European Review for Medical and Pharmacological Sciences 2019; 23: 1710-1721 Next-generation sequencing identifies a homozygous mutation in ACADVL associated with pediatric familial dilated cardiomyopathy

S.J. CARLUS1, I.S. ALMUZAINI2, M. KARTHIKEYAN3, L. LOGANATHAN3, G.S. AL-HARBI1, A.M. ABDALLAH4, K.M. AL-HARBI1

1Pediatrics Department, Cardiogenetics Unit, College of Medicine, Taibah University, Al-Madinah, Kingdom of Saudi Arabia 2Department of Pediatric Cardiology, Al-Madinah Maternity and Children Hospital (MMCH), Al-Madinah, Kingdom of Saudi Arabia 3Department of Bioinformatics, Alagappa University, Karaikudi, Tamil Nadu, India 4West Midlands Regional Genetics Laboratory, Birmingham Women’s NHS Foundation Trust, Birmingham, United Kingdom

Abstract. – OBJECTIVE: Pediatric familial di- Key Words lated cardiomyopathy (DCM) is a rare and se- Pediatric familial dilated cardiomyopathy, Targeted vere heart disease. The genetics of familial DCM sequencing, ACADVL, Saudi Arabia, Consanguinity, are complex and include over 100 known dis- Molecular docking, Molecular dynamics. ease-causing , but many causative genes are unknown. We aimed to identify the causative gene for DCM in a consanguineous Saudi Ara- bian family with affected family members and a history of sudden death. Introduction PATIENTS AND METHODS: Affected (two chil- dren) and unaffected (one sibling and the mother) Dilated cardiomyopathy (DCM) is a heart family members were screened by next-gener- ation sequencing (NGS) for 181 candidate DCM disease characterized by ventricular dilation and genes and underwent metabolic screening. Fif- impaired myocardial contractility. DCM affects ty-seven clinically annotated controls and 46 DCM 1 in 2500 of the general population and is one of cases were then tested for the identified mutation. the most common cardiomyopathies causing pe- In silico structural and functional analyses includ- diatric heart failure1-4. The prognosis for children ing modeling, structure prediction and with DCM is often poor5, with approximately dynamic simulations were performed. RESULTS: 40% undergoing cardiac transplantation or death A homozygous missense mutation 3 in exon 15 of the acyl-CoA dehydrogenase very within five years of diagnosis . long chain gene (ACADVL; chr17:7127303; G>A) DCM is known to be caused by a variety of mu- was identified in affected subjects that substitut- tations in sarcomere, mitochondrial, RNA-bind- ed histidine for arginine at codon 450 (p.R450H). ing, Z-disk, cytoskeletal, sarcoplasmic reticulum The variant was heterozygous in the mother and and nuclear envelope proteins6. In children, DCM unaffected sister. The mutation was absent in 57 is usually autosomal dominant (AD)7; however, clinically annotated controls and 48 pediatric DCM cases. The mutation was predicted to cause a 14% of pediatric DCM cases are associated with significant and deleterious change in the ACADVL a wide spectrum of extracardiac features such protein structure that affected drug binding, sta- as neuromuscular, metabolic, or malformation bility, and conformation. Metabolic screening con- syndromes with X-linked, mitochondrial or auto- firmed VLCAD deficiency in affected individuals. somal recessive (AR) inheritance8. About 20-50% CONCLUSIONS: The ACADVL R450H mutation of DCM patients have familial DCM (FDCM)9, is an uncommon cause of the DCM phenotype which is often characterized by genetic heteroge- that appears to be autosomal recessive. Targeted NGS is useful for identifying the causative muta- neity and high variability even between members tion(s) in familial DCM of unknown genetic cause. of the same family; the age of the onset and dis-

