Human aminoacyl-tRNA synthetases in diseases of the nervous system

ABSTRACT

Aminoacyl-tRNA synthetases (AaRSs) are ubiquitously expressed enzymes that ensure accurate translation of the genetic information into functional proteins. These enzymes also execute a variety of non-canonical functions that are significant for regulation of diverse cellular processes and that reside outside the realm of protein synthesis. Associations between faults in AaRS-mediated processes and human diseases have been long recognized. Most recent research findings strongly argue that 10 cytosolic and 14 mitochondrial AaRSs are implicated in some form of pathology of the human nervous system. The advent of modern whole-exome sequencing makes it all but certain that similar associations between the remaining 15 ARS genes and neurologic illnesses will be defined in future. It is not surprising that an intense scientific debate about the role of translational machinery, in general, and AaRSs, in particular, in the development and maintenance of the healthy human neural cell types and the brain is sparked. Herein, we summarize the current knowledge about causative links between mutations in human AaRSs and diseases of the nervous system and briefly discuss future directions.

Introduction

Aminoacyl-tRNA synthetases (AaRSs) are ancient ubiquitously expressed housekeeping enzymes that ensure faithful translation of the genetic information into functional proteins across domains of life. They couple proteinogenic amino acids to cognate tRNAs that carry specific anticodon sequences. The aminoacylation reaction is a 2-step transesterification reaction. In the first step, an AaRS activates an amino acid with ATP thus forming an aminoacyl adenylate and pyrophosphate. In the second step, the amino-acid residue is transferred from the adenylate conjugate onto the 3′-end of the tRNA. The acyl ester linkage between the amino acid and the 3′-OH group in the ribose ring establishes the aminoacyl-tRNA (aa-tRNA), which serves as a substrate for the translating ribosome. The error rate of the aminoacylation reaction correlates well with the overall error rate of protein synthesis, suggesting that the ability of AaRSs to select reaction substrates is pivotal for the accuracy of gene translation.¹ While quite efficient in sorting out different tRNAs, AaRSs often struggle when differentiating the cognate and the near-cognate amino acids. Thus, despite the pre- and post-transfer editing capabilities of AaRSs, occasional misactivation and mischarging reactions give rise to translational errors.

Based on their global structural features, AaRSs are divided into 2 major classes, class I and class II. The class I enzymes are further divided into a, b, and c subclasses based on the sequence conservation. AaRSs of both classes are multidomain enzymes that typically contain the catalytic and anticodon-binding domains. The class I enzymes are either monomers or dimers with a Rossmann fold catalytic domain that is composed of parallel  β-strands, whereas the class II AaRSs are either dimers or tetramers and they harbor 3 poorly conserved motifs and the catalytic domain built of antiparallel β-strands. With the exception of PheRS, the class I and class II AaRSs attach the amino acid to the 2′-OH and 3′-OH groups, respectively. After the transesterification the amino acid is always placed onto 3′-OH of the tRNA. Besides the catalytic and anticodon-binding domains, AaRSs often harbor editing domains that carry out post-transfer editing of the mischarged aa-tRNAs. In addition, as life forms evolved from unicellular to multicellular organisms, the corresponding AaRSs acquired a variety of appended domains that seem to be critical for non-canonical functions residing outside protein synthesis.^2,3 The newly acquired domains also increase the repertoire of interacting partners. One such example is a multi-aminoacyl tRNA synthetase complex (MSC), which is composed of at least 9 AaRSs and 3 auxiliary protein factors in humans (reviewed in ⁴).

The rise of compartmentalized eukaryotic cells that rely on oxidative phosphorylation has added another layer of complexity when AaRSs are considered. Mitochondria have a separate translational apparatus that guides synthesis of mitochondrial ribosomal proteins and 13 protein subunits of the mitochondrial respiratory chain (MRC) complexes I, III, IV, and V. The process relies on 17 exclusively mitochondrial and 2 bifunctional AaRSs (i.e. GlyRS and LysRS), and a set of mitochondrially encoded tRNAs. Given that mitochondrial GluRS (mt-GluRS) efficiently charges mt-tRNA^Gln and that the gene for mt-GlnRS has not been found, it is assumed that formation of mt-Gln-tRNA^Gln relies on transamidation. Also, while a single cytosolic GluProRS charges tRNA^Glu and tRNA^Pro, separate enzymes are needed for these reactions in mitochondria. Despite the compartmentalization, mt-AaRSs are nuclearly encoded but by distinct genes from the cytosolic isoforms. The accepted nomenclature states that AaRSs are represented by a single- or 3-letter amino-acid acronym and a number “2” added to the name in the case of the mitochondrial enzymes. Thus, GARS encodes cytosolic GlyRS or GARS, while GARS2 encodes mt-GlyRS or GARS2. In spite of its appeal, the endosymbiotic origin of mt-AaRSs is not universal because gene duplication and vertical and horizontal gene transfer events complicate the origin of both cytosolic and mitochondrial isoforms.^5,6

Given their central role during protein synthesis across domains of life, it is obvious that life would not be possible without AaRSs. It is also evident that any major alterations in the mode of action of any of these enzymatic systems would likely have serious consequence for the organism (reviewed in ^7-11). However, until the advent of modern genome sequencing techniques, the malfunctioning of human AaRSs was only linked to abnormal chronic immune responses designated as antisynthetase syndrome (reviewed in ^12,13). More recently, AaRSs are being increasingly implicated in disorders of the nervous system (Table 1). Just 4 y ago only a handful of ARS genes were connected to neuropathies, but the list is rapidly growing along with the diversity of clinical profiles presented. Not long ago it was assumed that pathogenic variants of cytosolic AaRSs are responsible for neuropathies of the peripheral nervous system and that variants of the mitochondrial isoforms are responsible for encephalopathies. As we shall see in the following paragraphs, the current evidence argues against this simplistic view. The burgeoning and intense investigations have intensified the discussion about the role of the translational machinery, in general, and AaRSs, in particular, in the development and maintenance of the healthy and diseased neural cell types, peripheral nervous system, and the brain. In this review, we shall attempt to summarize what is currently known about the role of AaRSs in pathologies of the human nervous system and we briefly discuss future directions and perspectives.

