Evidence by Expression Analysis of Candidate Genes for Congenital Heart Defects in the NF1 Microdeletion Interval


*Correspondence author: Paola Riva PhD, Department of Biology and Genetics - Medical Faculty, University of Milan, Via Viotti 3/5, 20133 Milan – Italy, Tel. +39 02 50315862, Fax. +39 02 50315864. E-mail: paola.riva@unimi.it


It was recently reported that congenital heart disease is significantly more frequent in patients with NF1 microdeletion syndrome than in those with classical NF1. The outcome of congenital heart disease in this subset of patients is likely caused by the haploinsufficiency of gene/s in the deletion interval. Following in silico analysis of the deleted region, we found two genes known to be expressed in adult heart, the Joined to JAZF1 (SUZ12) and the Centaurin-alpha 2 (CENTA2) genes, and seven other genes with poorly defined patterns of expression and function. With the aim of defining their expression profiles in human fetal tissues (15th–21st weeks of gestation), expression analysis by RT-PCR and Northern blotting was performed. C17orf40, SUZ12 and CENTA2 were found to be mainly expressed in fetal heart, and following RT-PCR on mouse embryos and embryonic heart and brain at different stages of development, we found that the orthologous genes C17orf40, Suz12 and Centa2 are also expressed in early stages of development, before and during the formation of the four heart chambers.

The presence of binding sites for Nkx2-5, a transcription factor expressed early in heart development, in all three mouse orthologous genes was predicted by bioinformatics, thus reinforcing the hypothesis that these genes might be involved in heart development and may be plausible candidates for congenital heart disease.


Neurofibromatosis type 1 (NF1) microdeletion syndrome is a condition caused by haploinsufficiency of the NF1 gene and contiguous genes. Most NF1 microdeleted patients carry a heterozygous deletion in 17q11.2 of 1.5 Mb (Riva et al. 2000; Jenne et al. 2001), which is caused by unequal homologous recombination of NF1 repeats (REPs) (Dorschner et al. 2000; Venturin et al. 2004a). This syndrome, in comparison with classical NF1 (MIM *162200) (Huson et al. 1999), is often characterized by the presence of a more severe phenotype displaying facial dysmorphisms, learning disabilities, mental retardation and cardiovascular abnormalities (Tonsgard et al. 1997; Riva et al. 2000; Venturin et al. 2004b), together with a higher lifetime risk for the development of MPNSTs (De Raedt et al. 2003).

We have recently reported evidence showing that congenital cardiovascular malformations are significantly more frequent in NF1 patients with microdeletion syndrome than in those with classical NF1 (20%vs 2.1%) (Venturin et al. 2004b; Lin et al. 2000). The outcome of congenital heart disease in NF1 microdeleted patients is probably caused by the haploinsufficiency of one or more of the genes in the deletion interval. It has been hypothesised that reduced gene levels may lead to less severe forms of congenital heart disease that allow the survival of newborns and adults (Epstein, 2000). The 1.5 Mb NF1 microdeletion comprises twelve genes and, on the basis of the known function of a few of them, there is no evidence that they include candidate genes for cardiovascular malformation. Only two genes, the Joined to JAZF1 (SUZ12) and Centaurin-alpha 2 (CENTA2) genes, are expressed in adult heart (http://www.kazusa.or.jp/huge/gfpage/KIAA0160/) (Whitley et al. 2002), whereas the expression pattern and function of seven of the twelve is poorly defined.

In order to select the genes in the 17q11.2 deletion interval that are specifically transcribed during heart development, as plausible candidates for congenital heart disease due to haploinsufficiency, we investigated their expression pattern in human fetal tissues by means of RT-PCR and Northern blotting, and in mouse embryos before and during heart development by means of RT-PCR.

Materials and Methods


The patients affected by the NF1 microdeletion syndrome and congenital cardiovascular malformations (Table 2) included in this study have been previously described in the literature (Venturin et al. 2004b; Riva et al. 2000; Dorschner et al. 2000; Tonsgard et al. 1997) and fulfil the NIH diagnostic criteria.