1710 Corresponding Author: S. Justin Carlus, Ph.D; e-mail: [email protected] ACADVL mutation in pediatric familial dilated cardiomyopathy ease progression also differ considerably between Study Participants and Clinical individuals with FDCM. To date, mutations in Evaluation over 100 genes have been reported to cause Peripheral blood samples were collected from DCM, accounting for 50% of affected individuals four individuals: two affected DCM subjects and but leaving half without a known genetic basis10. an unaffected normal child and their mother Acyl-CoA dehydrogenase very long chain from a consanguineous Saudi Arabian family (ACADVL) mutations are known to cause AR attending the Department of Pediatric Cardiol- DCM and result in very long-chain acyl-coenzyme ogy, MMCH, Al-Madinah, Saudi Arabia. The A dehydrogenase (VLCAD) deficiency, pediatric mother provided written informed consent for cardiomyopathy and sudden death11. ACADVL is study participation. There was a family history about 5.4 kb long, located on 17p13.1, of sudden death of the father, reported to be from and has 20 exons12. VLCAD deficiency is a rare AR heart disease. The patients reside in Al-Madinah metabolic disorder that affects mitochondrial fatty in western Saudi Arabia. acid β-oxidation, which in turn impacts heart, liver, DCM was diagnosed according to published and muscle function12. VLCAD deficiency can be guidelines17 using the family history, physical and classified into three clinical phenotypes, of which clinical examination, electrocardiograms (ECG), the early-onset form is severe with a high incidence chest x-rays, and 2D-echocardiography. Doppler of cardiomyopathy and high mortality13. echocardiography was performed using a Hewl- Our previous surveys of the registry of a ett-Packard 1500 or 5500 echocardiographic sys- pediatric cardiology clinic at the Maternity and tem. The diagnosis of DCM was characterized by Children Hospital (MMCH), Al-Madinah, Saudi the presence of left ventricular dilation <45% and Arabia, revealed a very high prevalence of DCM left ventricular end-diastolic dimension >117%. in the pediatric heart patient population, and the Other cardiac or systemic causes of DCM includ- DCM mortality rate is high in infants in central ing coronary and valvular diseases were absent. and eastern Saudi Arabia14. Affected children were more likely to die if they were the offspring Dilated Cardiomyopathy Gene Panel of a consanguineous marriage14. Saudi Arabia has We previously developed an in-house cus- one of the highest rates of consanguinity in the tom gene panel covering 181 genes confirmed world, as marriages between first cousins and or presumed to be involved in cardiomyopathy close relatives are widely accepted and frequently pathology10; the complete gene list is available occur15, and consanguinity may be a risk factor in Table I. Briefly, this gene panel covers cod- for some congenital abnormalities16. ing exons, exon-intron boundaries, and known In this paper, we aimed to identify novel mu- variants in the intronic regions of target genes. tation(s) and gene(s) responsible for familial DCM The total target region length is 1108523 base using targeted NGS technology in a consanguin- pairs (bp) involving 3852 target regions. Target eous Saudi Arabian family with affected family regions include exons, 250 bp of adjacent introns, members. In doing so, we identified a homozygous and 5′- and 3′-UTR regions. All annotated coding missense mutation in ACADVL in affected individ- regions were extracted from hg19 from Ensembl uals with concomitant VLCAD deficiency on met- (http://www.ensembl.org) and the UCSC genome abolic screening. To provide functional insights, in browser (http://genome.ucsc.edu). silico analyses were used to predict the deleterious effect of the mutation and to understand its effects Targeted Gene Sequencing on protein structure and drug binding. For targeted gene sequencing, DNA was ex- tracted from whole blood using the QIAamp DNA Blood Mini Kit (Qiagen, Hilden, Germany). DNA Materials and Methods quantity and quality were measured by reading the whole absorption spectrum (220-750 nm) with Ethics Statement a NanoDrop ND-2000 and calculating the DNA This study was fully in accordance with the eth- concentration and absorbance ratio at 260/280 ical standards of the Taibah University Ethics Re- nm. Each sample was quantified with a Qubit 3.0 search Committee, the Ethics Review Board of Ma- Fluorometer (Thermo Fisher Scientific, Waltham, dinah Maternity and Children Hospital (MMCH), MA, USA). Extracted DNA samples were evalu- and with the 1964 Helsinki Declaration and its later ated by loading 100 ng of DNA on a 0.8% agarose amendments or comparable ethical standards. gel for electrophoresis, which showed intact bands

1711 S.J. Carlus, I.S. Almuzaini, M. Karthikeyan, L. Loganathan, G.S. Al-Harbi, et al. of DNA without any degradation. Then, 2-3 µg of using Varminer (an in-house variant interpreta- genomic DNA per participant were processed for tion tool), and variant pathogenicity was assessed library preparation, with the library enriched using based on the 2015 American College of Medical a custom Nimblegen panel and sequenced as 2 x Genetics (ACMG) guidelines. 250 bp paired-end reads on an Illumina HiSeq 2500 (Illumina, San Diego, CA, USA) according Validation by Sanger Sequencing to the manufacturer’s protocol. Sanger sequencing was performed to confirm the presence of variants. Primer sequences were Bioinformatics Analysis designed using Primer3. The primer pairs used for Bioinformatics analysis was performed as pre- the Polymerase Chain Reaction (PCR) were: for- viously described18. After obtaining the data, we ward:5’-CCTGTGCTAGGAACCTGGAG-3’ and used the Genome Analysis Toolkit (GATK) best reverse: 5’CCCAGAGAGCTCCTTTCCTT-3’. The practice variant-calling pipeline to identify germ- amplicons were directly sequenced using BigDye line variants. Adapters from the raw reads were chain termination chemistry on an ABI 3730 DNA trimmed using fastq-mcf; trimmed reads were analyzer (Applied Biosystems, Foster City, CA, USA). mapped to the hg19 human reference genome using the Burrows-Wheeler Aligner (BWA); du- Metabolic Screening plicate reads were removed using Picard, and re- All study participants underwent metabolic alignment, recalibration and variant calling were screening in a certified private lab according to performed with GATK. The identified variants screening guidelines by flow injection analysis were annotated to genes using the variant effect tandem mass spectrometry (FIA-MS/MS). predictor (VEP) program against the Ensembl release 84 human gene model. Clinically rele- Additional Screening vant mutations were annotated using variants of the DCM Cohort and Controls published in the literature, and deep annotation The variant identified was assayed for in 57 of germline variants was performed using our in- clinically annotated controls and 46 DCM cases house VariMAT pipeline, annotating against the to evaluate its frequency in the Saudi Arabian latest Human Gene Mutation Database (HGMD), population and individuals with DCM. ClinVar, Online Mendelian Inheritance in Man (OMIM), SwissVar and population databases Structural and Functional (1000 Genome Phase 3, ExAC, National Heart, Analyses of ACADVL Lung, and Blood Institute (NHLBI), dbSNP, All computational studies were carried out on 1000 Japanese, Indian, and 500 internal Sau- a CentOS v7.0 Linux platform with Intel®Core™ di Arabia-specific database). Furthermore, we i7-4770 CPU 3.40 GHz processor using Mae- annotated variants against in silico prediction stro, version 11.4.011 (Schrodinger, LLC, New algorithms (PolyPhen-2, Sorting Intolerant from York, NY, USA) and GROningen MAchine for Tolerant (SIFT), MutationTaster2, and the likeli- Chemical Simulations (GROMACS) v5.0.5. The hood ratio test (LRT)). Variants were analyzed ACADVL crystal structure was extracted from

