Inhibition of Lipid Signaling Enzyme Diacylglycerol Kinase ϵ Attenuates Mutant Huntingtin Toxicity*

Background: A chemical screen using a kinase inhibitor library was carried out in a Huntington disease cellular model. Results: A target kinase, diacylglycerol kinase ϵ, when inhibited blocked mutant huntingtin toxicity. Conclusion: Inhibition of diacylglycerol kinase ϵ through pharmacological or siRNA knockdown prevents mutant Htt activation of caspase-3 and decreased levels of phosphoinositides. Significance: Diacylglycerol kinase ϵ is a novel therapeutic target for Huntington disease. Huntington disease (HD) is a dominantly inherited neurodegenerative disease caused by a polyglutamine expansion in the protein huntingtin (Htt). Striatal and cortical neuronal loss are prominent features of this disease. No disease-modifying treatments have been discovered for HD. To identify new therapeutic targets in HD, we screened a kinase inhibitor library for molecules that block mutant Htt cellular toxicity in a mouse HD striatal cell model, Hdh111Q/111Q cells. We found that diacylglycerol kinase (DGK) inhibitor II (R59949) decreased caspase-3/7 activity after serum withdrawal in striatal Hdh111Q/111Q cells. In addition, R59949 decreased the accumulation of a 513-amino acid N-terminal Htt fragment processed by caspase-3 and blocked alterations in lipid metabolism during serum withdrawal. To identify the diacylglycerol kinase mediating this effect, we knocked down all four DGK isoforms expressed in the brain (β, γ, ϵ, and ζ) using siRNA. Only the knockdown of the family member, DGKϵ, blocked striatal Hdh111Q/111Q-mediated toxicity. We also investigated the significance of these findings in vivo. First, we found that reduced function of the Drosophila DGKϵ homolog significantly improves Htt-induced motor dysfunction in a fly model of HD. In addition, we find that the levels of DGKϵ are increased in the striatum of R6/2 HD transgenic mice when compared with littermate controls. Together, these findings indicate that increased levels of kinase DGKϵ contribute to HD pathogenesis and suggest that reducing its levels or activity is a potential therapy for HD.


Huntington disease (HD) is a dominantly inherited neurodegenerative disease caused by a polyglutamine expansion in the protein huntingtin (Htt). Striatal and cortical neuronal loss are prominent features of this disease. No disease-modifying treatments have been discovered for HD. To identify new therapeutic targets in HD,
we screened a kinase inhibitor library for molecules that block mutant Htt cellular toxicity in a mouse HD striatal cell model, Hdh 111Q/111Q cells. We found that diacylglycerol kinase (DGK) inhibitor II (R59949) decreased caspase-3/7 activity after serum withdrawal in striatal Hdh 111Q/111Q cells. In addition, R59949 decreased the accumulation of a 513-amino acid N-terminal Htt fragment processed by caspase-3 and blocked alterations in lipid metabolism during serum withdrawal. To identify the diacylglycerol kinase mediating this effect, we knocked down all four DGK isoforms expressed in the brain (␤, ␥, ⑀, and ) using siRNA. Only the knockdown of the family member, DGK⑀, blocked striatal Hdh 111Q/111Q -mediated toxicity. We also investigated the significance of these findings in vivo. First, we found that reduced function of the Drosophila DGK⑀ homolog significantly improves Httinduced motor dysfunction in a fly model of HD. In addition, we find that the levels of DGK⑀ are increased in the striatum of R6/2 HD transgenic mice when compared with littermate controls. Together, these findings indicate that increased levels of kinase DGK⑀ contribute to HD pathogenesis and suggest that reducing its levels or activity is a potential therapy for HD.
Huntington disease (HD) 2 is caused by a polyglutamine expansion in the N terminus of the protein huntingtin (Htt) (1).
A pathological change in HD brain is the massive loss of medium spiny neurons in the striatum and loss of neurons in the cortex as the disease progresses. HD results in chorea, dementia, and eventually death. As there is no cure for HD, new therapeutic treatments are clearly needed. Protein kinases are the second most important group of drug targets, after G protein-coupled receptors. The kinase superfamily, whose members are related in the sequence and structure of their catalytic domain, includes CGMC kinases (including cyclin-dependent kinases (CDKs), glycogen synthase kinases (GSK), mitogen-activated protein kinases (MAP kinases), and CDK-like kinases), CaM kinases, serine-threonine kinases, tyrosine kinases, protein kinase A, G, and C families (PKA, PKC, PKG), and others. Because of their key roles in modulating signal transduction and gene transcription, protein kinases are considered to be important potential targets for drug therapy. We therefore evaluated kinase inhibitors in an HD cellular model and identified diacylglycerol kinases (DGKs) in our cellular screen.
