ATP6V0C variants impair V-ATPase function causing a neurodevelopmental disorder often associated with epilepsy

Abstract

The vacuolar H+-ATPase is an enzymatic complex that functions in an ATP-dependent manner to pump protons across membranes and acidify organelles, thereby creating the proton/pH gradient required for membrane trafficking by several different types of transporters. We describe heterozygous point variants in ATP6V0C, encoding the c-subunit in the membrane bound integral domain of the vacuolar H+-ATPase, in 27 patients with neurodevelopmental abnormalities with or without epilepsy. Corpus callosum hypoplasia and cardiac abnormalities were also present in some patients. In silico modelling suggested that the patient variants interfere with the interactions between the ATP6V0C and ATP6V0A subunits during ATP hydrolysis. Consistent with decreased vacuolar H+-ATPase activity, functional analyses conducted in Saccharomyces cerevisiae revealed reduced LysoSensor fluorescence and reduced growth in media containing varying concentrations of CaCl2. Knockdown of ATP6V0C in Drosophila resulted in increased duration of seizure-like behaviour, and the expression of selected patient variants in Caenorhabditis elegans led to reduced growth, motor dysfunction and reduced lifespan. In summary, this study establishes ATP6V0C as an important disease gene, describes the clinical features of the associated neurodevelopmental disorder and provides insight into disease mechanisms.

Keywords: ATP6V0C; V-ATPase; VMA3; epilepsy genetics; neurodevelopmental disorders.

Figure 1

V-ATPase structure. The peripheral domain (V₁, uppercase letters, in grey) is the site of ATP binding and hydrolysis. The integral domain (V₀, lowercase letters, in purple, red, blue and yellow) transports protons across membranes. The c-ring (red) is composed of nine c-subunits (encoded by ATP6V0C) and one c′′-subunit (encoded by ATP6V0B, not shown) and rotates after ATP hydrolysis to bring protons to ATP6V0A (blue). ATP6V0A possess two hemi-channels and a buried arginine residue (p.R735) that are required along with p.E139 in ATP6V0C for proton translocation.

Figure 2

Location and conservation of ATP6V0C variants. (A) Exon/intron structure of ATP6V0C. Boxes represent exons with black denoting coding regions. Scale bar = 100 bp length. (B) Protein alignment showing conservation of affected residues (highlighted in yellow). Glutamate residue (p.E139) required for proton transport is in bold. The following protein sequences were used in the alignments: H. sapiens, NP_001685.1; M. musculus, NP_001348461.1; D. rerio, NP_991117.7; D. melanogaster, NP_476801.1; C. elegans, NP_499166.1; S. cerevisiae, NP_010887.3. (C) Lollipop plot showing the transmembrane structure (green) and location of variants throughout ATP6V0C. Patient missense variants are indicated above in red. Missense (blue) and synonymous (grey) variants observed in gnomAD are shown below. Based on UniProt accession P27449. There is a significant enrichment of patient variants in TM4 (P = 0.006, Fisher’s exact test). (D) Plot showing tolerance of missense variants across ATP6V0C. The missense tolerance ratio (MTR) was calculated using 21 codon window sizes. A MTR score of <1 indicates intolerance to missense variation. Dashed lines on the plot denote ATP6V0C-specific MTRs: green = 5th percentile, yellow = 25th percentile and black = 50th percentile.

Figure 3

Knockdown of the Drosophila orthologue of ATP6V0C increases seizure duration. (A) Pan-neuronal (elaV-GAL4) RNAi-mediated knockdown of Dmel\Vha16–3 (CG32090) using RNAi (elaV > 102067 RNAi) is sufficient to increase the recovery time (RT) of third instar larvae to electroshock-induced seizure. Controls expressed GFP RNAi via elaV-GAL4 (elaV > GFP RNAi) or the RNAi (102067) without a driver (102067 RNAi). (B) Seizure induction due to expression of 102067 RNAi is preferentially rescued by pretreatment of larvae with levetiracetam (LEV) or topiramate (TOP). Lamotrigine (LAM) and valproate (VAL) were also effective, but not phenytoin (PHY). RT was normalized to a vehicle (dimethyl sulphoxide) only control. Data shown as mean ± SEM. One-way ANOVA with post hoc comparison (Dunnett’s); *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 4

