Mutations in DARS Cause Hypomyelination with Brain Stem and Spinal Cord Involvement and Leg Spasticity

Abstract

Inherited white-matter disorders are a broad class of diseases for which treatment and classification are both challenging. Indeed, nearly half of the children presenting with a leukoencephalopathy remain without a specific diagnosis. Here, we report on the application of high-throughput genome and exome sequencing to a cohort of ten individuals with a leukoencephalopathy of unknown etiology and clinically characterized by hypomyelination with brain stem and spinal cord involvement and leg spasticity (HBSL), as well as the identification of compound-heterozygous and homozygous mutations in cytoplasmic aspartyl-tRNA synthetase (DARS). These mutations cause nonsynonymous changes to seven highly conserved amino acids, five of which are unchanged between yeast and man, in the DARS C-terminal lobe adjacent to, or within, the active-site pocket. Intriguingly, HBSL bears a striking resemblance to leukoencephalopathy with brain stem and spinal cord involvement and elevated lactate (LBSL), which is caused by mutations in the mitochondria-specific DARS2, suggesting that these two diseases might share a common underlying molecular pathology. These findings add to the growing body of evidence that mutations in tRNA synthetases can cause a broad range of neurologic disorders.

Main Text

As recognized more than a decade ago, nearly half of childhood leukoencephalopathies remain unclassified and are a diagnostic problem for pediatric neurologists and neuroradiologists.¹ Many of these disorders are extremely rare, precluding conventional genetic linkage studies, but they are potentially amenable to whole-genome sequencing (WGS) or whole-exome sequencing (WES) analysis, particularly if unaffected family members are available and affected individuals can be nosologically clustered with MRI pattern recognition.¹,²

The first child examined in this study, subject 1 (Figure 1), was diagnosed with a tethered cord as a newborn and subsequently developed global developmental delay and severe leg spasticity (for clinical details, see Table S1, available online); extensive white-matter abnormalities involved the supratentorial brain white matter, brain stem, and spinal cord (Figure 1 and Table S2). A comprehensive panel of imaging, metabolic, and genetic studies failed to yield a definitive diagnosis, and he was subsequently enrolled in an ongoing study investigating the etiology of unclassified neurological disorders at the Murdoch Children’s Research Institute (the study was approved by the Royal Children’s Hospital Research Ethics Committee [HREC reference number 28097], and informed consent was obtained from all participating individuals). WGS was performed on the affected proband and both parents (approved by the University of Queensland Ethics Office, projects 2012000375 and 2012000376). Sequencing reads were aligned to the reference human genome (UCSC Genome Browser hg19) with the Burrows-Wheeler Aligner,³ and downstream processing was carried out with a pipeline composed of the Genome Analysis Toolkit,4, 5 SAMtools,⁶ Picard (see Web Resources), Annovar,⁷ and a set of custom analysis and visualization tools. Average genome coverage for this family exceeded 38× (Table S3), which yielded more than 3.2 million variants per genome. A custom-developed differential family-trio analysis yielded a single likely candidate gene—DARS (MIM 603084, RefSeq accession number NM_001349.2), encoding an aspartyl-tRNA synthetase (AspRS). In this gene, we found a compound-heterozygous mutation, c.1099G>T (p.Asp376Tyr) (paternal allele) and c.821C>T (p.Ala274Val) (maternal allele) (see Figure 2 and Table 1), which was validated by Sanger sequencing. Intriguingly, we noted that recessive mutations in the DARS mitochondrial paralog, DARS2 (MIM 610956), had been previously associated with another white-matter disorder, leukoencephalopathy with brain stem and spinal cord involvement and elevated lactate (LBSL [MIM 611105]).⁸

Figure 1.

