified DNA was calculated as described (26) using hypoxanthineguanine phosphoribosyltransferase mRNA as endogenous control.

Measurement of Mitochondrial Membrane Potential (⌬ m ) and of ROS
The TMRM "diffusion" method was used, which is adequate for the comparison of the ⌬ m between two populations of cells (27). The cells were loaded with 10 nM TMRM for 30 min at 37°C in KRB containing 1 mM Ca 2ϩ . TMRM fluorescence was registered at 510 nm using an Olympus F-View II CCD camera mounted on an Olympus IX-81 microscope equipped with ϫ40 Uplan FLN objective (Olympus, Tokyo, Japan). The TMRM fluorescence intensity was analyzed off-line using MetaMorph software (Universal Imaging). In the experiments with FCCP, ROIs were positioned across the peripheral cell area, and the standard deviation of TMRM fluorescence was analyzed using MetaMorph software before and after FCCP addition. To measure ROS, STHdh Q7 and STHdh Q111 cells were treated with 3-NPA (10 mM for 40 h), and loaded with 10 M 2Ј,7Ј-dichlorodihydrofluorescein diacetate acetyl ester (Invitrogen) for 30 min at 37°C in 100 l of KRB containing 1 mM Ca 2ϩ followed by 30 min of de-esterification. Fluorescence was measured in a Fluoroskan spectrophotometer (Ascent FL, Labsystems, Thermo Electron, Waltham, MA). After background subtraction, the data were normalized to the cell number obtained by counting of nuclei stained by Hoechst 33258 (Sigma).

Lentiviral Transduction with mitAEQ and Measurement of Mitochondrial Ca 2؉ Uptake
To produce plasmid pLV-mitAEQ-IRES-EGFP, mitochondrial AEQ was first subcloned in a pIRES2-EGFP (Clontech, Mountain View, CA) vector, then a cassette containing mitAEQ-IRES-EGFP was transferred to a pRRLsin.PPTs.hCMV.GFPpre lentiviral vector. The resulting construct was denominated as pLV-mtAEQ. Lentiviral particles were produced as described elsewhere (28). The desired lentiviral titer was obtained by infecting STHdh Q7 and STHdh Q111 cells with serial dilutions of the lentiviral stock. The minimal dilution adequate for 100% infection was used. 48 -96 h after the infection AEQ was reconstituted as described in a section above. The cells were permeabilized with an intracellular buffer (100 mM KCl, 1 mM KH 2 PO 4 , 5 mM sodium succinate, 1 mM MgCl 2 , 1 mM ATP, 20 mM HEPES; pH 7.0) supplemented with 10 M digitonin and 50 M EGTA. Cells were then perfused with Ca 2ϩ -EGTA buffers containing the indicated concentrations of free Ca 2ϩ (1.6 -71 M). FCCP (Sigma) was added in perfusion solutions at the indicated concentrations. A WebMaxLight software was used to prepare the Ca 2ϩ -EGTA buffers.

Caspase Activity Measurement
Cells were plated in 60-mm dishes at the density of 1.2 million cells per dish. 24 h after plating, the cells were treated with 10 M staurosporin for 6 h. The activity of caspase-3 was detected with the Caspase-3 Colorimetric Activity Assay kit (Chemicon International, Temecula, CA) according to the manufacturer's instructions. (29). They were challenged with ATP in a medium supplemented with 100 M EGTA to exclude Ca 2ϩ penetration through ionotropic P2X receptors (which are also present in the cells) and store-operated channels. ATP induced a rapid cytoplasmic Ca 2ϩ increase in STHdh Q7 (control) cells peaking at 1.62 Ϯ 0.17 M (Fig. 1A), and decaying to base line in about 50 s. In STHdh Q111 (mutant) cells the peak was much lower (0.67 Ϯ 0.15 M, p ϭ 61e-8, Fig.  1A). The dynamics of the mitochondrial Ca 2ϩ pool was explored directly with mtAEQ: the peak of the transient triggered by ATP was much higher in STHdh Q7 than in STHdh Q111 cells (45.10 Ϯ 16.76 M versus 11.24 Ϯ 1.81 M, p ϭ 6e-5, Fig. 1B).

