Mutations in the Key Autophagy Tethering Factor EPG5 Link Neurodevelopmental and Neurodegenerative Disorders Including Early-Onset Parkinsonism

Abstract

Objective: Autophagy is a fundamental biological pathway with vital roles in intracellular homeostasis. During autophagy, defective cargoes including mitochondria are targeted to lysosomes for clearance and recycling. Recessive truncating variants in the autophagy gene EPG5 have been associated with Vici syndrome, a severe early-onset neurodevelopmental disorder with extensive multisystem involvement. Here, we aimed to delineate the extended, age-dependent EPG5-related disease spectrum.

Methods: We investigated clinical, radiological, and molecular features from the largest cohort of EPG5-related patients identified to date, complemented by experimental investigation of cellular and animal models of EPG5 defects.

Results: Through worldwide collaboration, we identified 211 patients, 97 of them previously unpublished, with recessive EPG5 variants. The phenotypic spectrum ranged from antenatally lethal presentations to milder isolated neurodevelopmental disorders. A novel Epg5 knock-in mouse model of a recurrent EPG5 missense variant featured motor impairments and defective autophagy in brain areas particularly relevant for the neurological disorders in milder presentations. Novel age-dependent neurodegenerative manifestations in our cohort included adolescent-onset parkinsonism and dystonia with cognitive decline, and myoclonus. Radiological features suggested an emerging continuum with brain iron accumulation disorders. Patient fibroblasts showed defects in PINK1-Parkin-dependent mitophagic clearance and α-synuclein overexpression, indicating a cellular basis for the observed neurodegenerative phenotypes. In Caenorhabditis elegans, EPG5 knockdown caused motor impairments, defective mitophagic clearance, and changes in mitochondrial respiration comparable to observations in C. elegans knockdown of parkinsonism-related genes.

Interpretation: Our findings illustrate a lifetime neurological disease continuum associated with pathogenic EPG5 variants, linking neurodevelopmental and neurodegenerative disorders through the common denominator of defective autophagy. ANN NEUROL 2025;98:932-950.

Figure 1

Clinical characteristics of patients within the spectrum of EPG5‐related disorders. Left: Features of classical Vici syndrome. Right: Spectrum of features in EPG5‐related disorders with the phenotypic expansion of movement disorders presented in this study, including dystonia, parkinsonism, and myoclonus. [Color figure can be viewed at www.annalsofneurology.org]

Figure 2

EPG5 mutational spectrum and genotype–phenotype correlations in our cohort. (A) Protein overview with selected EPG5 variants. Truncating and splice site variants are shown above the schematic representation of the protein, missense and indel variants are pictured below (image created with IBS 2.0). (B) Kaplan–Meier survival curve of selected patients in the 3 subgroups: biallelic truncating (and splice site) variants, mixed truncating and missense variants, and biallelic missense variants. Median life expectancy noted for each curve. (C) Last reported age for each of the groups: biallelic truncating (and splice site) variants, mixed truncating and missense variants, and biallelic missense variants. (D) Overview and (E–I) specific effects of EPG5 missense variants in relation to the EPG5 protein model, indicating changes in protein structure. More detailed descriptions in Supplementary File 1. [Color figure can be viewed at www.annalsofneurology.org]

Figure 3

Neuroradiological spectrum in patients with EPG5‐related disorders. Brain magnetic resonance imaging from 6 individuals with EPG5‐related disorders including (A–C) one patient with Vici Syndrome and (D–R) five patients with atypical presentations. The arrow with the dotted border in (A) shows corpus callosum agenesis. Different degrees of degenerative involvement of the brain varying from mild forms with selective involvement of the fornix minors (short arrows; E,H), with or without mild corpus callosum atrophy and atrophy of the brain to more severe forms where there is an extremely thin and partially visualized corpus callosum (arrows; J–L) and cerebellar atrophy. Additional features included iron/micromineral deposition in the globus pallidi (black arrow; N), substantia nigra and red nuclei (black arrow; O), and also, as an isolated feature, in the pulvinar of the thalami (black arrows, Q).