Table 1.

– Human AaRSs and neurologic disorders.

Enzyme (gene)	Disorders
Cytosolic
AlaRS (AARS)	CMT2N with dHMN 31-33; mild axonal neuropathy 36; epileptic encephalopathy 85
AspRS (DARS)	HBSL 63,64
ArgRS (RARS)	Hypomyelination with severe spasticity and nystagmus 65
GlyRS (GARS)	CMTN with DHMN5A 14; CMTN 24,25
GlnRS (QARS)	PCH 75,77; fatal EOEE 78
HisRS (HARS)	CMTN with HMN 37; Usher syndrome 96; Late-onset sensory-predominant axonal peripheral neuropathy 97
LysRS (KARS)	Early-onset progressive microcephaly 80; Nonsyndromic hearing impairment DFNB89 98
MetRS (MARS)	CMT2N 30,35; HSP 101 (predicted)
TrpRS (WARS)	HMN 38;
TyrRS (YARS)	DI-CMTC 26,30
Mitochondrial
mt-AlaRS (AARS2)	(Ovario)leukodystrophy 62
mt-AsnRS (NARS2)	Epilepsy, microcephaly, developmental delay, cerebellar atrophy 84; Alpers syndrome 88; Leigh syndrome 90; Epilepsy, severe myopathy and intelectual disability 99
mt-AspRS (DARS2)	LBSL 42,46-50,52-54,57-59; ataxia 55
mt-ArgRS (RARS2)	PCH, PCH6 69-74; fatal EOEE 79
mt-CysRS (CARS2)	Epileptic encephalopathy 83; myoclonus epilepsy with ragged red fibers 102
mt-GluRS (EARS2)	LTBL 58,60; Fatal infantile multisystem disease 91
mt-HisRS (HARS2)	Perrault syndrome 93
mt-IleRS (IARS2)	Leigh syndrome 89
mt-LeuRS (LARS2)	Perrault syndrome 94
mt-MetRS (MARS2)	ARSAL 61; Perrault syndrome 95
mt-PheRS (FARS2)	Seizures, epilepsy, visual impairment, white matter loss 82; Alpers syndrome 87; HSP 100;
mt-ProRS (PARS2)	Epilepsy, microcephaly, developmental delay, cerebellar atrophy 84; Alpers syndrome 88
mt-ThrRS (TARS2)	Axial hypotonia, severe psychomotor delay 81
mt-ValRS (VARS2)	Microcephaly and epilepsy 81

ARSAL, autosomal recessive spastic ataxia with leukoencephalopathy; CMTN, CMT neuropathy; CMT2N, CMTN type 2; DI-CMTC, autosomal dominant intermediate CMTN type C; dHMN, distal hereditary motor neuropathy; DHMN5A, distal spinal muscular atrophy type V; EOEE, early-onset epileptic encephalopathy; HBSL, hypomyelination with brain stem and spinal cord involvement and leg spasticity; HMN, hereditary motor neuropathy; HSP, autosomal recessive hereditary spastic paraplegia; LBSL, leu-koencephalopathy with brain stem and spinal cord involvement and lactate elevation; LTBL, leukoencephalopathy with thalamus and brainstem involvement and high lactate; PCH, pontocerebellar hypoplasia; PCH6, PCH type 6;

Charcot-Marie-Tooth Neuropathy

Charcot-Marie-Tooth (CMT) disease is a hereditary peripheral neuropathy characterized by progressive deterioration of distal motor and sensory neurons that leads to muscle weakness and limb atrophy. The disease is typically categorized as either demyelinating type 1 or axonal type 2 based on its impact on the integrity of the myelin sheath or the axon of the peripheral nerve. CMT was the first neural disorder to be linked to mutations in genes encoding AaRSs. Antonellis and colleagues reported a causative link between the bifunctional and dimeric GlyRS with CMT neuropathy (CMTN) and distal spinal muscular atrophy type V (DHMN5A).¹⁴ Several pathogenic mutations map to the dimer interface ¹⁵ and they either weaken or strengthen the dimer formation. Other variants do not form granules in neurite projections ¹⁶ and/or have a localization defect in neuroblastoma cells.¹⁷ In a Drosophila model of the GARS-induced CMT2D severe dendritic defects in olfactory projections and both axonal and dendritic defects in mushroom body γ neurons were observed.¹⁸ Because only expression of the WT enzyme rescued the phenotype in flies, it was suggested that E71G and L129P are the loss-of-function variants. Analyses of 9 CMT-causing variants have shown that A57V, D146N, S211F, G240R, P244L, I280F, H418R, G526R, and G598A diminish the aminoacylation activity of GlyRS and cannot rescue the gars1Δ yeast strain.¹⁹ Using the mouse model of GARS-associated neuropathy it was shown that the L129P mutation causes a distribution defect of GlyRS in peripheral nerves.²⁰ Concurrently, it was shown that mutant Gars compromises development of neuromuscular junction before degeneration of synapses, implying that the neuromuscular synapse is an important site of early pathology in CMT2D mice. Most recently, it was shown that pathogenic variants E71G, L129P, G240R, and P234KY bind to neuropilin 1 receptor (Nrp1), which antagonizes the Nrp1-VEGF interaction that is important for motor neuron function.²¹ Using zebrafish, it was revealed that the WHEP domain-dependent neurotoxicity of the P234KY variant is non-cell autonomous.²² The authors argued that pathogenic GlyRS variants cause abnormal neuromuscular junction assembly, synaptic degeneration, and reduced viability. By contrast, the enzyme dimerization is required for the dominant neurotoxicity of pathogenic mutations of GARS in Drosophila.²³ Most recently, a novel heterozygous missense mutation E333D,²⁴ and compound heterozygous mutations D146Y and M238R ²⁵ were linked to CMT. E333D impacts a conserved residue in the catalytic domain, while D146Y and M238R affect residues located in insertion I and at the dimer interface, respectively, but the exact involvement of these variants in pathology is not clear.