Table I. List of 181 Genes screened in our cohorts. ABCC9, ACADVL, ACTA1, ACTC1, ACTN2, AKAP9, ALG10, AMPD1, ANK2, ANO5, B3GALNT2, B3GNT1, BAG3, BIN1, BMPR2, CACNA1C, CACNA1D, CACNA2D1, CACNB2, CALM1, CALM2, CALR3, CAPN3, CASQ2, CAV3, CCDC78, CFL2, CHKB, CLIC2, CNBP, CNTN1, COL12A1, COL6A1, COL6A2, COL6A3, COL9A3, CRYAB, CSRP3, CTNNA3, DAG1, DES, DMD, DMPK, DNAJB6, DNAJC19, DNM2, DPP6, DSC2, DSG2, DSP, DTNA, DYSF, EMD, EYA4, FHL1, FKRP, FKTN, FLNC, GATAD1, GJA1, GJA5, GLA, GMPPB, GNAI2, GNE, GPD1L, HCN4, HNRNPDL, HRAS, ISCU, ISPD, ITGA7, JPH2, JUP, KBTBD13, KCNA5, KCND3, KCNE1, KCNE2, KCNE3, KCNH2, KCNJ2, KCNJ5, KCNQ1, KLHL40, KLHL41, LAMA2, LAMA4, LAMP2, LARGE, LDB3, LMNA, LMOD3, MEGF10, MIB1, MTM1, MYBPC1, MYBPC3, MYF6, MYH2, MYH6, MYH7, MYL2, MYL3, MYLK2, MYOT, MYOZ2, MYPN, NEB, NEXN, NKX2-5, NPPA, NUP155, PABPN1, PKP2, PLEC, PLN, PNPLA2, POLG, POMGNT1, POMGNT2, POMK, POMT1, POMT2, PRDM16, PRKAG2, PSEN1, PSEN2, PTRF, RAF1, RBM20, RRM2B, RYR1, RYR2, SCN10A, SCN1B, SCN2B, SCN3B, SCN4B, SCN5A, SCO2, SDHA, SEPN1, SGCA, SGCB, SGCD, SGCG, SLC22A5, SLC25A4, SMCHD1, SNTA1, SPEG, STAC3, SYNE1, SYNE2, TAZ, TBX20, TCAP, TGFB3, TIA1, TMEM43, TMEM5, TMPO, TNNC1, TNNI2, TNNI3, TNNT1, TNNT2, TNPO3, TPM1, TPM2, TPM3, TRAPPC11, TRDN, TRIM32, TRPM4, TTN, TTR, VCL, VCP, VMA21.

1712 ACADVL mutation in pediatric familial dilated cardiomyopathy the (PDB ID: 2UXW; https:// a simple point charge (SPC), immersed in a rhom- www.rcsb.org/structure/2uxw) at 1.5 Å resolu- bic dodecahedron-shaped box, and the energy of tion. The crystal structure is a single chain with the complexes was minimized using the steepest 607 amino acids and co-crystallized ligands19. descent method of 1000 steps with a cutoff of 9 The protein sequence was obtained from the Un- Å for van der Waals and Coulomb forces. The iProtKB database (P49748) for sequence analysis solvated system was neutralized by adding ap- and was aligned using Geneious Pro software20. propriate ions to the simulation. Simulations were The mutant sequence was prepared by manually performed at constant pressure and temperature editing the fasta file. Both wild-type and mutant (NPT) ensemble using the Berendsen thermostat protein sequences were aligned using MAFFT with coupled temperature and pressure of 310 alignment v.7.017 with the Blosum62 matrix. K and 1 bar pressure, respectively24. Finally, the The aligned sequence was further subjected to protein, solvent, and ions inside the simulation box secondary structure prediction. Molecular docking were equilibrated, especially the heavy protein analysis was performed to assess structural chang- atoms that were controlled by a harmonic restraint es and ligand binding potency variations. The with a force constant equal to 1000 kJ mol−1 nm−2 protein structure was initially prepared using the using the NPT and NVT (constant-temperature, Protein Preparation Wizard in Maestro. We added constant-volume) ensembles. The MD simulation missing hydrogens and side chains, optimized the trajectories were saved at every time interval of 2 H-bonds, and performed energy minimization21. fs throughout the simulation period (10 ns), and the Energy minimization was performed to correct output files were further analyzed for root mean bad contacts and torsion angles and to stabilize square deviation (RMSD) calculations. RMSD cal- the structure in low potential energy. During en- culations are helpful for observing the structural ergy minimization, the structural deviation was deviations between wild-type and mutant constraint within 3.0 Å to gain a reliable structure. with respect to the backbone atoms in each residue and were calculated for the whole 10 ns data and Molecular Docking Studies plotted using Origin Pro 8 (OriginLab, Northamp- We studied two putative therapeutic drugs in ton, MA, USA). silico to test ACADVL binding affinity with and without the mutation: bezafibrate22, a hypolipid- Binding Free Energy Calculation emic drug and putative ACADVL replacement Prime/MM-GBSA is a module used to predict therapy, and L-carnitine23, a drug used to treat the free energy of binding for ligands and recep- in VLCAD-deficient patients. The tors25. Binding free energy calculations were carried drugs were prepared in the LigPrep module of out in Prime in Maestro. The binding free energy Maestro before proceeding to molecular docking (ΔGbind) was estimated as ΔGbind = ER:L − (ER analysis. The active site of the protein was de- + EL) + ΔGsolv + ΔGSA, where ER:L is the energy termined by the co-crystallized ligand available of the complex, ER + EL is the average energy of from the PDB structure. A grid was generated the ligand and apoprotein using the OPLS-3 force around the active site of wild and mutant-type field, and ΔGsolv (ΔGSA) is the difference between protein structures. The drugs were docked in the GBSA solvation energy (surface area energy) of the active site of proteins using the Ligand Docking complex and the sum of the corresponding energies module in Maestro in extra precision (XP) mode. for the ligand and apo-protein26. A docking conformation of a drug was considered better based on a more negative Glide XP score. In addition, the non-covalent interaction and glide Results energy were used to assess compound potency. Clinical Assessment of the DCM Family Molecular Dynamics Simulations The DCM cases were born of consanguineous The docked ACADVL protein complex was Saudi parents. There was a family history of sud- subjected to molecular dynamics (MD) simu- den death of the father, which was described as lations using GROMACS v5.0.5 and the GRO- “heart disease”. Due to a lack of clinical record MOS96 43a1 force field to help identify structural availability and religious constraints, the pedi- changes in wild-type and mutant proteins. Wild- gree was limited to two generations. The family type and mutant proteins were independently sim- pedigree was highly suggestive of a recessive ulated for 10 ns. The total system was filled with pattern of inheritance (Figure 1A).