Table 2.  Congenital heart defects in published NF1 patients with microdeletion
Patient IDCongenital heart defectReference
3Peripheral pulmonic stenosisTonsgard et al. (1997)
65Ventricular septal defect (upper part)Venturin et al. (2004)
72Patent ductus arteriosusVenturin et al. (2004)
93Hypertrophic cardiopathyVenturin et al. (2004)
116Mitral insufficiencyVenturin et al. (2004)
123-3Small atrial septal defectDorschner et al. (2000)
167-1Atrial septal defect, pulmonic stenosisDorschner et al. (2000)
169-1Dilated aortic valveDorschner et al. (2000)
183-1Pulmonic stenosisDorschner et al. (2000)
665Pulmonic stenosisRiva et al. (2000)
940Mitral valve prolapseRiva et al. (2000)
M.M.Pulmonic stenosisRiva et al. (2000)

Out of the 12 NF1 selected patients, ten were characterised by the typical 1.5 Mb deletion between the NF1 REPs (Venturin et al. 2004a; Riva et al. 2000; Dorschner et al. 2000); the remaining two, patients 3 and M.M., were found to carry a deletion involving the entire NF1 gene and flanking regions, but the deletion boundaries could not be precisely determined (Tonsgard et al. 1997; Riva et al. 2000).

The heart defects of the patients were pulmonic stenosis (5), atrial/ventricular septal defects (3), valve defects (3), hypertrophic cardiopathy (1) and patent ductus arteriosus (1). One patient had an atrial septal defect together with pulmonic stenosis.

Electronic Database Information

Information concerning gene expression patterns, the presence of specific functional domains in the protein products and their putative cellular role, and the existence of orthologous genes in model organisms, was obtained from the following websites: LocusLink (http://www.ncbi.nlm.nih.gov/LocusLink/), Ensembl Genome Browser (http://www.ensembl.org/) UCSC (http://genome.cse.ucsc.edu/), Human Unidentified Gene-Encoded Large Proteins Analyzed by Kazusa cDNA Project (HUGE) (http://www.kazusa.or.jp/huge/), SAGE (http://www.ncbi.nlm.nih.gov/SAGE/), BODYMAP (http://bodymap.ims.u-tokyo.ac.jp/) and, for the homologous murine sequences, Mouse Genome Informatics (http://www.informatics.jax.org/).

The functional domains of the proteins encoded by the studied genes were analysed using tools and links from the Expert Protein Analysis System (EXPASY) Molecular Biology Server (http://www.expasy.ch/).

The search for Nkx2-5 consensus binding sites was carried out using the MATCH (Matrix Search for Transcription Factor Binding Sites) web program (http://www.gene-regulation.com/cgi-bin/pub/programs/match/bin/match.cgi) with the minFP cut-off value that minimises false positive matches.

RT-PCR Analysis

The RNA analysis was carried out using human total RNAs isolated from fetal aorta, heart, skeletal muscle, liver and brain between the 16th and 21st weeks of gestation (Stratagene), human total RNAs from fetal aorta and heart at the 15th week of gestation (Bank of cells, tissues and DNAs, Telethon Project, Napoli), mouse RNAs extracted from eight pooled 7 and 8.5 dpc embryos using the RNAqueous-4PCR kit (Ambion), and mouse RNAs extracted from five pooled hearts and three pooled brains at 10.5, 12.5, and 14.5 dpc, using the TRIzol reagent (Life Technologies). All the mouse embryos, brains and hearts were pooled from individual animals. The time-mated CD1 mice were supplied by the Charles River Laboratory. The reverse transcriptions were carried out on 2.5 μg of total RNA using the ThermoScript RT-PCR System, and random primers for first-strand cDNA synthesis according to the instructions of the manufacturer (Invitrogen). The PCR primers were designed for different exons with each amplimer, thus making it possible to distinguish the cDNA-specific band from genomic PCR products; the primers used in this analysis are shown in Table 1. Five independent reverse transcriptions and gene-specific RT-PCR assays were set up for each analysed tissue, and the amplified products were tested by means of gel electrophoresis through 1.2% agarose. The PCR conditions were: 95°C for 5 min, 1 cycle and 95°C for 30 sec; specific annealing temperature for 30 sec and 72°C for 45 sec, for 35 cycles. The RT-PCR products were sequenced using the Big Dye Terminator kit (Applied Biosystem), and resolved on a 3100 ABI Prism Genetic Analyzer (Applied Biosystem).