The second AaRS associated with CMT was TyrRS (Table 1). In 2006 Jordanova and colleagues reported that the autosomal dominant intermediate CMT neuropathy type C (DI-CMTC) is caused by heterozygous missense mutations G41R and E196K, and by the de novo deletion I153-V156del in YARS in 3 unrelated families.²⁶ DI-CMT is a variant of classical CMT characterized by intermediate nerve conduction velocities and histological evidence of both axonal and demyelinating features. G41 is a highly conserved residue involved in formation of tyrosyl-adenylate, while E196 and V153-V156 are predicted to interact with tRNA^Tyr. These mutations reduced the first step of the aminoacylation reaction, yeast growth in the complementation assays, and enzyme localization in the axonal termini in differentiating motor neurons and neuroblastoma cells. Given that neuropathies have a major effect on axonal transport in the peripheral nervous system,²⁷ that intra-axonal translation facilitates regeneration of rat dorsal root ganglia and motor neurons,²⁸ and that inhibition of protein synthesis tampers regeneration of adult rat peroneal nerve,²⁹ it is tempting to speculate that TyrRS may have specific functions at the axonal termini. Hence, it was suggested that the YARS-related pathology starts at the axonal termini, that the loss-of-function affects synaptic plasticity leading to axonal degeneration and loss, and ultimately to a motor and sensory peripheral neuropathy. More recently, a missense mutation D81I in YARS was identified in a patient with the late-onset DI-CMTC.³⁰ The patient exhibited thenar muscle atrophy and asymmetric motor nerve conduction velocities that indicated demyelination in the left side, but axonal neuropathy in the right side. The mutation is predicted to affect the side chain in the catalytic domain of the enzyme, but it remains to be seen how this alteration leads to the disorder at the molecular level.

Recent findings have associated AlaRS, MetRS, HisRS, and TrpRS with CMT and Hereditary Motor Neuropathy (HMN) as well (Table 1). The R239H mutation in human AARS causes dominant axonal CMT2D with distal hereditary motor neuropathy (dHMN; CMT2N) in patients from the French,³¹ Australian,³² and UK cohort.³³ The patients presented sensory-motor distal degeneration secondary to axonal neuropathy and demyelination, and the Australian cohort presented the sensorineural deafness as well. The mutation affects the conserved residue important for aminoacylation of tRNA^Ala.³² In addition, a family in Ireland having CMT2N harbored another AlaRS variant, E688G,³³ while a Chinese family carried D893N mutant related to dHMN.³⁴ Further, R618C and P800T in MARS cause CMT2 with incomplete penetrance and late-onset CMT2, respectively.^30,35 The strictly conserved R618 is at the interface between catalytic and anticodon-binding domains and the R681C variant cannot rescue the mes1Δ allele in the yeast complementation assay. Hence, it is suggested that R681C is either the toxic gain-of-function or loss-of-function variant.³⁵ A heterozygous missense mutation G102R in AARS causes only mild axonal neuropathy and lower extremity hyperflexia in 5 siblings.³⁶ This variant could not complement the ala1 knockout mutant yeast strain, suggesting that Gly102 is important for function and/or structure of AlaRS. Furthermore, heterozygous mutations T132I, P134H, D175E, and D364Y in human HARS cause axonal and demyelinating motor and sensory neuropathies ranging from adult-onset CMT2, CMT1, and typical CMT to HMN. All mutations cause loss of function in yeast complementation assays, and D364Y is dominantly neurotoxic in C. elegans.³⁷ Most recently, a recurrent heterozygous missense mutation H257R in WARS was identified as a cause of autosomal dominant HMN.³⁸

In spite of many efforts, it is not clear how variants of AaRSs cause neuropathies. For instance, it is unknown whether mutations in GARS impact the cytosolic or the mitochondrial function, or both. Since mutations in other strictly cytosolic AaRSs cause similar neuropathies it is presumed that the deficiency of mitochondrial translation is not the primary cause of CMTN. Recapitulation of the pathogenic YARS mutations in Drosophila leads to cell-autonomous progressive deficit in motor neuron performance, neuronal dysfunction, and axonal degeneration, which are hallmarks of DI-CMTC. Hence, DI-CMTC might arise not due to haploinsufficiency of YARS, but due to a gain-of-function of the TyrRS variant and/or interference with some unknown function of the WT enzyme.³⁹ Studies on several GARS and YARS mutants in mouse models led to a proposal that, unless somehow the defects in translation are restricted to axons, peripheral neuropathies are not caused by a defect in their canonical functions.⁴⁰ Misfolding and aberrant localization/targeting of the analyzed variants were ruled out as the potential disease cause as well.⁴⁰ Taken together, further studies on mechanisms by which variants AaRS cause neuropathies are warranted.