InsP 3 -linked Ca 2ϩ Dynamics in Striatal Cells-The plasma membrane of striatal cells contains P2Y receptors
In contrast to ATP, the Ca 2ϩ response to BK was higher in STHdh Q111 (1.40 Ϯ 0.22 M with respect to 1.20 Ϯ 0.17 M in STHdh Q7 cells, p ϭ 0.01, Fig. 1C). In mitochondria the peaks were 35.44 Ϯ 9.94 M in control cells and 71.86 Ϯ 18.76 M, p ϭ 1.8e-8, in mutant cells (Fig. 1D).
The Ca 2ϩ content of the ER had no role in the differences between the cytoplasmic and mitochondrial responses in control and mutant cells, as the steady-state Ca 2ϩ level measured with erAEQ did not differ significantly between STHdh Q7 and STHdh Q111 cells (225.73 Ϯ 37.77 M and 202.65 Ϯ 42.71 M, respectively, p ϭ 0.1) (Fig. 1E).

Altered Expression of Purinergic and Bradykinin Receptors-
The simplest explanation for the differences in ATP and BKmediated Ca 2ϩ signaling would be the different amounts of receptors expressed in control and mutant cells. The differences, however, could instead have been due to the clonal nature of the cells. The expression level of the two receptors was thus analyzed in the clone of Fig. 1 and in two additional clones (STHdh Q111-2 and STHdh Q111-3 ).
The transcripts of the two P2Y receptor subtypes present in mouse (P2Y1 and P2Y2, Ref. 30) were indeed detected in the cells (Fig. 2, A and B). Q-RT-PCR with primers specifically designed to obtain amplicons of about 100 bp revealed a striking (about 20-fold) down-regulation of the transcription of the P2Y1 gene and a less pronounced down-regulation of the P2Y2 gene (Fig. 2, E and F) in all mutant cell lines.
The transcripts of the B1 and B2 kinin receptors were instead up-regulated (Fig. 2, C and D). In the Q-RT-PCR analysis, the B1 transcript was 3-12-fold higher in the three mutant cell lines, that of the B2 receptor 1.4 -1.6-fold higher (Fig. 2, G  and H).
Ca 2ϩ Homeostasis Dysfunction in STHdh Q111 Cells-The AEQ assay shown in Fig. 1 averaged Ca 2ϩ transients in the total population of cells expressing AEQ. The differences between control and mutant cells could thus have been due to the reduced (or enhanced) number of responding cells. Because the amount of AEQ consumed by released Ca 2ϩ was normalized to the total amount of expressed AEQ, if the population of cells responding to stimulation had been different in control and mutant cells, a correspondingly smaller portion of AEQ would have been used for the normalization. It was thus important to control the Ca 2ϩ changes in single cell.
The analysis of the images in single cells loaded with Fura-2 ( Fig. 3) revealed that ATP induced Ca 2ϩ transients in 88% of STHdh Q7 cells, but only in 54% of the mutant STHdh Q111 cells (Fig. 3D). The peaks of the transients in responding mutant STHdh Q111 cells were 18% lower than in control cells (0.5 Ϯ 0.23 versus 0.61 Ϯ 0.11 normalized Fura-2 ratios (n.r.), p ϭ   FEBRUARY 29, 2008 • VOLUME 283 • NUMBER 9 0.036, Fig. 3H). Thus, a portion of the larger (60%) decrease in the height of the Ca 2ϩ peaks detected in STHdh Q111 with AEQ had been due to the decreased number of responding cells. In the case of BK the number of cells-responders was instead essentially the same in control and mutant cells (76 and 74%, respectively). The increase in the height of the Ca 2ϩ peak in mutant cells (0.62 Ϯ 0.16 n.r.) with respect to control cells (0.51 Ϯ 0.12 n.r., p ϭ 0.03, 18%) was close to that seen with AEQ (16%). When cells were stimulated with both ATP and BK, all STHdh Q7 and STHdh Q111 cells in the population responded (Fig. 3L).

