ATAD3A oligomerization causes neurodegeneration by coupling mitochondrial fragmentation and bioenergetics defects

Abstract

Mitochondrial fragmentation and bioenergetic failure manifest in Huntington's disease (HD), a fatal neurodegenerative disease. The factors that couple mitochondrial fusion/fission with bioenergetics and their impacts on neurodegeneration however remain poorly understood. Our proteomic analysis identifies mitochondrial protein ATAD3A as an interactor of mitochondrial fission GTPase, Drp1, in HD. Here we show that, in HD, ATAD3A dimerization due to deacetylation at K135 residue is required for Drp1-mediated mitochondrial fragmentation. Disturbance of ATAD3A steady state impairs mtDNA maintenance by disrupting TFAM/mtDNA binding. Blocking Drp1/ATAD3A interaction with a peptide, DA1, abolishes ATAD3A oligomerization, suppresses mitochondrial fragmentation and mtDNA lesion, and reduces bioenergetic deficits and cell death in HD mouse- and patient-derived cells. DA1 treatment reduces behavioral and neuropathological phenotypes in HD transgenic mice. Our findings demonstrate that ATAD3A plays a key role in neurodegeneration by linking Drp1-induced mitochondrial fragmentation to defective mtDNA maintenance, suggesting that DA1 might be useful for developing HD therapeutics.

Fig. 1

ATAD3A binds to Drp1 in HD. a Affinity purification followed by tandem mass spectrometry analysis was conducted to identify Drp1-interacting  proteins in striatal neuronal cells derived from HD patient-iPS or normal subject-iPS cells (also see Supplementary Fig. 1A). The molecular and cellular functions of the Drp1 interactors in HD patient-derived neuronal cells are shown. b Upper: cellular location of the Drp1 interactors in HD neuronal cells. In particular, 32% of proteins are located on the mitochondria. ATAD3A ranked as a top candidate. Lower: Among mitochondrial candidates, mitochondrial nucleoid organization and energy production are enriched, and each account for up to 25% of all mitochondrial candidates. Immunoprecipitation (IP) of total protein lysates was performed in HdhQ7 and HdhQ111 cells (c), in wildtype striatal cells exposed to 3-NP (5 mM for 4 h) (d), in striatal extracts of YAC128 mice (6 months old) or R6/2 mice (12 weeks old) or age-matched wildtype mice (e), and in fibroblasts of HD patient (HD1, GM4208, 35 years old, Male) or normal subject (Con1, Huf1) (f). Data are mean ± SEM. Student t test. n = 3. g Total protein lysates of frozen postmortem caudate nucleus of normal subjects and HD patients with different disease severity (grade 1 to grade 4, slight to severe) were subject to IP analysis. Nor: normal subject 5214; G1: HD 4283, grade 1; G2: HD 4557, grade 2; G3: HD 3573, grade 3; G4: HD 2903, grade 4. Patient information shown in Supplementary Fig. 1F. The data below the blots indicate the relative density of Drp1 in ATAD3A immunoprecipitates (Drp1/ATAD3A in IP analysis). h GST-Drp1 (2 µg) and ATAD3A-Flag (2 µg) purified proteins were incubated in vitro for 30 mins. Left: IP with anti-Flag antibodies followed by western blot with the indicated antibodies. Right: GST pull down analysis. All shown representative blots are from at least 3 independent experiments