1713 S.J. Carlus, I.S. Almuzaini, M. Karthikeyan, L. Loganathan, G.S. Al-Harbi, et al.

A

B

C

Figure 1. A, Pedigree of the family, with proband indicated with an arrow. Males are depicted as squares, females as circles, and deceased individuals with a diagonal line. Blue shading represents individuals who are clinically affected with DCM, while open shading indicates unaffected individuals. Het indicates a heterozygous ACADVL p.R450H mutation and Hom indicates the homozygous ACADVL p.R450H mutation. B, Homozygous missense mutation in exon 15 of ACADVL (chr17:7127303; G>A) resulting in the substitution of histidine for arginine at codon 450 (p.R450) in affected DCM subjects. The mutation was present in the heterozygous state in the mother and unaffected sister. C, Sequence alignment shows the residue (p.R450) homology in closely related mammals. Amino acid conservation for the targeted region of ACADVL.

The unaffected mother was 38 years old and echocardiography revealed dilated left ventricles the unaffected female sibling was 1 year old. The in the presence of reduced left ventricular func- patients’ mother and sibling were healthy and had tion. The ejection fraction was 25%, with mild normal echocardiograms. DCM subject 1 was a mitral valve regurgitation without pericardial ef- six-year-old girl (disease onset at 3 years) and fusion. The primary diagnosis was DCM. There DCM subject 2 was a four-year-old boy (disease was no cerebellar ataxia, epilepsy, intellectual onset at 2 years). On chest x-ray, the patients disability, or inflammatory and infectious dis- had significant cardiomegaly. Electrocardiogram eases in either patient. Symptoms of metabolic (ECG) revealed prolonged QT intervals, while disorders were completely absent.

1714 ACADVL mutation in pediatric familial dilated cardiomyopathy

Genetic Analysis and Systematic score 827). Multiple sequence alignment revealed Prioritization of Candidate Variants high sequence similarity between closely related Targeted sequencing of 181 cardiomyopathy mammals (Figure 1C). On evaluating this muta- disease genes was performed on the two DCM tion frequency in other affected and unaffected patients, their mother and sibling. A mean of 0.6 healthy Saudi individuals, the identified variant Gb of sequence was generated per individual, of was completely absent in 57 clinically annotated which 94.1% of bases had ≥Q30 quality and over controls and 48 pediatric DCM cases. 98.3% of reads aligned to the . The average depth of coverage was 251x. An av- Metabolic Screening erage of 2345 exonic variants (range 2214-2645) On metabolic screening, the DCM patients was identified per individual, and 9383 were were VLCAD deficient and the unaffected nor- present in all four subjects. mal child and their mother had no evidence of Bioinformatics analysis detected a homozygous VLCAD deficiency. missense variant in exon 15 of ACADVL (chr17: 7127303; G>A; NM_000018.3; c.1349G>A; p.Ar- Protein Sequence and Structural Analysis g450His; rs118204016; depth: 109x) in the affected Secondary structure analysis was performed DCM individuals. This mutation resulted in a with the pair-wise aligned wild-type and mutant histidine to arginine substitution at codon 450 protein sequences. Differences in their secondary (p. R450H; ENST00000356839). The variant was structure were apparent, with loss of the helix and heterozygous in the mother and unaffected sister addition of a loop in the mutant protein (Figure (Figure 1A). This variant lies in the C-terminal do- 2). Energy minimization studies showed that the main of acyl co-A dehydrogenase, is not reported mutated protein demonstrated greater structur- in the 1000 genomes database, and has a minor al conformational changes than the wild-type allele frequency of 0.00004123 in the ExAC da- protein, with the superimposed view of both tabase. The variant was predicted to be “probably structures depicting the changes in the active damaging” by PolyPhen-2 (HumDiv and HumVar) site of the protein (Figure 3A; green protein and “damaging” by SIFT and Mutation Taster2. represents mutant, which dominates in the active Mutations were validated by Sanger sequenc- site, whereas the blue colored wild-type structure ing (Figure 1B). The p.R450H variant in ACADVL was masked). In the RMSD analysis, the wild- was in a highly conserved region (conservation type protein had an average RMSD of ~0.4 nm,

Figure 2. Secondary structure predicted by the pairwise aligned sequence of ACADVL. There is loss of the helix and addition of a loop in the mutant protein.