Leukoencephalopathies

Leukoencephalopathies designate a heterogeneous group of CNS disorders characterized by the loss of white matter and they represent the second disorder that was associated with mutations in ARS genes (Table 1). Interestingly, before any associations between human AaRSs and CNS disorders were known, Lee and colleagues reported that the sticky mouse phenotype is caused by a homozygous missense mutation A734E in the editing domain of cytosolic AlaRS.⁴¹ The mutant mice had rough, unkempt “sticky” appearance of fur, and with age, the rough coat was accompanied by follicular dystrophy and patchy hair loss. After 6 weeks of age, mild tremor progressed to ataxia accompanied by substantial loss of Purkinje cells, particularly in cerebellum. The A734E mutation compromises the proofreading activity of AlaRS, which leads to accumulation of misfolded proteins, cell death of terminally differentiated neurons, and neurodegeneration. The premise was that accumulation of misfolded proteins eventually overwhelms unfolded protein response and protein quality check leading to cell death. This was the first report suggesting that loss of canonical functions of AaRSs could lead to neurodegeneration via global protein misfolding. However, the study bore another layer of significance in that it had shown that faults in processes facilitated by AaRSs could dramatically affect CNS and not only the peripheral nervous system. Indeed, the following year, it was showed that pathogenic mutations in human DARS2 elicit leukoencephalopathy ⁴² (Table 1). Patients had slow progressive cerebellar ataxia, spasticity, dorsal column dysfunction, and mild cognitive decline and deficit, and they presented symmetric white matter lesions, stroke-like lesions of gray matter, and axonal neuropathy. The disorder was classified as Leukoencephalopathy with Brain stem and Spinal cord involvement and Lactate elevation (LBSL), an autosomal recessive disease most often manifested in early childhood. In total, 11 missense (S45G, C152F, R179H, Q184K, Q248K, R263Q, D560V, L613F, L626Q, L626V, Y629C), 2 nonsense (R263* and E425*), 3 deletional (A522_K588del, M134_K165del and A100_P132del), and 2 frame-shifting (R76SfsX5 and E424NfsX1) mutations were identified as the cause of LBSL.⁴² The recurrent frame-shifting mutation R76SfsX5 results in truncated enzyme, but because of its ‘leakiness’ a significant fraction of the gene product is functional. Mapping mutations onto the crystal structure of mt-AspRS ⁴³ reveals that they are well distributed across domains. The speculation was that levels of mt-tRNA^Asp might be significantly higher in the brain making this organ particularly vulnerable to changes in activity levels of mt-AspSR. However, targeting and co-localization of pathogenic variants is not altered ⁴⁴ and only R263Q and L626Q had reduced catalytic activity. R58G, T136S, and L626Q could not effectively form enzyme dimers,⁴⁴ while allosteric communication between monomers was impaired in Q184K and R263Q. Moreover, Q184K exhibited increased propensity toward aggregation.⁴⁵ Although the overall structure of variants was preserved, their stability was reduced when compared with the WT enzyme.⁴⁵ Because DARS2 mutations affect residues conserved only in mammals, it was suggested that the deficiency of non-canonical function(s) leads to the observed pathology.⁴⁵

Subsequent studies identified additional LBSL-causing mutations in DARS2. One patient was compound heterozygous for R76SfsX5 and C449_K521del,⁴⁶ while the other one carried R76SfsX5 and L239P.⁴⁷ Eight LBSL patients of the Finish cohort had the recurrent R76SfsX5, 7 harbored a deletion M134_K185del, and one had a missense mutation C152F.⁴⁸ The adult LBSL patients who were heterozygous for R76SfsX5 and M134_K185del developed information-processing and working memory impairment after adolescence.⁴⁹ Subsequently, a case of a 23-year old French Caucasian woman with LBSL and asymptomatic sister was reported.⁵⁰ The siblings were heterozygous for R76SfsX5 and L249I. Interestingly, both patients were asymptomatic in spite of typical extensive abnormalities detected by magnetic resonance imaging (MRI). A 17-year old Indian boy having LBSL without elevated lactate in white matter was compound heterozygous for R76SfsX5 and Q357*,⁵¹ while a Polish LBSL patient had only the frame-shifting mutation R76SfsX5.⁵² A consanguineous family having diffuse leukoencephalopathy, a variant of LBSL, was homozygous for an intronic change 22 base pairs upstream of exon 3 c.228–22T>A in DARS2 that causes exon 3 skipping and protein truncation.⁵³ Two siblings died at early age and one was alive at 23 y of age.⁵⁴ The severity of the disorder was thought to be a consequence of the homozygosity, but this was challenged by a rather benign episodic ataxia that was evoked by a homozygous R609W mutation.⁵⁵ It was revealed that mutations in intron 2 of DARS2 have a larger effect on exon 3 exclusion in neural cells than in non-neural cells.⁵⁶ Further, the severity of changes detected by MRI did not correlate with clinical profiles in 2 LBSL patients who had a mutation c.1395_1396delAA.⁵⁷ Additional LBSL-causing mutations–I139T, L250P, and R274SfxsX8 – were identified. Interestingly, in one case, the neurologic status deteriorated rapidly after surgery but then partially recovered, implying that the deterioration may be induced by stress and that it is not always irreversible.⁵⁸ Most recently, a case of an 8-year old male LBSL patient who was homozygous for a missense mutation R58G was reported.⁵⁹ In spite of the homozygosity, this patient presented episodic ataxia and adult-onset type of disease. Most intriguingly, the patient was presented at 9 months of age with severe neurologic deterioration after respiratory tract infection but then almost completely recovered by age 3. The symptom remission is most reminiscent to the EARS2-related cases that also recovered albeit only partly (see below) and reiterates that the clinical and genetic spectrum of DARS2-induced LBSL is broader than originally thought, and that stress-induced neurodegeneration may not be irreversible.

