EndophilinA-dependent coupling between activity-induced calcium influx and synaptic autophagy is disrupted by a Parkinson-risk mutation

Abstract

Neuronal activity causes use-dependent decline in protein function. However, it is unclear how this is coupled to local quality control mechanisms. We show in Drosophila that the endocytic protein Endophilin-A (EndoA) connects activity-induced calcium influx to synaptic autophagy and neuronal survival in a Parkinson disease-relevant fashion. Mutations in the disordered loop, including a Parkinson disease-risk mutation, render EndoA insensitive to neuronal stimulation and affect protein dynamics: when EndoA is more flexible, its mobility in membrane nanodomains increases, making it available for autophagosome formation. Conversely, when EndoA is more rigid, its mobility reduces, blocking stimulation-induced autophagy. Balanced stimulation-induced autophagy is required for dopagminergic neuron survival, and a variant in the human ENDOA1 disordered loop conferring risk to Parkinson disease also blocks nanodomain protein mobility and autophagy both in vivo and in human-induced dopaminergic neurons. Thus, we reveal a mechanism that neurons use to connect neuronal activity to local autophagy and that is critical for neuronal survival.

Keywords: Ca(2+) influx; Parkinson disease; endophilinA; neuronal activity; synaptic autophagy.

Figure 1

Ca²⁺ influx-induced synaptic autophagy is EndoA-dependent (A–E) Live imaging of non-stimulated and stimulated (30 min, 20 Hz) Drosophila larval NMJ boutons expressing Atg8^mCherry at endogenous levels in the absence of Ca²⁺ (A and A′), presence of Ca²⁺ (B and B′), presence of EGTA-AM (no Ca²⁺ in the buffer) (C), and presence of DMSO plus Ca²⁺ (D). Fluorescence intensities shown using scale (0–1292 gray value) indicated in (B). Quantification of the number of Atg8^mCherry dots (arrowheads) (E). Error bars: mean ± SEM; scale bar: 5 μm. Statistical significance: one-way ANOVA with Tukey’s multiple comparison test: ^∗∗ p < 0.01, ns, not significant, n ≥ 6 larvae (24 NMJs) per genotype. (F–H) Live imaging of genomically expressed Atg8^mCherry in NMJ boutons of w controls (F and F′) and of animals expressing RNAi against atg3 (under the control of the pan-neuronal driver nSyb-Gal4) (G and G′). Non-stimulated animals were incubated for 30 min in HL3 containing nefiracetam (Nefi) (10 μM) and post-synaptic glutamate receptor blocker 1-naphthylacetyl spermine trihydrochloride (NAS) (100 μM) to block muscle contractions (F and G). Stimulated animals were incubated for 30 min in HL3 with Nefi (10 μM), NAS (100 μM), and CaCl₂ (1 mM) (F′ and G′). Quantification of the number of Atg8^mCherry dots (arrowheads) (H). Error bars: mean ± SEM; scale bar: 5 μm. Statistical significance: two-way ANOVA with Šidàk multiple comparison test: ^∗∗ p < 0.01, ns, not significant, n ≥ 9 larvae (36 NMJs) per genotype. (I–K) Live imaging of genomically expressed Atg8^mCherry in NMJ boutons of w control animals (I and I′) and of endoA^−/− null mutant animals expressing phospho-dead endoA^S75A (genomic construct) (J and J′). Quantification of the number of Atg8^mCherry dots (arrowheads) (K). Error bars: mean ± SEM; scale bar: 5 μm. Statistical significance: two-way ANOVA with Šidàk multiple comparison test: ^∗∗ p < 0.01, ns, not significant, n ≥ 11 larvae (44 NMJs) per genotype. Full genotypes: see STAR Methods. Related to Figure S1.