Fig. 2

Enhanced ATAD3A oligomerization in HD models. a Western blot (WB) analysis of rat liver mitochondrial fractionations with the indicated mitochondrial antibodies. OM: mitochondrial outer membrane. CS: contact sites. IM: mitochondrial inner membrane. b HdhQ7 and HdhQ111 cells were stained with anti-ATAD3A and anti-Mitofilin (a mitochondrial contact site protein) or anti-Cyto C (a mitochondrial intermembrane space protein) antibodies. Nuclei were stained with DAPI. In situ Duolink proximity ligation assay (PLA) was performed. Histogram: quantitation of the number of PLA-positive signals (red). At least 200 cells/group were analyzed. 3 independent experiments, unpaired Student t-test. Scale bar: 10 µm. c Immunogold electron microscopy analysis with an antibody directed against ATAD3A N-terminus was performed in HdhQ7 and HdhQ111 cells. Bottom images are the images boxed. Scale bar: 100 nm. Blue dash line: OMM, mitochondrial outer membrane. Yellow dash line: IMM, mitochondrial inner membrane. Red arrowhead: ATAD3A. Histogram: the percentage of ATAD3A immunopositive gold particles on the contact site versus total number of gold particles. 3 independent experiments, unpaired Student t-test. d Left: ATAD3A protein level was determined by WB in the presence or absence of β-ME. Right: Protein lysates were incubated with the crosslinker BMH (1 mM). e HdhQ7 and Q111 cells were treated with BMH (1 mM, 20 min). f Total lysates of striatum from YAC128 (12 months old), R6/2 (12 weeks old), or age-matched wildtype mice were analyzed by WB with anti-ATAD3A antibody in the presence or absence of β-ME. n = 6–10 mice/group. g ATAD3A oligomers were analyzed in HD patient fibroblasts (HD1: GM21756, Female; HD2: GM04693, Male; HD3: GM21756, Female) and normal subjects (Con 1: nHDF, fibroblasts from juvenile; Con 2: HDF, fibroblasts from adult; Con 3: Huf1822, adult). h ATAD3A oligomers were analyzed in total lysates of HD patient postmortem brains (cortex: left; caudate nucleus: right) or normal subjects, under non-reducing condition. [Cortex: Normal X5302; HD X5298. Caudate nucleus: Normal 5214; HD Grade1 (G1): 4283; HD Grade2 (G2): 4557; HD Grade3 (G3): 3573]. See Supplementary Fig. 1F for patient information. The data are mean ± SEM. Shown representative blots are from at least 3 independent experiments

Fig. 3

ATAD3A oligomerization impairs mitochondrial fission and mtDNA stability. HdhQ7 and Q111 cells were transfected with control siRNA (NC) or ATAD3A siRNA (siA) for 3 days. a Downregulation of ATAD3A was validated by WB. Actin: a loading control. Drp1 polymerization was analyzed by WB with anti-Drp1 antibodies in the absence of β-ME. b Mitochondria were isolated from HD striatal cells (Upper) and cells exposed to 5 mM 3-NP for 4 h (Lower). Drp1 mitochondrial level was analyzed by WB. Mitochondrial loading control: VDAC. c HeLa cells were transfected with ATAD3A-GFP truncated mutants, shown in Supplementary Fig. 3E, for 48 h. Cells were stained with anti-Tom20 (red) and anti-DNA (cyan) antibodies. The co-localization of DNA and Tom20 was analyzed by confocal microscopy. Insert: the enlarged images. Scale bar: 10 µm. d The number of nucleoids immunopositive for both anti-DNA and anti-Tom20 was quantitated by NIH Image J software. 40–50 cells per group were analyzed. Three independent experiments, one-way ANOVA with Tukey’s post-hoc test. e Neuro2A cells were transfected with ATAD3A-Flag WT or mutants for 48 h, and total DNA was extracted for qPCR analysis to monitor the mtDNA lesion. Upper: Representative DNA agarose gel of the amplification of the 10 kb mtDNA fragment. Lower: The quantification of mtDNA lesion. Four independent experiments, one-way ANOVA with Tukey’s post-hoc test. f mtDNA content was analyzed by qPCR using primers from D-loop (left). TFAM mRNA level was analyzed by qPCR (right). At least 3 independent experiments, one-way ANOVA with Tukey’s post-hoc test. g Left: Cells were stained with mitoSOX fluorescent probe to evaluate mitochondrial superoxide production (mitoROS). Right: Cell death rate was measured by LDH release into cytosol. Four independent experiments, one-way ANOVA with Tukey’s post-hoc test. h The binding of TFAM and biotinylated mtDNA LSP probe was determined by biotin-streptavidin pull down in HD striatal cells transfected with ATAD3A siRNA (left) and in Neuo2A cells expressing ATAD3A truncated mutants (right). All shown representative blots are from at least 3 independent experiments. All data are mean ± SEM