Mutations in other AARS2 genes are associated with leukoencephalopaties as well (Table 1). A series of mutations in human EARS2 cause Leukoencephalopathy with Thalamus and Brainstem involvement and high Lactate (LTBL).⁵⁸ Patients presented abnormalities in cerebral and cerebellar white matter, thalamus, midbrain, pons, and medulla oblongata. Curiously, “mild” group partially recovered with age, while the “severe” group stagnated and displayed brain atrophy. Patients had mutations R168G and T426_R427insL in the maternal allele and R108W in the paternal allele, and a series of other mutations (see Table 1 in Steenweg et al. ⁵⁸). Another patient who was heterozygous for a missense mutation R412C and a frame-shifting mutation K471Nfs*14 in EARS2 exhibited only a mild case of LTBL.⁶⁰ Further, complex rearrangements of the MARS2 gene involving gene duplication give rise to Autosomal Recessive Spastic Ataxia with Leukoencephalopathy (ARSAL), which is typified by ataxia, severe cerebellar and some cerebral atrophy, dystonia, and leukodystrophy ⁶¹ (Table 1). Levels of mt-MetRS, mitochondrial translation, and respiration were decreased in patient's samples.⁶¹ Finally, 6 patients who had later onset (ovario) leukodystrophy carried mutations in AARS2 ⁶² (Table 1). Patients were heterozygous either for F50C and R521* or E405K and G965R mutations (other mutations: Q537* and Q784Sfs*9, R199C and V730M; A77V, R199C and V730M; F131del, R199C and V730M; and T871Nfs*21, R199C and V730M). F50C and E405K affect the catalytic domain, while other mutations impact the editing domain of mt-AlaRS. The F50C variant retained partial activity, whereas R521* was inactive. The patients presented signs of neurologic disorder in childhood and then evident cerebellar ataxia and spasticity, cognitive decline with frontal lobe decline and poor memory, inactivity and behavioral changes, and depression and other psychiatric features. Ovarian failure in female patients resembled Perrault syndrome caused by mutations in HARS2 and LARS2 (see Table 1 and under “Other sensorineural disorders”). Based on similarities with DARS2-caused LBSL and leukoencephalopathy with vanishing white matter, which is caused by mutations in the initiation protein factor eIF2B, the common pathophysiological mechanism for AARS2 and DARS2-related leukoencephalopathies was proposed.⁶²

Initially, a pattern emerged in which mutations in ARS elicit neuropathies of the peripheral nervous system and those in ARS2 cause CNS disorders. This dogma was first challenged by findings obtained from studies on the sticky mouse phenotype ⁴¹ and then followed by discoveries that mutations in exclusively cytosolic DARS and RARS cause leukoencephalopathies (Table 1). In particular, compound heterozygous and homozygous mutations in DARS cause leukoencephalopathy that was clinically characterized as Hypomyelination with Brain stem and Spinal cord involvement and Leg spasticity (HBSL), a disorder similar to DARS2-induced LBSL.⁶³ The patients were heterozygous for D367Y and A274V, homozygous for M256L, homozygous for R487C, heterozygous for R460H and R494G, or heterozygous for P464L and R494C. R487 and R494 are important for tRNA^Asp binding, while other residues are located in or near the active site. It was proposed that disrupted dimerization and/or inability of the pathogenic AspRS variants to participate in non-canonical functions may be the main driver(s) of HBSL. In another study, 3 individuals who were either heterozygous for S200C and S277F, homozygous for L426S, or heterozygous for H280L and D367H had HBSL.⁶⁴ S200, L426 and D367 are located in the catalytic domain, whereas S277 and H280 are in the flexible loop that alters its structure upon tRNA binding. Several patients with HBSL showed regression after viral illness or vaccination and 4 patients reacted positively to steroid therapy, hinting at the impairment of non-canonical functions of AspRS related to modulation of cytokine and immune responses in these cases. Concurrently, 4 patients harboring mutations in RARS were reported to suffer from hypomyelination characterized by severe spasticity and nystagmus⁶⁵ (Table 1). Two siblings were heterozygous for D2G and a mutation in intron 1 that causes aberrant splicing and a loss of canonical splice site. Another patient was heterozygous for D2G and a frame-shifting mutation C32Wfs*39 that truncates the enzyme. The fourth patient harbored Met1? and R512Q on separate alleles and exhibited microcephaly at early age, delayed development, transient nystagmus, axial hypotonia with poor head control, dystonic movements, and a pyramidal syndrome, which were suggestive of leukoencephalopathy with hypomyelination. D2G affects the N-terminal helix that is critical for formation of MSC, though in mammals there is an alternative start codon that yields soluble ArgRS devoid of the N-terminal domain. By contrast, R512Q affects the highly conserved residue in the core domain of the enzyme. Based on similarities with disorders caused by mutations in DARS, AIMP1,⁶⁶ and POLR3A ⁶⁷ and POLR3B,⁶⁸ it was proposed that mutations in RARS disrupt myelination through disruption of protein synthesis and tRNA homeostasis.

Severe early-onset brain disorders

Mutations in several ARS and ARS2 genes are associated with a severe brain disorder PontoCerebellar Hypoplasia (PCH), which is a group of related disorders affecting the development of the brain, particularly pons and cerebellum (Table 1). As the brain grows, severe microcephaly is manifested. The first report about a link between AaRSs and PCH was published over a decade ago.⁶⁹ Three patients from a consanguineous Sephardic Jewish family presented severe encephalopathy and multiple MRC defects.⁶⁹ A rapid white matter loss that led to brain volume decrease, and hypoplasia of cerebellum and vermis were consistent with PCH type 6 (PCH6). The 3 patients were homozygous for the IVS2+5 A>G substitution which results in skipping exon 2 during splicing and thus severely truncated mt-ArgRS. Activities of MRC I, III, and IV, and the levels of tRNA^Arg, Arg-tRNA^Arg, and mt-ArgRS transcripts were markedly reduced in patient samples. Subsequently, a female British patient born to nonconsanguineous parents developed profound mental retardation and severe microcephaly and she had edema of hands and feet.⁷⁰ Within 14 months after birth, cerebral atrophy, thinning of the pons, and atrophy of cerebellar hemispheres were evident and the disorder was classified as PCH6. The patient was compound heterozygous for missense mutations Q12R and M342V in RARS2. The former mutation impacts a highly conserved residue, while the latter affects exon splicing. Further, a patient who was heterozygous for a missense mutation M404K and K158del presented the profile consistent with PCH, and congenital lactic acidosis, and mild MRC IV deficiency.⁷¹ M404 is a highly conserved residue in helix α15 that interacts with the anticodon loop of tRNA. The deletion of K158 is difficult to predict but the residue is near the active site and immediately downstream of the HIGH motif, so it is plausible that the active site of the variant is disrupted. More recently, a report about 5 unrelated patients with PCH6 due to mutations in RARS2 was published.⁷² First family was heterozygous for I9V and R504-L528del, second for R245Q and R469H, and third for Q12R and W241R. W241R impacts the side chain in the core of the catalytic domain, and R245Q and R496H diminish the enzyme activity. The I9V mutation most likely modulates the enzyme targeting, while Q12R interferes with the splicing enhancer. The deletion variant, R504-L528del, is a product of aberrant splicing and skipping of exon 18. In spite of the truncation in the anticodon-binding domain, it is thought that this variant is functional. Somewhat milder case of 2 siblings who exhibited typical clinical features of PCH6 but with distinct neuroimaging features was presented around the same time.⁷³ The siblings were compound heterozygous for a missense mutation R258H and c.1651–2A>G. The siblings had supratentorial but not pontocerebellar atrophy, arguing that effects of RARS2 mutations are of broader spectrum than originally thought. Finally, a curios case of 2 siblings who presented PCH due to a homozygous mutation c.-2A>G in the 5′-UTR of the RARS2 gene was recently published.⁷⁴ The mutation is embedded in a CpG island within the RARS2 gene promoter and it markedly reduces the level of mt-ArgRS expression.

