Polyglutamine tracts regulate beclin 1-dependent autophagy

Abstract

Nine neurodegenerative diseases are caused by expanded polyglutamine (polyQ) tracts in different proteins, such as huntingtin in Huntington's disease and ataxin 3 in spinocerebellar ataxia type 3 (SCA3). Age at onset of disease decreases with increasing polyglutamine length in these proteins and the normal length also varies. PolyQ expansions drive pathogenesis in these diseases, as isolated polyQ tracts are toxic, and an N-terminal huntingtin fragment comprising exon 1, which occurs in vivo as a result of alternative splicing, causes toxicity. Although such mutant proteins are prone to aggregation, toxicity is also associated with soluble forms of the proteins. The function of the polyQ tracts in many normal cytoplasmic proteins is unclear. One such protein is the deubiquitinating enzyme ataxin 3 (refs 7, 8), which is widely expressed in the brain. Here we show that the polyQ domain enables wild-type ataxin 3 to interact with beclin 1, a key initiator of autophagy. This interaction allows the deubiquitinase activity of ataxin 3 to protect beclin 1 from proteasome-mediated degradation and thereby enables autophagy. Starvation-induced autophagy, which is regulated by beclin 1, was particularly inhibited in ataxin-3-depleted human cell lines and mouse primary neurons, and in vivo in mice. This activity of ataxin 3 and its polyQ-mediated interaction with beclin 1 was competed for by other soluble proteins with polyQ tracts in a length-dependent fashion. This competition resulted in impairment of starvation-induced autophagy in cells expressing mutant huntingtin exon 1, and this impairment was recapitulated in the brains of a mouse model of Huntington's disease and in cells from patients. A similar phenomenon was also seen with other polyQ disease proteins, including mutant ataxin 3 itself. Our data thus describe a specific function for a wild-type polyQ tract that is abrogated by a competing longer polyQ mutation in a disease protein, and identify a deleterious function of such mutations distinct from their propensity to aggregate.

Extended Data Figure 1. Ataxin-3 contributes to autophagosome formation.

a, Primary cultures of mouse cortical neurons were transduced with control or ataxin-3 lentiviral shRNAs and analysed for the levels of LC3-I under starvation condition (HBSS, 4 hr) or together with BafA1 (400 nM, 4 hr). Results are normalised to control cells (HBSS+BafA1). Mean ± s.e.m, n=5 replicates from two independent cultures. Two-way ANOVA (N.S not significant). b, HeLa cells were transfected with different ataxin-3 siRNA and scrambled siRNA used as a control. Ataxin-3 knockdown (KD) efficiency is presented as well as basal LC3-II levels. LC3-II levels in ataxin-3-depleted HeLa cells were normalised to control cells, n=4 independent experiments. One-way ANOVA (** P<0.01) with post-hoc Tukey’s test (* P<0.05, ** P<0.01). c, p62 levels in HeLa cells depleted of ataxin-3 by siRNA. p62 levels were normalised to control cells (n=4 independent experiments, ** P<0.01 one-tailed paired t-test). d, HeLa cells stably expressing mTagRFP-mWasabi-LC3 reporter were transfected with either scrambled or ataxin-3 siRNA and were analysed by the ThermoFisher cellomics system for assessing the number of autophagosomes and autolysosomes in the cells. Results are mean number of autophagosomes or autolysosomes per cell ± s.e.m in eight fields from a representative experiment out of three independent experiments (* P<0.05, ** P<0.01 one-tailed unpaired t-test). Representative images of the cells were taken by confocal microscopy (total 800 cells). Scale bar is 10 µm. e, Control and ataxin-3 KD HeLa cells were starved (HBSS, 4 hr) or kept in full media. The number of PI3P phospholipid dots were analysed by staining with anti-PI3P antibody. Results are mean dots per cell ± s.d from a representative experiment out of three independent experiments as well as representative confocal images of PI3P dots (red) for each condition (n=20 cells). Scale bar is 10 µm. Two-way ANOVA (column factor siRNA *** P<0.001, row factor starvation ** P<0.01, interaction P value * P<0.05) with Bonferroni’s post-test (** P<0.01, N.S, not significant).