MRI Patterns Characteristic of HBSL

MRI images of subjects 1, 2 and 4: (A), (B), and (D)–(L) show axial T2-weighted images, and (C) shows a sagittal T2-weighted image. In (A), (G), and (J), note the diffuse T2 hyperintense signal of almost the entire supratentorial white matter. In subject 1, only the anterior section of the posterior limb of the internal capsule and a narrow band of the subcortical white matter in the occipital lobe are spared. In all subjects, the superior (D, H, and K; red arrows) and inferior (E, I, and L; red arrows) cerebellar peduncles and the anterior brainstem (D, H, and K; blue arrows) are affected. In the spinal cord, the dorsal columns (C and F; blue arrows) and the lateral corticospinal tracts (F, red arrow) are hyperintense. We note that in several individuals, the white-matter signal on the T2-weighted images was more elevated than usual for hypomyelinating leukoencephalopathies, similar to what has recently been described for another disease, hypomyelination with congenital cataracts (MIM 610532).²

Figure 2.

DARS Mutations in Individuals with HBSL

(A) Genomic organization of DARS in humans (UCSC Genome Browser hg19).

(B) Mutations and their positions within the DARS cDNA. Note that all identified mutations fall in the 3′ end of DARS, which encodes the active site of the AspRS. No mutations were detected within the hinge domain or the anticodon-binding domain.

(C) Five DARS mutations affect amino acids that are conserved in S. cerevisiae (Table 1), and the orientation of these amino acids within the protein can therefore be visualized with the crystal structure of the yeast DARS homolog, DSP1 (Protein Data Bank ID 1ASZ), in complex with tRNA-Asp. In this image, p.Ala274 (yeast p.Ala327) is shown in red, p.Asp367 (yeast p.Asp421) is shown in green, p.Pro464 (yeast p.Pro521) is shown in cyan, p.Arg487 (yeast p.Arg543) is shown in orange, and p.Arg494 (p.Arg554) is shown in magenta. Note that four of the affected amino acids sit within regions that stabilize the tRNA (orange ribbon) in complex with DSP1. Two alterations, p.Ala274Val and p.Asp367Tyr, sit in the active-site pocket and resulted from a compound-heterozygous mutation in a single subject (Table 1). An animation of this structure can be found online (DARS animation in Web Resources). Visualization of the predicted human DARS structure and the HBSL mutations can be found in Figure S1.

Table 1.

DARS Mutations Associated with HBSL

Subject	Country of Origin	Genomic Position (hg19)	cDNAaMutation	Exon	Protein Alteration	Inheritance	Frequencyb	dbSNP ID	Conservationc	Damage Predictiond
1	Australia	chr2: 136,673,803	c.1099G>T	11	p.Asp367Tyr	paternal	0%	-	S. cerevisiae	damaging
		chr2: 136,678,161	c.821C>T	10	p.Ala274Val	maternal	0.001%	rs112205661	S. cerevisiae	damaging
2,e 3,e 5, 8	India and Pakistan	chr2: 136,680,399	c.766A>C	9	p.Met256Leu	homozygous	0%	-	C. elegans	damaging
4	India	chr2: 136,664,933	c.1459C>T	16	p.Arg487Cys	homozygous	0%	-	S. cerevisiae	damaging
6,e 7e	USA	chr2: 136,668,744	c.1379G>A	15	p.Arg460His	paternal	0%	-	C. elegans	damaging
		chr2: 136,664,912	c.1480C>T	16	p.Arg494Gly	maternal	0.001%	rs147077598	S. cerevisiae	damaging
9,e 10e	UK	chr2: 136,668,733	c.1391C>T	15	p.Pro464Leu	ND	0.0018%f	rs148806569	C. elegans	damaging
		chr2: 136,664,912	c.1480C>T	16	p.Arg494Cys	ND	0%	-	S. cerevisiae	damaging

The following abbreviation is used: ND, not determined.

^aDARS RefSeq NM_001349.2.

^bFrequency of occurrence according to the ESP6500SI release of the NHBLI Exome Variant Server (see Web Resources).

^cConservation was analyzed in 11 species, including human, chimp, rhesus, mouse, chicken, frog, zebrafish, tetraodon, Drosophila, C. elegans, and S. cerevisiae.

^dPlease see Table S4 for damage scores from six damage prediction algorithms.