1715 S.J. Carlus, I.S. Almuzaini, M. Karthikeyan, L. Loganathan, G.S. Al-Harbi, et al.

A B

Figure 3. A, The superimposed image of wild-type (blue) and mutant (green) ACADVL protein structures. The mutant dom- inates in the active site, whereas the wild-type structure was masked. B, RMSD plot of the wild-type and mutant proteins for the 10 ns MD run. whereas the mutant structure had an RMSD of No conformational changes in bezafibrate ~0.5 nm (Figure 3B), while MD analysis demon- binding were observed, but the protein confor- strated structural changes over the 10 ns run with mation was altered. For L-carnitine, both the a 1.0 Å difference. ligand and protein residue conformation were The docking results highlighted differences in identified. Exactly how these drugs interact with the binding capacitates of wild-type and mutated ACADVL is yet to be studied. In this study, we proteins (Table II). In the wild-type protein, be- report the first evidence to support the molecular zafibrate had the better Glide XP score of -5.844 mechanism that ACADVL directly binds to these kcal/mol, with a good binding free energy of drugs. Overall, the R450H SNP has a deleterious -43.371 kcal/mol. L-carnitine had a Glide XP effect in the view of conformational drift seen in score of -4.952 and free energy of -20.457 kcal/ both protein and drug molecules. mol. A similar pattern was observed for the mutant protein. This difference was due to the R450H mu- tation, as represented in the 2-D interaction view Discussion of the docked complex (Figure 4). Bezafibrate formed hydrogen bonds with Ser223 and a pi-pi Saudi Arabia has one of the highest rates of interaction with Phe461. In the case of the mutant consanguinity in the world, as marriages between structure, bezafibrate showed similar pi-pi stack- first cousins and close relatives are common and ing but it formed hydrogen bonds with Ser251 and widely accepted15. Consanguinity is a risk factor Gly463. L-carnitine formed two hydrogen bond in some congenital abnormalities16. In Al-Madi- with Trp249 and Ser251 in the wild-type protein nah, in western Saudi Arabia, the prevalence of but three hydrogen bonds with Trp249, Thr217, DCM is very high in the pediatric heart patient and Leu216 in the mutant protein. population.

Table II. Molecular docking results with bezafibrate and L-carnitine and the interacting residue. Drug Glide XP score Glide energy MMGBSA Bind Interacting (kcal/mol) (kcal/mol) (kcal/mol) Residues Bezafibrate Wild-type -5.844 -42.956 -43.371 Ser223, Phe461 Mutant type -6.380 -44.124 -41.140 Ser251, Gly463, Phe214 L-Carnitine Wild-type -4.952 -24.253 -20.457 Ser251, Trp259 Mutant type -4.608 -23.955 -28.955 Trp249, Leu216, Thr217

1716 ACADVL mutation in pediatric familial dilated cardiomyopathy

Figure 4. 2-D interaction representation of drugs in the active site of wild-type and mutant ACADVL.

We exploited the information obtainable from revealed significant changes in the helix and loop a pedigree to investigate the genetic cause of regions of the mutated protein, and molecular familial DCM in two children, an unaffected docking and structural analyses validated the normal sibling and their mother from a consan- structural differences and conformational chang- guineous Saudi family, using a 181-gene targeted es between wild-type and mutated protein com- NGS approach to identify potential pathogenic plexes. Bezafibrate alters oxidation in mutations and genes responsible for inherited myopathy complications and L-carnitine in rhab- DCM. domyolysis, and both drugs showed significant The two DCM patients were diagnosed clin- changes in binding in silico due to the mutation, ically with severe heart failure and a DCM phe- shedding light on their mechanism of action. notype. Their father had died suddenly, appar- ACADVL encodes very long-chain acyl-CoA de- ently of heart disease. Due to a lack of clinical hydrogenase (VLCAD), an that catalyzes record availability and religious constraints, the the dehydrogenation of long-chain acyl-CoA es- pedigree was only available for two generations. ters of 12 to 18 carbons at the mitochondrial inner The affected individuals were found to have a membrane28. During prolonged fasting, infectious homozygous missense mutation in exon 15 of illnesses, or over-exercise, patients with VLCAD the ACADVL gene (chr17:7127303; G>A) that deficiency cannot meet their energy demands resulted in the amino acid substitution of histi- due to reduced ketone body production, which dine for arginine at codon 450 (p.R450R), which produces various clinical symptoms. VLCAD was confirmed by Sanger sequencing. Our in deficiency is an autosomal recessive disorder of silico studies provided new knowledge about the fatty acid oxidation roughly classified into three structural impact of the p.R450H mutation and clinical phenotypes: 1) early onset, cardiac type, drug binding behavior at the active site of the a severe form often presenting in the neonatal pe- mutated protein. Secondary structural analysis riod or early infancy characterized by cardiomy-