Figure 4

Key features from the Epg5 ^Q331R knock‐in (KI) mouse model. (A) Epg5 mRNA levels are significantly reduced in the midbrain and brainstem of Epg5 ^Q331R KI mice. (B) A strong increase in LC3‐II is seen in the cerebellum and brainstem of the mutant mice. (C) A mild increase in p62 is seen in the cerebellum and a strong increase is seen in the brainstem of the mutant mice. (D) Epg5 ^Q331R KI mice show no motor phenotype at an early stage, but fall off the rotating rod much faster than wildtype (WT) mice at endstage (~12 months). (E) At endstage, mutant mice also show reduced rearing behavior in an open arena compared with WT mice. (F) Stride length is significantly shortened in KI mice at endstage on RotaRod analysis. (G) Female Epg5 ^Q331R KI mice spent significantly less time in the center zone of an arena than WT mice, whereas there is no significant difference between males. (H) Epg5 ^Q331R KI mice move through the virtually indicated center zone of an arena at higher speed than WT mice, whereas (I) Epg5 ^Q331R KI mice do not move faster through the outer zone, indicating an anxiety phenotype. (J) Representative TEM images of GFP‐Parkin expressing WT and Q331R mouse embryonic fibroblasts (MEFs) treated with oligomycin and antimycin (OA) and Baf A. Mitochondria are marked red. Scale bar, 1 μm. (K, L) Quantification of mitophagosomal number per square micron, and area in WT and Q331R MEFs stably expressing GFP‐Parkin. Plots represent the data from 3 independent experiments with n ≥ 20 TEM images. (M) FACS analysis of WT and Q331R MEFs stably expressing GFP‐Parkin and mt‐Keima. The P2 gated area encloses cells undergoing mitophagy and shows the percentage of cells within this gate of each plot. The percentages of cells within the mt‐Keima gate were much higher in wildtype cells (DMSO 1.14, OA 6 h 6.56, OA + BafA 6 h 0.25) than in EPG5 MEF cells (DMSO 0.40, OA 6 h 1.87, OA + BafA 6 h 0.73). (N) The percentage of cells undergoing mitophagy. Plots represent the data from 3 independent experiments. (O) Traces showing mitochondrial oxygen consumption rate measured in WT and Q331R MEFs using the Seahorse XFe96 extracellular flux analyzer. (P,Q) Quantitative analysis of (P) ATP‐linked respiration and (Q) spare respiratory capacity obtained from measurements in O. Data represents the mean ± SD of at least 3 independent experiments and analyzed by 1‐way ANOVA with Tukey's multiple comparisons test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. BS = brainstem; CB = cerebellum; FB = forebrain; MB = midbrain. [Color figure can be viewed at www.annalsofneurology.org]

Figure 5

Impaired autophagic flux and mitophagy in cells from a patient with EPG5‐related parkinsonism and SNCA overexpression. (A) Utilizing a fibroblast cell line from a 22‐year‐old patient with a homozygous EPG5 variant in c.6861_6862insTTTCCAACAGCAGAGTTC, p.Phe2287_Leu2288insPheProThrAlaGluPhe, we conducted immunoblots showing accumulation of p62 and LC3II in control and patient fibroblasts after 24 h of treatment with rapamycin (100 nM) and/or bafilomycin (200 nM). (B, C) Quantification of LC3II/β‐Actin ratio and p62 levels in A. (D) Immunoblot showing an increase of of PINK1 levels and its activity (phospho‐Ub) in control and patient fibroblasts in response to oligomycin (2.5 μM) and antimycin (2.0 μM; OA) treatment for indicated time. Ubiquitin levels (Ub) blot shows the total ubiquitin levels. (E) Quantification of the PINK1 levels in D. (F) Representative images of fibroblast cells cotransfected with mito‐Keima (ratiometric) and GFP‐Parkin (gray scale) before and after the treatment with OA for 24 h. Scale bars, 10 μm. (G) The proportion of the high ratio (561/458) signal area (red) to the total mitochondrial area plotted as mitophagy index. (H) Quantification of the number of GFP‐positive puncta per cell. Each experiment examined at least 12 transfected cells. (I) Representative immunofluorescence images of mCherry‐Parkin transfected fibroblasts immunostained for citrate synthase (CS) and LAMP1. Scale bar: overview, 10 μm; inset, 5 μm. Regions of interest (ROI) show line profile of Parkin‐CS puncta with the LAMP1‐positive vesicles, completely colocalizing in control fibroblasts. These structures were counted as “mitophagy events”, which were further plotted. (J) Quantitative analysis of the number of mitophagy events in H. Data in B, C, E, G, I, J, and K are represented as mean ± SD from 3 independent experiments. (K) Quantitative analysis of SNCA transcript in control and patient fibroblasts shows overexpression in cells from patients with the spectrum of EPG5‐related neurodevelopmental disorders (“attenuated VS spectrum”) in patient 1 (Q336R) and patient 2 (R1621Q) from Figure 6, as well as in the cell line from the patient 3 with EPG5‐related parkinsonism. *p < 0.05, **p < 0.005, ***p < 0.001, ****p < 0.0001 (2‐tailed unpaired Student t test or 1‐way ANOVA); ns = Not significant. [Color figure can be viewed at www.annalsofneurology.org]