Table 1.  Sequences of primers used for RT-PCR analysis

Northern Blot Analysis

Ten micrograms of RNA from each human fetal tissue were separated by means of gel electrophoresis through 1.2% agarose gel containing 2.2 M formamide in 10X MOPS buffer, according to standard procedures. The RNAs were transferred onto a Hybond membrane following the instructions of the manufacturer (Amersham), and hybridised using the RT-PCR-amplified segments as probes labelled by (α-32P)dCTP using the DNA Polymerase I Large (Klenow) Fragment Mini Kit (Promega). Following exposure of the membranes onto a Storage Phosphor Screen (Amersham Biosciences), the hybridisation pattern was obtained using a Typhoon 9200 Imager (Molecular Dynamics) and the intensity was volume quantitated using ImageQuant software (Molecular Dynamics). The relative expression profile for a given gene in each analysed tissue was determined by calculating the ratio between the volume of the gene and the volume obtained after β-actin hybridisation.


Expression Profile of 17q11.2 Genes in Human Fetal Tissues

We selected 12 patients with NF1 microdeletion syndrome and congenital cardiovascular malformations (Table 2). As 10 of them shared the same 1.5 Mb deletion between the NF1-REPs, we searched in this interval for candidate genes which, when haploinsufficient, may play a role in cardiovascular malformations.

In addition to the NF1 gene, the interval of interest includes 12 genes (Human July 2003 Assembly, http://genome.ucsc.edu/cgi-bin/hgGateway), of which we did not consider the oligodendrocyte-myelin glycoprotein (OMG) gene expressed in the central nervous system (Vourc'h et al. 2003), the cytokine receptor-like factor 3 (CRLF3) gene, which is expressed in the immune system (http://bioinfo.weizmann.ac.il/cards-bin/carddisp?CRLF3&search=crlf3&suff=txt) and the C17orf41 gene, which was only unambiguously mapped inside the deletion interval in the last version of the human genome assembly, available when selection of the genes for this study was already completed.

The remaining nine genes include SUZ12 (the human homologue of Suppressor of Zeste) (Birve et al. 2001), and CENTA2, which encodes a phosphatidylinositide-binding protein (Whitley et al. 2002), both of which are highly expressed in adult heart, although their expression pattern during embryonic and fetal development is still unknown in heart and other tissues. The functions and expression of the other genes (C17orf42, RNF135, EVI2B, EVI2A, RAB11FIP4, C17orf40 and LRRC37B) are poorly defined. We therefore used RT-PCR and Northern blotting analyses to study the expression of these nine genes in human fetal aorta, brain, heart, liver and skeletal muscle RNAs extracted between the 16th and the 21st weeks of gestation, and also used RT-PCR to study aorta and heart RNA extracted during the 15th week of gestation.

The RT-PCR results are summarised in Table 3. All of the genes were found to be expressed from the 16th to the 21st week in all tissues, with the exception of EVI2B (not expressed in aorta, heart, skeletal muscle and liver) and RAB11FIP4 (not detected in aorta). No specific transcripts of C17orf42, RNF135 and RAB11FIP4 were found in 15th week aorta or heart.

Table 3.  Expression pattern of human 17q11.2 genes by RT-PCR analysis
AortaHeartSkeletal muscle
18–19th wk
16th wk
19th wk
15th wk16–20th wk15th wk18–21st wk
  1. •expressed.

  2. ○not expressed.


Northern blotting analysis was used to estimate the expression of the above genes after normalisation by means of beta-actin hybridisation (Figure 1). Comparison of the expression profile of the same gene in all the analysed tissues revealed that SUZ12 is mainly expressed in the heart and LRRC37B in the aorta, with CENTA2 being mainly expressed in both, and C17orf40 in heart and skeletal muscle. RAB11FIP4 revealed two specific transcripts in brain, whereas the remaining genes (C17orf42, RNF135, EVI2A and EVI2B) did not reveal detectable hybridisation signals (data not shown).

Figure 1.

(A) Northern blot hybridisation of the five genes showing detectable transcripts in human fetal aorta (AO), brain (BR), heart (HE) liver (LI) and skeletal muscle (SM) (16th–21st weeks). The hybridisations with beta-actin are shown below the gene hybridisations of the same filter. Transcript sizes are indicated on the left. (B) Histograms showing the relative expression profile of each hybridisation after normalisation with beta-actin. The results are expressed in arbitrary units.

