Functional study of DAND5 variant in patients with Congenital Heart Disease and laterality defects

Heart morphogenesis is a complex process involving multiple cell types (cardiac and non-cardiac) thus requiring a precise control of all molecular and cellular mechanisms [1]. Subtle deviations in heart development lead to Congenital Heart Diseases (CHDs), which are the most prevalent form of birth defect (postnatal incidence of 0.8% of all newborns) and the leading non-infectious cause of death in the first year of life [2, 3]. The etiology of CHD is incompletely understood, thought to be multifactorial, involving multiple genetic and environmental factors, and the possible interactions between those factors [4]. Nevertheless, in the last years, CHDs are seen as disorders resulting from alterations in genes that control basic embryonic processes between the third and eighth week of gestation, when the majority of the cardiovascular structures develop and when the integrity of the left-right (LR) axis is established [5, 6]. In fact, in the majority of cases presenting laterality disorders, a complex heart malformation is also observed, suggesting that CHDs could be due to a laterality defect in the formation of the heart [7]. In others words, even in children that born with CHDs and that do not present others extracardiac laterality abnormalities, the etiology of these CHDs could arise from subtle faults in the formation of the left-right axis.

The molecular control of LR axis development is mainly achieved by the Nodal signaling pathway [8]. Studies in mice revealed that during gastrulation, Nodal, a growth factor from the TGF-ß family, is asymmetrically expressed in the node, and this expression is expanded and amplified in the left-lateral plate mesoderm (L-LPM), but inhibited in the right-lateral plate mesoderm (R-LPM) [9]. Nodal itself induces its intracellular signaling by binding to type I (ALK4 and ALK7) and type II (ActRIIa and ActRIIb) serine-threonine kinase receptors in the presence of one EGF-CFC (epidermal growth factor-Cripto-FRL1-Cryptic) family co-receptor, Cripto or Cryptic. This leads to the phosphorylation of regulatory Smads (Smad2 and Smad3), their association with Smad4, translocation into the nucleus, and to the interaction with the transcription factor FoxH1, which at the end activates the expression of Nodal target genes, in particular, Pitx2 [9]. Pitx2 is a transcription factor that regulates the fate of cells that will form the visceral organs, including the heart [8, 10]. The impairment of the Nodal signaling in the lateral plate mesoderm, by variants in the genes involved in the pathway such as ActRIIB (MIM #602730) [11, 12], LEFTY 1 (MIM #603037) [13, 14], LEFTY 2 (MIM #601877) [13, 14], CRYPTIC/CFC1 (MIM #605194) [15, 16], CRIPTO/TDGF1 (MIM #187395) [17, 18], FOXH1 (MIM #603621) [19, 20], NODAL (MIM #601265) [21, 22] and PITX2C (MIM # 601542) [23, 24] had already been associated to laterality disorders and/or CHDs in humans. For example, the heterozygous variant p.G260R in NODAL was associated with transposition of the great arteries (TGA), atrial and ventricular septal defects, double outlet right ventricle (DORV) and/or heterotaxy [21]. On the other hand, the NODAL p.S60I variant has been associated with the tetralogy of Fallot (TOF) phenotype [22]. Additionally, several variations in Nodal co-receptor CFC1 were linked to TOF, TGA, DORV and/or laterality defects [15, 17, 25].

Recently, we reported that a DAN family member, the Nodal antagonist Cerberus-like 2 (Cerl2) – DAND5 in humans (MIM #609068; refseq - NM_152654.2) – controls Nodal signaling at the mouse node [26, 27], and the transmission of LR asymmetry information to the left-lateral plate mesoderm in a precise time window [28, 29]. Furthermore, the absence of Cerl2 in Cerl2 KO embryos leads to a range of laterality defects and/or cardiovascular malformations such as incomplete atrial and ventricular septation, TGA, DORV, randomized positioning of the cardiac apex, ventricular hypertrophy and/or heterotaxy of the abdominal organs, and a significant mortality rate within a few hours after birth is observed [30, 31].

Here, in order to advance the understanding of DAND5 in the genetic etiology of laterality disorders and associated CHDs, we conducted a genetic screening to identify DAND5 allelic variants in children affected with CHDs and/or LR asymmetry defects. We identified a DAND5 allelic variant in two patients, which was absent in the referral chromosomes from an ethnically matched control population that reduces the inhibitory activity of DAND5. We suggest that this variant might contribute as a risk allele for congenital heart disease and/or laterality defects.

