Loss of Parkinson's disease-associated protein CHCHD2 affects mitochondrial crista structure and destabilizes cytochrome c

Mutations in CHCHD2 have been identified in some Parkinson's disease (PD) cases. To understand the physiological and pathological roles of CHCHD2, we manipulated the expression of CHCHD2 in Drosophila and mammalian cells. The loss of CHCHD2 in Drosophila causes abnormal matrix structures and impaired oxygen respiration in mitochondria, leading to oxidative stress, dopaminergic neuron loss and motor dysfunction with age. These PD-associated phenotypes are rescued by the overexpression of the translation inhibitor 4E-BP and by the introduction of human CHCHD2 but not its PD-associated mutants. CHCHD2 is upregulated by various mitochondrial stresses, including the destabilization of mitochondrial genomes and unfolded protein stress, in Drosophila. CHCHD2 binds to cytochrome c along with a member of the Bax inhibitor-1 superfamily, MICS1, and modulated cell death signalling, suggesting that CHCHD2 dynamically regulates the functions of cytochrome c in both oxidative phosphorylation and cell death in response to mitochondrial stress.

. The loss of dCHCHD2 leads to oxidative DNA damage. (a) Detection of protein oxidation by the Oxyblot assay was performed as in Fig. 8e. Thoraxes of 45-day-old flies were used for the Oxyblot assay. Protein oxidation signals were normalized based on Actin, and the signals were increased in dCHCHD2-deficient flies (n = 5, *p = 0.047 by two-tailed t-test). Signal specificity was confirmed in blots omitting 2-4dinitrophenyl hydrazine (DNPH) treatment. (b) The GSH/GSSG ratio was reduced in dCHCHD2-deficient flies. (n = 4, *p = 0.020, ***p = 0.0002 vs. age-matched dCHCHD2 +/+ by two-tailed t-test) (c) The levels of the OXPHOS proteins, SOD1 and DJ-1b in dCHCHD2deficient flies. Western blot analysis of the thorax tissues of 14-day-old adult flies (left) and embryos (right) used in Fig. 4 and Supplementary Fig. 4 are shown. DJ-1b -/--deficient flies (DJ-1b ∆93/DJ-1b ∆93) were also used as a control. Note that some of the OXPHOS proteins were not detected in embryos due to less-abundant expression. Asterisks indicate uncharacterized anti-complex I (NDUFS3)-detectable bands. (d) Whole brain tissues of 30day-old flies with the indicated genotypes were stained with anti-8-OHdG and anti-dTH antibodies. Scale bar = 75 µm. Intensity of anti-8-OHdG immunoreactive signals in whole brain (left graph, n = 6, *p = 0.020 by two-tailed student's t-test) and TH-positive regions (right graph, n = 19-51, **p = 0.0036, *p = 0.021 vs. Rev. by two-tailed student's t-test) were measured and graphed (mean ± s.e.m.). The numbers of samples analyzed are indicated in the graphs. Supplementary Fig. 3. Inactivation of subunits in the mitochondrial respiratory complexes and mitochondria-associated PD genes alters dCHCHD2 expression.
(a) Mutations in mitochondrial polymerase g lead to a loss of mitochondrial DNA (mtDNA). The ratio of mtDNA to genomic DNA of actin was measured by quantitative PCR, and values are normalized to those of the w-control (mean ± s.e.m., n = 9 in each group). ***P < 0.0001 vs. w-(one-way ANOVA with Tukey-Kramer test). (b) A series of UAS-RNAi fly lines against the respiratory complex subunits, the inhibition of which has been reported to affect lifespan extension and modulation of mitochondrial phenotypes in PD model flies 1, 2, 3, 4, 5 , were crossed with the MHC-GAL4. Endogenous levels of dCHCHD2 and ATP5a (Complex V, a-subunit) in the thorax muscles were analyzed by western blot. LacZ RNAi and dCHCHD2 RNAi lines served as negative and positive controls, respectively. dCHCHD2 H43 , revertant (Rev), and w-were also included as controls. The graph represents relative values (mean ± s.e.m.) of dCHCHD2 signals normalized to ATP5a signals from six independent samples. *p = 0.041, ***p = 0.0008 vs. LacZ (one-way ANOVA with Tukey-Kramer test). Control w-by two-tailed student's t-test. The graphs represent relative values (mean ± s.e.m., n = 4-6).

Supplementary Fig. 4. Complex IV regulation is not a primary role of CHCHD2.
(a) Measurement of the activities of the four respiratory chain complexes. Mitochondria isolated from thorax muscles were subjected to spectrophotometric analysis of the respiratory chain enzyme activities. All of the activities are presented as percentages of the values of dCHCHD2 +/+ (mean ± s.e.m., n = 12-16 from at least three independent experiments). There were no significant differences in the activity of each complex between the two genotypes (two-tailed t-test). The sample size is indicated in the graph. (b) CHCHD2 does not significantly bind to complex IV. CHCHD2-PA was purified from cultured cells using anti-PA tag beads (NZ1) and incubated with (Complex IV, +) or without (Complex IV, -) purified complex IV. NZ1 beads alone were also incubated with the same quantity of complex IV in parallel as a mock control. The samples eluted from the beads and input were subjected to SDS-PAGE. Coprecipitated complex IV was evaluated via CBB staining (left) and western blotting with anti-complex IV subunit IV (right). An asterisk indicates bands corresponding to CHCHD2-PA. Western blot analysis indicated that, at most, 0.1% of the input of subunit IV coprecipitated with CHCHD2 (lane 1). However, considering the observation that a similar quantity of subunit IV was detected in the absence of CHCHD2 (lane 3), the binding between CHCHD2 and subunit IV was very weak, if any binding occurred at all.