Extended Data Figure 2. Ataxin-3 regulates starvation-induced autophagy.

a-b, HeLa cells stably expressing GFP-LC3 were treated with control siRNA or ataxin-3 siRNA and incubated for 1 hr with carrier alone or carrier with 1 µM of PI3P phospholipid. Then, the control cells and the ataxin-3 KD cells with the different treatments were shifted to starvation condition (HBSS, 4 hr) or kept in full-media. a, The number of LC3 dots was analysed for each of the conditions and presented as mean LC3 dots per cell ± s.e.m. S.e.m. was determined from n=5 fields from a single representative experiment out of three independent experiments. Two-way ANOVA (column factor siRNA *** P<0.001, row factor starvation * P<0.05, interaction P value N.S) with Bonferroni’s post-test: for basal condition: N.S, for starvation: *** P<0.001. b, Representative confocal images of LC3 dots (green) from the different treatments are presented for the starvation condition. For a and b, the total number of cells analysed in basal control n=25; basal KD, basal KD carrier, basal KD carrier PI3P, n=30; HBSS control n=34; HBSS KD, HBSS KD carrier PI3P n=37; HBSS KD carrier n=32. Scale bar is 10 µm. c, Number of endogenous PI3P effector, WIPI2 dots in HeLa cells that were transfected with FLAG ataxin-3 or empty vector and starved (EBSS, 1hr) with or without the PI3P inhibitor Wortmannin (Wm, 20 nM). Data are presented as means ± s.e.m. of the number of WIPI2 dots per cell. S.e.m. was determined based on the total number of cells analysed using software described in methods from a representative experiment out of two independent experiments. Confocal images of WIPI2 dots (green) from the different treatments are shown. Number of cells analysed and used for the s.e.m. in Empty FLAG n=47; FLAG ataxin-3 n=45; FLAG ataxin-3/Wm n=37. Scale bar is 10 µm. One-way ANOVA (*** P<0.001) with post-hoc Tukey’s test (*** P<0.001). d-f, Mice were depleted of ataxin-3 in the liver by injection of ataxin-3 siRNA or control/scrambled siRNA formulations in the lateral caudal vein. The knockdown was left for 5 days with fasting on the 4th day. Livers from these mice were dissected, homogenised and proteins were resolved by SDS-page. d, Representative blots are shown, as well as in-vivo ataxin-3 knockdown efficiency. For the quantification of beclin 1 and LC3-II, see Fig. 1d,e. e, Quantification of p62 levels and f, Quantification of LC3-I levels in each group of mice (control fed n=9, ataxin-3 KD fed n=6, control fasted n=8, ataxin-3 KD fasted n=6, a.u arbitrary units). For e, two-way ANOVA (column factor siRNA * P<0.05, row factor fasting * P<0.05, interaction ** P<0.05) with Bonferroni’s post-test (** P<0.01, N.S). For f, two-way ANOVA (column factor siRNA P value N.S, row factor starvation * P<0.05, interaction P value N.S) with Bonferroni’s post-test (N.S). This suggests no obvious difference in LC3-I levels between the control and ataxin-3 KD groups.

Extended Data Figure 3. Ataxin-3 regulates beclin 1 stability and ubiquitination.

