2.1 Editorial policies and ethical considerations

All studies were performed in accordance with the Declaration of Helsinki protocols. The studies were reviewed and approved by the local institutional ethics board (University Medical Center Göttingen). Written informed consent was obtained from all affected subjects, parents, or legal representatives participating in this study.

2.2 Exome sequencing

DNA isolation from EDTA blood was carried out following standard protocols for all participants. In family 1, we performed exome sequencing (ES) for individual 1 and both her parents. Sequencing libraries were prepared using the Twist enrichment workflow (Twist Bioscience) and the Twist Core exome kit with spiked-in Twist RefSeq probes as well as a custom-design spike-in to enrich mitochondrial DNA. Library preparation and capture were performed according to the manufacturer's instructions and paired-end sequencing was performed on a NovaSeq. 6000 instrument (Illumina). Sequence data were aligned to GRCh37 (GCA_000001405.14) with bowtie2 (v2.4.3.1) (Langmead & Salzberg, 2012). Aligned data were further processed with GATK4 (Poplin et al., 2017) and joint calling was applied to generate vcf-files. Finally, resulting data were processed with custom scripts and uploaded for analysis with VarFish (Holtgrewe et al., 2020). In family 2, we performed ES of the index and both affected brothers. Exome capture was carried out with BGI Exome kit capture (59M) and the library was then sequenced on a BGISEQ-500, paired-end 100bp, at BGI laboratory in Shenzhen, China. Analysis of the raw data was performed using the software Varfeed (Limbus; Rostock) and the variants were annotated and prioritized using the software Varvis (Limbus; Rostock). In family 3, we performed ES for individual 5 and both her parents as previously described (Altmüller et al., 2016). ES data analysis and filtering of variants were carried out using the exome and genome analysis pipeline varbank (https://varbank.ccg.uni-koeln.de) of the Cologne Center for Genomics (University of Cologne, Cologne, Germany). We filtered for variants with a min. coverage of 10 reads, a minimum quality score of 10, and a MAF <1%. We have prioritized all potential protein-influencing variants with regard to their pathogenicity and clinical relevance according to all possible inheritance modes.

2.3 In silico prediction tools

2.4 Zebrafish model

Zebrafish (Danio rerio) were raised and maintained in an AALAC-approved facility at the Oklahoma Medical Research Foundation (OMRF) under standard conditions, and all experiments were performed as per protocol 20-07 approved by the Institutional Animal Care Committee of OMRF (IACUC). All zebrafish work was carried out in wild-type strain NHGRI-1 (LaFave et al., 2014). We designed four single guide RNAs (sgRNAs), targeting functional domains of wars1, and synthesized by in vitro transcription, from templates generated by annealing two oligos as described earlier (Varshney et al., 2016). We mixed the four sgRNAs with Cas9 protein (University of Berkley, CA, USA) to form the ribonucleoprotein complexes (RNPs) and injected into one-cell stage zebrafish eggs. Knock-down efficiency was confirmed by Quantitative reverse transcription polymerase chain reaction (RT-qPCR) (Supporting Information: Figure 1). Phenotyping was performed as described in previous methods (Lin et al., 2021). For size measurements, the mean values of head size and eye size of uninjected embryos were set as 100% and then comparative percentages were calculated for each other group (wars1 F0, F0 + WT, and F0 + Arg133Cys). Statistical analysis was performed using the one-way analysis of variance (ANOVA) with Tukey's multiple comparisons test. Variant-specific complementary DNA (cDNA) was generated from a wild-type WARS1 cDNA cloned into pCS2+ vector as a template using quick change site-directed mutagenesis kit (Agilent Inc.). mRNA was synthesized from a linearized template using SP6 mMessage kit (Thermo Fisher, Inc.), and purified by RNA clean & concentrator kit (Zymo Inc.).