Mutations in QARS also cause PCH, thus setting another example about links between cytosolic AaRSs and CNS disorders (Table 1). Children in 2 unrelated families displayed progressive microcephaly, severe seizures in infancy, atrophy of the cerebral cortex and cerebellar vermis, mild atrophy of the cerebellar hemispheres, enlarged subarachnoid space, enlarged lateral ventrices, and thin corpus callosum.⁷⁵ Affected individuals in one family were compound heterozygous either for G45V and R403W or Y57H and R515W. All mutations impacted highly conserved residues located in either the appended N-terminal (NTD) or the catalytic domains. Whereas mutations in NTD (G45V and Y57H) had only a slight effect on the enzyme activity, the R->W substitutions all but erased it. This effect was in part due to decreased solubility of the R->W variants, but also due to their inability to interact with ArgRS, which is at the core of MSC.⁷⁵ The homozygous pathogenic qars mutations caused decreased brain and eye size and extensive cell death in the brain of zebrafish.⁷⁵ Structural and biophysical analyses of these pathogenic mutants revealed that G45V and Y57H decrease the thermal stability of the enzyme, while not affecting the overall structure.⁷⁶ By contrast, the R->W mutations affect the stability and foldability of the enzyme. When overexpressed in E. coli these pathogenic variants form stable complexes with the bacterial chaperonin, hinting at misfolding problems. ⁷⁶ It may be that decreased solubility, stability, foldability, and increased propensity toward misfolding of the pathogenic variants cause loss of function that leads to the development of the disorder. Additional PCH-causing mutations in QARS have recently been identified. The affected patient was a compound heterozygous for R463* and Q742H and exhibited progressive microcephaly with diffuse cerebral atrophy, severely deficient myelination, intractable seizures, and developmental arrest.⁷⁷

Besides PCH, mutations in QARS and RARS2 cause Early-Onset Epileptic Encephalopathy (EOEE), a group of damaging diseases characterized by early-onset seizures, developmental delay, and poor prognosis (Table 1). In the first study, 2 siblings were compound heterozygous for Y57H and K496* in QARS.⁷⁸ Clinical profiles included intractable seizures in infancy, postnatal microcephaly, cerebral atrophy, white matter loss, hypomyelination, severe developmental delay, and intellectual disability. Overexpression in neuroblastoma 2A cells shows that expression of Lys496* severely reduced and that mutant protein may form aggregates.⁷⁸ In the second study, 2 siblings who were compound heterozygous for splice variants c878+5G>T and c110+5A>G in RARS2 displayed similar phenotype.⁷⁹ Although one of the mutations was recurrent,⁶⁹ patients did not present PCH6 but rather fatal EOEE. Both siblings exhibited profound global developmental delay, microcephaly, and progressive cerebral and cerebellar atrophy. It is not clear how these QARS and RARS2 mutations elicit even more severe pathology at the molecular level.