a, Beclin 1 levels in control siRNA-treated HeLa cells and ataxin-3 KD cells (t=0) and after cycloheximide (CHX, 50 µg/ml) treatment (t=8 hr). The percentage of beclin 1 degradation in control or ataxin-3 KD cells was compared and normalised to cells without CHX treatment (n=3 independent experiments, * P<0.05 one-tailed paired t-test). b, Beclin 1 levels in control siRNA-treated HeLa cells and ataxin-3 KD cells that were treated in the last 6 hr with proteasome inhibitor (MG132, 5 µM). n=3 independent experiments, one-way ANOVA (** P<0.01) with post-hoc Tukey’s test (*P<0.05, N.S, not significant). c, Beclin 1 levels in ataxin-3-depleted HeLa cells that were transfected with FLAG ataxin-3 wild-type (WT) or protease dead mutant, FLAG ataxin-3 C14A for 48hr. Results are normalised to control siRNA, n=3 independent experiments, one-way ANOVA (** P<0.01) with post-hoc Tukey’s test (** P<0.01). d, Control siRNA-treated and ataxin-3 siRNA-treated HeLa cells were transfected with the indicated vectors for 24 hr, treated in the last 6 hr with proteasome inhibitor (MG132, 10 µM) and endogenous beclin 1 was immunoprecipitated to detect beclin 1 ubiquitination. e, VPS34 levels in ataxin-3-depleted HeLa cells normalised to control siRNA, n=3 independent experiments, ** P<0.01 one-tailed paired t-test. f, HeLa cells were transfected with the indicated vectors for 24 hr, treated in the last 6 hr with proteasome inhibitor (MG132, 10 µM) and VPS34 was immunoprecipitated to detect VPS34 ubiquitination. The levels of the VPS34 components are co-ordinately regulated, and indeed decreased beclin 1 levels in ataxin-3-depleted cells were accompanied by decreased levels of VPS34. Still, no obvious change in VPS34 ubiquitination was observed in ataxin-3 over-expressing cells supporting a selective effect towards beclin 1. g, FLAG ataxin-3 WT and FLAG ataxin-3 C14A were co-expressed with GFP HTT exon 1 Q74 in HeLa cells for 48 hr. The number of aggregates was analysed by monitoring GFP fluorescence in 400 cells.. n=4 independent experiments. Results are normalised to control (empty vector). One-way ANOVA (** P<0.01) with post-hoc Tukey’s test (* P<0.05, N.S, not significant).

Extended Data Figure 4. Analysis of the beclin 1 lysine 402 modification.

a-b, HeLa cells were transfected with FLAG beclin 1 and HA-Ub for 24 hr, treated in the last 6 hr with proteasome inhibitor (MG132, 10 µM) and FLAG beclin 1 was immunoprecipitated for mass spectrometry analysis. Tryptic digests of ubiquitin-conjugated beclin 1 resulted in peptides that contain a ubiquitin remnant derived from the ubiquitin C-terminus (‘GG’ motif). a, Identification of a putative site of ubiquitination in beclin 1. Panel 1 shows the MSMS spectrum of the unmodified beclin 1 peptide spanning residues 401 to 416. Amino acids with corresponding y ions are shown in blue. Panel 2 shows the MSMS spectrum of an ion with a mass 114 Da greater than the unmodified peptide. The matching y ions and presence of a modified b2 ion indicate –GG modification of lysine 402. b, MSMS spectra filtered to high confidence covered 100% of the ubiquitin sequence. Tryptic peptide spanning residue 43 to 54 including lysine 48 was identified as the sole high confidence peptide with a modification corresponding to a –GG motif and the MSMS spectra of the peptide demonstrates fragments corresponding to a –GG modified lysine 48. c, Levels of FLAG beclin 1 and FLAG beclin 1 K402R in HeLa cells that were treated in the last 6 hr with proteasome inhibitor (MG132, 10 µM). Results are normalised to control (FLAG beclin 1 WT). n=3 replicates from two independent experiments. Two-way ANOVA (column factor K402R N.S, row factor MG132 *** P<0.001, interaction * P<0.05) with Bonferroni’s post-test (* P<0.05, N.S). d, HeLa cells were transfected with the indicated vectors for 24 hr and shifted in the last 4 hr to starvation media (HBSS). Beclin 1 and LC3-II levels were analysed and results are normalised to control (FLAG beclin 1 WT). For LC3-II levels, n=3 independent experiments, two-way ANOVA (column factor ataxin-3 * P<0.05, row factor K402R mutation ** P<0.01, interaction ** P<0.01) with Bonferroni’s post-test (*** P<0.001, N.S). For beclin 1 levels, n=4 independent experiments, two-way ANOVA (column factor ataxin-3 * P<0.05, row factor K402R mutation *** P<0.001, interaction P value N.S) with Bonferroni’s post-test (* P<0.05, N.S).