Figure 2

D265A and D265R mutations regulate EndoA flexibility (A) Protein alignment showing conservation of the negatively charged glutamic acid in position 264 of Rat ENDOA2 and the negatively charged aspartic acid in position 265 of Drosophila EndoA. (B and B′) Dynamics of native wild-type EndoA, at 4°C, as determined by HDX-MS experiments. HDX data (n = 2 biological × 3 technical repeats; see Table S2) were analyzed by PyHDX to obtain ΔG/residue values (see Table S3). Output ΔG values colored according to a gradient color map (as indicated) were plotted together with covariances across the linear sequence (B, top panel) or are shown in a linear box (B, bottom panel). ΔG values plotted on the AlphaFold 2.0 predicted 3D structure of dimeric EndoA (B′). (C and C′) Differences in the dynamics of the mutant proteins relative to the wild type. HDX-MS data (n = 2 biological × 3 technical repeats; Table S2) were analyzed by PyHDX to obtain the relative fractional uptake values (rfus), i.e., all deuterium uptake values were expressed relative to the maximally deuterated sample (FD; taken as 100%) and the undeuterated sample (taken as 0%) (see Table S4). The Δrfus(wt-mutant) for the HDXtime = 10 s were plotted on the 3D structure of dimeric EndoA. The color code used only indicates positive (purple; mutant is more rigid than wild type [WT]), negative (yellow; mutant is more flexible than WT), or no difference to WT (gray). The mutated site is indicated on either protomer with (C) blue (D265A) or (C′) red (D265R) spheres. Arrowheads indicate the portion of the unstructured loop that shows opposite effect between the mutants. Related to Figure S2 and Table S2. Average D-uptake and standard deviations of peptides analyzed by HDX-MS, related to Figure 2, Table S3. ΔG values per residue derived by HDX-MS data, related to Figure 2, Table S4. Relative fractional D-uptake per residue, related to Figure 2.

Figure 3

EndoA mutants alter Ca²⁺ influx-mediated synaptic autophagy induction (A–E) Live imaging of genomically expressed Atg8^mCherry at NMJ boutons of control (nSyb-Gal4/+) animals (A and A′) and of endoA^−/− animals expressing endoA^WT (B and B′), endoA^D265A (C and C′), and endoA^D265R (D and D′) (under the control of nSyb-Gal4). Non-stimulated animals were incubated for 30 min in HL3, Nefi (10 μM), and NAS (100 μM) (A–D). Stimulated animals were incubated for 30 min in HL3 containing Nefi (10 μM), NAS (100 μM), and CaCl₂ (1 mM) (A′–D′). Fluorescence intensities shown using scale (0–23645 gray value) indicated in (D). Arrowheads indicate Atg8^mCherry accumulations. Scale bar: 5 μm. Quantification of the number of Atg8^mCherry dots (arrowheads) per NMJ area (E). Error bars: mean ± SEM. Statistical significance: two-way ANOVA with Tukey’s multiple comparison test: ^∗p < 0.05, ^∗∗∗p < 0.001, ns, not significant, n ≥ 12 larvae (48 NMJs) per genotype. Significance levels displayed above “Nefi” columns refer to comparison with “Nefi” treatment (unstimulated) on control animals. (F–H) CLEM of boutons of endoA^−/− animals expressing endoA^D265R with nSyb-Gal4, as well as Atg8^mCherry expressed at endogenous levels. (F) Single confocal slice of an example NMJ displaying an Atg8^mCherry structure (arrowhead). Scale bar: 5 μm. (F′) Zoom out of the same NMJ shown in (F). Asterisks indicate branding marks, and arrowhead indicates Atg8^mCherry structure. Scale bar: 20 μm. (F′′) Electron micrograph of the same region as in (F′). Asterisks indicate branding marks, and arrowhead indicates Atg8^mCherry structure. Scale bar: 10μm. (F′′′) Overlay of confocal image in (F′) with electron micrograph in (F′′). Asterisks indicate branding marks, and arrowhead indicates Atg8^mCherry structure. Scale bar: 10 μm. (G and G′) Zoomed representation of the bouton containing the Atg8^mCherry structure shown as overlay (G) and electron micrograph only (G′). Red arrowhead indicates the structure correlating with the mCherry signal. Scale bar: 1 μm. (G′′) magnification. Scale bar: 200 nm. (H–H′′′′) Single TEM slices showing the putative autophagosomal structure visible in multiple consecutive slices (red arrowhead). Z = 1: 0, Z = 7: 42 nm, Z = 9: 560 nm, Z = 11: 700 nm, Z = 15: 980 nm. Scale bar: 1 μm. (I–K) Examples of additional Atg8-overlapping autophagosomal structures (red arrowheads) from endoA^D265R expressing animals. Scale bar: 1 μm. Related to Figures S3 and S4.

