Inhibition of mitochondrial protein import by mutant huntingtin

Abstract

Mitochondrial dysfunction is associated with neuronal loss in Huntington's disease (HD), a neurodegenerative disease caused by an abnormal polyglutamine expansion in huntingtin (Htt). However, the mechanisms linking mutant Htt and mitochondrial dysfunction in HD remain unknown. We identify an interaction between mutant Htt and the TIM23 mitochondrial protein import complex. Remarkably, recombinant mutant Htt directly inhibited mitochondrial protein import in vitro. Furthermore, mitochondria from brain synaptosomes of presymptomatic HD model mice and from mutant Htt-expressing primary neurons exhibited a protein import defect, suggesting that deficient protein import is an early event in HD. The mutant Htt-induced mitochondrial import defect and subsequent neuronal death were attenuated by overexpression of TIM23 complex subunits, demonstrating that deficient mitochondrial protein import causes mutant Htt-induced neuronal death. Collectively, these findings provide evidence for a direct link between mutant Htt, mitochondrial dysfunction and neuronal pathology, with implications for mitochondrial protein import-based therapies in HD.

Figure 1

Mutant Htt interacts with the TIM23 complex. (a) Caudate nucleus sections from human HD grade 2 and control brains, subjected to immunohistochemistry for indicated proteins. Mutant Htt aggregates detected by anti-Htt (EM48) antibody colocalize with mitochondrial proteins Tim23 and DRP1 in human HD caudate nucleus in deconvolved confocal images. (b) ST-Hdh cells transfected with mtGFP expression plasmid, subjected to immunofluorescence with anti-polyQ antibody (1C2) to label mutant Htt (red). Mutant Htt in ST-Hdh^Q111/Q111 cells partially colocalizes (yellow) with mitochondria in deconvolved confocal images. Scale bars (a,b), 10 µm. DAPI, 4′,6-diamidino-2-phenylindole. (c) Mouse forebrain mitochondria incubated with GST alone or GST-Httex1 proteins were subjected to GST pull-down assays. Bound proteins were identified by mass spectrometric analysis. Venn diagram represents the number of identified proteins. TIM23 complex components were identified among the Httex1-97Q–binding proteins. See also Supplementary Table 1. (d) Interaction between Tim23 and Httex1-97Q was verified by GST pull-down assays as in c, followed by immunoblotting with anti-Tim23 antibody. (e,f) Endogenous interaction between mutant Htt and Tim50 was found in ST-Hdh^Q111/Q111 cells (e) and 5-week-old R6/2 mouse forebrain (f) by coimmunoprecipitation (IP) with anti-Tim50 antibody followed by immunoblotting (IB). Normal IgG is a negative control. (g) Mitochondria isolated from ST14A cells were incubated with equimolar concentration of the indicated recombinant GST-fusion proteins and were subjected to GST pull-down assays. Bound proteins were analyzed by immunoblotting with indicated antibodies. Marker lane shows 20- and 15-kDa molecular mass standards. Right, Coomassie blue staining of GST fusion proteins used in assays. (h) A schematic representation of GST-Htt proteins used in c,d,g. The experiments (a,b,d–g) were successfully repeated three times. Full-length blots/gels (d–g) are presented in Supplementary Figure 9.

Figure 2

Mutant Htt directly inhibits mitochondrial protein import. (a) Coomassie blue staining of GST fusion proteins used in b,c. (b) Forebrain mitochondria prepared from adult WT mice were preincubated with GST or GST-Httex1 proteins (3 or 10 µM) on ice for 1 h in 10 µl and then subjected to import assays by adding 40 µl import reaction buffer containing [³⁵S]pOTC (60 min). Data are presented as mean + s.e.m. Inhibition of pOTC import by Httex1 proteins was dose dependent (^#P = 0.001 and *P = 0.0001, F_5,12 = 109.40). GST alone showed import activity similar to vehicle (phosphate-buffered saline), indicating no effect of GST on pOTC import (data not shown). (c) Kinetic analysis of pOTC import reaction after preincubation with indicated recombinant proteins as in b. Representative gel images used for quantification are shown. Import reaction times are indicated. Httex1 proteins decreased the import of pOTC into mitochondria in a polyQ length–dependent manner. ^# and * represent significant difference compared to GST and GST-Httex1-23Q at the given time point, respectively. 3 µM, 40 min: Httex1-97Q, ^#P = 0.008, F_2,6 = 7.404. 3 µM, 60 min: Httex1-23Q, ^#P= 0.009; Httex1-97Q, ^#P< 0.0001, *P= 0.002, F_2,6 = 43.28. 10 µM, 20 min: Httex1-23Q, ^#P= 0.002; Httex1-97Q, ^#P= 0.0002, F_2,6 = 31.29. 10 µM, 40 min: Httex1-23Q, ^#P< 0.0001; Httex1-97Q, ^#P< 0.0001, *P= 0.0006, F_2,6 = 153.73. 10 µM, 60 min: Httex1-23Q, ^#P< 0.0001; Httex1-97Q, ^#P< 0.0001, *P< 0.0001, F_2,6 = 1,111.73. Data are presented as mean ± s.e.m. Percentage pOTC import (b,c) represents the percentage of mOTC radioactivity compared to input (total [³⁵S]pOTC radioactivity added to each reaction). One-way ANOVA, Bonferroni t-test. n = 3 independent experiments (b,c). Full-length blots/gels (a,c) are presented in Supplementary Figure 9.

