3 DISCUSSION
The results obtained in this study showed that:
both somatic and germline mutations in GATA4 and NKX2.5 genes are uncommonly found in patients with CHD;
common genetic variants in the 3′UTR of GATA4 gene jointly interact to alter the disease susceptibility, likely by altering posttranscriptional microRNAs control;
haplotypes are fundamental to strengthen the genetic influence of disease manifestation;
a comprehensive panel of genes associated with the CHD phenotype,
assessed by high-throughput sequencing technology, may improve the ability to detect genetic mutations with high accuracy and cost- effectiveness manner.
Mutational screening in sporadic and familial CHDs and a comparison with previous results.
GATA4 and NKX2.5 are the most studied and well characterized cardiac transcription factors implicated in patients with congenital heart disease.
Several line of evidence suggested that GATA4 and NKX2.5 are co-expressed in the precardiac mesoderm during the earliest stages of its specification (Durocher et al., 1997; Brown et al.,2004). They are crucial to initiate cardiac differentiation and to activate the cardiac genetic program (Alison et al., 2005;
McCulley and Black, 2012).
Mutations in GATA4 and NKX2.5 genes have been related to various non- syndromic, isolated CHD, including familial and sporadic cases (Kodo et al., 2012; Kathiresan and Srivastava, 2012). However, detection frequency of these mutations is very low (Posch et al., 2008; Blue et al., 2012).
Therefore, other genes and molecular mechanisms are likely to be involved in the pathogenesis of CHDs. Interestingly, a high prevalence of somatic mutations in a number of transcription factors such as NKX2.5, TBX5, GATA4 and HAND1 has been reported in DNA extracted from an archive of formalin- fixed tissues stored for more than 20 years (Reamon-Buettner and Borlak, 2004a, 2004b, 2006; Reamon-Buettner et al., 2007). These mutations were primarily identified within malformed regions and not in unaffected regions taken from the same heart, suggesting a somatic hypothesis as a novel genetic mechanism for CHDs. Genetic studies in fresh frozen cardiac tissues were unable to confirm the results (Draus et al., 2009; Wang et al, 2011; Salazar et al., 2011; Esposito et al.,2011).
Nevertheless, the detection of somatic mutations in the connexin-40 and connexin-43 genes in sporadic atrial fibrillation strongly support that somatic or tissue-specific genetic variants can cause cardiac defects (Gollob et al., 2006;
Thibodeau et al., 2010). Similarly, a more recent study identified two novel mutations of GATA6 gene in malformed hearts tissue of CHD patients (Huang et al., 2013).
Our study showed no evidence of somatic NKX2.5 and GATA4 mutations in cardiac tissue of CHD patients. However, we cannot completely exclude that somatic mutations may rarely contribute to CHDs, especially given the low prevalence of NKX2.5 and GATA4 mutations in the CHD population (Posch et
al., 2008; Blue et al., 2012). Moreover, it is possible that low-level of somatic mosaicism can be undetected by using conventional Sanger-sequencing in small regions of affected tissue. More sensitive strategies, such as subcloning or next-generation sequencing, may be necessary to identify such mosaic somatic mutations.
Sequence variations in microRNA targets as genetic risks factors.
Evidence is accumulating about the essential role of the mRNAs 3’UTR in the regulation of gene expression. 3’UTR can specifically control the nuclear export, polyadenylation status, subcellular targeting and rates of translation and degradation of mRNA (Conne et al., 2000; Chen et al., 2008). Thus, modifications in this region may affect the expression of one or more genes and may contribute to the pathogenesis of different diseases. Accordingly, recent studies have shown that SNPs in the 3′UTR can affect miRNA gene regulation in different diseases (Cipollini et al., 2014; Haas et al., 2012). Variations in the 3′UTR may alter the sequence complementarity and the thermodynamics of miRNA/mRNA binding, creating or destroing binding sites in the seed region and resulting in gene target dysregulation (Chen et al., 2008; ).
In our study we found many known polymorphic variants in the 3′UTR of GATA4 gene with a good consonance between germline and somatic mutation sequences.