DGKs consist of a family of kinases that catalyze the phosphorylation of diacylglycerol (DAG) to produce phosphatidic acid (PA) (2). Both DAG and PA are important intracellular signaling molecules (2). DAG can activate protein kinase C (PKC), Ras guanyl nucleotide-releasing protein (RasGRP), and transient receptor potential channels (3,4). Transient receptor potential channels mediate store-operated calcium entry into HD medium spiny neurons, and recently, compounds that target these channels have been shown to block HD neurotoxicity (5). There are 10 human DGK isoforms, and each has unique biochemical properties, expression patterns, and subcellular localization. These enzymes have a DAG binding C1 and catalytic domain. DGK⑀ is unique in that it has a preference for substrate selection. DGK⑀ favors DAG with 1-stearoyl-2arachidonoyl (18:0, 20:4) acyl chains, which represent the prevalent acyl chain composition in phosphatidylinositol (PI) and its derivatives (6). The DGK⑀ knock-out mice showed reduced seizure in response to electroconvulsive shock (7). Also, DGK⑀ knock-out mice had lower content of PI with 1-stearoyl-2arachidonoyl acyl chains (8).
Phosphatidylinositol and phosphatidylinositol phosphates (PIPs) have been implicated in the pathological mechanisms of HD. A vesicle-binding assay has shown that Htt binds different forms of PIPs, whereas the mutant Htt bound to a subset of PIPs more strongly than wild type Htt (9). Mutant Htt has also been shown to sensitize the type 1 inositol (1,4,5)-triphosphate receptor in response to inositol (1,4,5)-triphosphate, which is a second messenger derived from phosphatidylinositol (4,5)-biphosphate (PIP 2 ) (10). When the C-terminal fragment of the type 1 inositol (1,4,5)-triphosphate receptor has been introduced to the striatum of a HD model mouse, neuroprotective effects have been observed (11). However, whether the level of 1-stearoyl-2-arachidonoyl PI and PIPs is related directly with mutant Htt-associated neurotoxicity has not been determined.
To identify potential HD therapeutic targets, we screened a kinase inhibitor library in striatal Hdh 111Q/111Q mouse HD cell model. The Hdh 111Q/111Q cells are immortalized striatal cells derived from a knock-in mouse model that expresses a mutant form of Htt with an expanded polyglutamine tract of 111 repeats (12). This cell line, together with the normal control line Hdh 7Q/7Q cells, represents a commonly used mouse HD cell model. Using multiple assays, we identified DGK inhibitor II, also known as R59949, as an effective compound decreasing toxicity associated with mutant Htt. We evaluated whether DGK inhibition affected lipid metabolism in our Hdh 111Q/111Q cellular assay. Indeed, DGK inhibition normalizes altered lipid metabolism in this HD cell model. Analysis of the 10 mouse DGK isoforms in striatal cells and tissue suggests that four family members are expressed highly in the striatum. One family member, DGK⑀, prevents cell death when knocked down in Hdh 111Q/111Q cells. Correspondingly, partial loss of function of the Drosophila homolog of DGK⑀ significantly improves Httinduced motor dysfunction in a fly model of HD in vivo. Our work provides novel mechanisms of neurotoxicity from mutant Htt in HD and identifies DGK⑀ as a therapeutic target for HD.

EXPERIMENTAL PROCEDURES
Cell Culture-All cell culture reagents were from Invitrogen unless otherwise stated. Striatal Hdh Q7/Q7 and Hdh Q111/Q111 cells were maintained in DMEM supplemented with 10% FBS at 33°C. Culture plates were coated with 50 g/ml rat tail collagen type-I (BD Biosciences) for 1 h at room temperature before immortalized striatal cells were seeded for experiments. For caspase-3/7 activity assay, Biocoat collagen I-coated 96-well plates from BD Biosciences were used. Knocking down DGK genes was achieved by transfecting corresponding Dharmacon siGENOME SMARTpool siRNAs (Thermo Scientific) via Amaxa Nucleofector kit L (Lonza).
Caspase-3/7 Activity Assay-The caspase activity assay was performed with Apo3 HTS kit (Cell Technology). Hdh 111Q/111Q cells cultured in 96-well plates were challenged with serum deprivation along with drug treatments. We used the EMD Calbiochem InhibitorSelect TM 96-well protein kinase inhibitor library I (catalog no. 539744). Inhibitors were dissolved in dimethyl sulfoxide, and the final concentration in the treatment well was 20 M for all kinase inhibitors except phosphatidylinositol 3-kinase ␥ inhibitor (10 M for this kinase inhibitor). DGK inhibitor I (R50922) was purchased from EMD Calbiochem. Control cells were treated with 0.2% dimethyl sulfoxide, the same as the kinase inhibitor-treated cells for 24 h. Medium was removed, and 50 l of 1ϫ lysis buffer was added to each well. After shaking on an orbital shaker at 700 rpm for 5 min, two 10-l aliquots of lysate were transferred to new 96-well plates for BCA protein concentration measurement, and 70 l of substrate mix (1ϫ lysis buffer with 1ϫ Apo3 HTS caspase-3/7 detection reagent and 20 mM DTT) was added into the remaining 30 l of lysate in each well. The plate was briefly shaken at 700 rpm for 30 s before reading. The Fusion-Alpha Universal Microplate Analyzer (PerkinElmer Life Sciences) was used for this fluorescence-based assay (excitation, 485 nm; emission, 530 nm) at 37°C. For each sample, the protein concentration was measured in duplicate with Pierce BCA protein assay kit (Thermo Scientific). The caspase activity was normalized against protein concentration for each sample.