Other AaRSs are being associated with severe childhood brain disorders as well. Two siblings with severe infantile visual loss, progressive microcephaly, developmental delay, seizures, and abnormal subcortical white matter were shown to be compound heterozygous for missense mutations R438W and E522K in KARS ⁸⁰ (Table 1). R438 and E522 are located in a highly conserved region of the catalytic domain of LysRS. Importantly, the KARS-induced syndrome was remarkably similar to the one caused by mutations in QARS and RARS2. Whereas similarities with RARS2-caused disorders could be rationalized by the presumed impairment of mitochondrial function, the alignment with the QARS-induced phenotype is more complex to unravel. A simple explanation is that human cytosolic GlnRS is, in fact, a bifunctional AaRS and that, yet again, faults in mitochondria could be blamed for the disorder. Alternatively, pathogenic variants of LysRS and GlnRS cannot participate in formation of MSC, which, in turn, would disturb proteostasis and tRNA homeostasis. Further, a homozygous missense mutation T367I in VARS2 was linked to microcephaly and epilepsy associated with deficiency of MRC I ⁸¹ (Table 1). Levels of mt-ValRS and aminoacylated tRNA^Val were decreased in patient samples and expression of the WT mt-ValRS rescued mitochondrial respiration in immortalized fibroblasts, suggesting loss of function as the disease cause. Also, a patient heterozygous for missense mutation D325Y and subtelomeric 6p25.1 deletion of 88kb in FARS2 had seizures, epileptic attacks, myoclonic jerks, visual impairment and minor dysmorphic features (Table 1). White matter lesions and the thin corpus callosum were evident. Since D325Y variant does not bind ATP and does not charge the cognate tRNA authors argued that haploinsufficiency of FARS2 may be responsible for the disease development.⁸² Further, heterozygous mutations E217del and P251L in CARS2 cause severe epileptic encephalopathy with the clinical profile similar to the FARS2- and VARS2-related cases ⁸³ (Table 1). The CARS2 mutations impact highly conserved residues in the catalytic domain and cause deficiency in MRC I, III, and IV, and incomplete assembly of MRC V. Furthermore, 2 variants in PARS2 (V95I and E203K) and 5 variants in NARS2 (RC51, H167R, F236C, L395R, and a mutation in a canonical splicing site) were found to cause epilepsy, developmental delay and progressive developmental regression followed by hypotonia, postnatal microcephaly, diffuse cerebral atrophy, and atrophy of the basal ganglia and cerebellum ⁸⁴ (Table 1). Lastly, Simmons and colleagues reported a case of 2 siblings and an unrelated individual with severe infantile epileptic encephalopathy, clubfoot, absent deep tendon reflexes, extrapyramidal symptoms, and persistently deficient myelination.⁸⁵ Surprisingly, these patients harbored mutations in a gene encoding an exclusively cytosolic enzyme, AlaRS (Table 1). The siblings were compound heterozygous for K81T and R751G, and the unrelated individual was homozygous for R751G. The catalytic activity of the AlaRS pathogenic variants was severely reduced, implying loss of function. These findings demonstrated that defects in mitochondrial and cytosolic AaRS isoforms converge to a common set of neuropathological phenotypes, as was initially hinted by studies on the sticky mouse.⁴¹

Fatal infantile Syndromes

Mutations in ARS and ARS2 genes also cause fatal neurodegenerative syndromes such as Alpers and Leigh syndrome (Table 1). The first report described 2 Finnish and one Saudi patient with fatal epileptic mitochondrial encephalopathy due to compound heterozygous missense mutations I329T and D391V ⁸⁶ or a homozygous Y144C mutation in FARS2.⁸⁷ All patients presented catastrophic epilepsy, lactic acidemia, and early lethality consistent with Alpers syndrome, a progressive neurodegenerative disorder that presents in infancy or early childhood and is characterized by diffuse degeneration of cerebral gray matter. The I329T mutation weakened ATP binding, D391V reduced affinity for L-Phe, and Y144C disrupted tRNA binding.⁸⁶ The stability and foldability of the I329T and D391V variants were also impaired, suggesting that loss of function of FARS2 leads to pathology.⁸⁶ Mutations in NARS2 and PARS2 also cause Alpers syndrome.⁸⁸ One patient was homozygous for a missense mutation P214L in NARS2. This mutation affects a residue in the catalytic domain important for enzyme dimerization. The other patient was compound heterozygous for Lys378* and S279L, which is located in a conserved motif of unknown function in mt-ProRS. Furthermore, mutations in IARS2 and NARS2 cause another fatal neurodegenerative syndrome, Leigh syndrome.^89,90 Symmetric, bilateral lesions in the basal ganglia, thalamus, and brain stem are the main features of this syndrome. One patient was heterozygous for W706* and E708K in IARS2, and the other one was compound heterozygous for Y323* and N381S in NARS2. E708K most likely affects the interface between the catalytic and anticodon-binding domains in mt-IleRS, while N381S decreases enzyme levels by disrupting dimerization of mt-AsnRS. However, the exact effect of these variants and their role in disease is not clear. Next, a homozygous mutation K65E in EARS2 causes fatal infantile multisystem disease involving the brain and liver in a child of related parents of Turkish decent ⁹¹ (Table 1). This was the first report describing multisystem effects of ARS2 mutations; such effects were typically seen in patients harboring mutations in genes encoding translation elongation factors and ribosomal proteins. Another report described mutations in TARS2 that cause axial hypotonia and severe psychomotor delay in 2 siblings who died within a few months after birth due to metabolic crisis ⁸¹ (Table 1). Patients were compound heterozygous for P282L and a nucleotide change in position +3 of intron 6 (g4255A>Gl; c.695+3A>G). Levels of mt-ThrRS and aminoacylated tRNAs were decreased, and expression of WT enzymes rescued mitochondrial respiration in immortalized fibroblasts thus suggesting loss of function as the cause of the disorder. It was also shown that the pathogenic mutation P282L decreases both aminoacylation and proofreading activities of mt-ThrRS.⁹² Taken together, fatal infantile syndromes seem to be caused primarily by the loss-of-function variants of AaRSs. Such deficiency is likely to cause the major disturbance of proteostasis, which ultimately leads to systemic failures and death.