^eSibling pairs are as follows: 2 and 3, 6 and 7, and 9 and 10.

^fFrequency specifically within the Africa-American (AA) ESP6500SI population.

Concurrently, as part of an ongoing study applying MRI pattern analysis² to unclassified leukoencephalopathies at VU University Medical Center (the study was approved by the VUMC Medical Ethics Committee, and informed consent was obtained from all participating individuals), four subjects—one pair of brothers and two single unrelated individuals from consanguineous Indian families—with a previously undescribed disorder were identified. All had diffuse hypomyelination of the cerebral white matter and striking abnormalities of specific brain stem structures (Figure 1 and Table S2). In those subjects who had a spinal MRI, the dorsal columns also showed abnormal signal. Clinically, all subjects developed nystagmus in the first year of life and had severe leg spasticity and mild cerebellar dysfunction. For identification of the responsible mutations, WES was performed on one subject and his mother and on another unrelated affected subject. Using the bioinformatic methods outlined above and in Table S3, we identified two homozygous DARS mutations: subject 2 was homozygous for c.766A>C (p.Met256Leu), and subject 4 was homozygous for c.1459C>T (p.Arg487Cys) (Figures 1 and 2 and Table 1). Sanger sequencing confirmed these mutations and a c.766A>C (p.Met256Leu) homozygous mutation in subject 2’s affected sibling (subject 3; see Tables 1 and S1). Both sets of parents were confirmed to be heterozygous carriers.

We then queried two biorepositories, the Centre for Children with White Matter Disorders and the Myelin Disorders Bioregistry Project (approved by the VUMC Medical Ethics Committee and the Children’s National Medical Center’s Ethics Program, respectively; informed consent was obtained from all participating individuals), and identified five more subjects who had the same MRI pattern and who were all confirmed to have mutations in DARS (Tables 1 and S1–S4). This included a five-member family that had been analyzed by ultra-high-coverage exome sequencing, which revealed a compound-heterozygous DARS mutation in two affected sisters (subjects 6 and 7; Tables 1 and S1–S4); this mutation was absent from their unaffected sibling. Taken together, these findings demonstrate that a previously uncharacterized leukoencephalopathy characterized by hypomyelination with brain stem and spinal cord abnormalities and leg spasticity (HBSL) is caused by recessive mutations in DARS.

To evaluate the potential effects of these DARS mutations, we assessed their relative incidence, conservation, and predicted pathogenicity (Table 1). Five were absent from SNPdb 135, the 1000 Genomes Project databases, and the National Heart, Lung, and Blood Institute (NHBLI) Exome Sequencing Project Exome Variant Server (ESP6500SI release), and the remaining three had minor allele frequencies of ∼0.001 (Table 1). All are conserved in the C. elegans DARS ortholog, drs-1, and five are conserved in the S. cerevisiae ortholog, DPS1. AspRS DARS is a fundamental component of the translational machinery, and a wide range of previous studies have shown that loss of AspRS activity in Drosophila and yeast is lethal (reviewed in Table S5) and that selective knockdown of AspRS in Drosophila neuroblasts leads to shorter lineages, alterations to cell shape, and defects in the ganglion mother cell.⁹ Consistent with these reports, we found that when assessed with a combination of six damage-prediction algorithms, all the HBSL DARS mutations were classified as pathogenic by at least two independent metrics (Table S4), strongly suggesting that these variants are the cause of HBSL.

The mutations identified in this study lie in the 3′ third of DARS, the region which corresponds to the C-terminal active-site domain of DARS. Two previous site-directed-mutagenesis studies in yeast have shown that alteration of any of the amino acids that reside within the active site’s hydrogen-bond network or the nearby side chains is detrimental to the enzyme¹⁰ and that any alteration in the C-terminal lobe of the protein leads to a minimum 3-fold reduction of enzyme activity.¹¹ To investigate how the mutations identified in this study might affect DARS activity, we mapped each affected amino acid to a predicted human DARS crystal structure (Figure S1) and all the highly conserved amino acids to the yeast DPS1 crystal structure (Figure 2). Strikingly, we found that two mutations alter amino acids that sit within the active-site pocket (p.Ala274Val and p.Asp367Tyr from subject 1), three alter amino acids that are likely to destabilize the tRNA-protein interaction (p.Arg487Cys, p.Arg494Cys, and p.Arg494Gly), and the remaining mutations alter amino acids that are likely to be fundamental to chains supporting the active site (Figure 2 and Figure S1).

