Background Truncus arteriosus (TA) is characterised by failure of septation of the outflow tract into aortic and pulmonary trunks and is associated with high morbidity and mortality. Although ranked among the least common congenital heart defects, TA provides an excellent model for the role of individual genes in cardiac morphogenesis as exemplified by TBX1 deficiency caused by point mutations or, more commonly, hemizygosity as part of the 22q11.2 deletion syndrome. The latter genetic lesion, however, is only observed in a proportion of patients with TA, which suggests the presence of additional disease genes.
Objective To identify novel genes that cause Mendelian forms of TA.
Methods and results We exploited the occurrence of monogenic forms of TA in the Saudi population, which is characterised by high consanguinity, a feature conducive to the occurrence of Mendelian phenocopies of complex phenotypes as we and others have shown. Indeed, we demonstrate in two multiplex consanguineous families that we are able to map TA to regions of autozygosity in which whole-exome sequencing revealed homozygous truncating mutations in PRKD1 (encoding a kinase derepressor of MAF2) and NRP1 (encoding a coreceptor of vascular endothelial growth factor (VEGFA)). Previous work has demonstrated that Prkd1−/− is embryonic lethal and that its tissue-specific deletion results in abnormal heart remodelling, whereas Nrp1−/− develops TA. Surprisingly, molecular karyotyping to exclude 22q11.2 deletion syndrome in the replication cohort of 17 simplex TA cases revealed a de novo hemizygous deletion that encompasses PRDM1, deficiency of which also results in TA phenotype in mouse.
Conclusions Our results expand the repertoire of molecular lesions in chromatin remodelling and transcription factors that are implicated in the pathogenesis of congenital heart disease in humans and attest to the power of monogenic forms of congenital heart diseases as a complementary approach to dissect the genetics of these complex phenotypes.
- Congenital heart disease
- Mendelian form
- cardiac morphogenesis
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Congenital heart diseases (CHDs) are the most common birth defects in humans with an incidence of around 1% and are the leading cause of birth defect-related infant mortality.1 This remarkably high frequency reflects the highly complex nature of cardiac morphogenesis, which is inherently vulnerable to a wide range of lesions. Some of the well-studied environmental risk factors include maternal diabetes and infections and exposure to chemical teratogens such as vitamin A derivatives.2 Unlike environmental factors, which can be many steps removed from a conceivable molecular mechanism, genetic determinants of CHD offer a direct window into the molecular circuitry of cardiac morphogenesis and as such lend themselves to the ultimate quest of developing molecularly inspired therapeutic and preventative strategies.3
The prevailing theory on the genetic predisposition to CHD is one that is invoked to explain the genetic risk of many other birth defects, that is, multifactorial inheritance, which suggests that an aggregate of risk loci in the setting of a permissive environment leads to CHD after exceeding a certain threshold.4 ,5 This theory is supported by empirical recurrence risk data that, in the majority of families, are incompatible with single risk loci that are both sufficient and necessary to cause CHD.6 Delineating poorly penetrant risk loci has been a major challenge in CHD genetics research since these are expected by definition to be shared by the healthy controls. A genome-wide association study (GWAS) is an approach that has been attempted to tackle this challenge but its success has been limited in terms of the proportion of explained heritability and the medical actionability of the highlighted loci.7 ,8 Clearly, alternative/complementary approaches are needed if we are to gain a more comprehensive knowledge about the contribution of genes to the causation of CHD.
