Bi‐allelic KARS1 pathogenic variants affecting functions of cytosolic and mitochondrial isoforms are associated with a progressive and multisystem disease

Abstract KARS1 encodes a lysyl‐transfer RNA synthetase (LysRS) that links lysine to its cognate transfer RNA. Two different KARS1 isoforms exert functional effects in cytosol and mitochondria. Bi‐allelic pathogenic variants in KARS1 have been associated to sensorineural hearing and visual loss, neuropathy, seizures, and leukodystrophy. We report the clinical, biochemical, and neuroradiological features of nine individuals with KARS1‐related disorder carrying 12 different variants with nine of them being novel. The consequences of these variants on the cytosol and/or mitochondrial LysRS were functionally validated in yeast mutants. Most cases presented with severe neurological features including congenital and progressive microcephaly, seizures, developmental delay/intellectual disability, and cerebral atrophy. Oculo‐motor dysfunction and immuno‐hematological problems were present in six and three cases, respectively. A yeast growth defect of variable severity was detected for most variants on both cytosolic and mitochondrial isoforms. The detrimental effects of two variants on yeast growth were partially rescued by lysine supplementation. Congenital progressive microcephaly, oculo‐motor dysfunction, and immuno‐hematological problems are emerging phenotypes in KARS1‐related disorder. The data in yeast emphasize the role of both mitochondrial and cytosolic isoforms in the pathogenesis of KARS1‐related disorder and supports the therapeutic potential of lysine supplementation at least in a subset of patients.

diagnostic or research investigations was obtained for all cases.
Clinical, biochemical, neuroimaging, and genetic studies were available for all cases. For each case, the mitochondrial diagnostic score was attributed according to Morava et al. (2006). Genomic DNA from each proband and both parents underwent exome sequencing (ES) using site-specific protocols and KARS1 variants were confirmed by Sanger sequencing. ACMG classification was used for standardizing variant interpretation (Richards et al., 2015). For consistency, variant nomenclature has been provided for the longer mitochondrial isoform (NP_001123561.1) for all variants except for the variant p.(Ala2Val), that was specific for the cytosolic isoform (NP_005539.1).
Predicted damaging effects of KARS1 variants were evaluated by SIFT, PolyPhen2, Mutation Taster, PROVEAN, M-CAP, CADD, and Phast-Cons. Multispecies sequence alignment of LysRS proteins was performed using CLUSTAL Omega. Amino acid residues affected by variants detected in cases were mapped on the crystal structure of the KARS-p38 complex (PDB: 4dpg) loaded into PyMOL or RasMol.
YPD medium (0.5% Yeast extract, 1% Peptone, and 2% glucose) was supplemented with either 200 µg/ml geneticine disulfate or 250 µg/ ml hygromycin B (Formedium). MSK1 and KRS1 encoding the mitochondrial and cytosolic isoforms of yeast LysRS respectively were polymerase chain reaction (PCR)-amplified (Table S1) and cloned under their endogenous promoter into the centromeric pFL38 plasmid that includes URA3 as selection marker. The pFL38MSK1 and pFL38KRS1 plasmids were separately transformed into the W303-1B strain, and one-step gene disruption (Wach et al., 1994) of MSK1 or KRS1 genes was performed in these strains because loss of MSK1 results in instability of mitochondrial DNA (mtDNA) (Gatti & Tzagoloff, 1991), whereas loss of KRS1 is lethal. For MSK1 disruption, the KanMX4 cassette was amplified from the BY4741 msk1Δ strain.
Transformation and selection on YPD medium supplemented with appropriate antibiotics were performed according to previous study (Gietz, 2014) to obtain W303-1B msk1Δ/pFL38MSK1 and W303-1B krs1Δ/pFL38KRS1. Human complementary DNAs (cDNAs) encoding the mtKARS1 (NM_001130089, purchased from OriGene) and cy-tKARS1 (NM_005548 purchased from OriGene) were PCR-amplified and cloned into the centromeric single copy expression vector pFL39-TEToff (Nolli et al., 2015), which contains TRP1 as selection marker and the TET-off cassette, which is made by the CYC1 promoter and seven repeats of the Tet operator (TetO). The cytKARS1 cDNA was also subcloned into multicopy expression vector YEplac112-TEToff obtained by subcloning the TET-off promoter in YEplac112 (Gietz & Sugino, 1988). Except for the p.(Ala2Val) variant, all KARS1 missense variants were inserted in both mitochondrial and cytosolic isoforms by PCR QuikChange™ (Agilent) using KOD Hot Start DNA Polymerase (Merck) and appropriate primers (Table S1).