Calcium Signaling and Mutated Huntingtin
Dynamics of InsP 3 Production in STHdh Q111 Cells-The Fura-2 experiments had also shown that basal Ca 2ϩ in nonstimulated STHdh Q111 cells was significantly lower than in control cells (43.3 Ϯ 5.32 fluorescence units (f.u.) versus 63.7 Ϯ 9.18 f.u., respectively, p Ͻ 0.001, Fig. 4A). This prompted a study of the InsP 3 levels. In line with the lower levels of basal Ca 2ϩ , the InsP 3 level in non-stimulated cells was significantly lower in STHdh Q111 than in STHdh Q7 cells (Fig. 4B, N/St). The dynamics of InsP 3 production was then explored in cells stimulated with either ATP or BK. As shown in Fig. 4B, the production of InsP 3 was delayed in STHdh Q111 with respect to STHdh Q7 cells. It was much lower 30 s after stimulation with either ATP or BK, the difference only disappearing after 60 s. The difference was lower with BK. When cells were stimulated with both agonists (Fig. 4C), the increase in InsP 3 level was clearly delayed in mutant cells, reaching a maximum at 20 s in control cells, but only at 60 s in STHdh Q111 cells.
The higher BK-induced Ca 2ϩ response may seem odd, considering the decreased rate of InsP 3 production in mutant cells. Possibly, the sensitivity of the InsP 3 receptors could have been changed in the mutants. Therefore, we explored the possibility directly by treating control and mutant cells with a membranepermeable derivative of InsP 3 (iInsP 3 /PM). The experiment showed that InsP 3 receptors in STHdh Q111 cells indeed were more sensitive then in STHdh Q7 cells (0.64 Ϯ 0.019 n.r., versus 0.47 Ϯ 0.018 n.r, respectively, p Ͻ 0.001, Fig. 4D). The result nicely confirmed previous data showing increased sensitivity of the InsP 3 receptors by mutant Htt (31).
Transcriptional Regulation of InsP 3 Controlling Enzymes-The slower rate of InsP 3 production in stimulated mutant cells prompted a study of the expression of the components of the PI cycle and of the InsP 3 receptors. The striatum of the KI-Hdh Q111 parent mouse was also studied. The transcripts of IMPA1 and inositol polyphosphatase, which are rate-limiting in the cycle, were down-regulated both in STHdh Q111 cells (by 40 and 20%, respectively, Fig. 5, A and C), and in the striatum (by 15 and 25%, respectively; Fig. 5, B and D).
Of the four PLC␤ isoforms that hydrolyze PIP 2 , only PLC␤3 was detected in the striatal cells, (in the striatum PLC␤1 was instead more abundant). The PLC␤3 transcript was up-regulated in STHdh Q111 cells, that of PLC␤1 in the striatum (Fig. 5,  E and F). The transcript of the neuronal type InsP 3 receptor (type 1) was down-regulated in both the mutant cells and the  striatum (Fig. 5, G and H), in line with previous studies showing that the expression of numerous signaling genes was dysregulated in mouse and cellular HD models, and in HD patients (32,33).
Ca 2ϩ Handling by Mitochondria in STHdh Q111 Cells-The driving force for mitochondrial Ca 2ϩ uptake (⌬ m ), was explored with the fluorescent probe TMRM (27) in peripheral mitochondria to avoid artifactual fluorescence changes due to variations in cellular thickness. Fig. 6D , p ϭ 2.1e-61). Ca 2ϩ uptake was monitored directly in mitochondria within control and STHdh Q111 cells, which were permeabilized with digitonin to expose them to concentrations of free Ca 2ϩ precisely controlled with EGTA. To increase the efficiency of AEQ expression and thus to improve the quality of the results, in these experiments the DNA of mtAEQ was transferred into the cells using lentiviral vectors pLV-mitAEQ. As shown in Fig. 6A, perfusion with an EGTA buffer generating 1.6 M free Ca 2ϩ induced a Ca 2ϩ transient which peaked at 103.6 Ϯ 33.5 M in STHdh Q7 cells (n ϭ 11) and at 86.7 Ϯ 27.7 M in STHdh Q111 cells (n ϭ 12). The difference was not significant (p ϭ 0.1). Perfusion with 11 M free Ca 2ϩ evoked a significantly higher mitochondrial Ca 2ϩ uptake transient in STHdh Q7 cells (300.6 Ϯ 28.4 M, n ϭ 6, Fig. 6B) than in STHdh Q111 cells (172.4 Ϯ 36.0 M, n ϭ 6, p ϭ 0.00012). At these two relatively low Ca 2ϩ concentrations, 3-NPA had no effect on the transients in either cell type. At 71 M Ca 2ϩ , however, the peak (453.8 Ϯ 18.6 M, n ϭ 6) was greatly reduced by 3-NPA (292.3 Ϯ 53.2 M, n ϭ 6, p ϭ 0.0001, Fig. 6C) in STHdh Q7 cells. In STHdh Q111 cells, the peak was much lower, (316.0 Ϯ 25.6 M, n ϭ 6, p ϭ 1e-5) and was decreased much more dramatically by 3- NPA (166.3 Ϯ 16.8, n ϭ 6, p ϭ 3.1e-6). Thus, mitochondria in mutant cells were about as efficient in taking up Ca 2ϩ as those in control cells when the concentration of the ion in the environment was low. As it increased, their ability to control Ca 2ϩ decreased progressively in mutant cells, the difference being exacerbated if complex II was inhibited by 3-NPA.