Figure 3

Mitochondria isolated from mutant Htt-expressing striatal cells and mouse brain exhibit decreased protein import. (a) Mitochondria isolated from the indicated striatal cells were subjected to pOTC import assays. Addition of a mitochondrial uncoupler, 2,4-dinitrophenol (DNP), to mitochondria before starting the import reaction confirmed that dissipation of mitochondrial membrane potential blocks import. Mitochondria from ST-Hdh^Q111/Q111 and N548mut cells showed significantly decreased pOTC import compared to that of control cell lines. Times are indicated in minutes in a,c. ST-Hdh^Q111/Q111, 8 min: *P = 0.0004, t = 5.90, d.f. = 8, n = 5 independent experiments; ST-Hdh^Q111/Q111, 16 min: ^#P = 0.0022, U = 0, n = 6 independent experiments; N548mut, 8 min: *P = 0.045, t = 2.29, d.f. = 10, n = 6 independent experiments; N548mut, 16 min: ^#P = 0.0022, U = 0, n = 6 independent experiments. (b) Schematic of synaptosomal and nonsynaptosomal mitochondria isolation protocol from mouse forebrain. Sup, supernatant. (c) Synaptosomal mitochondria isolated from 22- to 24-d-old, presymptomatic 150CAG R6/2 mice showed significantly decreased pOTC import compared to that of WT (*P = 0.029, t = 3.34, d.f. = 4, n = 3 independent experiments; 3 or 4 WT or R6/2 brains were pooled in each experiment). (d) Synaptosomal mitochondria isolated from 195CAG R6/2 showed significantly decreased pOTC import compared to that of WT (5–6 weeks, *P = 0.012, t = 3.59, d.f. = 6, n = 4 independent experiments; 10–11 weeks, *P = 0.029, U = 0, n = 4 independent experiments). Modest reduction of pOTC import was also found in 195CAG R6/2 nonsynaptosomal mitochondria at 5–6 weeks of age compared to that of control WT (*P = 0.0006, U = 0, n = 7 independent experiments), but not at 10–11 weeks of age (n = 9 independent experiments). (e) 195CAG R6/2 liver mitochondria showed significantly impaired pOTC import only in late-disease stage (13–14 weeks) (*P = 0.0044, t = 3.92, d.f. = 8; ^#P = 0.0079, U = 0, n = 5 independent experiments). (a,e) The data are scaled to pOTC import (equal to mOTC) in control mitochondria after the maximum reaction time (set equal to 1). (c,d) The data are scaled to mOTC in WT mitochondria at 60 min reaction time (set equal to 1). (a,c) Representative gel images of at least 3 independent experiments are shown. (a,c–e) Mann-Whitney U and unpaired t-tests (two-tailed). Data are presented as mean ± s.e.m. Full-length blots/gels (a,c) are presented in Supplementary Figure 9.

