Article Text
PDF
Abstract
Background A homozygous loss-of-function mutation p.(Arg316Gln) in the fat mass and obesity-associated (FTO) gene, which encodes for an iron and 2-oxoglutarate-dependent oxygenase, was previously identified in a large family in which nine affected individuals present with a lethal syndrome characterised by growth retardation and multiple malformations. To date, no other pathogenic mutation in FTO has been identified as a cause of multiple congenital malformations.
Methods We investigated a 21-month-old girl who presented distinctive facial features, failure to thrive, global developmental delay, left ventricular cardiac hypertrophy, reduced vision and bilateral hearing loss. We performed targeted next-generation sequencing of 4813 clinically relevant genes in the patient and her parents.
Results We identified a novel FTO homozygous missense mutation (c.956C>T; p.(Ser319Phe)) in the affected individual. This mutation affects a highly conserved residue located in the same functional domain as the previously characterised mutation p.(Arg316Gln). Biochemical studies reveal that p.(Ser319Phe) FTO has reduced 2-oxoglutarate turnover and N-methyl-nucleoside demethylase activity.
Conclusion Our findings are consistent with previous reports that homozygous mutations in FTO can lead to rare growth retardation and developmental delay syndrome, and further support the proposal that FTO plays an important role in early development of human central nervous and cardiovascular systems.
- FTO
- Developmental Delay
- Next Generation Sequencing
- Neonatal Intensive Care Unit
Statistics from Altmetric.com
Introduction
The fat mass and obesity-associated (FTO) gene encodes for a nuclear protein belonging to the AlkB homologue (ABH) subfamily of 2-oxoglutarate (2OG) and ferrous iron-dependent oxygenases.1 Although the physiological roles of FTO are not well understood, some of its bacterial and human homologues are involved in the repair of alkylated DNA and RNA damage by oxidative demethylation. Others, including FTO, are proposed to have regulatory roles.2–6 Two independent genome-wide association studies revealed that SNPs in the first intron of the FTO gene are strongly associated with body mass index (BMI) and predispose to childhood and adult obesity.7 ,8 These BMI and obesity-associated findings were subsequently replicated in multiple populations, including Europeans, Africans and Asians (reviewed in ref. 9). The mechanism(s) by which FTO affects BMI is unclear, and whether the associated SNPs lead to obesity through an alteration of the expression or function of FTO or through cis-acting gene(s) remains to be deciphered.10 Nonetheless, subsequent studies in mice further support an important role of the FTO gene in energy balance and body weight regulation, as loss of FTO leads to a significant reduction in adipose tissue and lean body mass,11 ,12 and overexpression of FTO leads to increased food intake and results in obesity.13
Intriguingly, while the association between FTO SNPs and obesity appears to be robust, several studies have failed to identify rare more penetrant FTO alleles in obese individuals,14–16 suggesting that rare variants in FTO alone do not confer the risk of obesity in the studied populations. In contrast, a homozygous loss-of-function (LOF) mutation in FTO was identified in a large Palestinian Arab consanguineous pedigree, in which, nine individuals presented with severe growth retardation and multiple congenital malformations, including microcephaly, severe psychomotor delay, cardiomyopathy and characteristic facial features (OMIM 612938).17 This finding suggests that FTO is required for normal development of human central nervous and cardiovascular systems and that homozygous LOF mutations in the FTO gene can lead to an autosomal recessive lethal disorder and multiple defects. However, no other disease-causing mutation in FTO has been reported to date. Here, we report a novel pathogenic mutation in FTO identified by next-generation sequencing (NGS) in a newborn with growth retardation and significant developmental delay as well as clinical features consistent with the syndrome previously reported.17
Material and methods
Clinical description of the patient
The patient was enrolled in a local research project to investigate the effectiveness of NGS to diagnose newborns with rare disease in the neonatal intensive care unit. Institutional research ethics board (Children's Hospital of Eastern Ontario) approval of the project was obtained, and free and informed consent was given prior to enrolment.