1717 S.J. Carlus, I.S. Almuzaini, M. Karthikeyan, L. Loganathan, G.S. Al-Harbi, et al. opathy, arrhythmias, liver dysfunction, and high The family pedigree was highly suggestive of mortality; 2) childhood onset, hypoglycemic type, a recessive pattern of inheritance. The p.R450H which is a milder form in which patients have re- variant identified in our study was homozygous peated episodes of hypoketotic hypoglycemia un- in affected DCM subjects and heterozygous in der hypercatabolic conditions such as long fasting, the mother and unaffected sister. This mutation infection, or diarrhea; and 3) late onset, myopathic was absent in 57 clinically annotated controls and type, another milder form which often presents 48 pediatric DCM cases. The variant lies in the during later childhood or adulthood and involves C-terminal domain of acyl co-A dehydrogenase muscular symptoms such as hypotonia, exercise is not reported in the 1000 genomes database, and intolerance, myalgia and rhabdomyolysis29. has a minor allele frequency of 0.00004123 in the The major long-chain fatty acids that accumu- ExAC database. late in VLCAD deficiency disrupt mitochondrial The mutation was in a highly conserved re- bioenergetics and Ca2+ homeostasis, acting as gion of the gene. Compound heterozygosity (dou- uncouplers and metabolic inhibitors of oxida- ble and triple mutations) has been reported to tive phosphorylation and inducers of MPT pore cause the hypertrophic cardiomyopathy (HCM) opening in the heart at pathologically relevant phenotype36. Our analysis revealed that none of concentrations. Disturbances in bioenergetics and these four individuals showed any disease-caus- Ca2+ homeostasis are therefore thought to con- ing mutations in the other 180 genes tested. The tribute to the cardiac manifestations observed in identified variant was completely absent in 57 VLCAD deficiency30. In animal models, VLCAD clinically annotated controls and 48 pediatric knockout (KO) mice recapitulate some features sporadic DCM cases. The ACADVL mutation of human VLCAD deficiency, with adult mice was highly likely to have caused the familial demonstrating cardiomyopathy with increased DCM phenotype in this family and be an uncom- numbers of degenerative fibers, collagen depo- mon cause of DCM. sition, vacuolated myocytes, and increased lipid The R450H mutation was first reported in a accumulation in cardiomyocytes31. 14-year-old Japanese girl with very mild manifes- VLCAD catalyzes the first step in the β-oxi- tations of VLCAD deficiency and compound het- dation spiral of fatty acid , a crucial erozygosity for two mutations in ACADVL: the pathway in cardiac energy production, and mo- R450H substitution and an A416T substitution. In lecular heterogeneity in very long-chain acyl- vitro functional expression studies showed that CoA dehydrogenase deficiency can cause auto- both mutant proteins retained residual activity at somal recessive pediatric cardiomyopathy and 30°C, concluding that the mild temperature-sen- sudden death. Mathur et al11 identified twenty-one sitive mutations resulted in the milder phenotype ACDAVL mutations in children, with 80% of in this patient37. mutations thought to cause infantile cardiomy- Ohashi et al38 reported a molecular analysis opathy. Infant cardiomyopathy can be either pri- of 13 Japanese VLCAD deficiency patients and mary or secondary; secondary causes include showed that the most prevalent mutation was mitochondrial disease, fatty acid oxidation de- a compound heterozygote p.R450H mutation. fects and Fabry’s disease. Energy production is Tajima et al39 identified the R450H mutation crucial for cardiac function and its deficiency in a Japanese VLCAD deficiency patient, also can cause cardiomyopathies. Mitochondrial fat- as a compound heterozygote, while Zhang et ty acid β-oxidation is an essential source of al40 reported seven Chinese VLCAD deficiency energy in infants after birth, and many studies patients with compound heterozygous p.R450H have reported that defects in β-oxidation of long accounting for 14.28% (2/14 alleles) of the total chain fatty acids can cause pediatric cardiomyop- mutations. Obaid et al41 reported the case of a athy32-35. An autosomal recessive inherited fatty three-year-old Saudi girl with VLCAD deficiency acid β-oxygenation disorder causes deficiencies and a compound heterozygote p.R450H mutation. in mitochondrial metabolism11, and inborn dis- p.R450H appears to be a relatively common mu- orders of fatty acid β-oxygenation and mito- tation in the Asian population. chondrial β-oxidation have been identified as a However, in these cases, cardiomyopathy was cause of different cardiomyopathies in infants32. completely absent. Here the variant was homo- β-oxidation enzyme deficiency in infants must zygous in both DCM subjects due to consan- be considered as a cause of the metabolic crisis, guinity. As noted above, DCM cases in Saudi cardiomyopathy, or sudden death34. Arabia often arise through autosomal recessive