In silico analysis predicted that specific 3′UTR variants in GATA4 gene could be
potentially of biological relevance influencing miRNA–mRNA interaction. To experimentally confirm the in silico prediction, a functional analysis of the +1521 C>G variant, which showed the highest |ΔMFEtot|, was performed by a
luciferase reporter assay. The analysis revealed that +1521 C>G SNP in the 3′UTR of GATA4 gene affected the binding ability of miR-583, supporting a functional significance for this polymorphism. To date, there are no evidence about the role of the miR-583 in the congenital heart disease pathogenesis.
The clinical significance of the +1521 C>G SNP was investigated in a group of healthy newborns and affected children, which showed that the presence of the +1521 G variant allele was associated with a significantly decreased CHD risk.
This finding is intriguing and it might be due to the relatively low binding of the miRNA to target the sequence containing the G allele, thus resulting in higher GATA4 protein levels.
A similar result was obtained for the +1158 C>T polymorphism, which demonstrated a strong LD with the +1521 C>G SNP (D’=0.82). Thus, the two SNPs are presumably co-inherited 80% of the time. The logistic regression analysis showed that the T allele was associated with a reduced risk for CHD compared to the CC genotype.
LD and haplotype association testing.
Since the genotype of only one SNP can predict the genotype of another SNP depending on their allelic association, SNPs whose alleles are in strong LD with one another can be grouped into haplotype blocks. Haplotype blocks are chromosomal regions containing a group of alleles of different genes that are strongly linked enough to be inherited as a unit and which tend to be in strong linkage disequilibrium with one another. Many studies have examined haplotypes structure and frequency to localize disease-causing loci and to
identify the most common, becoming powerful predictors of the disease (Tobin et al., 2004; Crawford et al., 2005; Fallin et al., 2001; Matsa et al., 2013).
The linkage disequilibrium analysis confirmed that the four loci +1158 C>T, +1256 A>T, +1355 G>A,+1521 C>G, having the highest |ΔMFEtots|, were in LD (D’ ≥ 80 %). Hence, it can be stated that the +1256 A>T and +1355 G>A polymorphisms are in perfect association with either of the polymorphisms studied and could exhibit their effect synergistically in conjunction with the other SNPs, although their effects on phenotype as single biomarker could be negligible.
The haplotype analysis in this study clearly showed that the T-T-G-C haplotype was associated with a decreased risk. Conversely, the C-A-A-C haplotype resulted able to 4-fold increase the risk of CHD, ascertaining that these four GATA4 polymorphisms were synergistically involved in the etiopathogenesis of CHD in a haplotype specific manner.
Novel and rare variants detection by using a targeted resequencing approach.
Recent studies suggested that the mostly sporadic nature of non-syndormic CHDs might be explained by low-penetrance, rare mutations leading to haploinsufficiency of the encoded proteins (Dorn et al., 2014). High-throughput sequencing technologies, established during the recent years, offer novel opportunities to further study the genetic background of complex human diseases. This powerful approach allows the detection of novel and rare mutations, whose frequency is too low to be detected by previous standard technologies.
Our preliminary NGS study has allowed an early detection of novel mutations in key cardiac genes with a high standard in terms of coverage and quality. In order to explore the presence of mutations that might contribute to the CHD susceptibility, we investigated 17 genes, selected on scientific evidence for a causative role in the development of CHD.
Notably, a large part of mutations detected in our study population are located within the NOTCH1 gene, which encodes for a large protein containing an extracellular domain and regulates many developmental pathways. The first mutation in the NOTCH1 gene, detected in patient number 2, consisted in a proline-threonine amino acid substitution in position 480. This hydrophobic- hydrophilic change might explain the damaging impact on the protein conformation predicted by SIFT and PolyPhen2 web tools. The second one was identified in patient number 12 and consists in a cysteine-tyrosine substitution.
The latter was observed in a patient affected by TOF (number 14) and involved a cysteine-arginine amino acid substitution, causing an electrically change in the protein structure.
The NOTCH signaling pathway is well known to be associated with the bicuspid aortic valve (BAV), the most common form of CHD (Garg V et al., 2005).
Consistent with the fact that the most severe forms of aortic valve obstruction in children are associated with a left-right asymmetry and an abnormal grow of the left ventricle, mutations within the NOTCH1 gene have been reported in patients with laterality defects, such as Tetralogy of Fallot, aortic coarctation and hypoplastic left heart syndrome (Hinton et al., 2009; Greenway et al., 2009;
Freylikhman et al., 2014).