Molecular modelling of patient and gnomAD variants. (A) Structure of part of the V₀ region of human V-ATPase (PDB 6wlw). Sites of patient (purple) and gnomAD (green) variants are shown superposed on ribbon backbone for two ATP6V0C subunits (gold), one next to the ATP6V0A subunit (cyan) and one on the opposite side of the c-ring. The back part of the c-ring is filled with grey, and the front part has been omitted for clarity. (B) Isolated view of the interaction between ATP6V0C variants and ATP6V0A. The functional amino acid p.E139 is also displayed (pink).

Figure 5

Patient variants show reduced V-ATPase function. (A) Quantification of average fluorescent intensity for each variant in the LysoSensor assay. Variants are grouped based on their location within or proximity to the nearest transmembrane (TM) domain. Data were normalized with mean of wild-type as 100% (denoted by dotted line) and mean of empty vector as 0%. Data shown as mean ± SEM (n = 71–132 cells per variant). Box and whisker plot of these data is presented in Supplementary Fig. 4. (B–E) Growth curves of vma3Δ S. cerevisiae expressing patient or gnomAD variants when grown in YPD, pH 5.5 with 5 mM CaCl₂. In all panels, wild-type is shown in black and the empty vector in grey. Mean of nine replicates per construct is shown with error bars omitted for clarity. Variants are grouped on the basis of their location within or proximity to the nearest TM domain. (F) eAUC was calculated using Growthcurver. Data were normalized within each plate with wild-type as 100% (denoted by dotted line) and empty vector as 0% and is shown as mean ± SEM. A one-sample t-test to a hypothetical mean of 100 was conducted with a Bonferroni correction (adjusted α level = 0.0003125). **P < 0.01, ***P < 0.001, ^#P < 0.0001.

Figure 6

Expression of patient variants in C. elegans exacerbate motor dysfunction and reduce lifespan. (A and B) WormLab analysis of body length and size at Day 1 of adulthood. All mutants are shorter and smaller than N2 controls. (C and D) Automated analysis of worm movement in liquid culture by WormTracker software. (C) In physiological M9 solution, all mutants show no motor deficits. (D) In presence of 350 mM NaCl concentration the p.G63A (P < 0.0001) and p.F137L (P < 0.0062) mutants show reduced movement scores in liquid culture over 270 min. Reduced movement was also observed with the p.L150F variant, but this difference was not statistically significant (P = 0.0869). (E) All mutant strains showed increased paralysis over 14 days compared to N2 controls (n = 313–317/strain, P < 0.0001). (F and G) In presence of osmotic stress (200 or 300 mM NaCl) the paralysis phenotype is exacerbated, leading to almost 100% paralysis after 8 days for the p.G63A strain (n = 246–260/strain, P < 0.0001). (H) All mutant strains exhibited reduced lifespan compared to N2 controls (n = 219–233/strain, P < 0.0001). (I and J) All mutant strains have reduced lifespans in presence of osmotic stress compared to N2 controls (200 mM NaCl: n = 182–228/strain. 300 mM, P < 0.0001) (300 mM NaCl: n = 200–244/strain, P < 0.0001). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 compared to N2 controls.

Figure 7

Patient mutations cause increased uncoordinated movement and neuronal signalling dysfunction in C. elegans. (A–C) Analysis of fine motor movement of worms after 30 min in 500 mM NaCl liquid culture. Mutants show increased activity index and wave initiation (A and B), but swimming speed was not significantly altered (C). (D) Synaptic transmission was evaluated by exposing Day 1 adult worms to aldicarb. Worms were scored over a 2-h period for paralysis. All mutants were hypersensitive to aldicarb treatment compared to N2 worms (n = 236–296/strain). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 compared to N2 controls.