1718 ACADVL mutation in pediatric familial dilated cardiomyopathy inheritance, with affected children more likely to 2) Raju H, Alberg C, Sagoo GS, Burton H, Behr ER. In- die as offspring of a consanguineous marriage14. herited cardiomyopathies. BMJ 2011; 343: d6966. Our results confirm previous data indicating that 3) Lipshultz SE, Sleeper LA, Towbin JA, Lowe AM, Orav ACADVL polymorphisms could lead to autoso- EJ, Cox GF, Lurie PR, McCoy KL, McDonald MA, Messere JE, Colan SD. mal recessive pediatric DCM and sudden death. The incidence of pediatric cardiomyopathy in two regions of the United The variant was present in the heterozygous States. N Engl J Med 2003; 348: 1647-1655. state in the mother and unaffected sister. Since 4) Towbin JA, Lowe AM, Colan SD, Sleeper LA, Orav there was a family history of sudden death from EJ, Clunie S, Messere J, Cox GF, Lurie PR, Hsu D, presumed heart disease, we advised clinical as- Canter C, Wilkinson JD, Lipshultz SE. Incidence, sessment of the unaffected child every 12 months. causes, and outcomes of dilated cardiomyopathy VLCAD deficiency can be reliably detected in children. JAMA 2006; 296: 1867-1876. through early metabolic screening but may be 5) Lipshultz SE, Cochran TR, Briston DA, Brown SR, Sambatakos PJ, Miller TL, Carrillo AA, Corcia L, San- missed if the child is asymptomatic and older chez JE, Diamond MB, Freunlich M, Harake D, Gayle than four days. Therefore, it is important to per- T, Harmon WG, Rusconi PG, Sandhu SK, Wilkinson form genetic analysis to avoid false-negative di- JD. Pediatric cardiomyopathies: causes, epidemi- agnoses of VLCAD deficiency in asymptomatic ology, clinical course, preventive strategies and newborn babies42. therapies. Future Cardiol 2013; 9: 817-848. 6) Parvari R, Levitas A. The mutations associated with dilated cardiomyopathy. Biochem Res Int 2012; Conclusions 2012: 639250. 7) Posafalvi A, Herkert JC, Sinke RJ, Van den Berg MP, Mogensen J, Jongbloed JD, Van Tintelen JP. Clini- Our data support the use of targeted NGS cal utility gene card for: dilated cardiomyopathy to identify the mutation(s) and novel gene(s) re- (CMD). Eur J Hum Genet 2013; 21: doi: 10.1038/ sponsible for familial DCM of unknown genetic ejhg.2012.276. cause, followed by metabolic screening for con- 8) Herkert JC, Abbott KM, Birnie E, Meems-Veldhuis firmation where possible. This is the first report MT, Boven LG, Benjamins M, du Marchie Sarvaas GJ, Barge-Schaapveld DQ, van Tintelen JP, van der Zwaag of the homozygous missense variation p.R450H PA, Vos YJ, Sinke RJ, van den Berg MP, van Langen IM, in ACADVL in association with a DCM pheno- Jongbloed JDH. Toward an effective exome-based type. Our findings suggest that p.R450H is asso- genetic testing strategy in pediatric dilated cardio- ciated with life-threatening, recessively inherited myopathy. Genet Med 2018; 20: 1374-1386. pediatric DCM. 9) Taylor MR, Carniel E, Mestroni L. Cardiomyopathy, familial dilated. Orphanet J Rare Dis 2006; 1: 27. 10) Carlus SJ, Al Harbi KM. Gene database for the Funding development of genetic testing for hypertrophic This research was supported by the Strategic Technologies cardiomyopathy. BMC Genom 2014; 15: P65. Programs of the National Plan for Science, Technology 11) Mathur A, Sims HF, Gopalakrishnan D, Gibson B, Rinal- and Innovation (MAARIFAH), Kingdom of Saudi Ara- do P, Vockley J, Hug G, Strauss AW. Molecular het- bia. Project No: 12-MED3174-05, through the Science and erogeneity in very-long-chain acyl-CoA dehydroge- Technology Unit (STU), Taibah University, Al Madinah Al nase deficiency causing pediatric cardiomyopathy Munawwarah, Kingdom of Saudi Arabia. and sudden death. Circulation 1999; 99: 1337-1343. 12) Aoyama T, Souri M, Ushikubo S, Kamijo T, Yamaguchi S, Kelley RI, Rhead WJ, Uetake K, Tanaka K, Hashimoto T. Conflict of Interests Purification of human very long-chain acyl-co- enzyme A dehydrogenase and characterization The authors declare that they have no conflict of interest. of its deficiency in seven patients. J Clin Invest 1995; 95: 2465-2473. 13) Andresen B, Olpin S, Poorthuis BJ, Scholte HR, Vian- eye-Saban C, Wanders R, Ijlst L, Morris A, Pourfarzam M, References Bartlett K, Baumgartner ER, deKlerk JB, Schroeder LD, Corydon TJ, Lund H, Winter V, Bross P, Bolund L, Gre- gersen N. Clear correlation of genotype with disease 1) Richardson P, McKenna W, Bristow M, Maisch B, Mautner B, O’Connell J, Olsen E, Thiene G, Good- phenotype in very-long-chain acyl-CoA dehydroge- nase deficiency. Am J Hum Genet 1999; 64: 479-494. win J, Gyarfas I, Martin I, Nordet P. Report of the 1995 World Health Organization/International 14) Al Jarallah AS, Al Abdulgader AA, Al Saadi AM, Society and Federation Of Cardiology Task Force Nasser AA, Zahraa JN. Outcome of dilated cardio- on the definition and classification of cardiomyop- myopathy (DCM) in Saudi children: a survey over athies. Circulation 1996; 93: 841-842. a decade. J Saudi Heart Assoc 2008; 20: 1-5.