Unlike other complex phenotypes such as diabetes and coronary artery disease that are believed to follow the multifactorial inheritance paradigm, CHDs are associated with substantial reduction in reproductive fitness. Until recently, babies born with severe CHD seldom lived long enough to reproduce. The mutation-selection balance in classical genetics dictates that a phenotype that has a low reproductive fitness will be introduced largely by de novo mutations.9 This de novo paradigm has proven fruitful when applied to severe intellectual disability, another complex phenotype that rivals CHD in frequency if one accepts it to represent a birth defect and is also associated with very low reproductive fitness, where around 60% of cases can be traced to de novo mutational events.10 Likewise, there is now increasing appreciation of the role played by de novo mutations (both at the chromosomal and single bp level) in the causation of CHD.11–13
It should be borne in mind, however, that de novo mutations are not the only mechanism by which mutations evade negative selection as a result of phenotypic lethality. Recessive mutations, including those causing lethal phenotypes, can successfully propagate in the population, albeit at low frequency, because carriers do not typically have adverse consequences on their reproductive fitness. Identifying CHD disease genes that act recessively is challenging in outbred populations where the frequency of patients who are homozygous or compound heterozygous for these low-frequency mutations tends to be exceedingly rare on the one hand, and because detecting these mutations in carriers will lack the necessary phenotypic context on the other. Consanguinity, however, facilitates the homozygous occurrence of even lethal recessive mutations by virtue of autozygosity.14 Therefore, the study of consanguineous families will offer a unique opportunity to unravel novel recessive CHD disease genes. With this in mind, we set out to study two consanguineous families that segregate a very rare severe form of CHD known as truncus arteriosus (TA) as part of our ongoing ‘Mendelian Phenocopy Project’. We show that we were able to map each family to a homozygous truncating mutation in a novel gene whose role in TA pathogenesis is supported by available mouse models. Furthermore, in the process of searching for additional mutations in these two genes in a replication cohort, we encountered a de novo hemizygous deletion encompassing a third novel gene whose mouse model also manifests TA.
Materials and methods
TA was diagnosed based on typical appearance on echocardiography. One consanguineous family was referred from Security Forces Hospital, whereas the other was referred from Prince Sultan Military Medical City. The simplex cases of TA (replication cohort) were referred from Prince Sultan Military Medical City, King Abdulaziz Medical City and the UK. Patients, parents and available siblings were enrolled after signing a written informed consent as part of an institutional review board-approved research protocol (KFSHRC RAC#2121053), and venous blood was collected in EDTA-containing tubes for DNA extraction.
DNA was extracted from the venous blood samples of available members of two consanguineous multiplex families followed by genome-wide single nucleotide polymorphism (SNP) genotyping on the Axiom SNP Chip platform according to the manufacturer's protocol. Determination of the entire set of autozygous intervals per individual (autozygome) was performed using runs of homozygosity as surrogates of autozygosity as described before.15 Autozygous intervals that are shared by the affected members of each family were considered as the likely intervals where a causal homozygous recessive mutation resides.16
Full exome capture was performed using TruSeq Exome Enrichment kit (Illumina) following the manufacturer's protocol. Samples were prepared as an Illumina sequencing library, and in the second step, the sequencing libraries were enriched for the desired target using the Illumina Exome Enrichment protocol. The captured libraries were sequenced using Illumina HiSeq 2000 Sequencer. The reads were mapped against UCSC hg19 (http://genome.ucsc.edu/) by BWA (http://bio-bwa.sourceforge.net/). The SNPs and Indels were detected by SAMTOOLS (http://samtools.sourceforge.net/). The resulting variants were considered for candidacy according to their position within the gene (coding/splicing), allele frequency in single nucleotide polymorphism database (dbSNP) (<0.001), absent in local database of 549 Saudi exomes and position within the genome (only those within the critical autozygous interval per family were considered).17
Detection of CNVs at the DNA level (molecular karyotyping) was performed using the CytoScan HD Array platform according to the manufacturer's instructions. This is a high-density chip that contains 2.6 million copy number probes with a resolution of 25 kb. Chromosome Analysis Suite (ChAS) software was used to call duplications and deletions using the default parameters. All candidate duplications >500 kb and deletions >200 kb were further tested in the healthy parents to determine their de novo status.
Total protein was extracted from patient and control lymphoblasts. Anti-neuropilin 1 antibody (EPR3113) was purchased from Abcam (ab81321). The membrane was blocked with 5% milk powder in phosphate buffered saline with tween 20 (PBST) for 1 h at room temperature and incubated with the primary antibody (dilution 1/1000) overnight at 4°C followed by stringency washes and treatment with secondary antibody for signal detection.