The p.(Ala2Val) variant was only inserted into the cytoplasmic isoform. Plasmids expressing wild-type and mutant alleles were transformed into the corresponding deleted strain. In msk1Δ strains expressing the mutant alleles, the pFL38MSK1 was lost through plasmid-shuffling on 5-FOA medium, as previously reported (Baruffini et al., 2010). Strains devoid of MSK1 were maintained in YP supplemented with 2% ethanol or 2% glycerol if respiratory proficient or in SC-W medium supplemented with 2% glucose if respiratory deficient. For oxidative growth analysis, strains were serially diluted and spotted on SC-W agar plates, with or without lysine 50 µg/ml, supplemented with 2% glycerol or 2% glucose as control. Plates were incubated at both 28°C and 37°C. Oxygen consumption rate (OCR) was measured at 30°C from 20 mg of wetweight yeast cell suspensions cultured under shaking in SC medium supplemented with 0.5% glucose until exhaustion of the latter (for 18 h at 28°C or for 16 h at 37°C) measured by a Clark-type oxygen electrode (Oxygraph System Hansatech Instruments) with 1 ml of air-saturated respiration buffer (0.1 M phthalate-KOH, pH 5.0) and 0.5% glucose. Values for OCR were normalized to the dry weight of the cells. In krs1Δ strains expressing mutant cytKARS1, yeast viability was evaluated by analyzing growth on 5-FOA medium supplemented with 2% glucose. For each cytKARS1 wild-type and mutant strain, a small patch of cells was at first inoculated in 200 µl of SC-W-U medium into a 96-well microtiter plate and grown for 24 h at 28°C without shaking thereafter. Then, 2 µl of this culture was inoculated in 200 µl of SC-W containing uracil to allow loss of the pFL38KRS1 plasmid and grown for 30 h at 28°C without shaking. For all strains, cell concentration was between 7.5 and 8.5 OD 600 . The 5 µl of undiluted culture and 1:10 dilutions were then spotted on the appropriate 5-FOA medium with or without lysine 50 µg/ml. Growth was assessed after incubation at 28°C or 37°C.
Western blots were performed on proteins extracted from deleted strains harboring MSK1 or KRS1 on pFL38 and mtKARS1 and cytKARS1 variants on pFL39-TEToff and YEplac112-TEToff, respectively. In the former case, MSK1 gene was retained to maintain normal mtDNA levels and respiratory proficiency in all mtKARS1 mutant strains; in the latter case, KRS1 gene was retained to maintain the viability of all strains. For msk1Δ strains, cells were grown under the same conditions used for the OCR assay. For krs1Δ strains, cells were grown under shaking in SC medium until OD 600 = 1.5-2.
Cells equivalent to 10 OD 600 were harvested, and proteins extracted by trichloroacetic acid precipitation, according to a previous protocol (Del Dotto et al., 2018). Proteins corresponding to 1.5 OD 600 of the initial cells were loaded on 15% sodium dodecyl sulfatepolyacrylamide gel electrophoresis, and electroblotted onto CAPPUCCIO ET AL.

| Statistical analyses
Statistical analyses were performed by using GraphPhPrism8, and Excel. p values below .05 were considered statistically significant.