The proband is female, the third child of a consanguineous (first cousins) couple of Tunisian origin. Amniocentesis was performed given a paternal balanced translocation: 46,XY, t(6;16)(q13;q24), and she was determined to carry the same translocation inherited from her asymptomatic father. Prenatal ultrasound indicated that she was small for gestational age, and her mother was induced at 38 weeks and 3 days due to apparent intrauterine growth restriction. She was born by an uncomplicated vaginal delivery, did not require resuscitation and had Apgar scores of 9 and 9 at 1 and 5 min, respectively. Her birth weight was 2.6 kg (−1.47 SD), length was 42.5 cm (−4 SD) and head circumference was 32 cm (−2 SD). She had feeding difficulties due to velopharyngeal insufficiency and reflux. At 28 days, her echocardiogram showed a patent ductus arteriosus and hypertrabeculation of her left ventricular apex. At 12 weeks, her weight was 4.25 kg (−3 SD), length was 49.5 cm (−5 SD) and her head circumference was 36.5 cm (−3 SD). Her facial features were characterised by periorbital fullness, a broad nasal bridge, anteverted nares, a smooth long philtrum, a thin upper vermillion border, a wide mouth, a normal palate and a small chin. She had cutis marmorata, and her tone was mildly increased. She had a G-tube inserted at 4 months due to continued feeding difficulties and poor weight gain. At 17.5 months, an MRI of her head showed decreased brain parenchyma, delayed myelination and a thin corpus callosum. A hearing assessment revealed mild-to-moderate hearing loss for 500 and 4000 Hz tone bursts bilaterally. At 24 months of age (figure 1), the time of her most recent assessment, she has severe global developmental delay, and is unable to sit unsupported. She smiles and laughs, but has no consistent sounds, and does not form words. She is being followed by ophthalmology for delayed visual development, but she does have a normal retinal examination. She has never had a seizure or any developmental regression. A summary of the clinical characteristics is presented in table 1.
Clinical characteristics of patients with p.(Arg316Gln) and p.(Ser319Phe) FTO substitutions
Photographs of the patient with the p.(Ser319Phe) FTO substitution. (A and B) Patient at 12 months of age and (C and D) 24 months of age. Facial features show arched eyebrows with mild synophrys, periorbital fullness, anteverted nares, a smooth long philtrum, a thin upper vermillion border and a small chin.
Targeted capture and NGS
Genomic DNA was extracted from peripheral blood lymphocytes using standard methods. DNA from the child and her parents were enriched for 4813 clinically relevant genes using the Illumina TruSight One Sequencing Panel enrichment kit. The three enriched libraries were pooled and sequenced on the MiSeq instrument (Illumina) in a trio approach according to the manufacturer's recommendations for paired-end 150 bp reads. The MiSeq Reporter Software (Illumina) was used for adaptor trimming, sample demultiplexing and fastq file generation. The NextGene software V.2.3.4.5 (SoftGenetics, State College, Pennsylvania, USA) was used for NGS data analysis and variant filtering. Briefly, fastq to fasta file conversion was performed using default parameters. Duplicate reads were removed before alignment to the reference human genome (build hg19). Single nucleotide variations and insertions/deletions were called and annotated using the default mutation filter settings. Data analyses were limited to coding sequences ±5 bp of exon-flanking intronic sequences. Synonymous and intronic variants other than the ones affecting the consensus splice sites with an allele frequency >1% in either the 1000 Genomes project (February 2012 data release) or the NHLBI Exome Sequencing Project (ESP) (January 2013 data release) were removed from the analysis.
2OG turnover assays
Nuclear magnetic resonance (NMR) spectra were recorded using Topspin 3.1 interfaced to a Bruker Avance AVIII 700 MHz instrument, optimised for 1H observation. Samples (50 μL) were prepared in Eppendorf tubes before transferring to 5 mm NMR tubes (Norell); time course data were then collected over a period of 45 min at 2 min intervals. The solvent deuterium signal was used as an internal lock signal, and the solvent signal was reduced by presaturation during a 2 s recovery delay. Samples were prepared in ammonium formate buffer (AFN), as adopted from a previously reported procedure.18 Enzyme stocks were in protiated 25 mm Tris buffer pH 7.5, which was diluted with AFN buffer according to online supplementary table S1.