2.5 RNA extraction and RT-qPCR from zebrafish

Total RNA was extracted from uninjected control, Cas9 protein injected, and wars1 F0 mutant embryos at 3 dpf using the TRIzol Reagent (Thermo Fischer Scientific) and purified by miRNeasy Mini kit (Qiagen), following the manufacturer's instructions. The mRNA was reverse-transcribed into cDNA using iScript cDNA-synthesis kit (Bio-Rad) according to the manufacturer's instructions. The cDNA was used as a template for RT-qPCR reactions using SYBR Green Supermix according to the manufacturer's instructions (Thermo Fisher Scientific) and the Light Cycler® 96 System (Roche). All relative quantifications were done using three independent biological replicates and three technical replicates, the values were normalized to the housekeeping gene 18S. The primer sequences for RT-qPCR are wars1: Forward: 5′-GCGAGTCCTGCCTTCTACAG-3′, Reverse: 5′-TAGCACACGGAATGAGGCAC-3′, 18S: Forward: 5′-TCGCTAGTTGGCATCGTTTATG-3′, and Reverse: 5′-CGGAGGTTCGAAGACGATCA-3′. The quantification was performed using the $${2}^{ \mbox{-} \unicode{x02206}\unicode{x02206}{C}_{{\rm{t}}}}$$ method and corresponding age-matched control samples as calibrators.

2.6 Structural mapping of disease-related variants in WARS1 and SARS1

Structure figures were prepared with PyMol 2.5 (Schrödinger LLC).

2.7 Cell culture

Fibroblasts were grown at 37°C in a 5% CO₂ humidified atmosphere in Dulbecco's modified Eagle's medium containing 4.5 g/L glucose, l-glutamine, sodium pyruvate, and 3.7 g/L NaHCO₃ (DMEM; PAN-Biotech GmbH), additionally supplemented with 10% fetal bovine serum (Merck KGaA; Darmstadt), penicillin-streptomycin (10,000 U/ml penicillin, 10 mg/ml streptomycin; PAN-Biotech GmbH), and 0.7 µg/ml amphotericin B (Merck KGaA).

2.8 Western blot

Fibroblasts were grown in 100 × 20 mm cell culture dishes (Greiner Bio-One GmbH) to a confluence of approximately 90%, before they were harvested by trypsinization and pelleted. Pellets were then incubated with 100 μl of protein lysis buffer (20 mM Tris/HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% w/v NP-40), and 1× Halt™ Protease Inhibitor Cocktail (Thermo Scientific) for 10 min on ice. After centrifugation at 4°C and 19,000g for 10 min, supernatants were collected, protein concentrations were determined using Pierce™ BCA Protein Assay Kit (Thermo Scientific) and 40 μg (33 µg for replicate 2) of protein was loaded onto a 4%–20% gradient precast Mini-PROTEAN TGX Stain-Free Gel (Bio-Rad Laboratories, Inc.). Following electrophoresis, the gel was blotted onto an Immun-Blot® PVDF membrane (Bio-Rad Laboratories, Inc.) using a Trans-Blot® Turbo™ Transfer System (Bio-Rad Laboratories, Inc.). Immunoblots were then blocked for 1 h at room temperature in 5% milk powder (blotting grade; Carl Roth Gmbh & Co. Kg) in tris-buffered saline with 0.1% TWEEN 20 (TBS-T) and subsequently incubated overnight at 4°C with horseradish peroxidase (HRP)-conjugated primary anti-WARS1 antibody (0.5 µg/ml, sc-374401 HRP; Santa Cruz Biotechnology, Inc.) in 2.5% milk/TBS-T. WesternBright Chemilumineszenz Substrat (Biozym Scientific GmbH) was used as HRP substrate for enhanced chemiluminescence detection of bands. Chemiluminescence was recorded using a ChemiDoc™ Touch Imaging System with Image Lab™ Touch Software (version 1.2.0.12; Bio-Rad Laboratories, Inc.). Membranes were then reprobed with rabbit anti-glyceraldehyde 3-phosphate dehydrogenase (α-GAPDH; 1:5000) monoclonal antibody (#2118; Cell Signaling Technology) followed by secondary mouse anti-rabbit IgG-HRP antibody (1:10000, sc-2357; Santa Cruz Biotechnology, Inc.) as loading control.

