The human FARS2 gene encodes mitochondrial phenylalanyl-tRNA synthetase, which contributes to the accuracy of the translation of mitochondrial proteins, avoiding ATP deficiency in the cells (Roy et al., 2005). To date, 41 pathogenic FARS2 variants, inherited in an autosomal recessive pattern, have been reported (Figure 1). The variants are related to three main phenotypic manifestations: infantile-onset epileptic mitochondrial encephalopathy, later-onset spastic paraplegia (SPG77), and juvenile-onset refractory epilepsy. Usually, the epileptic groups have a poorer prognosis, while SPG77 is associated with a less severe disease and prolonged survival (Shamseldin et al., 2012; Yang et al., 2016; Hotait et al., 2020). Here, we describe a patient affected by a complicated and rapidly progressive form of hereditary spastic paraplegia (HSP) with upper limb involvement, bilateral single palmar creases, clinodactyly, mild intellectual disability, and dysmorphic features, carrying compound heterozygous variants in FARS2, the known missense variant p.Pro361Leu, and a novel intragenic deletion involving the coding exons 2, 3, and 4. Herein, we provide the first detailed molecular characterization of the intragenic deletion, identifying the molecular mechanism underlying the genomic rearrangement.

A single heterozygous pathogenic variant was found in FARS2, c.1082C>T (p.Pro361Leu) (Figure 3A); the subsequent search for CNVs using the ExomeDepth tool led to the identification of an intragenic deletion spanning the region from intron 1 to 4 of FARS2. Amplification, subcloning, and sequencing of the junction fragment at the breakpoint revealed a deletion spanning 158,513 bp at the genomic level (g.5284218-5442731del). The absence of any repeats or homology regions suggested that a non-homologous recombination event had likely mediated the rearrangement (Figure 3B). At the cDNA level, the deletion involved 925 nucleotides (c.21_904del), eliminating the first three coding exons of the gene (exon 2 containing the ATG, and exons 3 and 4), corresponding to the first 319 amino acid residues of FARS2 protein. An in-frame ATG, 56 nucleotides downstream of the breakpoint, is still maintained, and a putative polypeptide containing the last 131 amino acids of FARS2 might still have been translated. This putative protein product (if any) is likely inactive because it lacks the catalytic functional domain (p.Met1_Ala320del), while it retains the FARS2 linker region and the anticodon-binding domain (Figure 3C). Segregation analysis showed that the missense variant was maternally inherited and transmitted to the proband’s unaffected sister, while the deletion was paternally inherited and segregated in the proband’s unaffected brother (Figure 3A). We confirmed these data by performing WES and array CGH on the trio. No additional pathogenic variants or CNVs were identified.

Several FARS2 patients have been described with clinical phenotypes ranging from different types of epilepsy to pure and complicated forms of HSP SPG77. To date, 41 pathogenic FARS2 variants were found, including missense, nonsense, splice-site variants, and deletions. Among all the variants, seven are deletions and four of them are associated with SPG77 (Vernon et al., 2015; Forman et al., 2019; Jou et al., 2019; Meszarosova et al., 2020) (Figure 1). Although the deletions were not molecularly characterized with breakpoint subcloning and sequencing, in one case (Meszarosova et al., 2020), the analysis of genomic sequences surrounding the deletion had revealed the presence of highly homologous ALU repeats in FARS2 intron 2 and in LYRM4 intron 1, likely mediating the rearrangement. Here, we describe a child affected by a complicated form of SPG77, carrying compound heterozygous variants in FARS2, the known p.Pro361Leu pathogenic variant, and a novel intragenic deletion of 158513 bp, involving the first three coding exons of the gene, including the start codon. Unlike the case reported by Meszarosova and colleagues, the breakpoint sequence analysis in our patient revealed that the rearrangement was not mediated by any homologous region, thereby indicating a non-homologous end joining (NHEJ) recombination event had likely occurred. The NHEJ pathway is well known to be the election pathway for repairing double-strand breaks because their misrepair can cause severe changes, such as deletions or chromosomal translocations (Shibata, 2017). Based on the putative effects of FARS2 variants, we hypothesize a dramatic decrease of the FARS2 protein function in our patient. Indeed, the maternal p.Pro361Leu allele may likely retain a residual activity since this variant was already reported as affecting the stability of the anticodon-binding domain, thereby leading to a decreased protein interaction with tRNAPhe (Vantroys et al., 2017), while the paternally deleted allele may generate a likely putative inactive protein (if any) without the functional catalytic domain. This genotype is associated with clinical features partially overlapping those of a family reported by Vernon and colleagues (Table 1), with an early-onset severe spastic paraplegia, global developmental delay, and several dysmorphisms. In addition, our patient showed mild intellectual disability and additional dysmorphic features, such as bilateral single palmar creases on the hands, and clinodactily that were not observed in patients reported by Vernon and colleagues. Furthermore, a unique aspect of this case is the disease that seems to progress in a biphasic pattern, with decreasing development in the first 8 years of life, when an autonomous gait was still possible, followed by a very rapid worsening of the motor function with a complete loss of ambulation by the age of 12 years. Considering the clinical features of all SPG77 patients described so far, including the novel variant reported here, genotype–phenotype correlations are difficult to be established, given that the variants, such as p.Arg419His, are associated with both HSP and epileptic encephalopathy (Jou et al., 2019; Barcia et al., 2021). Analogously, even when considering only the HSP-related variants, there is no clear correlation between the genetic defect and disease severity, with the p.Pro361Leu variant (also carried by the patient described here) found in both pure and complicated forms (Table 1). Furthermore, neither the presence of a deletion nor its size (large ones or single amino acid deletions) could be correlated with a specific phenotype. Indeed, the loss of a single residue (p.Val174del) is associated with a complicated HSP form with a very early onset and severe presentation (Vantroys et al., 2017), while a deletion of about 151.6 kb was found in a patient with a pure form of HSP (Meszarosova et al., 2020). Thus, the phenotypic differences observed may be due to different combinations of FARS2 variants, as previously postulated (Barcia et al., 2021). Overall, the clinical and genetic data presented here further confirm the lack of any genotype–phenotype correlations in FARS2 patients. This represents a major drawback in genetic counseling and clinical prognoses in these patients. Therefore, a better understanding of the molecular mechanism(s) of FARS2 variants should allow a better risk assessment for patients and their families. In addition, this may also lead to the development of novel therapies. To this aim, the Drosophila dFARS2 mutants, mimicking many disease features, might represent a useful tool to better define the pathogenic mechanisms underlying the variable degree of FARS2-related disease severity (Fan et al., 2021).

Our study widens the phenotypic spectrum and heterogeneity associated with FARS2 variants. Furthermore, we report the first molecular characterization of a FARS2 deletion and stress on the importance of combining different tools (i.e., sequencing and deletion/duplication studies) for variant screening in HSP patients, especially when the gene panels are inconclusive.