LEFTY2, as one of the important transforming growth factors in the Nodal/TGF- Lefty signaling pathway, is implicated in left-right asymmetry determination of organ systems during development and genetic evidence suggests that LEFTY2 acts as a feedback inhibitor of NODAL signaling (Deng et al., 2014).
Mice with targeted deletion of the Lefty2 asymmetric enhancer (which regulates left-right expression of Lefty2) show left isomerism, often associated with severe cardiac defects, such as transposition of the great arteries, double outlet of the right ventricle, TOF, a single (left) ventricle and a common atrioventricular canal (Meno et al., 2001; Mohapatra et al., 2009).
Consistent with this evidence, we found mutations in LEFTY2 gene in patients affected by ventricular septal defect and tetralogy of Fallot (number 2 and number 14, respectively). Patient number 2 showed a valine-glutamic acid substitution in position 57 in the LEFTY2 protein, while patient number 14 showed a substitution from a hydrophobic amino acid (proline) to a hydrophilic amino acid (serine), which might explain the predicted damaging impact on the protein structure.
Only one patient, affected by transposition of the great arteries, shows a mutation in the MYH6 gene (Tables 9), one of the most relevant gene involved in the cardiomyopathy development during the post-natal life. Specifically, the variant consisted in a isoleucine- phenylalanine amino acid substitution in position 324.
MYH6 encodes the alphamyosin heavy chain, a cardiac sarcomeric protein that
is part of the contractile unit of cardiovascular muscle and is expressed at high levels in the developing atria. Thus, MYH6 mutations are mainly associated with and atrial septal defects during embryogenesis (Posch et al., 2011).
Nevertheless, a more powerful study revealed functional variants within MYH6 gene correlated with other forms of CHD, like transposition of the great arteries, persistent truncus arteriosus, aortic stenosis, tricuspid atresia and persistence of foramen ovale (Granados-Riveron et al., 2010).
The only patient carried a novel variant in GDF1 gene is affected by tetralogy of Fallot. This variant led to an asparagine-serine amino acid substitution. GDF1 is a growth differentiation factor that belongs to the TGF-β superfamily and participates in the establishment and maintenance of left-right signals governing asymmetric organogenesis, including the heart and great vessel. Thus, mutations in GDF1 are likely to contribute to a spectrum of defects related to left-right axis formation, including right pulmonary isomerism and looping defects (transposition of the great arteries, TOF, interrupted aortic arch and double-out right ventricle) (Karkera JD et al., 2007).
Finally, the novel variant detected in patient number 13 involved a cysteine- serine substitution in the CFC1 gene. The CFC1 gene encodes Cryptic, which belongs to the EGF (epidermal growth factor)-CFC family of proteins that participate in early embryogenesis. The clinical phenotype associated with CFC1 mutations varied substantially and included defects typical of both left-
sided and right-sided isomerism, such as dextrocardia, double-out right ventricle, transposition of the great arteries, TOF and common atrioventricular canal anomalies (Goldmuntz et al., 2002; Selamet Tierney et al., 2007).
Further studies will be perform (by using Sanger sequencing or quantitative RT- PCR) in order to validate the association of detected mutations with the CHD risk in larger CHD and control populations. Nevertheless, it is evident that strong impact of NGS technologies in the discovery of mutations involved in
CHD. New high-throughput DNA sequencing technology represents a deeper and more reliable approach in detecting such common and rare variants with a low/intermediate penetrance.
4 CONCLUSIONS
Our data underline the great heterogeneity and pleiotropy among the genetic factors implicated in CHD. Indeed, both common variants in regulatory regions and novel mutations of genes with a major role in cardiogenesis could be implicated in the pathogenesis of CHD.
Further analyses in larger CHD and control populations are needed in order to confirm our results. Similarly, functional studies are required to analyze the true effects of each variant. Anyhow, the findings of this work highlighted novel molecular mechanism potentially implicated in the etiopathogenesis of CHD and open the research towards the development of new therapies able to reduce the incidence of CHD.