Liquid chromatography–mass spectrometry-based demethylation assays
For enzyme activity assays, a 25 μL reaction mixture containing final concentrations of 3 μM hFTO, 70 μM 3-methyl thymidine (3 meT) nucleoside, 160 μM 2OG, 500 μM L-ascorbate, 100 μM diammonium iron(II) sulfate and 25 mM Tris-HCl, pH 7.5, was incubated at room temperature for 1 h. After incubation, the reaction mixture was quenched with methanol (25 μL) and then centrifuged (14 000 rpm, 10 min) to remove precipitated protein. The supernatant was then dried using an Eppendorf Speedvac machine, then resuspended in (25 μL) MilliQ-purified water. The product (thymidine) and substrate (3 meT) were separated using a Waters Acquity ultra performance liquid chromatography Atlantis C-18 column (130 Å,1.7 μm, 2.1×100 mm) with a gradient of 95% A (H2O with 1% formic acid) to 80% B (methanol with 0.1% formic acid) over 6 min, at room temperature. The ultraviolet detection wavelength was set to 266 nm. Retention times for thymidine and 3 meT is 3.4 and 3.9 min, respectively. The masses for each peak were confirmed by electrospray ionisation-time of flight mass spectrometry (Waters EXACTIVE instrument). The area under the curve of the substrate (2OG) and product (succinate) peaks were integrated using Xcalibur Software (Thermo Scientific) in the positive ion mode, and the percentage conversion of 3 meT to thymidine was used to quantify the activity of wild type (WT) and Ser319Phe FTO. To determine the optimal enzyme concentration, assays were carried out in triplicate with a range of enzyme concentrations (2, 2.5, 3, 3.5 µM), assuming that they would be similar to those previously reported.
Computational modelling of the p.(Ser319Phe) FTO substitution
A crystal structure of WT FTO (PDB 4IDZ) was used to model the effect of the p.(Ser319Phe) substitution on protein structure. Bond orders and all hydrogen atoms were regenerated. The protonation states for Asp, Glu, His and Lys residues were assigned followed by geometric optimisation of the positions of the hydrogen atoms by restrained energy minimisation using the OPLS-2005 force field.19 Protonation states were predicted for pH 7.2 (range ±2.0) using PROPKA.20 Steric clashes were resolved with convergence of Root-Mean-Square Deviation (RMSD) to 0.3 Å. Energy minimisation was then performed using an all atom model with the OPLS-2005 force field after assigning partial charges and force-field parameters. The Generalized Born Surface Area (GBSA) effective water model21 was used as the solvation model, and non-bonded electrostatic interactions were truncated with a cut-off distance of 20 Å. The convergence process of energy minimisation was terminated after 1000 iteration cycles and 0.05 kcal/mol Å gradient thresholds.
Results
Identification of a pathogenic missense substitution p.(Ser319Phe) in FTO
We investigated a 21-month-old girl who presented with distinctive facial features, failure to thrive, global developmental delay, left ventricular cardiac hypertrophy, reduced vision and bilateral hearing loss. Her phenotype did not suggest a specific genetic condition, and thus, we performed targeted NGS of 4813 clinically relevant genes in the patient and her parents to search for a disease-causing mutation. On average, ∼16 million high-quality mappable reads were generated per individual, resulting in an average coverage of 90× and 97.5% of the targeted regions being covered by at least 10 reads. Given that the parents of the newborn were consanguineous, and the pedigree is consistent with an autosomal recessive mode of inheritance, we searched for genes with rare homozygous mutations. This strategy led to the identification of six candidate genes (LRP2, NHLRC1, FTO, HYDIN, CYP24A1, ITGB2), of these, only homozygous mutations in FTO were previously known to cause a multiple malformation syndrome in keeping with our patient's phenotype (table 1). This homozygous single nucleotide variant (c.956C>T) is located in exon 5 of the FTO gene (NM_001080432.2), and leads to the substitution of serine to phenylalanine at codon 319 (figure 2A). The variant information and associated clinical phenotype have been submitted to ClinVar (Accession No. SCV000211951). This mutation, as confirmed by Sanger sequencing (figure 2B), has not been previously described in dbSNP, the 1000 Genomes project, the NHLBI ESP or the Exome Aggregation Consortium databases, and it is predicted to be deleterious by Sorting Intolerant From Tolerant (SIFT), PolyPhen-2 and Mutation Taster. It affects a strictly conserved amino acid residue located in the eighth β-strand of the distorted double-stranded β-helix core fold that is conserved in 2OG-oxygenases, including FTO (figures 2C and 3).6
Identification of the p.(Ser319Phe) FTO substitution by next-generation sequencing. (A) Mutation visualisation in the Integrative Genomics Viewer browser: next-generation sequencing reads overlapping the homozygous mutations in the FTO gene. (B) Sanger sequencing validation: electropherograms showing the c.956C>T mutation in the affected child and healthy parents. (C) Multiple sequence alignment of FTO representative orthologs using Clustal Omega. The sequence above the alignment indicates the mutated residue (highlighted in red) identified in the patient described herein (P). The highly conserved affected residues Arg316 and Ser319 are highlighted in blue and red, respectively. The black arrows labelled with roman numerals identify three of the eight β strands that form the conserved double-stranded β-helix core fold of the 2-oxoglutarate oxygenases.28 B.t, Bos taurus; D.r, Danio rerio; E.c, Equus caballus; FTO, fat mass and obesity-associated gene; G.g, Gallus gallus; H.s, Homo sapiens; Ma.m, Macaca mulatta; M.d, Monodelphis domestica; Mu.m, Mus musculus; Or.a, Ornithorhynchus anatinus; R.n, Rattus norvegicus; X.t, Xenopus tropicalis.