ES was performed for all index cases. Their clinical and genetic findings are summarized in Table 1 and Figure 1. Mitochondrial scores were calculated for each case (Table S2) and none had a score less than 2 that is indicative of an unlikely diagnosis of mitochondrial disorder, whereas 22.2% had scores consistent with a possible diagnosis of mitochondrial disorder, 55.5% with a probable diagnosis, F I G U R E 2 (a) Distribution of mitochondrial disease diagnostic scores among the cases herein reported. Morphology studies were not performed in any of the cases. (b) Standard deviation score (SDS) of the occipitofrontal circumference (OFC) at birth (n = 8) and at last evaluation (n = 9). OFC at birth was not available for proband 9. OFC SDS at the latest available evaluation is statistically significantly different from birth OFC (p < .004) CAPPUCCIO ET AL. | 751 centile, −1.2 SDS). Vertical nystagmus was noted in the first few months of life and at 2.5 years, roving eye movements without tracking, unconjugated gaze, and nystagmus were also observed. He had swallowing difficulties and was also noted to have increased muscle tone, especially at the ankles. At 11 months, he developed infantile spams that were treated with ACTH, and later with vigabatrin and zonisamide. At the age of 3 years, he could roll over, sit independently for short periods, and could say two words. At his latest evaluation at 30 months of age, his OFC was 44.5 cm (<5th centile, −3.2 SDS) and at the age of 3.4 years his weight was 13.7 kg (20th centile) and his height 95.9 cm (19th centile).

| Family 4, Cases 5 and 6
Proband 5 was the second child of healthy consanguineous (firstcousins) parents from the Middle East with unremarkable family history ( Figure 1). Prenatal and perinatal history were uncomplicated. His birth weight was 2385 g (<5th centile, −2.6 SDS) and OFC was 30 cm (<5th centile, −4 SDS). Developmental milestones were delayed and at 2 years he offered eye contact, but without tracking and he had no intentional movements. His muscle tone was increased with brisk reflexes. He was found to have hearing loss. He had feeding difficulties and failure to thrive. At 25 months, his weight was 9.1 kg (<5th centile, −2.7 SDS), height 85.5 cm (19th centile), and OFC 39 cm (<5th centile, −6 SDS) (Figure 2b). At the age of 9 months, he developed infantile spams that were responsive to vigabatrin. Echocardiogram was normal. A brain MRI at 3 months revealed no abnormalities and showed mild cerebral atrophy with thin corpus callosum and lack of myelination at 2 years. On cerebral MRS lactate was slightly elevated. Proband 6 is the younger brother of proband 5 and his prenatal and perinatal history were also uncomplicated. His birth weight was 2930 g (8th centile) and his OFC was 30.5 cm (<5th centile, −3.7 SDS). He was more alert than his F I G U R E 3 Brain magnetic resonance imaging (MRI) images of case 1 (a)-(g): Axial T2-weighted images (a), (b), (f), axial FLAIR image (b), sagittal T1-weighted images (d) and (e), and coronal T2-weighted image (g) at 5 months (a) and (d) and 4 years (b), (c), (e)-(g). At 5 months the temporal horns and sulci were dilated but the signal of the unmyelinated white matter appeared normal (black arrows in (a)). Note the progressive thinning of the corpus callosum and of a diffuse cerebral tissue loss causing microcephaly between 5 months and 4 years (d) and (e). In the deep temporal, frontal and peritrigonal white matter some subtle hyperintensities appeared between 5 months and 4 years, with a more preserved signal in the posterior white matter (black and white arrows respectively in (b), (c), (f)). Brain MRI images and magnetic resonance spectroscopy (MRS) of case 2 at 12 months (i) and (k) and 17 months (h), (j), (l), and (m), (n): Axial T2-weighted images (i), (j), axial SWI phase image (h), coronal and sagittal T2-weighted images (k), (l), and (m). The first MRI scan highlighted bilateral and confluent T2 hyperintensities in basal ganglia, thalamic nuclei, and in capsular, deep, and peripheral white matter causing tissue swelling and sulcal effacement (i) and (k). Note also the prominent involvement of external capsules and of the white matter of temporal lobes (white arrows in (i) and (k)). Incomplete operculization of sylvian fissures was already present in the first MRI and such dilation worsened later (black arrows in (i) and (j)). The severe enlargement of frontal and temporal sulci and the thinning of corpus callosum due to the tissue loss are evident (black arrows in (l) and (m). There is also an arachnoid cyst located inferiorly to the vermis (black arrow in (m)). The reduction of N-acetyl-aspartate peak and the presence of a small peak of lactate are shown on MRS (TE 144) (p). Point-like hypointensities related to calcifications were bilaterally located in the frontal white matter (black arrows in (h)). MRI phase and magnitude SWI images of case 8 (o)-(p)) showed calcifications in the cortex of cerebellar hemispheres (black arrows). Brain MRI images of case 9 (q)-(u)): axial T2-weighted and FLAIR images (q) and (r), coronal and sagittal T1-weighted images (s), (t), and (u). Note the bilateral periventricular hyperintensities on T2 and FLAIR images, posteriorly containing little cavities (white arrows in (q), (r), and (s)). Mild enlargement of left frontal subarachnoid spaces and in the superior part of the vermis is also present (asterisks in (q), (r), and (u)) brother, offered eye contact and had eye tracking but as his older brother, he had random eye movements and nystagmus. He achieved no other developmental milestones. He had increased muscle tone with brisk reflexes. At the age of 9 months, he developed infantile spasms that did not respond to vigabatrin, phenobarbital, levetiracetam, and topiramate. He had feeding difficulties and failure to thrive. At the age of 12 months, his height was 76 cm (54th centile), weight 8.5 kg (12nd centile), and OFC 39 cm (<5th centile, −5.5 SDS).