Figure 4

Synaptic nanoscale organization of EndoA is Ca²⁺-influx-dependent (A–G) Representative images of Airyscan confocal sections of synaptic boutons of paraformaldehyde-fixed control (nSyb-Gal4/+) (A and A′) and endoA^−/− larvae expressing endoA^D265A (B and B′), endoA^WT (C and C′), and endoA^D265R (D and D′) (under the control of the pan-neuronal driver nSyb-Gal4) labeled with anti-EndoA antibody. Scale bar: 5 μm. Non-stimulated animals were incubated for 30 min in HL3, Nefi (10 μM), and NAS (100 μM) (A–D). Stimulated animals were incubated for 30 min in HL3 containing Nefi (10 μM), NAS (100 μM), and CaCl₂ (1 mM) (A′–D′). Zoomed in synaptic boutons (from B and D, respectively) showing quantification of EndoA intensity around the synaptic plasma membrane (magenta) and within the synaptic lumen (green) (E and F). Quantification of the EndoA-integrated intensity across genotypes indicated in (A–D) showing ratio of EndoA intensity at the membrane to that in the lumen (G). Error bars represent mean ± SEM; statistical significance: two-way ANOVA with Tukey’s multiple comparison test: ^∗ p < 0.05, ^∗∗ p < 0.01, ns, not significant, n ≥ 4 larvae (16 NMJs) per genotype. Significance levels displayed above “Nefi” columns refer to comparison with “Nefi” treatment (unstimulated) with control animals. (H and H′) Representative images of Airyscan confocal sections of individual synaptic boutons of paraformaldehyde-fixed control (nSyb-Gal4/+) in stimulated and non-stimulated conditions, labeled with anti-EndoA antibody and expressing Atg8^mCherry at endogenous levels. Scale bar: 5 μm. Only synaptic boutons with Atg8^mCherry puncta (dotted circles) indicative of autophagosomes were selected, and EndoA-integrated density in a radius of 100 nm from the Atg8^mCherry puncta was measured. (H′) Quantification of the normalized integrated EndoA intensity across genotypes indicated in (A–D) showing EndoA intensity 100 nm around Atg8^mCherry puncta. Data from all genotypes were normalized to control (nSyb-Gal4/+) unstimulated data. Error bars represent mean ± SEM; statistical significance: two-way ANOVA with Tukey’s multiple comparison test: ^∗ p < 0.05, ^∗∗ p < 0.01, ns, not significant, nd, no data, n ≥ 4 larvae (16 NMJs) per genotype. (I–K) Transgenic endoA^−/− larvae expressing endoA^WT∷mEos3.1, endoA^{D265A::mEos3.1}, or endoA^{D265R::mEos3.1} (nSyb-Gal4) were imaged using single molecule localization photoactivated localization microscopy (PALM) at 20 Hz. Representative images show cluster map color coded for cluster size and density distribution of endoA^::mEos3.1 generated by density-based spatial clustering of applications with noise (DBSCAN) analysis. Arrowheads indicate EndoA nanodomains. Fluorescence intensity shown using indicated scale (2–203). Scale bar: 2 μm. (I′–K′) Quantification of the mean cluster area of endoA^WT::mEos3.1, endoA^{D265A::mEos3.1}, and endoA^{D265R::mEos3.1} in non-stimulated and stimulated conditions. Error bars: mean ± SEM; statistical significance: Student’s t test two-tailed unpaired distribution: ^∗∗ p < 0.01, ns, not significant, n ≥ 5 larvae (20 NMJs) per genotype. (L–N) Transgenic endoA^−/− larvae expressing endoA^WT::mEos3.1, endoA^{D265A::mEos3.1}, or endoA^{D265R::mEos3.1} (nSyb-Gal4) were imaged using single-particle tracking photoactivated localization microscopy (sptPALM) at 20 Hz. Representative trajectories located within EndoA nanodomains in NMJ boutons of endoA^−/− larvae expressing endoA^WT::mEos3.1 (L), endoA^{D265A::mEos3.1} (M), and endoA^{D265R::mEos3.1} (N). (O) This is quantified for each genotype as mean square displacement (MSD) as a function of time. (P) Quantification of the area under the MSD curve (μm²s). Error bars: mean ± SEM; statistical significance: one-way ANOVA with Tukey’s multiple comparison test: ^∗∗ p < 0.01, n ≥ 4 larvae (n ≥ 14 nanodomains) per genotype. Related to Figures S5 and S6 and Video S1.

