Novel GFM2 variants associated with early-onset neurological presentations of mitochondrial disease and impaired expression of OXPHOS subunits

Mitochondrial disorders are a genetically and clinically heterogeneous group of diseases which can arise due to defects in the oxidative phosphorylation (OXPHOS) system. The OXPHOS system comprises five multi-subunit protein complexes—the electron transport chain (complexes I–IV) and ATP synthase (complex V)—which function together to generate cellular energy in the form of ATP. Mitochondria possess their own circular, double-stranded 16,569 bp genome, encoding 22 transfer RNAs (tRNAs), 2 ribosomal RNAs (rRNAs) and 13 polypeptides [1]. All OXPHOS components, with the exception of complex II, are under bi-genomic control and as such are the product of both nuclear DNA and mitochondrial DNA (mtDNA)-encoded proteins [2]. Nuclear genes also encode the proteins responsible for mtDNA maintenance, mitochondrial transcription and translation and all other mitochondrial processes. As a consequence, mutations in either mitochondrial or nuclear genes may compromise ATP synthesis and cause mitochondrial disease. Extensive clinical and genetic heterogeneity makes the identification, characterisation and diagnosis of mitochondrial disease challenging, because clinical features often overlap with other neurological or systemic diseases. The advent of next-generation sequencing, specifically whole exome sequencing (WES), has improved the identification of disease-causing pathogenic variants in many different genes resulting in a much greater diagnostic yield than previous candidate gene screening approaches [3].

Mutations in genes involved in the translation of mtDNA are well-documented causes of mitochondrial disease [4, 5]. These genetic variants characteristically lead to combined OXPHOS deficiencies. The spectrum of resulting clinical phenotypes is broad, often manifesting as multi-system disease with the heart, skeletal muscle, brain and liver commonly affected [6]. The translational machinery within mitochondria is distinct to that of the cytosol and has several features which are reminiscent of the system employed in prokaryotic protein synthesis. This is reflective of the evolutionary origins of mitochondria as an endosymbiont α-proteobacterium [7]. Mitochondria possess their own mitoribosome which is a combination of 12S and 16S rRNAs, encoded within the mitochondrial genome, and at least 80 nuclear genes encode the proteins that form the large and small mitoribosomal subunits [8].

Pathogenic variants have been described in several of the mitoribosomal subunits [9‐11]. Mitochondrial translation defects can also arise due to mutations in the mitochondrial tRNAs and in the mitochondrial protein translation factors [12]. Mitochondrial amino-acyl tRNA synthetases (mt-aaRSs) are a group of enzymes with an essential role within mitochondrial protein synthesis and are associated with a continuously expanding number of mitochondrial disease presentations. There are 19 mt-aaRSs, which utilise ATP to specifically catalyse the attachment of amino acids to their cognate tRNA [13]. Mitochondrial diseases resulting from mutations in genes encoding all 19 of the mt-aaRSs have now been described in the literature, with the most recently identified being WARS2 [14].

The human genes GFM1 and GFM2 encode the proteins mtEFG1 and mtEFG2, respectively, both of which are homologs of the highly conserved bacterial translation elongation factor G (EF-G) [15]. EF-G functions at multiple stages within prokaryotic protein synthesis, with roles in the translocation of the ribosome during elongation and ribosome recycling upon termination or stalling of translation [16]. In human mitochondria, the dual roles of prokaryotic EF-G are carried out by the two distinct human homologs. Human mtEFG1 acts as an elongation factor with mitoribosome translocation activity, while mtEFG2 functions at the termination step of translation to disassemble the mitoribosome and allow subsequent cycles of mitochondrial protein synthesis [17].

Seventeen mitochondrial disease patients with GFM1 mutations across 13 families have been described in the literature so far. Earlier cases were associated with very severe systemic disease with early or neonatal onset, resulting in death within the first 2 years of life. Some clinical features, such as microcephaly, liver disease and encephalopathy, were common to multiple cases. However, more recent publications have described a new, milder, disease progression associated with GFM1 mutations, with survival at 6 and 7 years of age [18, 19].

Two families with gene defects in GFM2, identified through whole exome sequencing, have previously been described in the literature associated with Leigh syndrome, microcephaly, simplified brain gyral pattern and insulin-dependent diabetes [20, 21]. Here we present two unrelated patients with previously unreported variants in GFM2, documenting OXPHOS deficiencies in different tissues and expanding the clinical phenotypes associated with GFM2-related mitochondrial disease.

All studies were completed according to local Ethical Approval of the Institutional Review Boards of Newcastle University (the National Research Ethics Service Committee North East—Newcastle & North Tyneside 1) and of the Technische Universität München. In agreement with the Declaration of Helsinki, all individuals or their guardians provided written informed consent before undergoing evaluation and genetic testing.