Modelling the three-dimensional structure of the p.(Ser319Phe) FTO substitution. (A) Ribbon representation of a crystal structure of FTO (PDB 4IDZ) with Fe(II), 2OG and model of the p.(Ser319Phe) substitution. The core DSBH is in purple and the eight strands of the conserved DSBH core fold are labelled according to convention.29 Substitution of either of two other residues positioned on DSBH strand βVIII has been shown to affect activity. Arg316 plays a key role in binding the 2OG C5 carboxylate: the Arg316Gln FTO is inactive. Substitution of Arg322 to Gln interferes with substrate binding. (B) Close-up sticks representation of the region around wild-type FTO Ser319 (salmon) and the modelled Phe319 (turquoise), indicating how neighbouring residues may accommodate the substitution. The p.(Ser319Phe) substitution results in the loss of two hydrogen bonds with Thr202 and His321. The bulky nature of the Phe side chain appears to distort this region, which is positioned near the 2OG-binding site, and thus, may affect 2OG binding. 2OG, 2-oxoglutarate; DSBH, double-stranded β helix; FTO, fat mass and obesity-associated gene.
In vitro characterisation of the FTO p.(Ser319Phe) substitution
Because of its close proximity to the 2OG-binding site of FTO (figure 3A), we proposed that the p.(Ser319Phe) substitution may alter the FTO catalysis/cosubstrate binding modes. To test this proposal, we made recombinant Ser319Phe FTO following established expression and purification procedures, and compared its activity with WT FTO (see online supplementary information). The Ser319Phe FTO was purified to near homogeneity as determined by SDS-PAGE (figure 4A). We first checked that Ser319Phe FTO was properly folded using circular dichroism (CD) spectroscopy, a method that can detect and quantify secondary structure content in proteins based on the interaction of circularly polarised light with the protein in solution. The spectra were the same, within limits of detection, indicating the presence of highly similar secondary structure characteristic of the same overall protein fold (see online supplementary figure S1A). Substitution of amino acids in proteins can affect protein stability. To analyse potential stability differences, we then compared WT and Ser319Phe FTO by CD measurements at 222 nm while increasing the temperature. The melting temperatures for WT and Ser319Phe FTO were almost indistinguishable, suggesting similar stability for the two proteins (see online supplementary figure S1B). However, the refolding results suggest that, unlike WT FTO (see online supplementary figure S1C), Ser319Phe FTO does not refold completely after thermal denaturation (see online supplementary figure S1D). Next, we monitored the demethylation activity of Ser319Phe FTO by liquid chromatography–mass spectrometry using 3-methylthymine (m3T); demethylation was observed, but again at a much reduced level compared with WT FTO (figure 4). We then used 1H NMR analysis to investigate if Ser319Phe FTO can catalyse the oxidation of the cosubstrate 2OG to the succinate product (figure 5A) in the presence of m6A primary substrate (2OG turnover, figure 5B). The results clearly demonstrate 2OG turnover activity for Ser319Phe FTO, though at a reduced level compared with WT FTO (figure 5C, D). Although further kinetic studies may reveal mechanistic details of the effect of the FTO p.(Ser319Phe) substitution, these initial results clearly reveal that the activity of recombinant isolated Ser319Phe FTO is reduced compared with WT FTO, and its refolding ability is altered.
(A) Recombinant Ser319Phe FTO has reduced demethylase activity. Sodium Dodecyl Sulfate Polyacrylamide gel electrophoresis (SDS-PAGE) showing purified WT and Ser319Phe-mutant FTO protein. (B) Comparison of activity by liquid chromatography–mass spectrometry analyses of 3-methylthymine (m3T) demethylation by WT and Ser319Phe FTO after 1 h incubation. Time course of (C) WT and (D) Ser319Phe FTO at different enzyme concentrations. FTO, fat mass and obesity-associated gene; WT, wild type.