Figure 5

Ca²⁺ and EndoA phosphorylation independently induce autophagy (A–F) Live imaging of genomically expressed Atg8^mCherry in NMJ boutons of control (nSyb-Gal4/+) animals (A and A′) and of endoA^−/− animals expressing endoA^S75A (B and B′), endoA^S75A+D265R (C and C′), endoA^S75D (D and D′), and endoA^S75D+D265A (E and E′) (nSyb-Gal4). Non-stimulated animals were incubated for 30 min in HL3, Nefi (10 μM), and NAS (100 μM) (A–E). Stimulated animals were incubated for 30 min in HL3 containing Nefi (10 μM), NAS (100 μM), and CaCl₂ (1 mM) (A′–E ′). Fluorescence intensities shown using scale (300–2441 gray value) indicated in (A′). Arrowheads indicate Atg8^mCherry puncta. Scale bar: 5 μm. (F) Quantification of the number of Atg8^mCherry puncta (arrowheads) per NMJ area. Error bars: mean ± SEM. Statistical significance: two-way ANOVA with Tukey’s multiple comparison test: ^∗ p < 0.05, ^∗∗ p < 0.01, ns, not significant, n ≥ 8 larvae (32 NMJs) per genotype. Significance levels displayed above “Nefi” columns refer to comparison with “Nefi” treatment (unstimulated) with control animals. (G–L) Representative images of Airyscan confocal sections of synaptic boutons of paraformaldehyde-fixed control (nSyb-Gal4/+) (G and G′) and endoA^−/− larvae expressing endoA^S75A (H and H′), endoA^S75A+D265R (I and I′), endoA^S75D (J and J′), and endoA^S75D+D265A (K and K′) (nSyb-Gal4) labeled with anti-EndoA antibody. Scale bar: 5μm. Non-stimulated animals were incubated for 30 min in HL3, Nefi (10 μM), and NAS (100 μM) (G–K). Stimulated animals were incubated for 30 min in HL3 containing Nefi (10 μM), NAS (100 μM), and CaCl₂ (1 mM) (G′–K′). Quantification of the EndoA-integrated intensity across genotypes (G–K) showing ratio of EndoA intensity at the membrane to that in the lumen (L). Error bars: mean ± SEM; statistical significance: two-way ANOVA with Tukey’s multiple comparison test: ^∗∗∗∗p < 0.0001, ns, not significant, n ≥ 4 larvae (16 NMJs) per genotype. Significance levels displayed above “Nefi” columns refer to comparison with “Nefi” treatment (unstimulated) with control animals.