Identification of consanguineous families with autosomal recessive forms of TA
We set out to identify multiplex consanguineous families in which the recurrence of TA may be attributed to a single causal mutation the recessiveness of which was unmasked by virtue of autozygosity. Two such families were recruited (family 1 and family 2) to test this hypothesis. The index in family 1 was born to a G7P6 mother and her first cousin husband. Pregnancy was complicated by polyhydramnios. Immediately after birth, multiple anomalies were noted in the form of short neck, cryptorchidism and a heart murmur necessitating neonatal intensive care unit admission where he was diagnosed with TA type I. Skeletal survey revealed the cause of the short neck as elevated scapulae. His karyotype was normal male. He died of hemodynamic instability at the age of 36 h. Family history is remarkable for two similarly affected siblings who died at the age of 10 days and 2 months (figure 1).
Index in family 2 is a 12-year-old Saudi girl who was born following an uneventful pregnancy to healthy half-first cousin parents. A heart murmur on her newborn examination prompted cardiology evaluation that showed cardiomegaly on chest X-ray and TA type I on echocardiography. She underwent uneventful surgical repair and has been asymptomatic since. Her most recent evaluation confirmed lack of dysmorphic or any other systemic manifestations. She has a 16-year-old sister with an identical history and six other siblings who are healthy (figures 1 and 3A–D).
Consistent with the hypothesis that TA in each of these two families was caused by a recessive mutation inherited as part of a common ancestral haplotype from the consanguineous parents, we have identified autozygous blocks shared by the affected members of each family (figure 1).
Identification of truncating mutations in NRP1 and PRKD1 in autosomal recessive TA
In order to identify the likely causal mutation within the candidate autozygous intervals in family 1 and family 2, we proceeded with whole-exome sequencing and only considered coding/splicing novel variants within the candidate intervals. Reassuringly, our filtering scheme highlighted a single variant in each family (figure 1). In family 1, six variants remained after applying the filtering steps comprising one splicing and five missense changes. In silico prediction of the pathogenicity of the missense changes was not compelling (see online supplementary table S1). However, the novel splicing variant affected a canonical donor splice site mutation in NRP1 (NM_003873.5:c.248+2T>G). Subsequent reverse transcription polymerase chain reaction (RT-PCR) sequence confirmed that this is indeed a truncating mutation that completely abolishes the donor site with resulting skipping of the whole exon (175 bp) predicting premature truncation of the protein (p.Asp25Gfs*25, figure 2B). Our access to patient lymphoblasts allowed us to further test the effect of this mutation at the protein level. Immunoblot analysis showed no detectable neuropilin 1 protein in the patient compared with controls (figure 2C). The novel variant in family 2 was a homozygous nonsense mutation in PRKD1 (NM_002742.2:c.1852C>T) that predicts premature truncation of the protein (p.R618*, figure 3).
Identification of de novo hemizygous deletion of PRDM1 in TA
In order to study the potential contribution of NRP1 and PRKD1 to the common sporadic form of TA, we attempted to fully sequence these two genes in simplex TA cases and were able to recruit 17 such cases as the basis of our ‘replication cohort’. Only 20% of the 17 cases are from consanguineous marriages. Since 22q11.2 deletion syndrome is seen in 40% of patients with TA,18 we performed molecular karyotyping on the six cases in which 22q11.2 had not be excluded molecularly prior to sequencing of NRP1 and PRKD1 (the other 11 had normal molecular karyotyping result prior to enrolment). Three cases with the classical breakpoints of 22q11.2 deletion syndrome were identified and excluded. Unexpectedly, our assay also identified a de novo microdeletion of 1654 kb on 6q21 (hg19, chr6:106 082 956–107 737 955) encompassing PRDM1 as well as seven other annotated genes (ATG5, AIM1, RTN4IP1, QRSL1, C6orf203, BEND3 and PDSS2; figure 4A). This patient presented in the immediate newborn period with a heart murmur. Echocardiography showed situs solitus, levocardia and TA type 2 with single large ventricular septal defect with over-riding aorta. The truncal valve was quadri-leaflet and dysplastic. Pulmonary artery branches arose separately from the ascending aorta. The aortic arch was left sided with normal branching pattern and no obstruction (figure 4B, C). In the remaining 13 simplex TA cases, no pathogenic variants in NRP1 or PRKD1 were identified.