WST Assay-Hdh 111Q/111Q cells were cultured in 96-well plates as described above. Then the medium was removed, and 100 l 1ϫ WST-8 reagent (Alexis) was added into each well. The cells were incubated for 2 h at 33°C before absorbance at 450 nm was measured on a SpectraMax 190 plate reader (Molecular Devices).
Western Blotting-Hdh 111Q/111Q cells were treated with serum starvation for 24 h in the presence or absence of drugs. Then cells were scraped off, pelleted, and washed once with PBS (Cellgro). Cell pellets were lysed by sonication in mammalian protein extraction reagent (from Thermo Scientific) containing protease inhibitors (one Complete mini tablet per 10 ml of mammalian protein extraction reagent, Roche Applied Science) and 1% phosphatase inhibitor mixture set II (Calbiochem) at 4°C. Protein concentration was measured with Pierce BCA protein assay kit to ensure equal sample loading. Protein samples (40 g) were run on 4 -12% Bis-Tris gel (Invitrogen) at 200 V for 1 h, transferred to nitrocellulose membrane (Whatman) at constant 20 V for 14 h at 4°C, probed with anti-Htt (1:500, MAB2166, Millipore), anti-HttNeo513 (1:750, polyclonal generated as described previously (13) Open Biosystems), and anti-GAPDH (1:10,000, Fitzgerald).
Immunocytochemistry-Treated cells on glass coverslips (Bellco Glass) were fixed by 4% paraformaldehyde for 20 min at room temperature. Fixed cells were permeabilized with 0.1% Triton X-100 (Sigma) for 10 min and blocked with 5% donkey serum (Millipore) in PBS for 1 h. Primary antibodies were diluted with 1% BSA (Roche Applied Science) in PBS. The primary antibody (anti-PIP, 1:200 from Echelon) was incubated overnight at 4°C. After three washes with PBS, the secondary antibody (Alexa Fluor 488-conjugated donkey anti-mouse IgG, 1:1000 from Invitrogen) was added for 1.5 h incubation at room temperature. Coverslips with stained cells were finally mounted with Prolong Gold Antifade reagent with DAPI (Invitrogen) onto glass slides (VWR) for microscopy.
Lipid Extraction-Cells in a 15-cm plate were placed on ice and then washed once with ice-cold PBS, scraped in PBS on ice, and centrifuged. The pellet was suspended in 1200 l of ice-cold 1:1 chloroform (CHCl 3 )/methanol (MeOH) (14). After vortexing for 1 min to mix, cells were centrifuged, and the supernatant was discarded. To the pellet, 600 l of 2:1 ice-cold CHCl 3 /MeOH with 0.25% 12 N hydrochloric acid (HCl) was added followed by 10 min of vortexing at 4°C. Then, 120 l of 1 N HCl was added, and the mixture was vortexed for 15 s followed by centrifuge at 3500 rpm for 5 min. The lower phase was collected and dried under the protection of nitrogen. The pellet was redissolved in 200 l of 1:1 CHCl 3 /MeOH with 30 mM piperidine before multiple reaction monitoring (MRM)-MS analysis.
Mass Spectrometry Profiling-To profile the lipids in the extracts, samples were first analyzed by MALDI-MS using a vMALDI-LTQ platform (Thermo Scientific) operating in the negative ion mode. Briefly, 1 l of matrix (2,5-dihydroxybenzoic acid) dissolved in methanol at 10 mg/100 l) was gently mixed with 1 l of extracted phospholipid in 1:1 CHCl 3 /MeOH solution with piperidine, and the mixture was loaded on an MALDI plate specially marked with ImmEdge Hydrophobic Barrier Pen (Vector Laboratories, Inc.). The laser power was adjusted to optimize the MALDI signal, and spectra were acquired over m/z 300 -2000.