Figure 6

EndoA D265 mutants induce neurodegeneration (A–G′′) Representative electroretinogram (ERG) traces recorded from control (cn bw; longGMR Gal4/+) and endoA^+/− mutants expressing endoA^WT, endoA^D265A, or endoA^D265R (longGMR-Gal4). Cn bw mutations remove the protective eye pigmentation. Prior to ERG recordings, animals were exposed to 1–3 (A, C, E, and G) or 7 days of constant dark (A′, C′, E′, and G′) or 7 days of constant light (A′′, C′′, E′′, and G′′). Average traces are depicted in black. Gray arrowhead: on and off transients and black arrowhead: depolarization. (B–H′′) Histological sections of retinas of flies exposed for 1–3 (B, D, F, and H) or 7 days to constant dark (B′, D′, F′, and H′) or constant light (B′′, D′′, F′′, and H′′) stained with toluidine blue. Arrowheads indicate morphologically abnormal ommatidia. Scale bar: 10 μm. (I) Quantification of ERG depolarization amplitude recorded upon a 1 s light pulse in flies exposed to dark or light. Bars show mean ± SEM. Statistical significance: one-way ANOVA with Tukey’s multiple comparison test: ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗∗p < 0.0001. Significance level refers to the difference to the control (cn bw; longGMR Gal4/+) of the indicated condition (light or dark). Recorded flies per condition ≥ 8. (J) Quantification of the number of intact ommatidia (expressed in percenatge of the total) where all 7 rhabdomeres are visible. Bars show mean ± SEM. Statistical significance: one-way ANOVA with Tukey’s multiple comparison test: ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗∗p < 0.0001. Significance level refers to the difference to the control (cn bw; longGMR Gal4/+) of the indicated condition (light or dark). Single data points represent the average percentage of intact ommatidia of three histological sections of three animals per condition. (K) Schematic of dopaminergic neuron organization in clusters in the fly brain (posterior view). (K′) Representative maximum projection confocal image of a control brain stained with anti-TH antibody revealing dopaminergic neurons. Scale bar: 100 μm. (L and L′) Quantification of the number of TH⁺ neurons in the indicated clusters in aged brains. Experimental time points were defined based on the survival of the mutant animals: endoA^D265A expressing flies and respective controls were tested at 23–25 days old; endoA^D265R expressing flies and respective controls were tested at 1–2 days old. Bars show mean ± SEM. Statistical significance indicated above each column refers to the comparison to the control (nSybGal4/+) of each cluster and calculated with a two-way ANOVA: ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001, ns not significant.