Fig. 4

ATAD3A deacetylation. a Total cell lysates of HdhQ7 and HdhQ111 cells or fibroblasts of HD patient and control subjects were subject to IP with anti-ATAD3A antibodies followed by WB with anti-acetyl lysine antibodies. Asterisks (*) indicates the acetylated ATAD3A. b ATAD3A K135 in peptide 130-AQYADLLAR-138 in human (K134 in mouse) was identified as an acetylated site (also see Supplementary Fig. 6A). The K135 labeled in pink is present on the surface of ATAD3A N-terminus simulated structure. HEK293 cells were transfected with the indicated Flag-tagged K135 mutants or ATAD3A WT for 48 h. c ATAD3A dimers were analyzed by WB in the presence or absence of β-ME. Histogram: the quantification of relative density of ATAD3A dimer in the absence of β-ME versus total protein level under reduced conditions. At least 6 independent experiments, one-way ANOVA with Tukey’s post-hoc test. d Immunoprecipiates with anti-Flag antibodies were analyzed with anti-Myc antibodies. Histogram: the quantification of relative density of Myc-ATAD3A/ATAD3A-Flag in the immunoprecipitates. Five independent experiments, one-way ANOVA with Tukey’s post-hoc test. e IP of total protein lysates was performed with the indicated antibodies. f HeLa cells were transfected with Flag-ATAD3A WT or K mutants for 48 h. Cells were stained with anti-Tom20 (green) and anti-Flag (red) antibodies. Mitochondrial morphology was imaged. Scale bar: 10 µm. Histogram: the percentage of cells with fragmented mitochondria to total number of cells. Three independent experiments, one-way ANOVA with Tukey’s post-hoc test. g The binding of TFAM and biotinylated mtDNA LSP probe was determined by biotin-streptavidin pull down. Histogram: the relative density of TFAM in the Biotin-LSP precipitates. Three independent experiments, one-way ANOVA with Tukey’s post-hoc test. h HeLa cells were stained with anti-Tom20 (red) and anti-DNA (blue) antibodies at the indicated groups. The co-localization of DNA and Tom20 was analyzed by confocal microscopy. Insert: the enlarged images. Scale bar: 10 µm. The number of nucleoids immunopositive for both anti-DNA and anti-Tom20 was quantitated by NIH Image J software. Three independent experiments, one-way ANOVA with Tukey’s post-hoc test. All shown blots are from at least 3 independent experiments. Data are mean ± SEM

Fig. 5

DA1 peptide blocks Drp1/ATAD3A binding. a Sequence of homology between Drp1 (human, NP_036192) and ATAD3A (human, NP_001164006). Columns (:) indicate identical amino acids; single dot (·) indicate high similarity between amino acids. b Stick drawings of ATAD3A and Drp1 main domains. Highlighted in red are the two regions of homology between the two proteins, region DA1 in Drp1 and the corresponding region DA2 in ATAD3A. c Mapping DA1 on crystal structure of Drp1 GTPase-GED fusion construct and DA2 on the N-terminus (residues 1–210) of ATAD3A simulated structure. d Upper: HdhQ7 and HdhQ111 cells were treated with TAT or peptide DA1 (1 µM/day for 4 days). Lower: R6/2 or wildtype mice were treated with TAT or DA1 (1 mg/kg/day) from 6 to 12 weeks. The total lysates of cells or mouse striatum were subject to IP analysis. Three independent experiments. e DA1 or TAT was incubated with recombinant proteins in vitro for 30 mins. GST pull down followed by WB analysis was carried out. 3 independent experiments. f Biotin-conjugated DA1 or TAT (10 µM, each) was incubated with total lysates of HdhQ7 and HdhQ111 cells. Immunoprecipitates were analyzed by WB. 3 independent experiments. g HEK293 cells were transfected with ATAD3A-Flag wildtype (WT) or ΔN50 mutant plasmids followed by TAT or DA1 treatment (1 µM, each) for 48 h. ATAD3A oligomerization was analyzed by WB. Three independent experiments. h Left: HD cells were treated with TAT or DA1 (1 μM/day for 4 days), 4 independent experiments. YAC128 or wildtype mice from 6 months of age (Middle, 6–7 mice/group), and R6/2 or wildtype mice from 12 weeks of age (Right, 4 mice/group) received either TAT or DA1 (1 mg/kg/day). Drp1 mitochondrial levels were determined by WB. i Drp1 oligomerization was analyzed using total striatal lysates of mice (Upper: YAC128-12 months old; Lower: R6/2 mice-12 weeks old). n = 6 mice/group. j Drp1 oligomerization was analyzed by WB at the indicated groups. Three independent experiments. All data are mean ± SME. One-way ANOVA with Tukey’s post-hoc test