Mass Spectrometry MRM-MS Analysis-Samples were analyzed by MRM-MS on a 4000 QTRAP hybrid triple quadrupole/ linear ion trap mass spectrometer (AB Sciex) equipped with a Turbo IonSpray source in negative ion mode. Lipid samples were introduced by direct infusion using a syringe pump (Mode 11 series, Harvard Apparatus) at a flow rate of 10 l/min. A total of 18 ion transition pairs from six base phosphoinositides (34:1, 36:1, 36:4, 38:4, 38:5, and 40:6) with one, two, or three phosphate moieties (PI, PIP, and PIP 2 ) were quantified by MRM-MS. The negatively charged parent ion was used as the Q1 mass, and the diagnostic fragment ion of the dehydrated headgroup was chosen for Q3 (15), which differentiates phosphoinositides with different fatty acid compositions (supplemental Fig. 1). The Q1 m/z, Q3 m/z, dwell time, declustering potential, collision energy, and collision cell exit potential for each transition are listed in supplemental Table 1. Inositol 1,4,5-trisphosphate species were not included in the study due to their extremely low abundance. Data were collected with an ion spray voltage of Ϫ4500 V, curtain gas of 10 psi, nebulizer gas of 25 psi, and an interface heater temperature of 600°C. MRM-MS transitions were acquired and monitored at unit resolution in both Q1 and Q3. Each sample was typically infused for 3-6 min, and each sample was analyzed two times with the average signal of the runs reported. Quantification was performed using Analyst (version 1.5), and each transition was integrated individually. The quantification results were normalized against cell numbers counted by Z1 Coulter Particle Counter (Beckman Coulter).
Drosophila Motor Performance Analysis-Tests were carried out using 15 age-matched virgin females. Animals were placed in an empty vial and tapped down. The number of animals able to climb 9 cm after 15 s was recorded as a percentage of the total. This was repeated 10 consecutive times, and the average of the ten observations was plotted for each day shown in the chart. Two replicates were tested in parallel for each genotype. Animals were raised at 26.5°C. The nervous system driver line elav-GAL c155 was obtained from the Bloomington Drosophila Stock Center at University of Indiana. The inducible shRNA line targeting the Drosophila homolog of DGK⑀ (4659GD) was obtained from the Vienna Drosophila RNAi Center. NT-Htt128Q animals express an N-terminal Htt fragment comprising exons 1-4 (first 336 amino acids with 128Q polyglutamine expansion) have been described previously (16).

RESULTS
Kinase Inhibitor Library Screening Identifies DGK Inhibitor II as Therapeutic Target for HD-We screened a kinase inhibitor library that contained 80 different inhibitors (see supplemental Table 2) in the mouse Hdh 111Q/111Q cell model utilizing caspase activity and WST-8 as endpoints during serum withdrawal (Fig.  1). As reported previously, serum withdrawal results in a dramatic increase in caspase activity in Hdh 111Q/111Q when compared with Hdh 7Q/7Q cells (13,17). We found several inhibitors that reduce caspase-3/7 activity when Hdh 111Q/111Q cells undergo serum withdrawal. These include Akt inhibitor X, PDK1/Akt/Flt dual pathway inhibitor, chelerythrine chloride, DGK inhibitor II, IGF-1R inhibitor II, Lck inhibitor, PDGF RTK inhibitor III, PKC␤ II/EGFR inhibitor, rapamycin, SU11652, VEGF receptor 3 kinase inhibitor, VEGF receptor 2 kinase inhibitor III, and VEGF receptor 2 kinase inhibitor IV (Fig. 1A). However, when coupled with WST-8 assay, which measures bioreduction of WST-8 mostly dependent on the glycolytic NAD(P)H production of viable cells, DGK inhibitor II (R59949) and rapamycin were the only kinase inhibitors that protected FIGURE 1. Screening of the kinase inhibitor library identified DGK inhibitor as a potential drug candidate. A, the screening of the kinase inhibitor library in Hdh 111Q/111Q cells during serum withdrawal using caspase activity assay as an end point. Several kinase inhibitor hits, including DGK inhibitor II, decreased caspase-3/7 activity. B, screening the same library using viability WST-8 assay library in Hdh 111Q/111Q cells during serum withdrawal. The WST-8 assay and caspase activity assay identified DGK inhibitor II along with rapamycin, a known protective molecule, as a hit. See supplemental Table 2 for further details of kinase inhibitors used in this screen. neg contr, negative control; pos contr, positive control; DMSO, dimethyl sulfoxide. The compounds are defined in supplemental Table 2. JUNE 15, 2012 • VOLUME 287 • NUMBER 25

JOURNAL OF BIOLOGICAL CHEMISTRY 21207
Hdh 111Q/111Q cells against serum withdrawal (Fig. 1B), thus indicating that these compounds act specifically on Hdh 111Q/111Q -induced toxicity. Rapamycin has been reported to reduce HDassociated toxicity through induction of autophagy (18). The identification of DGK inhibitor II as blocking mutant Htt toxicity is novel, and the mechanism of action for this effect is not known. To confirm that other DGK inhibitors were effective in this assay, we tested another DGK inhibitor, DGK inhibitor I (also known as R59022). As shown in Fig. 2A, both DGK inhibitors showed reduced caspase activity in a dose-dependent manner, suggesting a role for DGKs in toxicity caused by the polyglutamine mutation in Htt. Hdh 7Q/7Q cells generally show very modest caspase activation after serum starvation in 24 h, and DGK inhibitors do not alter the caspase-3/7 activity in serum starved Hdh 7Q/7Q cells (supplemental Fig. 2). Given this result, we utilized Hdh 111Q/111Q cells for most of our studies.