2.9 Quantification and statistical analysis of Western blot data

Densitometric quantification of western blot bands on raw 16-bit images was performed using Gel Analyzer within Fiji/ImageJ 2.1.0/1.53c (Schindelin et al., 2012). Results are presented as boxplots (center line corresponds to median, upper and lower hinge correspond to 25th and 75th percentiles, upper and lower whisker extend to largest/smallest value within 1.5× inter-quartile range of the respective hinge, outliers beyond the whiskers are presented individually). For statistical analyses, ANOVA was performed (significance level α = 0.05), followed by post hoc Tukey analysis. Results are presented as compact letter display of all pair-wise comparisons in increasing order.

2.10 Aminoacylation assay

Aminoacylation was assessed by measuring WARS1 activity in cultured fibroblasts. Fibroblast lysates (cytosolic fraction) were incubated in triplicate at 37°C for 10 min in a reaction buffer containing 50 mmol/L Tris buffer pH 7.5, 12 mmol/L MgCl₂, 25 mmol/L KCl, 1 mg/ml bovine serum albumin, 0.5 mmol/L spermine, 1 mmol/L ATP, 0.2 mmol/L yeast total tRNA, 1 mmol/L dithiothreitol, 0.3 mmol/L [¹³C₄, ¹⁵N]-tryptophan, [D₅]-phenylalanine, and [D₂]-tyrosine. The reaction was terminated using trichloroacetic acid. After sample washing with trichloroacetic acid, ammonia was added to release the labeled amino acids from the tRNAs. [¹³C]-phenylalanine and [¹³C₉]-tyrosine were added as internal standards and the labeled amino acids were quantified by liquid chromatography with tandem mass spectrometry (LC-MS/MS). Phenylalanyl-tRNA synthetase and tyrosyl-tRNA synthetase activities were simultaneously detected as control enzymes.

2.11 Cycloheximide chase assay

Patient and control fibroblasts were seeded into six-well plates (CytoOne®; Starlab) at 2.2 × 10⁵ cells/well and grown to a confluence of approximately 90%, before 10 mg/ml cycloheximide (≥94% TLC-grade; Merck) in H₂O at a final concentration of 250 μg/ml was added to the medium. Cells were treated for 8, 4, 2, or 0 h and all fibroblasts were harvested at the same time by trypsinization. Proteins were lysed using 25 µl of protein lysis buffer and protein levels were analyzed by immunoblot densitometry as described before, again with GAPDH as loading control. Protein levels were normalized to their respective 0 h treatment.

3.1 Clinical reports

Individual 1 is the index patient of family 1 (Figure 1a). She is the second child of a healthy consanguineous couple from Syria. The parents are first cousins and they also have a healthy son (Figure 2a). The girl was presented to the genetics clinic at the age of 5 years for microcephaly, autism, and intellectual disability. Pregnancy and birth were reported as uncomplicated, but there are no detailed records. Her birth measurements were reported as follows: birthweight 2200 g (1st percentile, −2.5 SD) and head circumference 33 cm (11th percentile, −1.2 SD). Developmental delay was noticed at 9 months when she failed to reach motor milestones. She learned to walk at 26 months. The parents reported that she had learned a few words of Arabian, but stopped speaking altogether at the age of 3 years, either as a symptom of early childhood autism or of developmental regression. At age 5 years she is not toilet trained. She received the diagnoses of autism, speech-language disorder and, tentatively, mild intellectual disability. Physical examination revealed a height of 100 cm (1st percentile, −2.4 SD), a weight of 15 kg (3rd percentile, −1.8 SD), and a head circumference of 46 cm (<1st percentile, −4.2 SD) (Table 1). Besides the severe microcephaly, she showed no striking dysmorphic features. Her gait was ataxic, but otherwise she showed no obvious neurological abnormalities, especially no signs of neuropathy. Apart from secondary loss of active speech, she showed no signs of a progressive neurodegenerative disorder at the age of 5 years. Cranial MRI showed a normal configuration of the cerebral cortex, the inner and outer cerebrospinal fluid spaces, the basal ganglia, and the pituitary gland. The corpus callosum was described as thin and periventricular white matter hyperintensities were reported (images not shown).