DNA from 38 unrelated children with congenital heart defects arising from possible left-right defects and 40 ethnicity- and geographic-matched controls was direct sequenced for the detection of variants in the two exons of DAND5 gene. Sequence analysis of our patient’s cohort revealed a non-synonymous c.455G > A heterozygous DAND5 variant in two patients (Fig. 1a), not detected in in the control population. This alteration was also present has heterozygous in the apparently normal mother of patient 2. Unfortunately, the father of patient 2 and the parents of patient 1 were not available for sequencing. For validation, all exons were sequenced using forward and reverse internal primers flanking the missense variant. Clinically, patient 1 presented ventricular septal defect with overriding aorta, pulmonary atresia and left isomerism, while patient 2 presented overriding aorta, ventricular septal defect, right ventricular hypertrophy and pulmonary atresia (a case of extreme tetralogy of Fallot phenotype; Table 2). The c.455G > A variant in exon 2 of DAND5 results in a substitution of an Arginine residue for a Histidine (p.R152H) localized in a highly conserved region of the cysteine-rich DAN domain (Fig. 1b). Interestingly, the DAND5 p.R152H variant had been already annotated in the NCBI (http://​www.​ncbi.​nlm.​nih.​gov) and Ensemble (http://​www.​ensembl.​org) databases as Single Nucleotide Variation (SNV - rs45513495), with a minor allele frequency of ~1% (low-frequency variant; Table 3) and described in ExAC database with an allele frequency of 0.01043 in all populations but never reported to be associated with disease. To decide whether the detected missense mutation is pathogenic or not, we use in silico analysis tools (PolyPhen2, Mutation Taster, PROVEAN and I-Mutant 3.0) and the results estimated differing functional effect of this variant in DAND5 (Table 4). Using the up-to-date PolyPhen-2 algorithm (http://​genetics.​bwh.​harvard.​edu/​pph2/​index.​shtml), we observed that the DAND5 p.R152H variant had no effect on the DAND5 protein (classified as benign). Mutation Taster algorithm (www.​mutationtaster.​org) predicted that this variant is a polymorphism that might affect DAND5 protein features, namely alterations in the cysteine-rich domain. Moreover, PROVEAN software (http://​provean.​jcvi.​org/​) predicted a deleterious effect (score − 2.84) on the proper function of the DAND5 protein, and I-Mutant 3.0 predicted that the a.a. alteration will largely reduce the stability of the protein (a score of −1.28 a reliability index of 9; disease reliability index of 3). To clarify the function of this protein modification, we reproduced the p.R152H variant in an expression vector containing the human DAND5 cDNA, and we evaluated the in vitro effect of p.R152H in the regulation of Nodal activity using a well-established Nodal-dependent luciferase assay as readout. This assay consists of a luciferase reporter gene under the control of three activin-responsive elements promoter, pAR3-lux, which is transcriptionally activated by Nodal signaling [21]. In addition, we used the human version of the main intervenient genes, NODAL, CRYPTIC/CRIPTO, FOXH1, all necessary for a reliable transduction of the levels of Nodal signaling in this type of assay. The results, from triplicates of three independent experiments, showed that the p.R152H DAND5 variant causes a significant reduction in the normal NODAL-inhibitory activity of DAND5 (Fig. 2). Moreover, a dose-response at three different concentrations for the wild-type vs. mutated protein confirms the reduced NODAL-inhibitory activity of DAND5 (Additional file 1: Figure S1 and Additional file 2: Figure S2, being the two forms equally produced upon transfection, Additional file 3: Figure S3).

The DNA sequencing analysis of our 38 unrelated children and 40 control cohorts, revealed two patients, patient 1 and patient 2, displaying the same variant in DAND5 gene, not detected in the control population. One of the patients presented ventricular septal defect with overriding aorta, pulmonary atresia and left isomerism, while the other presented overriding aorta, ventricular septal defect, right ventricular hypertrophy and pulmonary atresia (a case of extreme tetralogy of Fallot phenotype). The phenotypically normal mother of patient 2 also presented heterozygosity the DAND5 c.455G > A variant, suggesting incomplete penetrance. No others variants were detected neither in the DAND5 gene of the 38 patients nor control population. This suggests that other variants in DAND5 might have a more severe or lethal phenotype, or that due to its small length DAND5 has less probability to randomly gain a new alteration.