Mutant Huntingtin Proteolysis Is Reduced in Cells Treated
with DGK Inhibitor II-The proteolysis of mutant Htt by caspases to generate the toxic N-terminal fragments is linked directly to the cell death mechanisms in HD. In Hdh 111Q/111Q cells, proteolysis of Htt was detected upon serum withdrawal (Fig. 2B, lane 2) using Western blot analysis. Addition of either DGK inhibitor I or II reduced the level of Htt proteolysis (Fig.  2B, lanes 3, 4, and 6). Using an antibody specific to the caspase-3 cleavage site at the 513 amino acid of Htt (NeoHtt513), we found that Htt proteolysis at this site was induced by serum starvation (Fig. 2B, lane 2) and reduced by treatment with DGK inhibitors (Fig. 2B, lanes 3, 4, and 6). We did not detect cleavage of Htt at amino acids 552 or 586 using neoHtt552 or neoHtt586 under these conditions (data not shown). Immunocytochemistry revealed that cleaved active caspase-3 was induced in Hdh 111Q/111Q cells upon serum withdrawal (data not shown). Treatment with DGK inhibitor II prevented the generation of cleaved caspase-3 (data not shown).
Protective Effect of DGK Inhibitors in HD Does Not Require PKC-The reaction catalyzed by DGKs is the phosphorylation of DAG into PA. DAG is a well known PKC activator, and the inhibition of DGKs would likely lead to decreased DAG consumption and increased PKC signaling. To test whether this could be the mechanism underlying the protective effects of DGK inhibitors, we treated cells with two PKC inhibitors, bisindolylmaleimide I and Gö6983. Both bisindolylmaleimide I and Gö6983 inhibit PKC isoforms belonging to conventional and novel PKC subfamilies, the two subgroups of PKCs responsive to DAG. Caspase activity, WST-8 and mutant Htt proteolysis were evaluated during serum withdrawal conditions with these two inhibitors. The protective effect achieved by DGK inhibitor II treatment in Hdh 111Q/111Q cells was not altered by co-treatment of bisindolylmaleimide I or Gö6983 using caspase activity (Fig. 3A), WST-8 (Fig. 3B) or the proteolysis of Htt (Fig.  3C) as end points. DGK inhibitor I did not affect Htt proteolysis in Hdh 7Q/7Q cells in the presence or absence of PKC inhibitors as expected (supplemental Fig. 3). Both inhibitors showed strong inhibition of PKCs as measured by their effects on phorbol ester-induced PKC activation (data not shown). Therefore, the protective effect of DGK inhibitors on Hdh 111Q/111Q cells does not require PKC signaling as a downstream mediator.

Altered Phosphoinositides in Hdh 111Q/111Q Cells and DGK Inhibition Normalizes Lipid Metabolism During Serum
Withdrawal-DGKs are critical enzymes involved in lipid metabolism. When certain DGK family members are knocked out in mice, a decrease in polyphosphoinositides is observed (8). Therefore, a critical mechanism for rescue of HD cytotoxicity may be through altered lipid metabolism. Lipid profiling of cellular phosphatidylinositol monophosphate/phosphatidylinolsitol biphosphate (PI/PIP/PIP 2 ) was carried out using mass spectrometry in total extracts from Hdh 7Q/7Q and Hdh 111Q/111Q . Representative MALDI-MS spectra are shown in supplemental Fig. 4 -8. The predominant species is PI, but we were also able to detect PIP and PIP 2 . We also noted that the distribution and pattern between the Hdh 7Q/7Q and Hdh 111Q/111Q extracts is distinct for these lipids (data not shown).
Having established that the PI, PIP, and PIP 2 could be detected by MALDI-MS, we quantified the levels in Hdh 7Q/7Q FIGURE 2. Both DGK inhibitor I (R59022) and II (R59949) reduced caspase activity and mutant Htt proteolysis in Hdh 111Q/111Q cells undergoing serum withdrawal. A, retesting DGK inhibitor II and the evaluation of related compound DGK inhibitor I confirmed the protective effect of these kinase inhibitors in Hdh 111Q/111Q cells. We found a dose-dependent reduction of caspase-3/7 activity in Hdh 111Q/111Q cells after serum withdrawal with DGK inhibitor I and II. DGK inhibitor treatment conditions were compared with FBS withdrawal only condition by one-way analysis of variance (*, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.005 for all figures). B, Hdh 111Q/111Q cells were treated with different doses of DGK inhibitor I or II during serum starvation. Western analysis demonstrates that serum withdrawal produces huntingtin product detected by Htt 2166 and neoHtt513 antibody, and this fragment was reduced by DGK inhibitors at 10 M. Control levels of GAPDH suggest equal protein loading. Quantification of the Htt fragment bands was also shown.