Through the application of whole exome sequencing, we have identified previously unpublished recessive GFM2 variants in two unrelated patients with clinical features of mitochondrial disease and biochemical evidence of respiratory chain dysfunction. Western blot analysis of patient fibroblasts showed differential effects at the protein level for each of these patients. While Patient 2 shows decreased steady-state levels of mitochondrial-encoded OXPHOS subunits NDUFB8 (of complex I), COXI/COXII (of complex IV) and CYTB/CORE2 (of complex III), indicating a combined OXPHOS deficiency, steady-state OXPHOS protein levels were relatively unchanged in Patient 1 when compared to controls (Fig. 3a).

The gene GFM2, located at 5q13.3, encodes the protein mtEFG2 which has an essential role following the termination of mitochondrial translation as GTP hydrolysis is necessary for the recycling of the mitoribosome [17]. To date, GFM2 variants have been identified in four patients (two sets of siblings), with phenotypes described as microcephaly, simplified gyral pattern and insulin-dependent diabetes in the first report and Leigh syndrome complicated by arthrogryposis multiplex congenita in the most recent family [20, 21]. While Leigh syndrome is a familiar mitochondrial phenotype, there are clinical aspects to these four cases that are rather atypical for mitochondrial disease, simplified gyral pattern and arthrogryposis multiplex. In contrast, the two cases reported here have a consistent clinical phenotype that is entirely compatible with mitochondrial disease and includes ‘red flag’ clinical features such as neurodevelopmental regression.

The steady-state levels of COXI and COXII in the skeletal muscle of Patient 1 were decreased (Fig. 3b), demonstrating a complex IV deficiency. These results are in accordance with the original diagnostic assessment of individual respiratory chain enzyme activities in Patient 1, which detected a complex IV activity of around 40% of controls. The residual levels of mtEFG2 in skeletal muscle from Patient 1 were approximately 50% of the two controls (Fig. 3b). The c.636delA p.(Glu213Argfs*3) GFM2 mutation in Patient 1 causes a frame shift and a premature stop codon and is, therefore, likely to encode a non-functional truncated version of mtEFG2 that is likely to undergo nonsense-mediated mRNA decay. The 50% residual level of steady-state mtEFG2 seen in the skeletal muscle of Patient 1 can be explained by the p.(Arg190Gln) variant occurring in a region of the gene with only moderate sequence conservation. This is supported by the in silico pathogenicity predictions suggesting the missense change would not have major impacts on protein stability, consistent with the loss of only 48% normal mtEFG2 levels.

Steady-state OXPHOS subunit deficiency in Patient 1 is observable in skeletal muscle, but not fibroblasts, indicating tissue-specific differences. The deficiency of complex IV activity was most severe in the muscle (Fig. 1k) and less so in fibroblasts (Fig. 1l). This tissue-specific discrepancy in severity of OXPHOS defect is not uncommon in defects of mitochondrial translation and can arise due to the differences in ATP reliance in either tissue or differential levels of gene expression in the two tissue types. A number of deficiencies due to mt-aaRS mutations, such as YARS2 encoding the mitochondrial tyrosyl-tRNA synthetase, show a much clearer OXPHOS defect in myoblasts than fibroblasts. In the case of pathogenic YARS2 variants, a marked decrease in steady-state levels of NDUFB8 was observed only in myoblasts [26]. Muscle relies much more heavily on the generation of ATP through oxidative phosphorylation, and so variants in mitochondrial disease genes can on occasion present a much clearer effect at the steady-state protein level [27].

Despite mtEFG2 playing an important role in the termination step of mitochondrial translation, our [³⁵S] translational assay showed no marked decrease in [³⁵S] methionine/cysteine incorporation in growing fibroblasts as a measure of de novo mitochondrial protein synthesis, which may have been expected for Patient 1, due to a lack of clear phenotype in fibroblast cells. A possible explanation for the signal for Patient 2 being similar to controls can be found when the stage of translation at which mtEFG2 exerts its function is considered. In the majority of cases, mitoribosome recycling occurs upon completion of synthesis of a nascent peptide [28]. If mtEFG2 is dysfunctional, it will not prevent the incorporation of amino acids throughout the synthesis of a peptide, but will prevent the proper release of this product. Therefore, the signal coming from incorporated methionine/cysteine will not differ between control and patient cells.

In summary, we present two patients with developmental delay, dystonia, dysarthria and neuroimaging abnormalities in the putamen and caudate nuclei, along with subcortical white matter involvement who harbour previously unreported GFM2 variants, identified through the application of WES. The histochemical, biochemical and genetic investigations, together with the functional data presented in this report, form compelling evidence that the GFM2 variants identified through WES are pathogenic and causative of mitochondrial disease in both patients.