The functional effect of the variant, which leads to a substitution of an arginine for a histidine in the position 152 of the highly conserved cysteine-rich DAN domain, was inconclusive in the predicted analysis performed with four different prediction algorithms. Nevertheless, these algorithms have only an accuracy of 65–80% in the prediction of known disease missense variants, being even less accurate when examining missense variants with milder effect [32]. Taking this in attention, we further evaluated the effect of the DAND5 variant using a Nodal-dependent luciferase assay. The results showed a substantial decreased in the function of the DAND5 variant protein when compared to the wild-type DAND5 protein. Since p.R152H is localized in the functional domain of the DAND5 protein and because DAND5 inhibits NODAL by direct binding [31], we hypothesize that the protein alteration caused p.R152H variation might interfere with the interaction efficiency of DAND5-NODAL. Therefore, we further extrapolate that, most probably, during the embryonic development of patients with p.R152H DAND5, which has a DAND5 molecule with reduced activity, the overall activity of NODAL on the node is above the proper levels. This abnormal NODAL activity on the node might be reflected then in excess of NODAL activity on the left-lateral plate mesoderm and/or an ectopic expression of NODAL and PITX2 on the right-lateral plate mesoderm of the embryo. Interestingly, the phenotypes observed in these two patients, are consistent with an ectopic expression of NODAL and PITX2 on the right side of the embryo. Tetralogy of Fallot is a malformation of the cardiac outflow tract (OFT) that comprises ventricular septal defect, overriding of the aortic valve, right ventricular outflow tract obstruction (pulmonary stenosis or in its most severe form, pulmonary atresia), and right ventricular hypertrophy [33]. The OFT forms during heart looping as a result of a coordinated process uncompressing proliferation, differentiation, and migration of the pre-cardiac cells of the secondary heart field (SHF). These cells are known to be sensitive to the asymmetrically left expression of the NODAL and PITX2 in the SHF [34]. Any imbalance in the LR patterning that might lead to an ectopic right-sided expression and loss-of-function of PITX2 results in reversed heart looping and abnormal shifting of the OFT [35, 36], suggesting that anomalous LR body patterning has the tendency to develop severe congenital heart defects. Moreover, cases of entirely bilateral expression of NODAL and PITX2 in the cells of both left and right lateral plate mesoderm lead to left isomerism phenotypes, in which both atria present a left morphology, commonly associated with atrioventricular septal defects [37]. In agreement, it has been observed that the absence of Cerl2 (DAND5 mouse homolog) in Cerl2 KO mice leads to several cardiovascular malformations, including incomplete atrial and ventricular septation, conotruncal defects, ventricular hypertrophy, and/or to laterality defects like left isomerism and thoracic heterotaxy or situs inversus [30, 31].

ExAC database describes DAND5 p.R152H as a SNV, detected in the heterozygous state in more than 1200 individuals, and present in 16 apparently normal homozygotes. Indeed, the mother of patient 2 is phenotypically normal but displays the DAND5 p.R152H variant in hererozygosity. Nevertheless, we cannot exclude the presence of a mild or undiagnosed heart defect on the mother of patient 2, or on the subjects of the genome projects, since detailed examination of these subjects is rarely available. In addition, the ExAC database and the American College of Medical Genetics and Genomics warns to the fact that in these population databases the presence of individuals with subclinical diseases among the supposed healthy individuals leading to minor discrepancies in some variant frequencies [32]. For example, in a study conducted in Newcastle, 30% of the patients (198/669 infants) were undiagnosed at the time of discharge from the hospital, and in 30/198 the diagnosis was only made after death [38]. Moreover, some phenotypes of CHD such as small atrial septal defects or ventricular defects may be asymptomatic and undetected throughout life [39]. Also, there are cases of apparently normal individuals whose laterality defects were only detected during adulthood [40]. Despite that, differences in the overall level of functional Nodal pathway might be explained by factors like differential allelic expressivity, an incomplete penetrance, the presence of genetic modifiers, and/or environmental factors. Interestingly, all Cerl2 heterozygous mice are phenotypically normal, and ~40% of the Cerl2 homozygous KO mice become normal adults, without any CHD and/or laterality defect [31]. Therefore, we are not in the presence of a Mendelian disorder. Moreover, the fact that 16 apparently normal homozygous individuals carry the DAND5 variant suggests incomplete penetrance and sustains the multifactorial etiology and complex trait of CHD. This indicates that the distribution and strength of the Cerl2 inhibitory signal, and ultimately the imbalance of the dose-sensitive Nodal signaling, might be sensitized by additional perturbations during LR patterning likewise it was observed in other genes of the NODAL signaling cascade [15, 17, 21]. For example, cryptic (CFC1 mouse homolog) full-mutant mice presents randomization of abdominal situs and complex cardiac malformations such as atrial and septal defects, dextrocardia, DORV and TGA, but a normal configuration of the OFT was seen in one mutant [41]. In humans, it was identified and classified as disease variant due to its loss-of-function effect, a heterozygous single-base-pair deletion (G174 del1) in CFC1 in two patients both with dextrocardia and TGA but one of the patients presented right isomerism, whereas the other patient presented left isomerism. This CFC1 variant was also reported in one subject with sporadic DORV, ventricular septal defect, aortic arch hypoplasia [15] and in a patient with tetralogy of Fallot [17]. Interestingly, likewise we observed in p.R152H DAND5 variant, the mothers of these CFC1 variant patients were found to carry the G174del1 alteration but none presented any clinical phenotype, suggesting incomplete penetrance. Accordingly, similar cases of reduced penetrance have been reported in CHD by others [25, 42].

Although the DAND5 p.R152H did not attest a specific genotype-phenotype link, our functional results support a model in which the proper levels of Nodal signaling, on both left- and right-lateral plate mesoderm, are particularly sensitive to gene dosage effects for all genes involved in the pathway. The imbalance in dosage-sensitive Nodal signaling is a final common way for laterality defects and associated CHDs.