To gain further insight into the pathogenic role of these mutations in HBSL, we performed an investigation of DARS expression in humans and mice by using a curated set of publicly available data. Consistent with the fact that DARS is a core component of the translational machinery, we found that it is diffusely localized in cytoplasm and broadly expressed across tissue types (by both RNA-seq and array-based measures of expression), including the central nervous system (Figures S2 and S3). In mice, Dars shows specific expression in sagittal brain sections with staining in the hippocampus, the dentate gyrus, and the molecular layer of the cerebellum, which is strikingly similar to the expression patterns of glycyl-tRNA synthetase (Gars) and lysyl-tRNA synthetase (Kars), but shows little overlap with glial and oligodendrocyte markers (Figure S4). For example, Dars, Kars, and Gars are clearly expressed in the cerebellar molecular layer (which houses Purkinje neurons), whereas the oligodendrocyte markers Mog, CNPase, and Olig2 are highly expressed in the granular layer (Figure S4). DARS expression in the developing and adult human brain shows similar patterns—it is highly expressed in the ventricular and subventricular zones, including hippocampal subfields, the midlateral temporal cortex, and the frontal polar cortex (Figure S5). Likewise, DARS immunostaining of the cerebellum, cerebral cortex, hippocampus, and lateral ventricle all show preferential neuronal staining, and staining of peripheral neurons is evident in the colon (Figure S6). Overall, these data are consistent with HBSL’s clinical presentation.

It is becoming increasingly apparent that mutations in genes encoding both cytoplasmic and mitochondrial tRNA synthetases (ARSs) can cause a wide range of neurological and multisystem disorders (reviewed in Table 2). For example, leukoencephalopathy with thalamus and brain stem involvement and high lactate (LTBL) is caused by mutations in EARS2⁹ (MIM 612799), encoding a mitochondrial tRNA synthetase, and several forms of Charcot-Marie-Tooth disease (CMT [MIM 613287, 601472, 608323, and 613641]), a peripheral neuropathy, are caused by damaging mutations in AARS (MIM 601065), GARS (MIM 600287), YARS (MIM 603623), and KARS (MIM 601421) (Table 2), encoding cytoplasmic aminoacyl-tRNA synthetases. Similarly, recessive mutations in DARS2, encoding the mitochondria-specific aspartyl-tRNA synthetase, cause LBSL.⁸

Table 2.