The mitochondrial defect in HD neurons has been suggested to concern complex II. Even if no unambiguous proof of it has yet been obtained in the model cells used here, it was interesting to study whether the inhibition of the complex was specific in inducing mitochondrial damage (⌬ and Ca 2ϩ uptake); i.e., whether other means of deenergizing mitochondria had the same effect as 3-NPA. STHdh Q7 and STHdh Q111 cells were thus treated with increasing concentrations of the uncoupler FCCP. As shown in Fig. 6E 0.8 M FCCP decreased the ⌬ m by 33% in STHdh Q111 cells, but only by 6% in STHdh Q7 cells (p 2e-7, n ϭ 6). Fig. 6F shows that the uncoupler decreased Ca 2ϩ uptake more in the mitochondria of mutant cells than in those of the controls (65% versus 44%, p ϭ 0.0049, n ϭ 5 at 0.8 M FCCP). Thus, mitochondria of mutant cells were more sensitive to deenergization in general, not only to that caused by inhibitors of complex II.
Mutant huntingtin, and 3-NPA, have been proposed to make the permeability transition pore (PTP) more sensitive to Ca 2ϩ (16, 34 -36). The preferential opening of the pore could explain the decrease of the membrane potential in the mitochondria of mutant cells challenged with high Ca 2ϩ concentrations (Fig. 6, B and C). To test this possibility, the experiment of Fig. 6B was thus repeated in the presence of the PTP blocker CsA. When cells permeabilized with digitonin were perfused with a buffer containing 20 M free Ca 2ϩ the decrease of the ⌬ was significantly more pronounced in mutant with respect to control cells (50%, p Ͻ 0.0001 versus 14%, p Ͻ 0.025, respectively). However, the ⌬ in the mitochondria of mutant cells pretreated with CsA was not different from that of CsA-treated control cells, indicating that the greater decrease of ⌬ in the mitochondria of mutant cells challenged with high Ca 2ϩ concentration, was indeed due to the preferential opening of the PTP (Fig. 6G). Then experiments were performed to assess whether the preferential opening of PTP could have a role in the reduced mitochondrial Ca 2ϩ uptake in STHdh Q111 (see Fig. 6B). Permeabilized cells transduced with the lentiviral vector pLV-mtAEQ were pretreated for 1 h with CsA (10 M), and then perfused with a buffer containing 20 M free Ca 2ϩ . As shown in Fig. 6H, the pretreatment completely abolished the decrease in mitochondrial Ca 2ϩ uptake in mutant cells when challenged with a high Ca 2ϩ concentration. Interestingly, the height of the peaks of the CsA-treated STHdh Q7 and STHdh Q111 cells (421.88 Ϯ The data represent the mean Ϯ S.D. for at least four PCRs from two independent retrotranscriptions for each condition. The differences were significant at p Ͻ 0.05, *; p Ͻ 0.01, **; and p Ͻ 0.001, ***. HPRT, hypoxanthine-guanine phosphoribosyltransferase. Transcriptional Regulation of ROS-scavenging Enzymes -Huntingtin generates ROS in the striata of a HD mouse model, and 3-NPA, which generates ROS, does so in PC12 cells (37,38). The level of ROS was thus explored in control and mutant striatal cells. Because a first set of experiments on the mutant cell clone employed in most of the work had failed to reveal differences in ROS level between mutant and control cells, the study was extended to two other mutant cell clones (STHdh Q111-2 and STHdh Q111-3 ). In all mutant clones, the concentration of ROS did not differ from the controls. However, when the cells were exposed to 3-NPA, the concentration of ROS increased as expected both in control and mutant cells. However, the increase was much more significant (even if variable) in the latter (Fig.  7A, p ϭ 0.012, n ϭ 4).