A bran MRI at 7 months showed cerebral atrophy with thin corpus callosum and lack of myelination. Cerebral MRS was normal. On physical exam, he was not noted to have murmurs and he has not been evaluated by echocardiogram. weeks of life, she was noted to have hypotonia, feeding difficulties, and failure to thrive. Intermittent strabismus, visual upward fixation, random eye movements, and lack of eye contact were noted (Supplementary Video). She achieved head control at 5 months of age. At 10 months, her length was 67.0 cm (<5th centile, −2.2 SDS), weight 6.3 kg (<5th centile, −3.0 SDS) and OFC 39 cm (<5th centile, −5.2 SDS) (Figure 2b). Echocardiogram revealed an atrium septal defect. Brain SWI at 9 months identified two hypointense areas in the cortex of the cerebellar hemispheres whose signal inverted on magnitude map, and therefore were related to calcifications (Figure 3 o,p). Increased lactate was detected on cerebral MRS and blood. Initial auditory brainstem response was normal but re-testing at 10 months showed bilateral sensorineural hearing loss with thresholds of 25-45 dB.

| Family 7, Case 9
The proband is the only child of healthy nonconsanguineous Caucasian parents with unremarkable family history (Figure 1). He was born after 40 weeks of uncomplicated gestation. At birth, his weight was 3855 g (85th centile, +2 SDS), her length 60 cm (>97th centile, >+3 SDS). Birth OFC measurement was unavailable. At 6 months of life, he was noted to have hearing loss, hypotonia, and developmental delay. At age 6 years, he gradually lost the acquired gross motor skills, and became unable to crawl, walk, and speak.

| KARS1 variants
Twelve KARS1 variants were identified in the nine cases herein re- with LysRS showed that mutated most amino acids are either highly conserved (Arg108, Phe291, His401, Pro499, and Pro533) or semiconserved (Arg205, Ile346, Arg438, Phe585, and Asn591) in mammals, fungi, and plants. The Ala57 residue is conserved in most animals, whereas the Ala2 is conserved in most tetrapods but not in fishes and invertebrates, in which the program failed to align the N-terminal region of cytosolic LysRS (Figure 4a).
Pathogenicity scores predicted KARS1 variants to be predominantly damaging (Table S3) and all of them except for p.(Ala2-Val) were classified as pathogenic or likely pathogenic according to ACMG criteria (Richards et al., 2015) (Table S3). LysRS protein consists of the anticodon binding domain and catalytic domain, link by a hinge (Figures 1 and 4b).  Previous studies have shown that human cDNA encoding KARS1

| Functional studies in yeast
isoform 1 (mtKARS1, accession number NM_001130089.1) complements MSK1 deletion in yeast (Sepuri et al., 2012). Consistent with previous findings, a transformation of wild-type human mtKARS1 cloned in a single copy plasmid under the TET-off promoter restored growth defect of the msk1Δ strain. The mtKARS1 mutant alleles were introduced into msk1Δ, carrying a wild-type MSK1 on a URA3-bearing vector, which is lost upon exposure to 5-FOA. Oxidative growth was then determined through spot assay on glycerol medium and most decrease compared with wild-type ( Figure S1A). These results indicated that the partial or total mitochondrial function impairment was mainly caused by deficiency of enzyme activity rather than reduced protein levels.