Extended Data Figure 5. Analysis of the interaction of the polyQ domain with beclin 1.

a, Empty FLAG, FLAG beclin 1 evolutionary conserved domain (ECD) alone, FLAG beclin 1 ΔECD, FLAG beclin 1 full length and GFP Q35 were transfected in HeLa cells for 24 hr and the cell lysates were immunoprecipitated with anti-FLAG antibody. Immunocomplexes were analysed using anti-GFP antibody. b, Superimposition of human beclin 1 ECD (pdb 4DDP) and Vps30 (pdb 5DFZ), the yeast orthologue of beclin 1. Structures reveal that the N-terminal helix (dark blue helix) of the human structure is displaced, most likely due to protein truncation for crystallographic purposes. The yeast structure suggests that this helix is part of the coiled-coil CC2 of beclin 1 instead of the ECD. c, Two binding-sites in human beclin 1 ECD reveal high docking scores for polyQ₇ (the docking scores for site 1 and site 2 are -10.394 and -10.721, respectively). Sites comprising the region adjacent to the N-terminal helix (dark blue) were not considered for the docking. d-e, Surface charge illustrations of human beclin 1 ECD with the two sites of polyQ interaction. Site 2 is in close proximity to a protruding hydrophobic loop (aromatic finger) composed by Phe359, Phe360 and Trp361 (top right e - cartoon view), which are thought to be implicated in beclin 1 anchorage to lipid membranes.

Extended Data Figure 6. Expression of polyQ tracts impairs beclin 1-dependent starvation-induced autophagy.

a, HeLa cells were transfected with empty GFP or GFP Q35 with or without FLAG ataxin-3 Q22 for 24 hr and were shifted to starvation condition (HBSS) in the last 4 hr. LC3-II and beclin 1 levels were analysed from the cell lysates. Results are mean ± s.e.m normalised to control (empty GFP), n=5 independent experiments, One-way ANOVA (for LC3-II ** P<0.01, for beclin 1 * P<0.05) with post-hoc Tukey’s test (* P<0.05, ** P<0.01, N.S, not significant). b, HeLa cells were transfected with empty GFP or GFP Q35 for 24 hr and were shifted to starvation condition (HBSS) in the last 4 hr. p62 levels were than analysed from the lysates. Results are normalised to control (empty GFP), n=3 independent experiments, * P<0.05 one-tailed paired t-test. c, HeLa cells were treated with 20 nM beclin 1 siRNA or scrambled siRNA (control) for 3 days. Beclin 1 KD efficiency is presented. d-e, Control and beclin 1 KD HeLa cells were transfected with empty GFP or GFP Q35 for 24 hr and were shifted to starvation condition (HBSS) in the last 4 hr. The number of endogenous LC3 dots (red) was analysed in the GFP-expressing cells (green). Results are mean number of LC3 dots per cell in four fields from a representative experiment out of three independent experiments, as well as confocal images for each condition (number of cells analysed in control GFP n=32, control GFP Q35 n=27, beclin 1 KD GFP n=25, beclin 1 KD GFP Q35 n=23). Scale bar is 10 µm. Two-way ANOVA (column factor GFP Q35 ** P<0.01, row factor beclin 1 KD *** P<0.001, interaction ** P<0.01) with Bonferroni’s post-test (*** P<0.001, N.S).