Calcium Signaling and Mutated Huntingtin
The transcription of the genes involved in the breakdown of ROS (Gpx1, catalase, SOD1, and SOD2) was analyzed next in the three independently generated STHdh Q111 clones (Fig. 7, B-D). The transcripts were up-regulated by 30 -100% in all three mutant clones (most significantly in clone STHdh Q111-2 ), with the exception of SOD2, which was instead unchanged (not shown). Pilot experiments on the transcription of the ROS genes in the striata of the KI-Hdh Q111 mice have so far failed to yield sufficiently reproducible results (not shown).  The recent observation that the transcription of the PGC-1␣ gene was specifically repressed in a STHdh Q111 cell clone (21) prompted an investigation of the interplay between the transcription of the PGC-1␣ gene, and that of the genes of ROSscavenging enzymes. The study was performed in the three independently generated STHdh Q111 cell clones. The results (Fig. 7E) showed great variability in the amounts of PGC-1␣ transcripts in the three clones. No convincing correlation was found between the levels of transcription of the PGC-1␣ gene and those of the 3 genes for the ROS-scavenging enzymes. In the mutant clone used in most of the experiments described here the transcript of PGC-1␣ increased markedly with respect to control cells (by 1.92 Ϯ 0.27-fold, p ϭ 0.0045), as did those of the ROS-scavenging enzymes (see Fig. 7, B-D). In another clone the amount of the PGC-1␣ transcript increased only marginally (by 1.23 Ϯ 0.15-fold, p ϭ 0.056), whereas those of the ROS-scavenging enzymes instead increased markedly. In the third clone the amount of PGC-1␣ transcript decreased sharply with respect to control cells (by 0.48 Ϯ 0.083-fold, p ϭ 0.0004), whereas those of the ROS-scavenging enzymes either increased or remained essentially unchanged. Direct measurement of ROS (Fig. 7A) showed that the increase in their concentration induced by 3-NPA did not correlate convincingly with the levels of the transcripts of the scavenging enzymes nor with that of PGC-1␣. Evidently, the dynamics of ROS generation/degradation in the mutant cells was not exclusively controlled by transcriptional effects.
Susceptibility of STHdh Q111 Cells to Apoptosis-HD neurons have particular propensity to succumb by apoptosis (39,40), and the model cell line used in this contribution has been claimed to be selectively vulnerable to treatments that cause non-apoptotic death (16). The vulnerability of three mutant cell lines to apoptotic treatments was thus investigated. Fig. 8 shows that the tendency to undergo apoptosis after treatment with the classical apoptotic agent staurosporin (10 M for 6 h) was significantly greater in all three mutant cell lines (by 2.28 Ϯ 0.59-, 2.84 Ϯ 0.66-, 2.26 Ϯ 0.37-fold in STHdh Q111-1 , STHdh Q111-2 , and STHdh Q111-3 cells, respectively) in respect with STHdh Q7 cells (p Ͻ 0.05 for all cases).

DISCUSSION
The accumulation of the expanded poly-Q fragments of Htt in the cytoplasm, nuclei, and axons is a hallmark of HD. As mentioned in the Introduction, the reasons for the preferential vulnerability of GABAergic MSNs to Htt, which is widely expressed in the brain and the body, is still an open problem. Mutant Htt aggregates are widely discussed with respect to their role in causing cell damage, and have even been claimed to decrease cell death risk in striatal neurons (41). Thus, they may have a number of functions, the dysregulation of the transcription of genes necessary for the function and eventual survival of striatal neurons now emerging as the most important (32,33,(42)(43)(44). Transcription in HD neurons, however, could also be dysregulated by other agents, e.g. by ROS, the effect of the poly-Q aggregates being possibly linked to the late stages of the disease.
ROS have long been known to directly damage mitochondria: they dissipate the ⌬ m , impairing the production of ATP, and preferentially damage mitochondrial DNA (18,(45)(46)(47). The mitochondrial dysfunction mediated by ROS has received new impulse from recent work showing that in HD striata, and even in the model cells used in this work, a transcriptional coactivator of the genes of ROS-scavenging enzymes (PGC-1␣) is transcriptionally down-regulated (21). The down-regulation has been suggested to be important in the damage to striatal neurons: however, the experiments presented here have shown that the level of PGC-1␣ transcript is but one of the factors active in the regulation of the level of ROS, as it does not correlate with those of the ROS-scavenging enzymes in the model cells used here.