Figure 4

Mutant Htt-expressing primary neurons show impaired mitochondrial protein import. (a) Primary cortical neurons from eight R6/2 and eight littermate WT embryos (E15.5) were individually plated into two dishes and treated at day in vitro (DIV) 7 with or without sublethal H₂O₂ (10 µM) for 2 h. Isolated mitochondria were then subjected to pOTC import assay (30 min). R6/2 neurons treated with sublethal H₂O₂ showed a significant decrease in pOTC import compared to vehicle-treated R6/2 neurons (unpaired t-test; *P = 0.022, t = 2.58, d.f. = 14, n = 8 cultures per condition prepared from 8 different embryos from 3 independent experiments). Data points are presented, with mean ± s.e.m. (b) Primary cortical neurons prepared from WT embryos were transduced with Httex1-25Q (Htt25Q), Htt72Q or control empty-vector lentivirus (Vec) at DIV 5. MTS assays were performed at DIV 14. Expression of Htt72Q, but not Htt25Q, decreased MTS-reducing activity, indicating that mutant Htt decreases mitochondrial metabolic activity (*P < 0.0001 compared to Htt25Q or vector control, n = 17 (vector), 12 (Htt25Q) or 12 (Htt72Q) cultures from 3 independent experiments). (c) Primary cortical neurons were transduced as in b. Mitochondria were isolated from neurons at DIV 10 before mutant Htt-induced neuronal death and subjected to pOTC import assays (30 min). Mitochondria isolated from neurons expressing Htt72Q exhibited decreased protein import compared to those expressing Htt25Q or vector control (*P < 0.0001, F_2,18 = 24.42, n = 5 (vector), 10 (Htt25Q) or 6 (Htt72Q) mitochondria samples from 3 independent experiments). Htt25Q–expressing neurons showed no impairment of mitochondrial import compared to control neurons transduced with empty vector. Data (b,c) are presented as mean + s.e.m. One-way ANOVA, Bonferroni t-test.

Figure 5

Global and TIM23-driven mitochondrial protein import is necessary for survival of primary neurons. (a) Knockdown of Tom40 was confirmed using COS cells transfected with human Tom40-GFP expression plasmid and Tom40 RNAi (U6-Tom40.2, U6-Tom40.3) or control U6 plasmid. (b) DIV 5 cortical neurons were cotransfected with U6-Tom40.2 RNAi or U6 plasmid and with β-galactosidase (β-gal) expression plasmid and subjected to immunocytochemistry with anti–β-gal antibody at DIV 8. Arrows indicate representative β-gal⁺ control (top) and Tom40 RNAi (bottom) neurons, the latter of which show condensed or fragmented nuclei. (c) Neurons treated as in b were quantified for cell death by scoring nuclear morphology. Neuronal death was significantly increased in Tom40 knockdown cortical and striatal neurons compared to control neurons (one-way ANOVA, Fisher least significant difference; cortical neurons, *P = 0.0002, ^#P < 0.0001, F_2,12 = 41.65, n = 5 coverslips from 4 independent experiments; striatal neurons, *P = 0.018, ^#P < 0.0001, F_2,14= 19.46 ; n = 6 (U6 and U6-Tom40.3) or 5 (U6-Tom40.2) coverslips from 4 independent experiments). (d) Cortical neurons cotransfected with U6-Tom40.3 RNAi or U6 plasmid and with a β-gal plasmid were subjected to immunocytochemistry with anti–β-gal and anti-active caspase-3 antibodies and nuclear labeling with DAPI at the indicated time points after transfection. Active caspase-3⁺ neurons among β-gal⁺ transfected neurons were quantified. Inset: representative image of Tom40 knockdown neurons with active-caspase-3 immunoreactivity. Tom40 knockdown neurons showed increased active caspase-3 compared to control neurons (unpaired t-test; *P = 0.0005, t = 7.71, d.f. = 5, n = 6 slides per group from 3 independent experiments). (e) Transfected cortical neurons as in d were treated with Q-VD-OPh (20 µM) or vehicle (DMSO) and quantified for cell death as in c. qVD-OPh decreased Tom40 RNAi-induced cell death. *P < 0.0001 compared to U6 (DMSO), ^#P < 0.0001 compared to U6-Tom40.3 (DMSO), **P < 0.0001 compared to U6 (qVD-OPh), F_3,20= 94.96, n = 6 coverslips per condition from 3 independent experiments. (f) Knockdown of Tim23 using shRNA lentiviruses (Tim23.a, Tim23.b and Tim23.c) was confirmed in cortical neurons by immunoblotting. Ctrl: luciferase shRNA lentivirus. (g) DIV 5 neurons infected with Tim23 shRNA or ctrl shRNA lentiviruses were subjected to MTS assays at DIV 12. Tim23 knockdown decreased mitochondrial metabolic activity in cortical and striatal neurons (cortical neurons: *P < 0.0001 compared to ctrl, F_3,66 = 121.96, n = 13 (ctrl), 19 (Tim23.a, Tim23.b, Tim23.c) cultures per group from 6 independent experiments; striatal neurons: *P < 0.0001 compared to ctrl, F_3,69 = 57.93, n = 16 (ctrl), 19 (Tim23.a, Tim23.b, Tim23.c) cultures per group from 6 independent experiments). (h,i) Neurons transduced as in g were fixed at DIV 12 and subjected to immunofluorescence with indicated antibodies and nuclear staining (Hoechst 33342). Cell death was assessed by nuclear morphology. Representative images of transduced cortical neurons (h). Tim23 knockdown significantly increased the percentage of dead cells in cortical (*P < 0.0001 compared to ctrl, F_3,24= 60.98, n = 7 cultures per condition from 3 independent experiments) and striatal neurons (*P = 0.0004 and ^#P = 0.0013 compared to ctrl, F_3,28= 7.623, n = 8 cultures per condition from 3 independent experiments) (i). (j) Knockdown of Tim23 was confirmed using N2a mouse neuroblastoma cells transfected with Tim23 RNAi plasmid (U6-Tim23/CMV-GFP) or control scrambled plasmid (U6/CMV-GFP) by immunoblotting. (k,l) DIV 5 cortical neurons were transfected with U6-Tim23/CMV-GFP or U6/CMV-GFP plasmid. At DIV 8, loss of mitochondrial membrane potential and cell death were assessed using TMRM and nuclear dye RedDot2 by live confocal imaging. Representative images (z projection) at indicated time points demonstrate Tim23 knockdown induces mitochondrial depolarization and subsequent cell death (k). The percentage of TMRM⁺ neurons and live neurons among GFP⁺ transfected neurons was quantified (l). Tim23 knockdown significantly decreased the number of TMRM⁺ neurons and live neurons compared to control transfection (log-rank test, *P = 0.002; n = 49 neurons from 3 wells per group). Experiments (a,f,j) were successfully repeated three times and full-length blots/gels are presented in Supplementary Figure 9. (c,e,i) 100–200 neurons per coverslip were counted. Scale bars: 10 µm (b,d,k), 20 µm (h). Data are presented as mean + s.e.m. (c,e,g,i) or mean ± s.d. (d). (e,g,i) One-way ANOVA, Bonferroni t-test.