Extended Data Figure 7. Impaired starvation-induced autophagy and reduced beclin 1 levels in cells expressing expanded polyQ forms of huntingtin.

a, Empty FLAG, FLAG huntingtin (HTT) N-terminal fragment (1-350) Q17, FLAG HTT (1-350) ΔQ and beclin 1 were transfected in HeLa cells for 24 hr and cell lysates were immunoprecipitated with anti-beclin 1 antibody. Immunocomplexes were analysed using anti-FLAG antibody. b, GFP ataxin-3 Q28 and FLAG full-length HTT Q138 were transfected in HeLa cells for 24 hr and endogenous beclin 1 was immunoprecipitated. Immunocomplexes were analysed using anti-ataxin-3 antibody (detect GFP-ataxin-3) and anti-FLAG antibody (detect HTT). The ratio of the bound ataxin-3 to beclin 1 is presented. c, stable-inducible HEK293 cells were switched on for 48 hr with doxycycline (Dox) to express GFP-HTT wild-type exon 1 (GFP-HTT Q23) or mutant GFP HTT exon 1 (GFP-HTT Q74). In the last 4 hr cells were starved (HBSS) and beclin 1 levels were analysed in each cell type. Results are normalised to control HTT Q23 cells no Dox (n=4 independent experiments). Two-way ANOVA (column factor Dox ** P<0.01, row factor HEK cells N.S, interaction P value N.S) with Bonferroni’s post-test (** P<0.01, N.S). d-e, Quantification of the number of LC3 dots in the starved cells. Results are mean dots per cell in four fields of a representative experiment out of three independent experiments. Representative confocal images of endogenous LC3 dots (red) and GFP-HTT (green) in each of the conditions (number of cells analysed in GFP-HTT Q23 no Dox n=41; GFP-HTT Q23 with Dox n=34; GFP-HTT Q74 no Dox n=39; GFP-HTT Q74 with Dox n=43). Scale bar is 10 µm.. Two-way ANOVA (column factor Dox *** P<0.001, row factor HEK cells * P<0.05, interaction P value N.S) with Bonferroni’s post-test (*P<0.05, ** P<0.01).

Extended Data Figure 8. Impaired starvation-induced autophagy in striatal cell lines and in brain derived from mouse models of Huntington’s disease.

a, Striatal cell lines derived from HTT (Q7/Q111) heterozygous knock-in mouse and HTT (Q7/Q7) “wild-type” knock-in mouse were kept in full media or starved (EBSS, 1hr). In each experiment, cells were analysed for WIPI2 dots in different condition. We could not detect WIPI2 dots in full media in these cells as dots became apparent after starvation-induced autophagy. The number of WIPI2 dots per cell is presented normalised to control HTT (Q7/Q7) cells. n=3 independent experiments, * P<0.05 one-tailed paired t-test. Representative confocal images of WIPI2 (red) in each of the conditions are presented (n=80 cells analysed). Scale bar is 10 µm. b, HTT (Q7/Q111) and HTT (Q7/Q7) striatal cells were treated with bafA1 (400 nM) in full media or starved with HBSS together with bafA1 (400 nM) for 4 hr and analysed for LC3-II levels. Results are normalised to control (HTT (Q7/Q7) in full media). n=3 independent experiments, two-way ANOVA (column factor mut HTT *** P<0.001, row factor starvation ** P<0.01, interaction ** P<0.01) with Bonferroni’s post-test (*** P<0.001, N.S, not significant). c, beclin 1 levels in the starved HTT (Q7/Q111) and HTT (Q7/Q7) striatal cells. Results are normalised to control HTT (Q7/Q7) cells. n=3 independent experiments. ** P<0.01 one-tailed paired t-test. d, Sections of brains from Huntington’s disease (HD) HD-N171-N82Q transgenic young and adult mice (6 and 12 weeks old, respectively) were analysed for neuronal aggregates in the motor cortex. For each brain, 400 cells were counted in at least three sections. Results are mean percentage of cells with aggregates from three brains, ** P<0.01 one-tailed unpaired t-test. e, Young wild-type (WT) mice and HD transgenic mice were fed or fasted. Brains from these mice were dissected, homogenised and proteins were resolved by SDS-page for analysing the levels of endogenous beclin 1, LC3-I, LC3-II and p62 in each group. PolyQ levels were analysed using anti-polyQ antibody showing the expression level of the polyQ HTT exon 1. Representative blots are shown that were used to generate the data in Fig 4 c,d. f-g, Quantification of p62 and LC3-I levels in each group (WT fed n=7, HD fed n=5, WT fasted n=7, HD fasted n=6, a.u arbitrary units). For LC3-I levels, two-way ANOVA (column factor HD *** P<0.001, row factor fasting N.S, interaction P value N.S) with Bonferroni’s post-test (* P<0.05, *** P<0.001). For p62 levels, two-way ANOVA (column factor HD *** P<0.001, row factor fasting N.S, interaction P value N.S) with Bonferroni’s post-test (* P<0.05, *** P<0.001).