The discovery that inhibitors of mitochondrial complex II induce cell death in striatal neurons, and produce neuronal degeneration in the striatum in vivo (11) was an important development. Complex II inhibitors, as all other mitochondrial damaging agents, also generate ROS, emphasizing their potential role in the damage to HD neurons. The finding that the complex II inhibitor 3-NPA mimics the effects of mutant Htt is an indication that somehow links the inhibition of the complex to the latter. The matter, however, still has unclear facets: recent work on cultured striatal neurons transfected with an expanded poly-Q tract have shown no down-regulation of the mRNA of two components of complex II, but have instead documented a down-regulation of their proteins (48). It thus appears probable that complex II is defective in HD neurons due to some post-transcriptional (proteolytic) effect. Whether the complex is also defective in the cell model used here is an open question. In Htt neurons the defect of complex II would not only generate ROS, but would also make mitochondria less able to respond to requests for increased activity of the respiratory chain. The defect would of course be exacerbated if the function of complex II would be further depressed by inhibitors. The canonical endogenous inhibitor of the complex is malonic acid, which has frequently been used to induce neuronal degeneration (49 -51). Work now in progress in our laboratory explores the possible increase of malonic acid in Htt neurons.
The work described here has shown that the mitochondrial damage in cells expressing mutant Htt only becomes evident  FEBRUARY 29, 2008 • VOLUME 283 • NUMBER 9 JOURNAL OF BIOLOGICAL CHEMISTRY 5787 when mitochondria are exposed to stressing insults. Mitochondria in HD neurons thus appear to be in a borderline situation: this is also indirectly shown by experiments in which the parenteral administration of doses of 3-NPA that would not produce striatal damage it rats, induced instead striatal neurodegeneration when combined with sub-toxic doses of amphetamine (52). ROS would be likely candidates as the stressing agent(s) that would make the silent mitochondrial damage fully expressed, but Ca 2ϩ could be an equally plausible candidate, given its accepted role as a mediator of neuronal damage (12,13). Ca 2ϩ signaling defects have indeed been a recurring theme in HD research (17,(53)(54)(55). Recent work has shown that the mitochondrial ⌬ in cells expressing mutant Htt was particularly vulnerable to Ca 2ϩ (17,34), and poly-Q constructs have been found to preferentially deenergize mitochondria exposed to Ca 2ϩ (17). However, the work presented here has shown that the damaging effect of Ca 2ϩ was not due to the overloading of mitochondria with it. The damage appears instead to be linked to the continuous leakage of accumulated Ca 2ϩ , incompletely compensated by the activity of the uptake uniporter made insufficient by the decrease of the ⌬. The ⌬ of mutant mitochondria would be able to sustain the accumulation of a Ca 2ϩ pulse only lasting a short time, but not that of Ca 2ϩ persistently increased in their environment. As suggested by others, and directly confirmed in the present work (Fig. 6, G  and H), it appears likely that the increased sensitivity of the PTP to Ca 2ϩ is critical to the inability of mitochondria to retain Ca 2ϩ , thus triggering the increased Ca 2ϩ cycling and eventual ATP deprivation. Naturally, the drainage of ATP would have dire consequences for cell life. Peri-mitochondrial Ca 2ϩ could also increase by an additional mechanism because mutant Htt and the huntingtin-associated protein HAP1A form a ternary complex with the InsP 3 receptor, potentiating its Ca 2ϩ releasing activity (31,56) and exposing mitochondria to persistently high Ca 2ϩ concentrations (57).

Calcium Signaling and Mutated Huntingtin
In summary, HD mitochondria appear to be more vulnerable to stress. Once stressed, they become peculiarly deenergized, with obvious deleterious consequences for the life of the neurons. Ca 2ϩ and ROS are the most plausible inducers of such mitochondrial stress. The alterations of the transcription of the enzymes that control their homeostasis are important actors in their modulation in HD neurons. In the cell model described here the transcriptional alterations in the proteins that control the Ca 2ϩ signal are likely to be operational on a longer time scale. The resulting alterations of Ca 2ϩ homeostasis could reflect a compensatory mechanism developed by the cells to prevent, or at least to delay, Ca 2ϩ from increasing to the levels that would fatally harm their mitochondria, making their silent damage fully evident.