Figure 6

Augmentation of mitochondrial protein import rescues neurons from mutant Htt-induced death. (a) DIV 5 cortical neurons were transduced with WT Httex1-25Q (Htt25Q) or mutant Httex1-72Q (Htt72Q) lentivirus and at DIV 6 were cotransduced with lentivirus expressing three subunits of the TIM23 complex, Tim23, Tim50 and Tim17a. Mitochondria isolated from transduced neurons at DIV 10 were subjected to pOTC import assay (30 min). Overexpression of the TIM23 complex subunits increased pOTC import in Htt72Q neurons (*P = 0.007 compared to GFP-expressing Htt72Q neurons, F_2,27 = 14.05, n = 10 samples per condition from 5 independent experiments). (b) Primary cortical neurons transduced as in a were subjected to MTS assays at DIV 14. Overexpression of the TIM23 complex subunits partially but significantly metabolic activity (*P = 0.006 compared to GFP-expressing Htt72Q neurons, F_2,45 = 59.39, n = 17 wells per group for GFP-expressing Htt25Q and GFP-expressing Htt72Q neurons, n = 14 wells for TIM23-expressing Htt72Q neurons from 4 independent experiments). (c) Primary cortical neurons transduced as in a were assessed for cell death by scoring nuclear morphology at DIV 14. Htt72Q–expressing neurons showed increased cell death compared to Htt25Q–expressing neurons (^#P < 0.0001, GFP-expressing Htt72Q compared to GFP-expressing Htt25Q neurons). Overexpression of the TIM23 complex subunits in Htt72Q neurons inhibited neuronal death (*P < 0.0001 compared to GFP-expressing Htt72Q neurons). F_3,51 = 29.47, n = 16 (GFP-expressing Htt25Q neurons), 15 (GFP-expressing Htt72Q neurons) and 12 (TIM23-expressing Htt72Q and TIM23-expressing Htt25Q neurons) wells per group from 3 independent experiments; 200 neurons were counted per well. (a–c) Data are presented as mean + s.e.m. One-way ANOVA, Bonferroni t-test.