Individual 2 is the index patient of family 2 (Figure 1b). He is one of eight children of a consanguineous couple from Iraq. He has two similarly affected younger brothers (individuals 3 and 4), and a healthy brother and sister (Figure 2b). Three siblings were already deceased. One sister passed away at age 13 years, she reportedly had severe intellectual disability and was nonverbal. She likely suffered from the same condition as her affected brothers. Another sister and another brother passed away in their second year of life and were reported to have had poor growth. It is unclear, whether the latter two children had the same condition as their affected siblings. Individual 2 was 13 years and 6 months old at last presentation. He has global developmental delay and intellectual disability. He learned to sit at 2.5 years, learned to crawl in the second year of life, and learned to walk with assistance at 3 years, but he never learned to walk independently. He is not toilet-trained and he is nonverbal. He cannot eat independently but is orally fed. Physical examination revealed a height of 144 cm (1st percentile, −2.2 SD), a weight of 24.8 kg (<1st percentile, −3.9 SD), and a head circumference of 49 cm (<1st percentile, −4.3 SD) (Table 1). Cardio-pulmonary status was normal. He showed muscular hypotonia, his gait was unsteady, and walking only possible with assistance. He did not react to his own name or show any interest in his siblings. He showed the following dysmorphic features: a microcephalic aspect with a sloping forehead, a large nose with a high nasal bridge, synophrys, long palpebral fissures, and large ear lobes (Figure 1b). Cranial MRI revealed reduced cerebral volume with enlarged internal and external cerebrospinal fluid spaces, especially supratentorial, a thin corpus callosum, and supratentorial leukoencephalopathy (Figure 1d). Individual 2 has a history of childhood malignant melanoma of the occiput with multiple pulmonary, lymphatic, and osseous metastases. Individual 3 is the younger brother of individual 2. He was 8 years and 2 months old at last presentation. He has a history of developmental delay. He learned to sit at approximately 1 year of age, crawled at about 2 years of age, and walked at 3 years. He shows a steady but broad-based gait. He was toilet trained at 4 years but needs assistance in all aspects of his daily life. He interacts with other children but is unable to communicate verbally. He has a height of 121 cm (4th percentile, −1.8 SD), a weight of 18.9 kg (<1st Percentile, −2.9 SD), and a head circumference of 48 cm (<1st percentile, −3.8 SD). He showed dysmorphic features, similar to his elder brother (Table 1). Individual 4 is the youngest affected child from family 2. He was 7 years and 2 months old at last presentation. He has a history of global developmental delay similar to his elder brothers. He is unable to stand or walk unassisted. He does not speak, except for one word, and communicates with noises. He is unable to feed himself independently. Neurological examination revealed a movement disorder that was classified as mainly dystonic. He makes eye contact, but does not follow any instructions. Cardio-pulmonary status, and kidney and liver function were normal. Audiological and ophthalmological examinations were normal. EEG showed no signs of epilepsy. Cranial MRI revealed slightly enlarged subarachnoid spaces compared to healthy peers indicative of reduced cerebral volume, white matter T2-hyperintensities, and a thin corpus callosum (Figure 1e). He has a height of 113 cm (1st percentile, −2.3 SD), a weight of 18 kg (1st percentile, −2.4 SD), and a head circumference of 48 cm (<1st percentile, −3.5 SD) (Table 1). In none of the affected children from family 2 there was evidence of degenerative neurological disease.