To further analyze alterations in lipid metabolism, we also performed immunocytochemistry on Hdh 7Q/7Q and Hdh 111Q/111Q cells using a pan-lipid antibody that recognizes PIPs. As shown in Fig. 4C, we detected less PIP staining in Hdh 111Q/111Q when compared with Hdh 7Q/7Q cells. Furthermore, the levels of PIP increased during serum withdrawal and were reduced by DGK inhibitor II treatment (Fig. 4C).
Identification of DGK⑀ as Therapeutic Target for HD-There are 10 DGK mammalian isoforms, and therefore it is possible that one or several of these family members are possible therapeutic targets of the DGK inhibitors in the context of HD toxicity. Five DGK isoforms, ␣, ␤, ␥, ⑀, and have been shown to be expressed in the brain (19). To identify the isoforms most abun-dant in the striatum, we used RT-qPCR to quantify the isoforms in this region of the brain. ␤, ␥, ⑀, and DGK isoforms were expressed in the striatum (Fig. 5). We then knocked down these four DGKs in striatal Hdh 111Q/111Q cells individually with siRNA and performed the caspase activity assay after serum starvation. Only the siRNA specific to DGK⑀ resulted in decreased caspase-3/7 activity, similar to treatment of DGK inhibitors (Fig. 6A, data not shown for other DGKs). We confirmed the knockdown of DGK⑀ using RT-qPCR (Fig. 6B), and the level was similar in magnitude to that found for the reduction in caspase-3/7 activity (Fig. 6A).
We also investigated whether the levels of DGK⑀ are altered in response to mutant Htt. RT-qPCR for DGK⑀ showed that Hdh 111Q/111Q cells have higher levels of DGK⑀ than Hdh 7Q/7Q cells (Fig. 6C), indicating that DGK⑀ levels are increased in response to mutant Htt.
In Vivo Evidence for DGK⑀ Reduction in Levels of DGK ⑀ Homolog Improves Neuronal Function in Drosophila Model of HD-We next sought to validate this observation in vivo. We found that in the striatum of R6/2 HD transgenic mice, a HD mouse model with overexpression of N-terminal fragment of Htt containing ϳ120 polyglutamines, the levels of DGK⑀ are higher than in the striatum of littermate control mice (Fig.  7A). These results suggest a specific role of DGK⑀ in HD toxicity. . DGK inhibitor did not require PKC to take effects. A, the reduced caspase activity in DGK inhibitor II-treated Hdh 111Q/111Q cells after serum withdrawal was not altered by PKC inhibitors bisindolylmaleimide I (BIM1) or Gö6983. After seeding, Hdh 111Q/111Q cells were cultured in normal 10% FBS (column 1) or serum-free medium (columns 2-5) for 24 h. Dimethyl sulfoxide (DMSO) control or 20 M DGK inhibitor II with or without 1 M bisindolylmaleimide I or 1 M Gö6983 were added simultaneously with serum withdrawal. Caspase activity assay was performed after the treatment. B, increased WST-8 by DGK inhibitor II in Hdh 111Q/111Q cells after serum withdrawal was not affected by PKC inhibitors. Hdh 111Q/111Q cells were treated the same as in A, and WST-8 assay was used instead of caspase activity assay. One-way analysis of variance was used for statistical analysis for both A and B (ns, not significant). C, decreased Htt fragment in Hdh 111Q/111Q cells after serum withdrawal by DGK inhibitor was not affected by PKC inhibitors shown by Western blotting. Serum-starved Hdh 111Q/111Q cells were co-treated with 10 M DGK inhibitor II (DGKi) with or without 1 M bisindolylmaleimide I or 1 M Gö6983. Western blot using the antibody Htt2166 against Htt was performed. rel., relative.
The above data suggest that DGK⑀ is involved in HD and may function as a suppressor of Htt-induced toxicity in vivo. To investigate this possibility, we studied the effect of decreasing DGK⑀ levels in a well established Drosophila HD model (16). Expression of N-terminal Htt with a 128Q (NT-Htt128Q) expansion in the Drosophila nervous system (using the neuronal elav-GAL4 driver) leads to progressive motor performance deficits when compared with normal animals (Fig. 7B, compare blue discontinuous line with black dotted lines). Consistent with Hdh 111Q/111Q cell data, decreasing the levels of the Drosophila DGK⑀ in the nervous system by means of an inducible shRNA line (NT-Htt128Q/DGK⑀ shRNA 4659GD), significantly ameliorates the toxic effects of expanded Htt (Fig. 7B, compare solid red lines with black dotted lines). NT-Htt128Q/ DGK⑀ shRNA 4659GD show a delay in disease onset, and they perform better than animals expressing NT-Htt128Q with normal levels of DGK⑀. These data, together with the observation in the striatum of R6/2 mice, confirm that the Htt-DGK⑀ interaction is relevant in vivo.

