Biallelic mutations in valyl-tRNA synthetase gene VARS are associated with a progressive neurodevelopmental epileptic encephalopathy

Aminoacyl-tRNA synthetases (ARSs) function to transfer amino acids to cognate tRNA molecules, which are required for protein translation. To date, biallelic mutations in 31 ARS genes are known to cause recessive, early-onset severe multi-organ diseases. VARS encodes the only known valine cytoplasmic-localized aminoacyl-tRNA synthetase. Here, we report seven patients from five unrelated families with five different biallelic missense variants in VARS. Subjects present with a range of global developmental delay, epileptic encephalopathy and primary or progressive microcephaly. Longitudinal assessment demonstrates progressive cortical atrophy and white matter volume loss. Variants map to the VARS tRNA binding domain and adjacent to the anticodon domain, and disrupt highly conserved residues. Patient primary cells show intact VARS protein but reduced enzymatic activity, suggesting partial loss of function. The implication of VARS in pediatric neurodegeneration broadens the spectrum of human diseases due to mutations in tRNA synthetase genes.

encephalopathy and primary or progressive microcephaly. Overall, the discovery of the link between VARS and the complex developmental syndrome is of great interest for Nature Comms readership. However, some functional link between low enzymatic activity and neurological developmental abnormalities need to be presented to fully make the case for causality. - In Figure 1b, the MRI panels need to highlight the aspects affected compared to normal human MRI at these ages. Nature comms scientific readership is broad and the images as presented now are not very meaningful to most readers. A similar set of axial and sagittal views of normal brains (at least for 1 and 4 years) should be shown for comparison. - The enzymatic activity of the patients cell extracts should show control levels in Figure 2. I would prefer to see Suppl. -Some basic evaluation of cell cycle and neurogenic potential of the iPSCs from the patients would be necessary to show that human embryos can't develop normally with 20% activity of their VARS proteins.
Reviewer #3 (Remarks to the Author): Friedman et al. provide some evidence, suggesting that mutations in VARS the only known valine cytoplasmic-localized aminoacyl-tRNA synthetase may be responsible for a patient phenotype characterized by neurodevelopmental deficits and epilepsy. They expand the known range of mutations in this gene that are connected with this kind of phenotype. The first association of VARS with NEURODEVELOPMENTAL DISORDER WITH MICROCEPHALY, SEIZURES, AND CORTICAL ATROPHY was presented by Karaca et al. in 2015 (https://omim.org/entry/192150).
The genetic findings presented in this paper are based on WGS or WES. These technologies frequently lead to the discovery of more than one variant that co-segregate with the disease and may have pathogenic potential. This is why filtering and prioritization of variants are extremely important and the stragegy used should be addressed in detail. In this respect the manuscript in its present form is in my view not specific enough. I would like to see more information about the sequencing results and the bioinformatic analysis, including the most prominent variants/genes that were filtered out in the different individuals. This is interesting because the phenotype varies somewhat (see also below) and there might be putative additional factors among the filtered out variants. This should also be addressed in the discussion. Also, since different effect prediction algorithms often result in different predictions concerning the effects of specific sequence variants, I suggest that the authors use a variety of effect prediction programs (e.g. SIFT, PolyPhen, VEP, SNAP2, MutationTaster) for interpretation of their results, presented perhaps in a table for easy comparison by the reader. This is all the more important because, even though the patient phenotype is more or less well overlapping among the investigated individuals, there are also differences. For example in patient 3007, who does not show microcephaly but also differences concerning the type and frequency of seizures, and it would be interesting if there are also differences at the genetic level i.e. severity of the impairment or other features of the mutation that could explain this. Particularly, since for this patient there is no experimental evidence for a loss of VARS function presented either.
Experimental findings concerning VARS activity should be presented with proper information on the statistics performed and the level of significance should be indicated. It is not clear in which way the intra-assay variation is taken into account. Also, it is not clear whether technical or biological replicates were investigated.