Individual 5 is the index patient of family 3 (Figure 1c). She is the second child of healthy, unrelated, German parents (Figure 2c). She was born via spontaneous vaginal delivery at 38+4 weeks of gestation with a birth weight of 3360 g (54th percentile, +0.1 SD), a length of 52 cm (65th percentile, +0.4 SD), and a head circumference of 33 cm (11th percentile, −1.2 SD). Postnatal adaptation was good, APGAR was 9/10/10 at 1/5/10 min and the arterial cord blood pH was normal with 7.28. Early routine examinations were unremarkable, however at 2 months of age she was hospitalized for bronchopneumonia and at this time point microcephaly was noted, as well as pulmonary valve stenosis and esotropia. Cerebral ultrasound revealed pachygyria and an enlargement of the inner and outer cerebrospinal fluid spaces. TORCH serology was negative. She received physiotherapy for hypotonia. At the age of 7 months she reportedly had a first febrile seizure and she continued to have focal seizures with secondary generalization regularly and was hospitalized several times for status epilepticus. Her development is severely delayed, she neither walks nor speaks. She has a history of eating her own hair. Post-ictal EEGs revealed sharp-slow-wave complexes temporally on both sides, and a right temporal focal point was identified after status epilepticus. At 11 years of age, cMRI showed reduced cerebral volume compatible with the severe microcephaly, likely secondary dysgenesis of the corpus callosum, reduced cerebellar volume, mesial temporal sclerosis on the left side and parahippocampal white matter intensities. Physical examination at 11 years revealed increased tendon reflexes and spasticity of all four limbs, most prominently of the legs. The patient was friendly, but quite restless and tried to put objects in her mouth and to bite her caregivers, out of curiosity rather than aggressiveness. Her height was 129 cm (1st percentile, −2.5 SD), weight 17 kg (<1st percentile, −5.0 SD), and her head circumference was 46.5 cm (<1st percentile, −6.2 SD) (Table 1). She exhibited the following dysmorphic features: a microcephalic aspect with a sloping forehead, deep-set eyes, a long nose with a high nasal bridge, long palpebral fissures, and low-set ears (Figure 1c). Before ES, she received a diagnostic next-generation sequencing panel for epilepsy-related genes, as well as chromosome analysis and array-CGH, all with normal results.

3.2 Molecular findings

We performed trio ES on DNA of individual 1 and her parents. The data set was initially analyzed for variants in known genes for microcephaly. As this analysis did not reveal a pathogenic variant, we then analyzed the data for homozygous variants, due to the consanguineous background, and identified the homozygous missense variant c.397C>T, p.(Arg133Cys) (chr14:100826916G>A; GRCh37/hg19) in the WARS1 gene (NM_004184.4; Figure 2a). We also checked for relevant de novo variants but found none. The variant c.397C>T is present in one heterozygous allele in the gnomAD browser. It affects a highly conserved residue within the N-terminal domain and is predicted to be damaging by the prediction programs SIFT (Score: 0.000), PROVEAN (Score: −7.12), MutationTaster (Score: 180), and PolyPhen2 (HumDiv Score: 1.000; HumVar Score: 0.999). Both parents were shown to be heterozygous carriers of the variant c.397C>T (Figure 2a).

In family 2, we performed ES for three affected brothers and identified the identical homozygous variant c.397C>T, p.(Arg133Cys) in WARS1 that had also been identified in individual 1, in all three of them (Figure 2b). Targeted Sanger sequencing in the two healthy siblings showed that they did not carry the variant in the homozygous state (data not shown). There was no parental DNA available for segregation analysis.

The identification of the identical missense variant in the WARS1 gene in two families prompted us to ask whether there might be a shared haplotype. We, therefore, compared rare and common SNVs at the WARS1 locus and in a 1-Mb region flanking WARS1 (Figure 3). We found that the homozygous haplotypes of individual 1 and individuals 2–4 shared no rare variants with a MAF <1% except WARS1 c.397C>T in the analyzed region. Identical homozygous variants in individuals 1 and 2–4 all had a MAF >50%. This analysis likely excludes a common ancestral allele in families 1 and 2. The fact that the same mutation arose independently in two patients with very similar phenotypes lends weight to the supposed pathogenicity of the variant.

Family 3 was analyzed via patient-parent trio ES. We identified the compound-heterozygous variants c.904C>T, p.(Arg302Cys) (chr1:109777988C>T; GRCh37/hg19), and c.1168C>T, p.(Arg390Cys) (chr1:109779081C>T; GRCh37/hg19) in the SARS1 gene (NM_006513.4). The variant c.904C>T was inherited from the healthy mother and the variant c.1168C>T was inherited from the healthy father (Figure 2c). The variant c.904C>T, p.(Arg302Cys) is absent from the gnomAD browser, while c.1168C>T, p.(Arg390Cys) is present in two heterozygous alleles. Both variants are predicted to be damaging by the prediction programs SIFT (Scores: 0.007 and 0.000, respectively), PROVEAN (Scores: −7.53 and −7.91, resp.), MutationTaster (Scores: 180, resp.), and PolyPhen2 (HumDiv Scores: 1.000, resp.; HumVar Scores: 1.000, resp.) and both variants affect highly conserved residues within the catalytic domain of SARS1 (Figure 2d).