DISCUSSION
To identify new therapeutic targets in HD, we screened a kinase inhibitor library for molecules that block mutant Htt cellular toxicity. We found that DGK inhibitor II decreased caspase-3/7 FIGURE 4. Quantification of phosphoinositides in Hdh 7Q/7Q7 and Hdh 111Q/111Q cells using mass spectrometry. A, Hdh 7Q/7Q cells have higher levels of phosphoinositides (38:4) than Hdh 111Q/111Q cells. Total lipids were collected from either Hdh 7Q/7Q or Hdh 111Q/111Q cells grown in normal 10% FBS condition. Lipid samples were subjected to MRM-MS for quantification. Intensities of each lipid species was normalized against cell numbers. B, in Hdh 111Q/111Q cells serum deprivation caused increase in 38:4 PIP and PIP 2 , whereas DGK inhibitor II blocked this increase. Hdh 111Q/111Q cells were treated with serum-free medium with or without indicated concentration of DGK inhibitor II (DGKi) for 24 h before total lipid extraction. All samples were subjected to MRM-MS for quantification. Intensities of each lipid species were normalized against cell number. One-way analysis of variance was used for statistical analysis. C, immunocytochemistry of Hdh 7Q/7Q and Hdh 111Q/111Q cells using a pan-PIP antibody. Cells were subjected to 24 h serum-free medium treatment (SFM) in the presence or absence of 20 M DGK inhibitor II. Images are at 20ϫ magnification activity after serum withdrawal in striatal Hdh 111Q/111Q cells. Our screen also identified rapamycin as a kinase inhibitor that blocked mutant Htt-mediated cell death. The DGK inhibitor II decreased the level of cleaved caspase-3, the accumulation of the 513-amino acid N-terminal Htt fragment processed by caspase-3, and blocked alterations in lipid metabolism during serum withdrawal. We also tested the DGK inhibitor I, which was not in the initial library of 80 compounds, and this compound was effective as well.
There are 10 distinct mammalian isoforms of DGKs. To identify the possible diacylglycerol kinase family member that was responsible for rescuing striatal Hdh 111Q/111Q -mediated toxicity, we knocked down all four DGK isoforms expressed in the brain (␤, ␥, ⑀, and ) using siRNA. Only the knockdown of the family member, DGK⑀, blocked striatal Hdh 111Q/111Q -me- FIGURE 5. Expression of DGK isoforms in mouse striatum. Total RNA sample from mouse striatum tissue was subjected to RT-qPCR for all 10 mammalian DGK isoforms. Actb stands for ␤-actin, a housekeeping gene that we used for normalization. ␤, ␥, ⑀, and DGKs showed relatively higher expression levels. rel., relative. FIGURE 6. DGK⑀ in immortalized Hdh 111Q/111Q cells. A, DGK⑀ was knocked down by siRNA in Hdh 111Q/111Q cells before serum withdrawal. Caspase activity showed decreased levels of caspase-3/7 activity with siRNA specific to DGK⑀ (siDGK⑀) compared with non-targeting (NT) siRNA in serum-starved Hdh 111Q/111Q cells. B, RT-qPCR confirmed knockdown of DGK⑀ in Hdh 111Q/111Q cells. Student's t test was used for statistical analysis. C, RT-qPCR showed higher expression level of DGK⑀ in Hdh 111Q/111Q cells than in normal Hdh 7Q/7Q cells. Serum starvation further increased DGK⑀ expression. One-way analysis of variance was used for statistical analysis. rel., relative. FIGURE 7. DGK⑀ in mouse and Drosophila HD models. A, RT-qPCR showed higher DGK⑀ level in transgenic HD R6/2 striatum than in wild type (WT) mouse striatum tissue at the age of 10.5 weeks (Student's t test, p Ͻ 0.001). B, reduction in DGK⑀ homolog improves neuronal function in a Drosophila model of HD. Chart represents motor performance as a function of age in control flies and flies expressing NT-Htt128Q either alone or together with an inducible shRNA targeting the Drosophila DGK⑀ homolog (CG8657), in the nervous system. Test measures the ability of the animals to climb 9.5 cm in 15 s (see "Experimental Procedures" for details). Control animals (blue dashed line) perform well in the motor performance test for the complete duration of the experiment. Animals expressing NT-Htt128Q in the nervous system show progressive impairment of motor performance starting at day 12 (black dotted line). Animals expressing NT-Htt128Q in the nervous system together with an shRNA targeting the Drosophila DGK⑀ homolog show a suppression of the Htt-induced motor performance impairment (solid red lines). Notice that the Htt128Q/DGK⑀ shRNA 4659GD animals show delayed disease onset (impairments begin at day 14) and