Damaging de novo missense variants in EEF1A2 lead to a developmental and degenerative epileptic-dyskinetic encephalopathy

Abstract

Heterozygous de novo variants in the eukaryotic elongation factor EEF1A2 have previously been described in association with intellectual disability and epilepsy but never functionally validated. Here we report 14 new individuals with heterozygous EEF1A2 variants. We functionally validate multiple variants as protein-damaging using heterologous expression and complementation analysis. Our findings allow us to confirm multiple variants as pathogenic and broaden the phenotypic spectrum to include dystonia/choreoathetosis, and in some cases a degenerative course with cerebral and cerebellar atrophy. Pathogenic variants appear to act via a haploinsufficiency mechanism, disrupting both the protein synthesis and integrated stress response functions of EEF1A2. Our studies provide evidence that EEF1A2 is highly intolerant to variation and that de novo pathogenic variants lead to an epileptic-dyskinetic encephalopathy with both neurodevelopmental and neurodegenerative features. Developmental features may be driven by impaired synaptic protein synthesis during early brain development while progressive symptoms may be linked to an impaired ability to handle cytotoxic stressors.

Keywords: EEF1A2; de novo; dyskinesia; epilepsy; yeast complementation assay.

Figure 1.

A) Distribution of EEF1A2 de novo pathogenic variants described previously or in this study (bold) in patients with neurodevelopmental disorders. The number of patients observed with recurrent variants are shown in parentheses (bold, this study), and vertical lines in the bottom panel denote missense variants from the gnomAD dataset seen more than once. The majority of pathogenic variants tend to cluster around the Switch I and II domains and these functional domains tend to be devoid of missense variants in the general population. B) Structure of eEF1A2 complexed with GDP showing locations of modeled variants. The locations and orientation of Domain I (green cartoon), domain II (teal cartoon), domain III (peach cartoon), switch I (red), switch II (yellow) and bound GDP (teal stick) and Mg⁺⁺ are shown. Locations of observed variants are shown as sticks with magenta spheres. Positions of ExAc variants are shown in orange.

Figure 2:. Structural brain images from patients with EEF1A2 mutations showing cerebral atrophy and abnormal myelination.

Sagittal T1-weighted (A) and axial T2-weighted (B,C) of Patient #11 performed at 1.5 months of age show mild diffuse prominence of the supratentorial CSF spaces and absent myelination in the posterior limbs of internal capsules (arrows, B). (D-F) Corresponding MR images performed in the same patient at 5.5 years of age reveal cerebral and cerebellar atrophy with predominant white matter volume loss, ex vacuo ventricular dilatation (asterisks, E, F), thin corpus callosum (arrowhead, D) and diffuse enlargement of the sulci. The myelination has slightly progressed (arrows, E), but there is persistent diffuse hazy elevated T2 signal in the deep and subcortical cerebral white matter. Note the severe vermian atrophy (empty arrow, D). Sagittal T1-weighted (G), axial T2-weighted (H) and coronal FLAIR (I) images of Patient #12 performed at 2 years of life reveal diminished white matter volume with thin corpus callosum (arrow, G), mild ex vacuo ventriculomegaly and enlarged CSF spaces, associated with markedly reduced myelination. Mild enlargement of vermian and cerebellar hemispheric sulci (empty arrows, G, I) is also present. (J-L) Corresponding images in the same patient at 7 years of age demonstrate progression of the cerebral and cerebellar atrophy (empty arrows, J, L) and minimal interval deposition of myelin in the posterior limbs of internal capsules and deep white matter in the temporo-occipital regions (arrows, K).

Figure 3:. Complementation studies of EEF1A2 variants using yeast.

(A) Representative image from protein extracts obtained from 2x10⁷ cells and analyzed by western blot using antibodies recognizing the V5 epitope or cytosolic protein Pgk1 and graph summarizing normalized protein levels. Some variant steady-state protein levels were significantly diminished (p<0.05) compared to wild-type. Similar trends were found when TEF2 variants were expressed in all yeast strains used un this study. Experiments were performed at least three times. (B) Plasmid shuffling findings using the MC214 strain transformed with TEF2 variants in p416GPD. Fresh cultures in SC medium lacking uracil were grown overnight and equal numbers of cells were deposited on SC-uracil plates with or without 0.5 g/l fluoroanthracilic acid (FAA). FAA medium select for colonies that have lost pTEF2-TRP1 and only have p416GPD derived plasmids. All TEF2 variants assayed except P333L showed failed growth on FAA plates, indicating a deficit in protein synthesis. All experiments were performed in triplicate; representative image shown. Note that amino acids 159–160 are not found in the yeast TEF2 sequence. For consistency, all variants are listed as human residue equivalents. (C) Yeast tef2-Δ strain shows sensitivity to the drug rapamycin that is reversed after transformation with the plasmid p416GPD carrying a wild-type copy of the TEF2 gene. A N-terminal V5-tagged version of TEF2 also rescued growth in presence of rapamycin. Only P333L TEF2 variant was unable to complement the rapamycin sensitivity phenotype. Yeast strains were cultured overnight in liquid Synthetic Complete (SC) medium lacking uracil and equal numbers of cells were deposited on YPD plates with or without Rapamycin (40 ng/ml) and incubated at 30°C for 48 hours. Representative image taken from three independent biological replicates. (D) In order to test for a possible dominant-negative effect, the same experiment described in (C) was performed using a wild type strain (BY4742).