To investigate the functional consequences of KARS1 variants on the activity of cytLysRS, cDNA of KARS1 isoform 2 (cytKARS1; NM_005548.2) was first cloned into the same centromeric expression vector and inserted in a krs1Δ strain expressing the wild-type KRS1 on a URA3-bearing vector. This construct was unable to rescue growth and OCR were investigated, but they were found to be both similar to wild-type (data not shown).
For most mutants, the growth defects were not improved by supplementation of lysine in medium (Figure 5d). However, the p.(Phe585Cys) and to a lesser extent the p.(Pro533Arg) showed partial growth improvement upon lysine supplementation, suggesting that cytoplasmic lysine levels are a limiting factor for the activity of these two cytKARS1 mutant alleles.
Steady-state levels of mutant LysRS normalized for Pgk1 were all similar to wild-type cytLysRS (between 80% and 120%) except for the p.(Ala2Val) that showed a 2-fold increase ( Figure S1b). As for their mitochondrial counterparts, these results indicated that the growth impairment of cytLysRS mutants was mainly caused by deficiency of enzyme activity rather than protein levels. We speculate that the improved growth of p.(Ala2Val) mutant can be due to the higher steady-state protein levels.   (Schrier & Falk, 2011) and have been reported in individuals with KARS1 defects (Ardissone et al., 2018;Fuchs et al., 2019;Itoh et al., 2019;Joshi et al., 2016;Kohda et al., 2016;Lieber et al., 2013;McMillan et al., 2015). They appear to be relatively frequent also in our series (6/8) and might be a diagnostic clue.
LysRS affects the expression of several genes involved in the Patients carrying pathogenic KARS1 variants have been treated with mitochondrial cocktail or idebenone (Avula et al., 2014;Verrigni et al., 2017) that is expected to have some effects only on manifestations dependent on mitochondrial dysfunction rather than on the defect of the cytosolic isoform. For therapy of patients with various ARS defects, supplementation of the corresponding amino acid or high protein intake has been proposed (Casey et al., 2015;Fuchs et al., 2019) based on in vitro studies showing a variable degree of biochemical improvements (Friederich et al., 2018;Hadchouel et al., 2015). Similarly, supplementation with lysine might provide benefit in patients carrying KARS1 pathogenic variants.
In the present study, the detrimental effects of two cytKARS1 variants located in the catalytic domain, namely p.(Phe585Cys) and p.(Pro533Arg) expressed in yeast, were partially improved by lysine supplementation. Interestingly, the Pro533 is near the Glu529 corresponding to the Glu428 in E. coli LysRS that is directly involved in the binding to lysine (Onesti et al., 1995). Moreover, the Phe585 is in close proximity to the Arg581 corresponding to the Arg480 in E. coli LysRS that binds through an ionic interaction the gamma-phosphate of ATP during the synthesis of the intermediate lysyl-adenylate (Desogus et al., 2000). Therefore, variants affecting Pro533 and Phe585 could alter the local structure of the lysine binding domain, resulting in decreased binding affinity for lysine, thus explaining the growth improvement after lysine supplementation. It remains to be determined whether lysine supplementation would be effective also in higher eukaryotes . Nevertheless, lysine supplementation that is currently administered to patients with lysinuric protein intolerance is sufficiently safe to be investigated in clinical trials.
In conclusion, KARS1-related disorder is a multi-system mitochondrial disease with congenital progressive microcephaly and cerebral tissue loss, white matter anomalies, epilepsy, oculomotor dysfunction, and immune-hematological dysfunctions. Therefore, we expand the spectrum of clinical abnormalities associated with KARS1 pathogenic variants and emphasize the importance of mitochondrial and cytosolic LysRS dysfunction in the pathogenesis of this disorder.

ACKNOWLEDGMENTS
We are grateful to patients and their parents for their participation to the study. This study was supported by Telethon