Figure 6

Impaired autophagic flux and mitophagy defects in EPG5‐mutated patient cells. (A) Immunoblot showing accumulation of p62, NDP52, and LC3II in control and patient fibroblasts after 24 h of treatment with rapamycin (100 nM) and/or bafilomycin (200 nM). Quantifications of (B) p62 levels in, (C) NDP52 levels, and (D) LC3II/β‐Actin ratio, respectively. (E) Immunoblot showing increase of PINK1 levels and its activity (phospho‐UbS65) in relation to total ubiquitin levels (Ub) in both Q336R and R1621Q fibroblasts in response to oligomycin (2.5 μM) and antimycin (2.0 μM; OA) treatment for indicated time, (F) with quantification of PINK1 levels. (G) Representative images of fibroblast control and Q336R patient cells cotransfected with mito‐Keima (ratiometric) and GFP‐Parkin (gray scale) under vehicle treatment with DMSO (left) and after treatment with OA for 24 h (right). Scale bars, 10 μm. (H) Proportion of the high ratio (561/458) signal area (red) to the total mitochondrial area plotted as mitophagy index. (I) Pearson coefficient indices between GFP‐Parkin and mito‐Keima (high ratio area) show the accumulation of GFP‐Parkin on the mitophagosomes/mitolysosomes over time under normal and OA‐treated condition. (J, K) Quantification of the number of GFP‐positive puncta per cell in DMSO and OA 24 h treatment, respectively. Each experiment examined n ≥ 20 transfected cells. Scale bar: 10 μm. (L) Representative immunofluorescence images of mCherry‐Parkin transfected fibroblasts immunostained for citrate synthase (CS) and LAMP1. 3D reconstruction of inset in Control‐1 shows Parkin‐CS puncta inside the LAMP1‐positive vesicles (white arrows). These structures were counted as “mitophagy events” and quantitatively demonstrated in (M). Scale bar: overview, 10 μm; inset, 2 μm. Data in (B–D and F) are represented as mean ± SD from 3 independent experiments. *p < 0.05, **p < 0.005, ***p < 0.001, ****p < 0.0001 (2‐way ANOVA). ns = Not significant. [Color figure can be viewed at www.annalsofneurology.org]

Figure 7

Caenorhabditis elegans model of epg‐5/EPG5 dysfunction. Motor phenotype, mitochondrial oxygen consumption, and mitophagy flux in Caenorhabditis elegans knockdown of epg‐5/EPG5. (A) Analysis of stretch showed significantly decreased maximum differences in curvature per stroke in epg‐5i/EPG5i, ccz‐1i/CCZ1i, and pdr‐1i/PRKNi, indicating flatter body bends, demonstrated by pooled data from 3 biological replicates of n = 10 worms. (B) Analysis of curling showed significantly decreased percentage of time spent in bent‐over shapes in epg‐5i/EPG5i, rab‐7i/RAB7i, ccz‐1i/CCZ1i, and pdr‐1i/PRKNi worms, indicating much flatter movements, demonstrated by pooled data from 3 biological replicates of n = 10 worms. (C) Measurement of oxygen consumption rate by Seahorse Respirometer revealed an increase in spare capacity in epg‐5i/EPG5i, rab‐7i/RAB7i, ccz‐1i/CCZ1i, and pdr‐1i/PRKNi, indicating an excess of uncoupled capacity of the respiratory electron transport chain not being used in basal respiration. (D) Microscopy of mitophagy flux markers revealed an increase in LGG‐1:RFP and DCT‐1:GFP puncta in body wall muscle cells when compared with luci control. Quantification of punctae per 100 μm² visual counting, shown as pooled data from 3 biological replicates of n = 10 worms. Statistical significance levels: *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001 as determined by ANOVA. [Color figure can be viewed at www.annalsofneurology.org]