Supplementary Fig. 5. hCHCHD2 interacts with MICS1 in human cultured cells.
(a) (Upper) Amino acid sequence of human MICS1. We purified binding proteins using human CHCHD2 WT, T61I and R145Q as probes. Experiments were independently repeated twice, and subsequent mass spectrometry analysis was repeated twice per experiment. A peptide fragment shown in red was detected three and four times in CHCHD2 WT and T61I samples, respectively. A peptide fragment shown in blue was detected once in T61I samples. (Middle and lower) Representative MS/MS spectra assigned to EAALEPSMEK from CHCHD2 WT and T61I samples are shown. (b) HeLa cells transfected with hCHCHD2-Myc in combination with Cyto c-FLAG or with MICS1-HA were co-stained with anti-Myc, anti-FLAG or anti-HA antibodies. Mitochondria were visualized with MitoTracker Deep Red (Thermo Fisher Scientific). Merged images of green and red channels are also shown. Note that the mitochondria were fragmented following MICS1 overexpression, which is consistent with a previous report 6 . Scale bars = 10 µm. (c) HEK293T cells were transfected with MICS1-HA with or without hCHCHD2-Myc or hCHCHD2-FLAG. Cells treated with or without 0.5 µg ml -1 DSP in PBS for 20 min at RT were lysed and subjected to immunoprecipitation using anti-FLAG or anti-Myc magnetic beads (Medical & Biological Laboratories). MICS1-HA was specifically co-precipitated with hCHCHD2-FLAG or hCHCHD2-Myc. (d) HEK293T cells were transfected with MICS1-HA with or without hCHCHD2-FLAG WT or its mutants. Immunoprecipitation using anti-FLAG was performed as in (c). The amounts of precipitated R145Q were lower than WT or T61I probably due to its impaired dimerization activity (see also Fig. 7e). (e) PD-associated mutations do not affect the binding of CHCHD2 to Cyt c. HEK293T cells were transfected with Cyt c-FALG with or without hCHCHD2-Myc WT or its mutants and treated with 0.5 µg ml -1 DSP. Immunoprecipitation using anti-Myc was performed as in (c). Supplementary Fig. 6. CHCHD2 suppresses Cyt c release upon oxidative stress treatment.
(a) MEFs were prepared from CHCHD2 +/+ and CHCHD2 -/mouse embryos. An empty vector, hCHCHD2 WT, T61I or R145Q was virally introduced into CHCHD2 -/-MEFs. CHCHD2 expression levels were comparable in WT and PD mutant cells. (b) Human CHCHD2 is localized in mitochondria of CHCHD2 -/-MEFs. Cells were stained with anti-CHCHD2 and anti-Hsp60 antibodies. Scale bars = 50 µm. (c) CHCHD2 suppresses Cyt c release caused by peroxide treatment. CHCHD2-deficient MEFs harboring empty vector (KO), hCHCHD2 WT, T61I or R145Q expression plasmid were used, as in Fig. 5c and d. Cells were treated with 50 nM peroxide for the indicated periods of time. The levels of cytosolic Cyt c (mean ± s.e.m.) were determined via western blotting and were normalized to lactate dehydrogenase (LDH) levels. *p < 0.05, **p < 0.01, ***p < 0.001 vs. the same periods in WT from seven independent experiments (one-way ANOVA with Tukey-Kramer test). Supplementary Fig. 7. dMICS1 rescues motor defects in dCHCHD2 -/flies. (a) Knockdown efficiency of dMICS1. The levels of dMICS1 transcripts were measured using quantitative RT-PCR, which were then normalized to housekeeping rp49 levels. Expression of dMICS1 RNAi and LacZ RNAi was induced by the MHC-GAL4 driver. dMICS1 RNAi reduced the dMICS1 transcript levels by more than 50% compared with LacZ RNAi. The graph represents relative values (mean ± s.e.m.). n = 3 from three independent samples. (b) LacZ or dMICS1 was driven by Da-GAL4 at 22˚C. The body length of larvae was measured 6 days post hatch. dMICS1 overexpression (OE) led to growth retardation (***p < 0.0001 by one-way ANOVA with Tukey-Kramer test) while dCHCHD2 expression did not affect this phenotype and the hatching efficiency. The sample size is indicated in the graph (n = 9-21). Scale bar = 250 µm. (c) Climbing defects in dCHCHD2 -/flies were rescued by dMICS1 overexpression. dMICS1 was expressed in the muscle tissues using the MHC-GAL4 driver. Thirty-day-old flies were analyzed (mean ± s.e.m.). Twenty trials with 49-50 flies from three independent experiments. ***p = 0.0001.