Fig. 6

DA1 treatment reduces mitochondrial damage and cell death in HD cell cultures. HdhQ7 and HdhQ111 striatal cells were treated with control TAT or peptide DA1 (1 µM/day for 4 days). a Mitochondrial morphology was determined by staining cells with anti-Tom20 antibody. Scale bar: 10 µm. The percentage of cells with fragmented mitochondria relative to the total number of cells was quantitated. Three independent studies, one-way ANOVA with Tukey’s post-hoc test. b TFAM and PGC1α protein levels were analyzed by WB. Histogram: the relative density of TFAM and PGC1α to Actin. Three independent studies, one-way ANOVA with Tukey’s post-hoc test. c HdhQ7 and HdhQ111 cells were transfected with control siRNA (siCon) or ATAD3A siRNA (siAD) followed by treatment with TAT or DA1 (1 µM/day for 4 days). TFAM and ATAD3A protein levels were analyzed by WB. Histogram: the relative density of TFAM to Actin. Four independent studies, one-way ANOVA with Tukey’s post-hoc test. d mtDNA-encoded or nuclear-encoded mitochondrial electron transport proteins were analyzed by WB with the indicated antibodies. Actin: a loading control. VDAC: a mitochondrial loading control. Histogram: the quantitation of relative mtCO2 protein level to Actin. Three independent studies, one-way ANOVA with Tukey’s post-hoc test. e TFAM and mtDNA LSP interaction was analyzed by biotin-streptavidin pull down. Histogram: the relative density of TFAM in the Biotin-LSP precipitates. Three independent experiments, one-way ANOVA with Tukey’s post-hoc test. f mtDNA lesion was measured by qPCR. Representative DNA gel is showed in Supplementary Fig. 7J. Five independent studies, one-way ANOVA with Tukey’s post-hoc test. g mitoROS was evaluated by mitoSOX fluorescent probe. At least 100 cells per group were analyzed. At least 3 independent studies, one-way ANOVA with Tukey’s post-hoc test. h Mitochondrial respiratory activity was determined by a seahorse analyzer. Three independent studies, one-way ANOVA with Tukey’s post-hoc test. All data are mean ± SEM