Disease-Associated Genes Encoding tRNA Synthetases

Gene	Gene MIM	Cognate Amino Acid	Localization	Disease	Hallmarks	Phenotype MIM	Inheritance	Reference
AARS	601065	alanine	C	CMT 2N	axonal neuropathy	613287	AD	Latour et al.12
AARS2	612035	alanine	M	combined oxidative phosphorylation deficiency 8	muscular hypotonia, fatal hypertrophic cardiac myopathy	614096	AR	Gotz et al.13
DARS	603084	aspartate	C	HBSL	leukoencephalopathy (hypomyelination) with brain stem and spinal cord involvement	-	AR	this study
DARS2	610956	aspartate	M	LBSL	leukoencephalopathy with brain stem and spinal cord involvement and elevated lactate	611105	AR	Scheper et al.8
EARS2	612799	glutamine	M	LTBL	leukoencephalopathy with thalamic and brain stem involvement and elevated lactate	614924	AR	Steenweg et al.14
FARS2	611592	phenylalanine	M	combined oxidative phosphorylation deficiency 14	neonatal encephalopathy with seizures and liver involvement	614946	AR	Elo et al.15 and Shamseldin et al.16
GARS	600287	glycine	C	CMT 2D; distal hereditary motor neuropaty type V	axonal neuropathy	601472 and 600794	AD	Antonellis et al.17
HARS	142810	histidine	C	Usher syndrome type 3B	retinitis pigmentosa, sensorineural deafness, ataxia	614504	AR	Puffenberger et al.18
HARS2	600783	histidine	M	Perrault syndrome 2	sensorineural deafness, primary amenorrhea, ovarian dysgenesis	614926	AR	Pierce et al.19
KARS	601421	lysine	C	recessive intermediate CMT	neuropathy with axonal and demyelinating features	613641	AR	McLaughlin et al.20
LARS2	604544	leucine	M	Perrault syndrome	sensorineural deafness, premature ovarian failure	-	AR	Pierce et al.21
MARS2	609728	methionine	M	spastic ataxia 3	autosomal-recessive spastic ataxia with leukoencephalopathy (arsal)	611390	AR	Bayat et al.22
RARS2	611524	arginine	M	pontocerebellar hypoplasia type 6	pontocerebellar and cerebellar atrophy, epilepsy	611523	AR	Edvardson et al.23
SARS2	612804	serine	M	hyperuricemia, pulmonary hypertension, renal failure, and alkalosis	renal failure in infancy, metabolic alkalosis, pulmonary hypertension	613845	AR	Belostotsky et al.24
YARS	603623	tyrosine	C	dominant intermediate CMT	neuropathy with axonal and demyelinating features	608323	AD	Jordanova et al.25
YARS2	610957	tyrosine	M	mitochondrial myopathy	sideroblastic anemia and myopathy with exercise intolerance	613561	AR	Riley et al.26

Abbreviations are as follows: C, cytoplasm; M, mitochondria; AD, autosomal dominant; and AR, autosomal recessive.

Intriguingly, despite the fact that DARS and DARS2 have mutually exclusive subcellular localizations (Table 2), LBSL and HBSL are strikingly similar in their presentation. For example, they both manifest with abnormalities in the same central nervous system structures, including the superior and inferior cerebellar peduncles, the medial lemniscus and pyramidal tracts in the brain stem and the dorsal columns, and the lateral corticospinal tracts in the spinal cord—structures that are rarely affected in other leukoencephalopathies and the combination of which appears to be unique to HBSL and LBSL. The similarity in MRI abnormalities is such that one of the individuals (subject 7) had been initially classified as “variant LBSL” (Figure S7). Consistent with other reports,¹⁰ we find it unlikely that HBSL and LBSL are caused by amino acid mischarging alone (although this might be an important factor). Indeed, recent work has demonstrated that mutations in DARS2 decrease catalytic activity by impairing dimer formation or by reducing protein expression27, 28 but that activity of the respiratory chain is normal in individuals with LBSL.⁸ Many ARSs are known to have alternative functions beyond translation,²⁹ including AspRS, which is involved in asparagine biosynthesis.³⁰ This suggests that the disease etiology and progression of HBSL and LBSL, and possibly other ARS-associated illnesses, might be driven by processes and biological relationships that are not yet fully understood.

Published: May 2, 2013

Web Resources

The URLs for data presented herein are as follows:

• 1000 Genomes Project, http://www.1000genomes.org/

• Annovar, http://www.openbioinformatics.org/annovar/

• Burrows-Wheeler Aligner, http://bio-bwa.sourceforge.net/

• DARS animation, http://goo.gl/nwj60

• Genome Analysis Toolkit, http://www.broadinstitute.org/gatk/

• NHBLI Exome Sequencing Project (ESP) Exome Variant Server, http://evs.gs.washington.edu/EVS/

• Online Mendelian Inheritance in Man (OMIM), http://www.omim.org

• Picard, http://picard.sourceforge.net/

• SAMtools, http://samtools.sourceforge.net/

• SNPdb, http://www.ncbi.nlm.nih.gov/projects/SNP

• UCSC Genome Browser, http://genome.ucsc.edu/