Recombinant Ser319Phe FTO displays reduced 2OG turnover activity compared with WT FTO as shown by 1H NMR studies. (A) Reaction scheme for FTO demethylation activity. FTO consumes 2OG and molecular oxygen as cosubstrates for the oxidative demethylation of N-methylated nucleosides (ie, 3 meT and m6A) and also produces formaldehyde, carbon dioxide and succinate as by-products. (B) 1H NMR analysis of WT and Ser319Phe FTO. Time course NMR spectra demonstrating WT FTO catalyses turnover of 2OG (triplet at 2.35 ppm) to succinate (single peak near 2.30 ppm) in the presence of m6A substrate. Time course of succinate production and 2OG depletion (C) in the absence and (D) in the presence of m6A substrate: WT FTO depletion of 2OG (turquoise), Ser319Phe FTO depletion of 2OG (red), WT FTO production of succinate (blue) and Ser319Phe FTO production of succinate (olive). 2OG, 2-oxoglutarate; FTO, fat mass and obesity-associated gene; m6A, N6-methyladenine; NMR, nuclear magnetic resonance; WT, wild type.
Discussion
Targeted NGS of clinically relevant genes in a trio approach led to the identification of a homozygous missense mutation (c.956C>T; p.(Ser319Phe)) in the FTO gene in a newborn presenting with growth retardation, significant developmental delay and clinical features overlapping with the syndrome described by Boissel et al in 2009.17 Although five other homozygous missense mutations were identified in this patient after variant filtering (data not shown), four were predicted to be benign, and one (c.619C>G; p.(Leu207Val)) in CYP24A1 (cytochrome P450, family 24, subfamily A, polypeptide 1) was predicted to be damaging, but unlikely to be relevant to our patient's phenotype. The FTO p.(Ser319Phe) substitution affects a strictly conserved amino acid residue (figure 2C) located in the catalytic domain of the FTO protein (figure 3A). Interestingly, the p.(Ser319Phe) substitution is located two residues downstream of the previously identified LOF substitution p.(Arg316Gln), crucially involved in binding 2OG,22 that was shown to inactivate FTO enzymatic activity in vitro.17 A third missense p.(Arg322Gln) substitution, located in the same conserved double-stranded β helix β-strand VIII, and important for N-methyl-nucleotide substrate binding, was identified at the heterozygous state following a mutational analysis of FTO in lean and obese individuals and reported to severely impair the enzymatic activity of FTO in vitro.14 Consistent with our predictions, the biochemical analyses reveal that the p.(Ser319Phe) substitution reduces, but does not ablate, FTO activity.
The FTO p.(Ser319Phe) substitution identified in our patient has not been previously reported in the publicly available SNP databases. Moreover, inspection of the NHLBI ESP database for variants in the FTO gene revealed only one homozygous missense substitution (c.1214C>T; p.(A405V)) in four different individuals out of 12 996 alleles from individuals with European, American and African–American ancestries. However, this missense substitution, which has a minor allele frequency (MAF) of 0.0095, affects a non-conserved residue located in the COOH-terminal helical-bundle domain,14 and is predicted to be benign by different in silico prediction programmes. All the other non-synonymous (NS) mutations (46 missenses and 1 nonsense) were only found at the heterozygous state, and appear to be rare in the general population as none exceed 0.005 of MAF. Interestingly, only 9 missense mutations out of 47 heterozygous NS mutations were found in the catalytic domain of FTO, and none exceeds 0.001 of MAF. Altogether, this suggests that protein alterations affecting the catalytic domain of FTO are rare and strongly supports the proposal that the p.(Ser319Phe) substitution is disease-causing.
Cultured skin fibroblasts of a patient carrying the homozygous p.(Arg316Gln) substitution in FTO were previously reported to have decreased proliferative ability and premature senescence-like phenotype when compared with controls.17 Our patient's skin fibroblasts did not manifest these phenotypes (data not shown). This difference may reflect our observation of reduced, but not ablation of activity for the p.(Ser319Phe) substitution, whereas, the p.(Arg316Gln) substitution ablates activity since arginine 316 is a strictly conserved 2OG and Fe-binding residue in FTO, and is crucial for the demethylase activity. It is also possible that the impaired proliferation and accelerated senescence phenotype observed for the p.(Arg316Gln) substitution can be attributed to the fact that the assay was performed using fibroblasts extracted from only one patient out of nine individuals carrying the p.(Arg316Gln) substitution.17 This senescence-like phenotype could also be due to a second mutation located outside of the 6 Mb homozygous region shared between the nine affected individuals, which would also explain the absence of the senescence-like phenotype in our patient.