Fig. 7

DA1 treatment reduces HD-associated neuropathology. a Neurons were stained with anti-GAD67 antibodies to indicate striatal neurons and anti-Tom20 antibodies to label mitochondria. See Supplementary Fig. 7K for a cluster of neurons. Scale bar: 10 μm. Mitochondrial length along the neurite of GAD67-immunopositive neurons was quantitated. At least 50 neurons/group were analyzed. b Neurons were stained with anti-MAP2/anti-GAD67 (left) or anti-Tau/anti-DARPP32 (right) to indicate dendritic and axonal morphology, respectively. MAP2⁺ dendrite and Tau⁺ axon length in neurons were quantitated. At least 50 neurons/group were analyzed. Scale bar: 10 μm. c mitoROS was determined by mitoSOX probe. d Cell death induced by BDNF withdrawal for 24 h was determined by the release of LDH. Three independent experiments. HD R6/2 and wildtype mice were treated with either TAT or DA1 from 6 weeks (see Supplementary Fig. 8A). e Body weight was recorded. **p < 0.001, *p < 0.05 vs. HD mice treated with TAT. Student t-test. TAT: 21 mice; DA1: 15 mice. f Survival of R6/2 mice from the age of 6 to 21 weeks was analyzed by Log-rank (Mantel-Cox) test. g locomotion activity of R6/2 mice (12 weeks old) was determined (WT/TAT: 16 mice; WT/DA1: 10 mice; R6/2/TAT: 13 mice; R6/2/DA1: 9 mice). Shown is horizontal activity. h YAC128 and wildtype mice were treated with TAT or DA1 from the age of 3 to 12 months (see Supplementary Fig. 8A). Locomotion activity was monitored (n = 12–20 mice/group). Shown is vertical activity. Two-way ANOVA with Tukey’s post-hoc test. #p < 0.05 vs. HD mice treated with TAT; *p < 0.05 vs. wildtype mice treated with TAT. i DARPP-32 immunodensity was assessed in the dorsolateral striatum of TAT-treated or DA1-treated R6/2 mice (12 weeks old). Scale bar: 100 µm. n = 6 mice/group. j Golgi-Cox staining of mouse brain was shown. The total dendritic length/neuron was quantitated. 20 neurons/group was analyzed. Scale bar: 10 µm. k Striatum volume was assessed. n = 5–7 mice/group. l DARPP-32 and PSD95 protein levels in YAC128 mouse (12 months old) striatal extracts were analyzed by WB. n = 7 mice/group. All data are mean ± SEM. One-way ANOVA with Tukey’s post-hoc test in a–d, g, i–l

Fig. 8

DA1 treatment reduces mtDNA damage in HD mice. Mitochondria were isolated from striatum of R6/2 or age-matched wildtype mice (12 weeks old) (a), and YAC128 or age-matched wildtype mice (6 months old) (b). Protein levels of mtCO2 and TFAM were determined by WB. VDAC: a mitochondrial loading control. Histogram: the relative density of mtCO2 and TFAM to VDAC. n = 3–6 mice/group. One-way ANOVA with Tukey’s post-hoc test. c Plasma samples were harvested from YAC128 mice or wildtype mice (6 months old). The mtDNA content in plasma was examined by qPCR. One-way ANOVA with Tukey’s post-hoc test. n = 6 mice/group. d Brain coronal sections (10 µm) of R6/2 (left, 12 weeks old) or YAC128 (right, 6 months old) mice were stained with anti-Iba1 antibody to determine microglial activity. Boxed images were enlarged on the bottom. Scale bar: 10 μm. Histogram: the quantitation of Iba1 immunodensity/mm². n = 3 mice/group. ANOVA with Tukey’s post-hoc test. All data are mean ± SEM. e A summary scheme. Under stressed conditions, such as HD, ATAD3A forms oligomers due to K135 deacetylation via its coiled-coil domain, and recruits fission protein Drp1 to the mitochondria where ATAD3A and Drp1 form a complex. This leads to mitochondrial fragmentation. ATAD3A oligomerization impairs mtDNA maintenance by disrupting the binding between TFAM and mtDNA, resulting in the loss of mtDNA and subsequent mitochondrial bioenergetics defects. As a result, ATAD3A oligomerization simultaneously causes mitochondrial fragmentation and mitochondrial bioenergetics defects, which lead to mitochondrial dysfunction and neuronal cell death. DA1 peptide directly binds to ATAD3A to block ATAD3A and Drp1 interaction. Treatment with DA1 quenches ATAD3A oligomerization which therefore reduces excessive mitochondrial fission and the loss of mtDNA. Consequently, treatment with DA1 both in HD cultures and in HD animals reduces HD-associated neuropathology