3.3 Zebrafish studies

Zebrafish have a single ortholog of human WARS1. First, we investigated the effect of loss of wars1 function on animal development by generating a wars1 knockout using CRISPR/Cas9-mediated targeted mutagenesis and analyzed phenotypes in the F0 (founding generation) generation. Injected embryos were raised, and the phenotypes scored within the first 5 days of post-fertilization (dpf) (F0 generation). Wars1 F0 mutant embryos did not show any obvious phenotypes until 2 dpf. At 3 dpf, mutants show a smaller head and eyes, as well as heart edema compared to control animals (Figure 4). The smaller head phenotype is similar to the microcephaly observed in individuals affected by the homozygous WARS1 variant. To confirm whether the phenotypes in the zebrafish model arise from the loss of wars1 function, we used an in vitro synthetic mRNA rescue method by using wild-type (WT) human WARS1 mRNA, co-injected with wars1 sgRNA/Cas9 RNP. mRNA rescue restored phenotypes shown by F0 mutants, confirming the specificity of the phenotypes (Figure 4). To test the pathogenicity of the p.Arg133Cys variant in WARS1, we compared the efficiency of variant WARS1 mRNA to rescue the F0 mutant phenotypes to WT for both head and eye size. mRNA carrying the variant failed to rescue the eye and head size phenotypes (Figure 4), suggesting the variant significantly impacts protein function.

3.4 Structural mapping

To gain further insight into the potential functional consequences of the variants identified, we mapped them onto previously determined structures of human WARS1 (Shen et al., 2006) and SARS1 (Xu et al., 2013) (Figure 5a).

In the case of WARS1, we used the structure of the enzyme in complex with its target (Shen et al., 2006). WARS1 functions as a homodimer, and the anticodon binding domain of one monomer interacts with the anticodon loop of the tRNA to position the substrate in the active site of the other monomer (Shen et al., 2006). Structural mapping shows that Arg133 is not in vicinity of the active site and does not appear to be directly involved in interactions with the tRNA substrate (Figure 5a). Instead, it is part of a basic patch in the N-terminal domain (comprised of Arg127, Arg133, and Arg134), which forms a contact interface with an acidic patch (comprised of Glu408, Asp410, Asp411) in the anticodon-binding domain. Arg133 may form a salt bridge with Asp410, and its substitution with cysteine would abolish this salt bridge.

For structural mapping of the two SARS1 variants, we used the structure of the enzyme in complex with the substrate analog Ser-SA (Xu et al., 2013). This reveals that both Arg302 and Arg390 are located within the active site of the enzyme (Figure 5a). Arg302 is highly conserved among seryl-tRNA synthetases and forms direct contacts with the substrate Ser-SA (Xu et al., 2013). Thus, substitution of this residue with cysteine is likely to affect substrate binding by the enzyme. Arg390 is located at the edge of the active site and may thus be involved in tRNA binding. This residue is conserved as a basic residue (arginine or lysine) from human to Escherichia coli (C. Wang et al., 2015), suggesting that its substitution to cysteine would likely affect proper function of the enzyme.

3.5 Functional analyses in WARS1 mutant patient fibroblasts

Western blot analysis showed that the abundance of WARS1 was significantly reduced in WARS1 mutant patient fibroblasts from individual 2 compared with three controls. Protein abundance was determined as pixel intensity of WARS1 immunoblot band, relative to β-actin loading control, on six replicates. ANOVA was performed (significance level α = 0.05, [F(3,20) = 13.49, p = 4.85e−05]), followed by post hoc Tukey analysis (Figure 5b). A cycloheximide chase assay did not show significantly faster protein degradation in patient fibroblasts than in controls (Supporting Information: Figure 2). Aminoacylation activity was normal in patient fibroblasts when compared with controls and endogenous YARS1 and FARS1 activity (Figure 5c).