TA rivals Ebstein anomaly as one of the least common CHD with incidence of <0.01%, accounting for 1% of all CHD.19 During cardiac morphogenesis, the common outflow tract of the primitive heart tube undergoes significant patterning that results eventually in its septation and the formation of two distinct muscular trunks for the aortic and pulmonary arteries.20 Our understanding of the molecular network that governs this process was greatly aided by the identification of 22q11.2 microdeletion syndrome as the single most common identifiable cause of TA, seen in 40% of cases.18 This microdeletion syndrome is characterised by the hemizygous deletion of a number of genes, but compelling evidence supports TBX1 as the most relevant gene for the TA phenotype of this multisystem syndrome.21 TBX1 is a major transcription factor in cardiac morphogenesis. One major role it plays is in the migration of neural crest cells (although these cells do not express TBX1) to the outflow tract, failure of which leads to complete failure of septation with resulting TA.22
Just like TBX1 discovery was a major milestone in the quest to understand the molecular pathogenesis of TA, we aimed in this study to identify other monogenic causes of this CHD. We hypothesised that our highly consanguineous Saudi population will facilitate the occurrence of autosomal recessive forms of CHD, which have rarely been reported in the literature, and that our approach of combined exome/autozygome analysis will reveal novel disease genes as it did for several other complex conditions.23 Indeed, our approach highlighted NRP1 and PRKD1 as such novel genes and we report the first truncating mutations in these genes in the context of TA.
NRP1 encodes neurophilin-1, a dual coreceptor that mediates the signalling of VEGFA and semaphorin when coupled to VEGFR2 and Plexin receptors, respectively.24 However, unlike semaphorin, NRP1 is necessary to mediate the signalling of VEGFA, which explains the nearly identical vascular phenotype observed in Vegfa−/− and Nrp1−/− mouse models.25 ,26 The early embryonic lethality in the Nrp1−/− prompted the engineering of an endothelial-specific knockout model (Nrp1 is highly expressed in the endothelium). The resulting conditional knockout mouse developed TA with 100% penetrance.27 Of note, a recent report suggests that plexin-D1 gene mutation in human may cause TA,28 further supporting the link we propose between TA and NRP1. Despite the apparent similarity between Tbx1-related and Nrp1-related TA, the mechanism appears to be different. In Tbx1−/− there is an arrest in the migration of neural crest cells that form the cushions and eventually mediate the septation of the outflow tract, in addition to reduction in the population of the second heart field cells, which results in impaired elongation and subsequent looping of the heart tube, essential patterning events for the successful septation of the outflow tract. In contrast, these cells appear normal in the conditional Nrp1−/− mouse but the cushions fail to form properly, perhaps reflective of a VEGFA-mediated mechanism that is distinct from TBX1.27 Interestingly, this mouse model also displays additional features of 22q11.2 microdeletion syndrome in humans such as hypoplastic thymus and cleft palate, prompting those authors to propose NRP1 as an attractive candidate in human patients with features of DiGeorge syndrome but lack 22q11.2.27 We note that our patient with homozygous truncation of NRP1 does indeed phenocopy the heart defect of DiGeorge syndrome but lacks the other features so it will be interesting to observe the spectrum of phenotypes associated with NRP1 mutations in future patients. Of note, a SNP in NRP1 (rs2228638) was found to significantly increase risk of another outflow tract anomaly (Tetralogy of Fallot) in a GWAS, raising the intriguing possibility that both rare and common variants in this gene have the potential to influence heart modelling during development.29
PRKD1 is a similarly compelling candidate in the pathogenesis of TA. It encodes a potent kinase that phosphorylates class II histone deacetylases (HDACs 4, 5, 7,and 9).7 Phosphorylation of HDAC relieves their negative regulation of the myocyte enhancer factor 2 (MEF2) class of transcription factors (MEF2A-D) and their target genes.30 In other words, PRKD1 depresses MEF2, a molecular switch that is relevant here since MEF2 family is expressed both in cardiogenic precursor cells and in differentiated cardiomyocytes during cardiac morphogenesis.31 More relevant to the TA phenotype in the family with homozygous PRKD1 truncation is the observation that Mef2c−/− mice have a fully penetrant aberrant heart tube looping with grossly abnormal outflow tract and defective cushion formation (with dramatic reduction in the expression of VEGF) leading to embryonic lethality by E10.32 ,33 Consistent with a non-redundant role of PRKD1 in derepressing MEF2, the Prkd1−/− mouse is also characterised by embryonic lethality. Unfortunately, the heart phenotype is in this mouse model is as yet uncharacterised, and the subsequent engineering of a tissue-specific deletion under the control of α-myosin heavy chain (α-MHC) may not have faithfully recapitulated the role of endogenous PRKD1.34 The TA phenotype in our patients with homozygous PRKD1 truncation argues for the early embryonic lethality in Prkd1−/− to be likely due to abnormal cardiogenesis, possibly similar to that seen in Mef2c−/−. Clearly, detailed phenotyping of Prkd1−/− and/or engineering a different conditional knockout will be needed to clarify this as will future patients with PRKD1 mutations.