Other sensorineural disorders

It is prudent to briefly mention that variants of AaRSs cause various sensorineural disorders. Mutations in HARS2, LARS2, and MARS2 cause Perrault syndrome (ovarian dysgenesis and sensorineural loss), developmental delay, and poor growth ^93-95 (Table 1). Also, a homozygous missense mutation Y454S in human HARS causes Usher syndrome,⁹⁶ which is characterized by progressive vision and hearing loss in early childhood years, delayed gross motor development, hyperactive patellar tendon reflexes, mild truncal ataxia, and a wide-base gait. The side chain of Y454 resides at the interface between catalytic and anticodon-binding domains but the Y454S mutation only mildly impacts the activity of the enzyme. Further, the R137Q variant of human HisRS causes late-onset sensory-predominant axonal peripheral neuropathy.⁹⁷ The mutation impacts the highly conserved residue at the dimerization interface and this is both the loss-of-function and neurotoxic variant.⁹⁷ Further, homozygous missense mutations D377N and Y173H in KARS cause autosomal-recessive nonsyndromic hearing impairment DFNB89 in unrelated families ⁹⁸ (Table 1). These mutations may affect tetramer formation, interactions with tRNA^Lys and p38, and the aminoacylation activity. A homozygous mutation inducing aberrant splicing of exons 7 and 8 in NARS2 elicits mild intellectual disability, epilepsy and severe myopathy.⁹⁹ A missense homozygous mutation D142Y in FARS2 causes autosomal-recessive Hereditary Spastic Paraplegia (HSP),¹⁰⁰ a heterogeneous group of neurodegenerative disorders characterized by spasticity of the lower limbs due to pyramidal tract dysfunction. The D142Y mutation severely disrupts both steps of the aminoacylation reaction and it lowers the enzyme activity by ∼90%.¹⁰⁰ It is important to mention that heterozygous mutations V5M and R702W in MARS were predicted to cause HSP as well.¹⁰¹ Lastly, a homozygous mutation c665G>A in exon 6 of human CARS2, which causes aberrant splicing and removal of 28 amino acids that are involved in binding of the acceptor end of tRNA^Cys, evokes clinical symptoms closely resembling myoclonus epilepsy with ragged red fibers.¹⁰² The affected individuals presented myoclonic epilepsy, progressive spastic quadriplegia, progressive loss of vision and hearing, and progressive cognitive decline most likely due to the loss-of-function of the mt-CysRS variant.

Concluding remarks

A decade ago, a handful of ARS and ARS2 genes were associated with the nervous system disorders. The initial pattern suggested that whereas faults in cytosolic isoforms were responsible for the peripheral neuropathies, malfunctioning of mitochondrial enzymes were blamed for more severe CNS disorders. However, variants of 10 cytosolic and 14 mitochondrial AaRSs give rise to quite heterogeneous, and often overlapping, clinical profiles (Table 1). When effects of mutations in distinct cytosolic and mitochondrial enzymes converge to a common phenotypic picture, the shared, yet to be identified, pathological mechanism is more than suggestive. As more patients having neurologic disorders of unknown etiologies are subjected to whole-exome and whole genome sequencing, it is all but certain that links between the remaining AaRSs and neurologic disorders will soon be revealed. A number of studies analyzed how specific mutations might impact the structure and function of particular AaRSs at the molecular level, and yielded proposals that the loss of (canonical or non-canonical) function, neurotoxicity, gain of function, mislocalization, mistargeting, misfolding, and/or aggregation of the enzyme may be the culprit behind pathologic conditions. When the loss of function was evident, typically severe disorders with fatal outcome would appear. In other instances, affected patients often carried a combination of mutant gene products in which one of the mutations was ‘mild’ and the other was damaging. Such a situation could lead to accumulation of misfolded proteins that eventually overwhelm both the unfolded protein response and protein quality check. Alternatively, or perhaps concurrently, the neural cell types with the pathogenic genotypes somehow become more vulnerable under duress. Several reports mentioned that patients deteriorated after respiratory tract infection, vaccination, surgery, and in many cases the disorder would stampede after birth. Hence, the compromised translational apparatus could reveal fragility of the fully differentiated neural cell when exposed to a stressor. Along the same lines, in a few instances a partial and even full recovery of patients was recorded, thus arguing that the damage may not be always irreversible and that these enzymatic systems could be targets for pharmacological treatments. In spite all of this, the role of AaRSs and mt-AaRSs in the development of both the diseased and healthy human nervous system remains poorly understood.

We have mentioned how activity of particular variant AaRSs could be altered and/or lost, but one would expect the universal impact of these variants on all cell types. Surprisingly, in most patients no other organ system besides the nervous system was affected, at least not as dramatically. So, why are the neural cells particularly vulnerable to faults in maintaining proteostasis? Perhaps, the chaperone machinery, the unfolded protein response, and protein quality check operate in a distinct manner in neural cells, which makes the entire system extremely fragile when a component or components do not operate at certain capacity. Also, because of the unique cellular morphology that relies on the proteinaceous sheath covering the axonal extension, neurons could pose significant challenges to malfunctioning gene translation apparatus. Errors that may accumulate in the myelin itself could hamper the sheath formation and they could simply overwhelm both the protein folding and degradation systems. Furthermore, specific requirements for timely transport, targeting and recycling of proteins may make these processes significantly more impacted in neural cells. To add a layer of complexity, these events are likely to be spatially constrained during development of brain structures, the brain itself, and the nervous system as a whole. Hence, the deficient proteostasis could impair cellular cross talk and the critical developmental events. Moreover, as already alluded above, it is plausible that fragility of proteostasis reveals itself at some critical, yet undefined, juncture during neurologic development and/or under duress. Lastly, it is attractive to suggest that, at least in some instances, non-canonical functions of AaRSs that may regulate neurodevelopment could be impaired by mutations. This is especially relevant for those enzymes that participate in formation of MSC, a large assembly whose structure and function are still not fully understood.

However, mutations in other enzymes and protein factors associated with gene translation (e.g. EIF4B, AIMP1, TSEN, POL3A, POL3B, CLP1, EXOSC3, and SEPSECS) cause similar neurologic disorders and the question about the role of the intact gene translation machinery in the development and maintenance of neural cell types, peripheral nervous system, specific brain structures and the brain is positioned front and center. It could well be that the translational apparatus, and AaRSs as its integral part, play a unifying role during these processes. To define this role and address the fundamental questions, comprehensive studies using cell- and animal-based model systems in which pathogenic genotypes and phenotypes are faithfully recapitulated are urgently needed. With the advent of modern CRISPR-based and induced pluripotent stem cell technologies, engineering of such systems is within reach. Concurrently, more structural studies on human AaRSs, mt-AaRSs, and MSC are warranted. These combined efforts guarantee an avalanche of new and exciting discoveries about AaRSs and their role in the human nervous system.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Funding

This work was supported by the National Institute of General Medical Sciences grant (GM097042 to MS).

Author contributions

JO and MS wrote the manuscript.