continue to perform better than the NT-Htt128Q for the remainder of the experiment. Genotypes are as follows: normal control, Elav-GAL4/w1118; NT-Htt128Q, Elav-GAL4/ w1118;ϩ;UAS-NT-Htt128Q(f33A)/ϩ. Htt128Q/DGK⑀ shRNA 4659GD: Elav-GAL4/w1118; DGK⑀ shRNA 4659GD/ϩ;UAS-NT-Htt128Q(f33A)/ϩ. Error bars represent S.E. Two experimental replicates are shown for the NT-Htt128Q and the Htt128Q/DGK⑀ shRNA animals. diated toxicity. This isoform DGK⑀ is unique in that it has a hydrophobic domain that facilitates attachment to membranes (20), and furthermore, it has specificity for diacylglycerol containing arachidonyl chains (21,22). Loss of function of the Drosophila DGK⑀ homolog was evaluated in an HD Drosophila model, and we found significant improvement of Htt-induced motor dysfunction. In addition, the levels of DGK⑀ were higher in the striatum of R6/2 HD transgenic mice when compared with controls, confirming the relevance of the interaction in vivo.
We have discovered that inhibition of the enzyme DGK is protective to immortalized striatal cells expressing mutant Htt. Because the enzymatic activity of this protein is to catalyze the phosphorylation of DAG into PA, we expect that inhibition would lead to a rise in DAG production and decreased PA production. Classically, DGKs are known as regulators of PKC family members. However, we tested PKCs inhibitors and did not find the protection imparted by DGK inhibition was dependent upon PKCs for mutant Htt toxicity.
Because generation of PA through DAG phosphorylation also is the first step in PI resynthesis, we evaluated whether the inhibition of DGK could impact lipid metabolism in Hdh 7Q/7Q and Hdh 111Q/111Q cells. First, we found that the levels of PI, PIP, and PIP 2 in Hdh 7Q/7Q cells are higher when compared with Hdh 111Q/111Q cells. It is possible that the lower levels of PIs in the Hdh 111Q/111Q cells reflect a compensatory mechanism to block striatal cell death. Consistent with this notion, we found that both Hdh 7Q/7Q and Hdh 111Q/111Q cells undergoing serum withdrawal had elevated levels of PIs. When we evaluated the impact of DGK inhibition on lipid levels in Hdh 111Q/111Q cells undergoing serum withdrawal, we found the levels of PIP and PIP 2 were reduced significantly. In addition, immunocytochemistry demonstrated an increase in PIP signaling during serum withdrawal, which was blocked by DGK inhibition. This may be relevant to the protective effect of DGK inhibition.
A number of alterations in lipid metabolism have been noted before in HD. This includes lower cholesterol levels in affected areas of the brain (23) and reduced ganglioside (GM1) synthesis (24). Further modulation of GM1 restores normal motor behavior in HD mouse model YAC128 (25). Our findings further emphasize the role of altered lipid metabolism in HD, particularly with respect to phosphatidylinositols, which are already known to have distinct binding to the mutant Htt protein. This may be particularly important as the normal function of Htt appears to be involved in vesicular trafficking, and many protein complexes involved in this process require PIPs. Htt binds different forms of PIPs, whereas the mutant Htt bound to a subset of PIPs more strongly than wild type Htt in vesiclebinding assays (9). Therefore, inhibition of DGKs may alter the interaction of mutant Htt with PIPs. DGK⑀ levels as measured by RT-qPCR appear to be modulated by polyglutamine expansion in the Htt protein. We found higher levels of DGK⑀ in immortalized striatal cells expressing mutant Htt and in the mouse model of R6/2 HD transgenic mouse model. Furthermore, cells undergoing cell death had increased levels of DGK⑀. This may increase the levels of PIP 2 , which is known to regulate cell death through P2X7 receptors (26).
In conclusion, we have identified DGK⑀ as a therapeutic target for HD. We have shown that pharmacological inhibition of DGK activity can improve Htt-induced toxicity in cells. In addition, we show that the abnormal phospholipid levels induced by mutant Htt are restored upon decreasing DGK activity. Finally, we also have observed in vivo that DGK⑀ levels are increased in response to Htt and that decreasing DGK⑀ levels rescues Htt toxicity in an animal model. There are DGK⑀ mutant mice available (7,8), and our future work will be aimed at how modulation of this enzyme in mammalians affects disease progression and neuropathology in HD mouse models.