REVIEWER #1 -With interest I read the manuscript by Friedman et al reporting on bi-allelic VARS mutations causing a progressive neurodevelopmental epileptic encephalopathy. The authors provide a concise and well-written manuscript containing both clinical and molecular genetic data of patients with mutations in VARS.
• Response. We thank Reviewer #1 for these positive comments.
-I have only some minor comments that the authors could consider to improve their manuscript. -the mutation nomenclature is not according the HGVS • Response. Corrected  • Response: Figure 1b has been modified to highlight the MRI abnormalities compared with normal child brain as a control. We have included images from a single 3 year old child as normal imaging are similar for the ages of affected patients presented.
-The enzymatic activity of the patients cell extracts should show control levels in Figure  2. I would prefer to see Suppl. Figure 2 data on heterozygous incorporated in the actual Figure 2.
• Response. Done -Some basic evaluation of cell cycle and neurogenic potential of the iPSCs from the patients would be necessary to show that human embryos can't develop normally with 20% activity of their VARS proteins.
• Response. We did not generate patients' iPSCs in this study and are therefore unable to study their cell cycle and neurogenic potential. We agree with Reviewer #2 that these experiments would be interesting in the future but are beyond the scope of this paper. In this study, our efforts were focused on demonstrating the pathogenicity of the patient missense mutations, by identifying enzymatic defects in patient cells, as demonstrated in the text and figures. • Response. We now present specific information about how variants were prioritized and filtered. Filtering and prioritization of variants was performed in three independent labs. The issue of variant prioritization is simplified in our consanguineous families, because there were very few homozygous rare potentially damaging variants to consider that fell into regions of homozygosity, which was the case for 3 of the families. Nevertheless, in every family, the VARS gene emerged as the top candidate based upon heuristic variant prioritization, which we now describe in the manuscript. Details regarding the bioinformatics analysis including description of the most prominent genes/variants that were considered for each individual are summarized in the supplemental data section. A table summarizing in silico predictions for each variant using several prediction programs (SIFT, PolyPhen, VEP, SNAP2, MutationTaster) has been added as a Supplemental Table. • We agree that etiology of variable phenotypic features and severity in our cohort is of great interest. There may be genetic and/or environmental factors influencing phenotypic severity in our patients. Interestingly, as seen in several neurodegenerative diseases, the age of onset of the disease is inversely proportional to the rapidity of neurodegeneration. Though we are unable to provide any definitive explanation for the variable clinical features in our cohort, the influence of genetic modifiers is of great interest. Unfortunately, our cohort is too small to draw conclusions regarding possible interactions. We have nevertheless, added a statement regarding possible genetic or environmental modifiers to the discussion.

-Experimental findings concerning VARS activity should be presented with proper information on the statistics performed and the level of significance should be indicated. It is not clear in which way the intra-assay variation is taken into account. Also, it is not clear whether technical or biological replicates were investigated.
• Response. Greater details of the experimental conditions are now provided in the methods section. Skin punches were collected at a single time from each patient studied here, then expanded and used for enzymatic analysis. We have performed all transferase activity assays according to standard clinical procedures, in triplicate technical replicates, but without biological replicates. The intra-assay variation of an analytical method displays which part of the variation between samples can be explained by the method itself. Hence any difference between control and patient that is larger than the intra-assay variation (15%) must be due to an altered VARS activity of the patient sample. Since technical triplicates were measured for this study, this analytical variation is already taken into account.  Hum Mutat. 2017 Jun;38(6):621-636. doi: 10.1002/humu.23205. Epub 2017 are not mentioned.
• Response. We agree and have incorporated these recent results into our manuscript.
I would like to start by thanking the authors for the revised version of their manuscript. Taking into account the comments of all reviewers, I believe the manuscript has been improved substantially.
I have two small remarks remaining. Firstly, the HGVS nomenclature uses '(' instead of '['. Please adjust. Secondly, in figure 3c, the authors use p-values to indicate their statistical significant values obtained. Please use p<0.0001 (or provide the actual obtained scientific value) instead p=0.000 for scientific accuracy.
Reviewer #2 (Remarks to the Author): I am happy with the response to my comments.
Reviewer #3 (Remarks to the Author): I appreciate that in response to my earlier comment, the authors have now presented the range of variants found in the respective pedigrees.
However, I am still not convinced by the presented evidence that (functional) VARS-deficiency is the sole cause of the observed phenotype in each of the presented families.
While it is of course true that VARS shows up in all families, in each there are variants in other genes that are considered damaging by some or even a majority of prediction tools. In one case a nonsense mutation in a different gene was found but completely disregarded. The change affects SCO2 in 3007-III-1, the patient with the most divergent phenotype. The mutation has the potential to be functionally severe and in my opinion cannot be disregarded solely based on the absence of a reported second mutation for an autosomal recessive condition or poor phenotypic overlap with previous findings. Particularly the latter appears a little ironic since the phenotypic overlap of this patient with the others is also far from perfect and no results concerning VARS activity are shown for this patient. Experiments addressing the molecular effects of this change and their combination with the impact of the VARS alteration would tremendously enhance the novelty and originality of this study. As it stands, however, neither the clinical nor the molecular basis for including this patient are very convincing.
In Family 3439 a duplication was found in GPR88 and one splice region variant in MLKL. Not surprisingly both do not have a score from the prediction tools used, however, these changes might well have pathogenic potential. Both manuscript and supplement, however, fail to address these specific variants and why they were discarded without further analysis. I would expect at least an in silico prediction of the effect of the insertion on the affected protein structure, an experimental test of the potential splice mutation to rule out its functional impact and proper discussion of their exclusion.
Specific points: The article merely reports the coincidence of VARS variants and a specific phenotype without functional data that are sufficient to warrant a statement such as "VARS deficiency results in progressive cortical atrophy". This needs to be toned down e.g. "VARS deficiency is associated with progressive cortical atrophy".