Figure 7

SH3GL2 Parkinson disease coding variant impairs Ca²⁺-induced synaptic autophagy (A) Protein alignment showing evolutionary conservation of glycine 276 in the unstructured loop region of Human ENDOA1 and the proximity to the aspartic acid at position 265 (corresponding to glutamic acid 264 in Human ENDOA1). (B–C′) Representative live confocal images of NMJ boutons of endoA^−/− animals expressing Atg8^mCherry at endogenous levels and SH3GL2^WT or SH3GL2^G276V (nSyb-Gal4). Non-stimulated animals were incubated for 30 min in HL3, Nefi (10 μM), and NAS (100 μM) (B and C). Stimulated animals were incubated for 30 min in HL3 containing Nefi (10 μM), NAS (100 μM), and CaCl₂ (1 mM) (B′ and C′). Fluorescence intensities shown using scale (0–65535 gray value) indicated in (B′). Arrowheads indicate Atg8^mCherry positive autophagosomes. Scale bar: 5 μm. (D) Quantification of the number of Atg8^mCherry puncta per NMJ area. Error bars: mean ± SEM. Statistical significance: two-way ANOVA with Tukey’s multiple comparison test: ^∗ p < 0.05, ns, not significant, n ≥ 8 larvae (32 NMJs) per genotype. Significance level displayed above “Nefi” column refers to comparison with “Nefi” treatment (unstimulated) on control animals. (E–F′) Representative images of Airyscan confocal single slice sections of synaptic boutons of endoA^−/− larvae expressing SH3GL2^WT (E and E′) and SH3GL2^G276V (F and F′) (nSyb-Gal4) labeled with anti-ENDOA1 antibody. Scale bar: 5 μm. (G) Quantification of the ENDOA1 integrated intensity across genotypes showing ratio of ENDOA1 intensity at the membrane to that within the cytosol. Error bars: mean ± SEM; statistical significance: two-way ANOVA with Tukey’s multiple comparison test: ^∗∗∗∗p < 0.0001, ns, not significant, n ≥ 6 larvae (24 NMJs) per genotype. Significance levels displayed above “Nefi” column refer to comparison with “Nefi” treatment (unstimulated) on control animals. (H and I) Representative trajectories located within ENDOA1 nanodomains in NMJ boutons of endoA^−/− larvae expressing SH3GL2^WT::mEos3.1 (H) and SH3GL2^{G276V::mEos3.1} (I). (J) This is quantified for each genotype as mean square displacement (MSD) as a function of time. (K) Quantification of the area under the MSD curve (μm²s) represented in (J). Error bars mean ± SEM; statistical significance: unpaired t test with Welch’s correction: ^∗∗∗∗p < 0.0001, n ≥ 4 larvae (n ≥ 20 nanodomains) per genotype. Related to Figure S7.

Figure 8

Expression of SH3GL2^G276V in differentiated dopaminergic neurons recapitulates findings from Drosophila synapses (A) Schematic representation of gene editing strategy to knock in the G276V mutation in SH3GL2. (A′) Sanger sequencing showing successful homozygous editing. (B) Representative maximum projection confocal images of terminally differentiated (60 days) SH3GL2^WT ventral midbrain dopaminergic neurons (vmDAns) stained with the ventral midbrain marker FOXA2, dopaminergic marker TH, and neuronal marker MAP2. Scale bar: 100 μm. (B′) Quantification of the number of dopaminergic neurons over the total number of neurons in the field of view expressed as percentage of TH⁺/MAP2⁺ neurons. Error bars: mean ± SEM. Statistical significance: one-way ANOVA with Tukey’s multiple comparison test: ns, not significant. Single data points representing single confocal images. Data from three independent vmDAn differentiations (circles and empty/filled triangles represent independent differentiations). (C and C′) Western blot with anti-ENDOA1 from induced vmDAn. (C′) Quantification of ENDOA1 intensity relative to β-actin expression. Experiments performed with three biological replicates (neurons from three independent differentiations, displayed as circle and full/empty triangles). Error bars: mean ± SEM. Statistical significance: Student’s t test: ns not significant. (D–D′′) Calcium imaging on induced vmDAns transfected with GCaMP6f. (D) Example traces of single SH3GL2^WT (black) and SH3GL2^G276V (blue) neurons. (D′) Quantification of the average activity of neurons in one field of view (peak activity) and (D′′) quantification of the average peak amplitude of the calcium responses. Error bars: mean ± SEM. Statistical significance: Student’s t test: ns, not significant (E) Representative maximum projection confocal images of induced vmDAn (60 days) SH3GL2^WT vmDAn stained with anti-TH and anti-LC3B antibodies. Arrowheads: LC3B⁺ autophagosomes within TH⁺ neurites. Scale bar: 20 μm. Insert: zoom of the region indicated, where multiple LC3B⁺ puncta are visible. (E′) Quantification of the number of LC3B⁺ autophagosomes within TH⁺ neurites normalized to the total TH⁺ neurite length in the field of view. Error bars: mean ± SEM. Statistical significance: Student’s t test. ^∗∗∗p < 0.001. Single data points represent single confocal images. Data from three independent vmDAn differentiations (circles and empty/filled triangles represent independent differentiations). Related to Figure S8.