Extended Data Figure 9. Expansion of the polyQ domain in ataxin-3 decreased deubiquitinase activity and increased its interaction with beclin 1.

a, Beclin 1 was purified from proteasome inhibitor-treated cells that co-expressed HA-Ub and was incubated in-vitro with recombinant ataxin-3 Q22 or ataxin-3 Q80 for 30 min in deubiquitination buffer and samples were analysed for beclin 1 ubiquitination using anti-HA antibodies. b, Number of LC3 dots in ataxin-3 KD HeLa cells that were transfected with GFP ataxin-3 Q28, GFP ataxin-3 Q84 and GFP ataxin-3 ΔQ and starved with HBSS in the last 4 hr. Results are normalised to control siRNA-treated cells from n=4 independent experiments. One-way ANOVA (*** P<0.001) with post-hoc Tukey’s test (** P<0.01, *** P<0.001, N.S, not significant). c, Representative confocal images are presented for each of the conditions from b (LC3 dots in red and ataxin-3 staining in green, n=20 cells analysed).Scale bar is 10 µm. d, HeLa cells were transfected with empty vector, FLAG ataxin-3 Q22, FLAG ataxin-3 Q80 and HA Ub for 24 hr, treated in the last 6 hr with proteasome inhibitor (MG132 10 µM). Endogenous beclin 1 was immunoprecipitated from the lysates for analysis of different polyUb linkage using K48 polyUb or K63 polyUb antibodies, and for detection of bound ataxin-3 using anti-ataxin-3 and anti-polyQ antibodies.

Extended Data Figure 10. Effect of different disease proteins with polyQ expansion on beclin 1 ubiquitination, beclin 1 levels and starvation-induced autophagy.

a, HeLa cells were transfected with empty vector, GFP atrophin-1 (ATN-1) Q19, GFP ATN-1 Q71 and HA Ub for 24 hr, treated in the last 6 hr with proteasome inhibitor (MG132 10 µM). Endogenous beclin 1 was immunoprecipitated from the lysates for ubiquitination analysis and for detection of bound ATN-1 using anti-ATN-1 antibody. b, HeLa cells were transfected with empty vector, HA androgen receptor (AR) Q25, HA AR Q120 and HA Ub for 24 hr, treated in the last 6 hr with proteasome inhibitor (MG132 10 µM). Endogenous beclin 1 was immunoprecipitated from the lysates for ubiquitination analysis and for detection of bound AR using anti-AR antibody. c, Primary fibroblasts derived from healthy controls (n=5) and from HD patients (n=7) were starved with HBSS together with bafA1 (400 nM) for 4 hr and analysed for LC3-II levels. Results are mean ± s.e.m. * P<0.05 one-tailed Mann Whitney test. d-f, Primary fibroblasts derived from healthy controls and from patients with different polyQ diseases were kept in full media or starved with HBSS for 4 hr and analysed for LC3-II levels (LC3-II/actin ratio is presented). d, Spinocerebellar ataxia type 3 (SCA3) samples. e, Dentatorubral-pallidoluysian atrophy (DRPLA) samples. f, HD samples. The BafA1 experiments for these sets of patients are presented in Fig. 4. g, Beclin 1 levels (beclin 1/actin ratio is presented) in the starved control cells vs. SCA3, SCA7, DRPLA and HD patient cells. We only had one SCA7 patient sample and thus we have not analysed it further.