Although our study specifically aimed to identify autosomal recessive monogenic causes of TA, we have serendipitously identified an interesting de novo event in the course of excluding 22q11.2 deletion in the replication cohort. Chromosomal aberrations are not infrequent in CHD with estimated 3%–10% (higher when CHD is part of a syndrome) harbouring likely pathogenic CNVs.35–39 The contribution of CNVs to TA is particularly impressive with estimated 40% of patients having detectable 22q11.2 deletion, which is consistent with our finding that three of six simplex TA cases in whom no molecular karyotyping had been performed had this classical microdeletion syndrome.18 Of 14 simplex TA cases in which 22q11.2 deletion was excluded, we have identified a de novo deletion involving seven annotated genes including PRDM1, which encodes PR domain-containing protein 1 (also known as BLIMP1). PRDM1 is a zinc finger protein that exerts a transcriptional repressor role through binding to DNA and recruiting such corepressors as histone methyltransferases and deacetylases.7 We believe that PRDM1 is the most compelling candidate for a haploinsufficient gene that caused TA in this patient in view of the expression profile of Prdm1 in the progenitors of the second heart field cells that give rise to the right ventricle and outflow tract.40 More importantly, while Prdm1−/− is embryonic lethal, conditional Sox2-Cre-driven deletion of Prdm1 results in a number of anomalies including TA, the same CHD we observe in the patient with haploinsufficiency for PRDM1 that we report here.41 Furthermore, a patient with TA was reported to have de novo deletion on 6q with the following coordinates 106 543 543–119 109 372.42 When combined with our deletion, we can determine the critical deletion interval for TA phenotype to be 106 543 543–10 773 7955, which still includes PRDM1 (among others). In addition, there are no heterozygous splicing/nonsense or frameshift variants reported in this gene in the Exome Variant Server (http://evs.gs.washington.edu/EVS/). Despite these various lines of evidence, we submit that future patients will be needed to unequivocally establish PRDM1 as a bona fide TA disease gene.
In summary, our positional mapping strategy of monogenic forms of TA led to the identification of two recessively acting and, serendipitously, one dominantly acting disease gene whose candidacy is strongly supported by animal model data. This study corroborates others in demonstrating the tremendous power of consanguineous populations in revealing monogenic forms of complex phenotypes. Although rare, such monogenic forms are invaluable to our pursuit of full molecular understanding of the pathogenesis of CHD and the ultimate quest to develop molecularly inspired therapeutic and preventative strategies.
We are grateful to the study families for their enthusiastic participation. We thank the staff at the Genotyping and Sequencing Core Facilities at KFSHRC for their technical help and our research coordinators (Mais Hashem and Firdous Abdulwahab) for their assistance.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online supplement
Twitter Follow Fowzan Alkuraya at @alkuraya, and Agaadir Almoisheer at @AgaadirA1988
Contributors RS and FSA: Collected and analysed data and wrote the manuscript. AH, MHA, MZS, SW, KD, BK, JG, SA, FMA, KA and AA: Collected and analysed data.
Funding This work is funded in part by KACST 13-BIO1113-20 (FSA), and a British Heart Foundation Personal Chair (BK).
Competing interests None.
Patient consent Obtained.
Ethics approval KFSHRC IRB.
Provenance and peer review Not commissioned; externally peer reviewed.