The molecular mechanisms through which mutations in FTO can lead to a severe phenotype in humans are yet to be discovered. FTO is highly expressed in brain, and can demethylate 3-methylthymine, and to a lesser extent 3-methyluracil, in both single-stranded DNA and RNA.1 ,23 FTO and a related enzyme, ALKBH5,24 have also been shown to actively demethylate m6A in mRNA.6 m6A at specific mRNA consensus sequences appears to play an important role in regulating gene expression, mRNA splicing and transport.25 ,26 The m6A containing consensus sequence is highly conserved between human and mouse, and their mRNA levels are regulated from early development through adulthood in a dynamic and tissue-specific manner.25 ,26 Specifically, m6A levels in mouse brain mRNAs were found to be low during embryonic development and to increase dramatically by adulthood. The fact that FTO is the first identified animal m6A mRNA demethylase, and that it is present in the nucleus,1 supports the proposal that m6A is a physiological substrate of FTO, and further suggests that mutations affecting the catalytic machinery of FTO would alter the m6A mRNA landscape, which may, in turn, affect gene expression and mRNA splicing. FTO has also been reported to play a role in the coupling of amino acids to mammalian target of rapamycin complex 1 (mTORC1) signalling as cells lacking FTO display a decreased activation of the mTORC1 pathway and an overall decrease in mRNA translation.27 Interestingly, this signalling role was disrupted in cells expressing Arg316Gln FTO that lacks demethylase activity, suggesting that FTO enzymatic activity is crucial to its role in the cellular sensing of amino acids.27 Together, our results suggest that homozygous LOF mutations in FTO may lead to a severe developmental phenotype through a dysregulation of gene expression via altered levels of specific sets of mRNA. The identification of the downstream targets specifically regulated by FTO in the future would shed more light on the pathophysiological mechanisms leading to disease, and may open new avenues for therapeutic interventions.
In summary, we report the second patient with a novel pathogenic mutation in FTO. This patient presents with distinctive facial features, failure to thrive, global developmental delay, decreased brain volume, left ventricular hypertrophy and bilateral hearing loss. Our findings further confirm that homozygous mutations in FTO lead to a rare syndrome of severe growth retardation and developmental delay along with variable other clinical features, and further support that FTO plays an important role in the early development of human central nervous and cardiovascular systems. Phenotypic variability may exist due to residual activity of the FTO enzyme, and ‘early death’ may not be as universal a feature. Given the functional data presented herein, we propose that this syndrome be referred to as ‘FTO-deficiency syndrome’ instead of the moniker, ‘Growth retardation, developmental delay, coarse facies, and early death’ currently used in OMIM. Finally, we show that NGS of a ‘disease-relevant’ targeted panel can be used as the first diagnostic approach for rare genetic conditions, and may lead to improve the care of newborns by providing an accurate molecular diagnosis.
Acknowledgments
The authors would, first and foremost, like to thank the patient and her family, without whose participation, this work would not have been possible. DAD is supported by the CIHR Institute of Genetics Clinical Investigatorship Award. The University of Oxford thanks the Wellcome Trust, the Biotechnology and Biological Research Council and Cancer Research UK for financial support.
References
Footnotes
Twitter Follow Fiona McMurray at @FionaMc2005
Contributors Conception and design of the work: HD, DAD and CMA. Acquisition of data: HD, DZ, FMM, AY, JV, JW, SS and HC. Analysis and interpretation of data: all authors. Drafting of the manuscript: HD, DAD and CMA. Critical revision of the manuscript: all authors. Final approval of the version to be published: all authors.
Funding This work was supported by the CHAMO Academic Health Science Centres Alternate Funding Plan (AHSC_AFP) Innovation Fund (#9436) and the Care4Rare Canada Consortium, which includes funding from Genome Quebec and Genome Canada (OGI-064), the Canadian Institutes of Health Research (GPH-129346), the Ontario Genomics Institute (REG1-4404), Ontario Research Fund (REG1-4404) and Children's Hospital of Eastern Ontario Foundation.
Competing interests None declared.
Patient consent Obtained.
Ethics approval Children's Hospital of Eastern Ontario, Ottawa, Ontario, Canada.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement There are no additional unpublished data from this study.