Figure 1. Ataxin-3 contributes to autophagosome formation by regulating the levels of beclin 1.

Mouse cortical neurons were transduced with control or ataxin-3 lentiviral shRNAs and analysed for: a, LC3-II levels with/without BafA1. b, LC3-II levels in starvation (HBSS) with/without BafA1. c, Beclin 1 levels in the starved cells. Results are mean ± s.e.m. n=5 replicates from two independent cultures. d-e, Control mice and mice depleted of liver ataxin-3 were fasted (24 h). Liver samples were analysed for beclin 1 (d) and LC3-II (e). Control fed n=9, ataxin-3 knockdown fed n=6, control fasted n=8, ataxin-3 knockdown fasted n=6, a.u arbitrary units. Extended statistical analysis in Supplementary Table 1. Two-way ANOVA with Bonferroni’s post-test (a, b, d, e). One-way ANOVA with post-hoc Tukey’s test (c). For representative blots and in-vivo ataxin-3 knockdown efficiency, see Extended Data Fig. 2d. Gel source data in Supplementary Fig. 1.

Figure 2. Beclin 1 deubiquitination by ataxin-3.

a, Endogenous ataxin-3 was immunoprecipitated from HeLa cell lysates and blots probed for endogenous beclin 1. b, Ubiquitinated beclin 1 was incubated in-vitro with recombinant ataxin-3 or ataxin-3 C14A for 2 h and analysed for beclin 1 ubiquitination using anti-HA antibodies. c, Evolutionary conservation of region around beclin 1 K402. d, Control and ataxin-3 depleted HeLa cells were transfected as indicated (24 h), incubated for last 6 h with proteasome inhibitor (MG132, 10 µM). Wild-type (WT) FLAG beclin 1 and mutant FLAG beclin 1 K402R were immunoprecipitated with anti-FLAG antibody for ubiquitination analysis. Gel source data in Supplementary Fig. 1.

Figure 3. Ataxin-3 polyQ domain contributes to ataxin-3 interaction with beclin 1.

a,b,c, Constructs were transfected in HeLa cells for 24 h. a, Cell lysates were immunoprecipitated with anti-GFP antibody and immunocomplexes analysed with anti-FLAG antibody. b,c, Cell lysates were immunoprecipitated with anti-FLAG antibody and immunocomplexes analysed with anti-GFP antibody. Gel source data in Supplementary Fig. 1. d, Structural docking modelling reveals two interactions sites between beclin 1 ECD and PolyQ₇. Surface charge illustration of human beclin 1 ECD showing high scored docking pose of polyQ₇ stretch (docking scores for site 1 and site 2 are -10.394 and -10.721, respectively). Full structural analysis in Extended Data Fig. 5.

Figure 4. Expanded polyQ tracts inhibit ataxin-3-beclin 1 interaction, decrease beclin 1 levels and impair starvation-induced autophagy.

a,b HeLa cells were transfected with indicated constructs for 24 h and immunoprecipitated for endogenous beclin 1. a, Cells treated for last 6 h with proteasome inhibitor (MG132 10 µM) were analysed for beclin 1 ubiquitination and for beclin 1-bound polyQ using anti-GFP antibody. b, Immunocomplexes were analysed for beclin 1-bound ataxin-3 using anti-FLAG antibody. Bound ataxin-3/beclin 1 ratio is presented. c-d, Brain samples from wild-type (WT) mice and Huntington’s disease (HD) transgenic mice were analysed for beclin 1 (c) and LC3-II (d). WT fed n=7, HD fed n=5, WT fasted n=7, HD fasted n=6, a.u arbitrary units. Extended statistical analysis in Supplementary Table 1. Two-way ANOVA with Bonferroni’s post-test. Representative blots in Extended Data Fig. 8e. e-g, Primary fibroblasts from healthy controls or patients with different polyQ diseases treated with bafA1, in full media or starved/HBSS analysed for LC3-II/actin ratio. Gel source data in Supplementary Fig. 1.