1719 S.J. Carlus, I.S. Almuzaini, M. Karthikeyan, L. Loganathan, G.S. Al-Harbi, et al.

15) El-Hazmi M, al-Swailem AR, Warsy AS, al-Swailem AM, QM/MM, and free-energy simulations. J Comput Sulaimani R, al-Meshari AA. Consanguinity among Aided Mol Des 2016; 30: 707-730. the Saudi Arabian population. J Med Genet 1995; 27) Ashkenazy H, Abadi S, Martz E, Chay O, Mayrose I, 32: 23-26. Pupko T, Ben-Tal N. ConSurf 2016: an improved 16) Abdulrazzaq Y, Bener L, al-Gazali LI, al-Khayat AI, methodology to estimate and visualize evolu- Micallef R, Gaber T. A study of possible delete- tionary conservation in macromolecules. Nucleic rious effects of consanguinity. Clin Genet 1997; Acids Res 2016; 44: 344-350. 51: 167-173. 28) Izai K, Uchida Y, Orii T, Yamamoto S, Hashimoto T. 17) Mestroni L, Maisch B, McKenna WJ, Schwartz K, Novel fatty acid beta-oxidation in rat Charron P, Rocco C, Tesson F, Richter R, Wilke A, liver mitochondria. I. Purification and properties Komajda M. Guidelines for the study of familial di- of very-long-chain acyl-coenzyme A dehydroge- lated cardiomyopathies. Collaborative Research nase. J Biol Chem 1992; 267: 1027-1033. Group of the European Human and Capital Mo- 29) Gregersen N, Andresen BS, Corydon MJ, Corydon bility Project on familial dilated cardiomyopathy. TJ, Olsen RK, Bolund L, Bross P. Mutation analysis Eur Heart J 1999; 20: 93-102. in mitochondrial fatty acid oxidation defects: ex- 18) Geetha TS, Lingappa L, Jain AR, Govindan H, Mandloi N, emplified by acyl‐CoA dehydrogenase deficien- Murugan S, Gupta R, Vedam R. A novel splice variant cies, with special focus on genotype-phenotype in EMC1 is associated with cerebellar atrophy, visual relationship. Hum Mutat 2001; 18: 169-189. impairment, psychomotor retardation with epilepsy. 30) Cecatto C, Amaral AU, da Silva JC, Wajner A, Schimit Mol Genet Genomic Med 2018; 6: 282-228. MD, da Silva LH, Wajner SM, Zanatta Â, Castilho 19) McAndrew RP, Wang Y, Mohsen AW, He M, Vock- RF, Wajner M. Metabolite accumulation in VLCAD ley J, Kim JJ. Structural basis for substrate fatty deficiency markedly disrupts mitochondrial bio- acyl chain specificity crystal structure of human energetics and Ca2+ homeostasis in the heart. very-long-chain acyl-CoA dehydrogenase. J Biol FEBS J 2018; 285: 1437-1455. Chem 2008; 283: 9435-9443. 31) Exil VJ, Roberts RL, Sims H, McLaughlin JE, Malkin 20) Kearse M, Moir R, Wilson A, Stones-Havas S, Cheung RA, Gardner CD, Ni G, Rottman JN, Strauss AW. M, Sturrock S, Buxton S, Cooper A, Markowitz S, Very-long-chain acyl-coenzyme a dehydrogenase Duran C, Thierer T, Ashton B, Meintjes P, Drummond deficiency in mice. Circ Res 2003; 93: 448-455. A. Geneious basic: an integrated and extendable 32) Kelly DP, Strauss AW. Inherited cardiomyopathies. desktop software platform for the organization N Engl J Med 1994; 330: 913-919. and analysis of sequence data. Bioinformatics Wallace D. 2012; 28: 1647-1649. 33) Diseases of the mitochondrial DNA. Annu Rev Biochem 1992; 61: 1175-1212. 21) Muthusamy K, Chinnasamy S, Nagarajan S, Sivaraman Strauss AW, Powell CK, Hale DE, Anderson MM, T, Chinnasamy S. Isolation and characterization of 34) Ahuja A, Brackett JC, Sims HF. bioactive compounds of Clematis gouriana Roxb. Molecular basis of ex DC against snake venom phospholipase A2 human mitochondrial very long chain acyl-CoA (PLA2) computational and in vitro insights. J dehydrogenase deficiency causing cardiomyop- Biomol Struct Dynamics 2017; 35: 1936-1949. athy and sudden death in childhood. Proc Natl Acad Sci U S A 1995; 92: 10496-10500. 22) Gobin-Limballe S, Djouadi F, Aubey F, Olpin S, Andre- sen BS, Yamaguchi S, Mandel H, Fukao T, Ruiter JP, 35) Hale DE, Bennett MJ. Fatty acid oxidation disor- Wanders RJ, McAndrew R, Kim JJ, Bastin J. Genetic ders: a new class of metabolic diseases. J Pediatr basis for correction of very-long-chain acyl-coen- 1992; 121: 1-11. zyme A dehydrogenase deficiency by bezafibrate 36) Ingles J, Doolan A, Chiu C, Seidman J, Seidman C, in patient fibroblasts: toward a genotype-based Semsarian C. Compound and double mutations in therapy. Am J Hum Genet 2007; 81: 1133-1143. patients with hypertrophic cardiomyopathy: impli- 23) Watanabe K, Yamada K, Sameshima K, Yamaguchi S. cations for genetic testing and counselling. J Med Two siblings with very long-chain acyl-CoA de- Genet 2005; 42: e59. hydrogenase (VLCAD) deficiency suffered from 37) Fukao T, Watanabe H, Orii KE, Takahashi Y, Hirano A, rhabdomyolysis after l-carnitine supplementation. Kondo T, Yamaguchi S, Aoyama T, Kondo N. Myopa- Mol Genet Metab Rep 2018; 15: 121-123. thic form of very-long chain acyl-CoA dehydroge- 24) Gharaghani S, Khayamian T, Keshavarz F. Dock- nase deficiency: evidence for temperature-sen- ing, molecular dynamics simulation studies, and sitive mild mutations in both mutant alleles in a structure-based QSAR model on cytochrome Japanese girl. Pediatr Res 2001; 49: 227-231. P450 2A6 inhibitors. Structural Chem 2012; 23: 38) Ohashi Y, Hasegawa Y, Murayama K, Ogawa M, Hase- 341-350. gawa T, Kawai M, Sakata N, Yoshida K, Yarita H, Imai 25) Genheden S, Ryde U. The MM/PBSA and MM/GB- K, Kumagai I, Murakami K, Hasegawa H, Noguchi S, SA Methods to estimate ligand-binding affinities. Nonaka I, Yamaguchi S, Nishino I. A new diagnostic Expert Opin Drug Discov 2015; 10: 449-461. test for VLCAD deficiency using immunohisto- 26) Misini Ignjatović M, Caldararu O, Dong G, Mu- chemistry. Neurology 2004; 62: 2209-2213. noz-Gutierrez C, Adasme-Carreno F, Ryde U. Bind- 39) Tajima G, Sakura N, Shirao K, Okada S, Tsumura M, ing-affinity predictions of HSP90 in the D3R Nishimura Y, Ono H, Hasegawa Y, Hata I, Naito E, Grand Challenge 2015 with docking, MM/GBSA, Yamaguchi S, Shigematsu Y, Kobayashi M. Development

1720 ACADVL mutation in pediatric familial dilated cardiomyopathy

of a new enzymatic diagnosis method for very- 41) Obaid A, Nashabat M, Alfadhel M, Alasmari A, Al long-chain Acyl-CoA dehydrogenase deficiency by Mutairi F, Alswaid A, Faqeih E, Mushiba A, Alban- detecting 2-hexadecenoyl-CoA production and its yan M, Alalwan M, Marsden D, Eyaid W. Clinical, application in tandem mass spectrometry-based biochemical, and molecular features in 37 Saudi selective screening and newborn screening in patients with very long chain Acyl CoA Dehydro- Japan. Pediatr Res 2008; 64: 667-672. genase deficiency. JIMD Rep 2018; 40: 47-53. 40) Zhang RN1, Li YF, Qiu WJ, Ye J, Han LS, Zhang HW, 42) Boneh Al, Andresen BS, Gregersen N, Ibrahim M, Lin N, Gu XF. Clinical features and mutations in Tzanakos N, Peters H, Yaplito-Lee J, Pitt JJ. VLCAD seven Chinese patients with very long chain acyl- deficiency: pitfalls in newborn screening and con- CoA dehydrogenase. deficiency. World J Pediatr firmation of diagnosis by mutation analysis. Mol 2014; 10: 119-125. Genet Metab 2006; 88: 166-170.

1721