All of the transcripts of the analysed genes were of the size reported in the public NCBI database, with the exception of CENTA2, which had a transcript of 5.3 Kb instead of 1.8 Kb, and LRRC 37 B, for which two transcripts of 1.6 and 2.5 Kb were found instead of the reported 3 Kb transcript (Figure 1).

Expression Profile of Orthologous Genes in Mouse Embryo

As heart development is complete before the human gestation periods amenable to our analysis, and no human fetal tissues of earlier gestation periods were available, we determined the expression profile of the genes investigated in humans during heart embryogenesis in mouse. The known orthologous murine genes were investigated by means of RT-PCR analysis of 7 dpc and 8.5 dpc total embryos, and 10.5 dpc, 12.5 dpc and 14.5 dpc embryonic heart and brain. The Suz12, C17orf40, Rab11Fip4 and Centa2 genes were expressed in all of the analysed embryonic tissues, and there were faint bands of Evi2a, Evi2b and Rnf135, but not at all of the analysed embryonic stages (Figure 2).

Figure 2.

RT-PCR of seven orthologous murine genes in mRNA from total embryo (TOT), embryonic brain (BR) and embryonic heart (HE) at different dpc.

Given their early expression during heart development, Centa2, Suz12 and C17orf40 meet the minimal requirements for consideration as candidate genes for the outcome of congenital heart defects in NF1 microdeleted patients. As mutations in the NKX2-5 gene (which encodes a transcription factor) cause a pattern of cardiac defects similar to those seen in NF1 microdeleted patients (Schott et al. 1998), we verified whether these genes encode a NKX2-5-specific binding site. This binding site is known for mouse genes regulated by Nkx2-5, and so we used the MATCH tool (http://www.gene-regulation.com/cgi-bin/pub/programs/match/bin/match.cgi) to search for the consensus sequence (G/TYAAGTG) in a region of 9 Kb upstream of the three genes, and found one at −8637 bp from Centa2, −2960 bp from C17orf40 and −7567 bp from Suz12 (in all three cases with the maximum score of 1).


Patients with microdeletion syndromes often have congenital heart defects (Gelb, 2001), and we have recently reported that this is also significantly found in NF1 microdeleted patients (Venturin et al. 2004b; Riva et al. 2000; Dorschner et al. 2000; Tonsgard et al. 1997). The structural heart malformations in the NF1 deleted patients included pulmonic stenosis, atrial/ventricular septal defects, valve defects, hypertrophic cardiopathy and patent ductus arteriosus. The idea of candidate genes for heart defects was suggested by the fact that at least 10 out of our 12 patients had the same 1.5 Mb deletion interval (Venturin et al. 2004a; Riva et al. 2000; Dorschner et al. 2000). The heterogeneous nature of the above structural heart defects led to hypothesise that haploinsufficiency of one or more of the 17q11.2 genes might affect the heart development during both embryogenesis and fetal development.

We used RT-PCR to analyse the nine genes included in the deletion interval for which there is no detailed information concerning their developmental expression profiles and/or functions. The findings obtained in human fetal tissues at different gestational times indicated the expression of CENTA2, EVI2A, C17orf40, SUZ12 and LRRC37B in all of the analysed tissues throughout the considered gestation period (Table 1). The gene expression profile of 16th–21st weeks tissues, revealed by means of Northern blot analysis, indicated that SUZ12, CENTA2 and C17orf40 are mainly expressed in heart, and LRRC37B and CENTA2 in aorta. Specific heart expression has been reported only for SUZ12 in human adult tissue (http://www.kazusa.or.jp/huge/gfpage/KIAA0160/) and CENTA2 in rat (Whitley et al. 2002), whereas no data concerning human fetal heart (or adult heart for C17orf40) have been previously reported. Our findings suggest that these genes may play a role in heart formation during fetal development and growth. Northern blot experiments detected transcripts of sizes other than those reported by the sequence of cDNA clones specific for CENTA2 (5.3 Kb instead of 1.8 Kb), and LRRC37B for which two transcripts of 1.6 and 2.5 Kb were observed instead of the reported 3 Kb transcript (Figure 1). This study allowed us to identify the size of CENTA2 mRNA, which had previously only been predicted by sequencing cDNA clones. In the case of LRRC37B the 1.6 Kb mRNA may be related to the KIAA0563 transcripts, comprising 10 gene/pseudogenes including LRRC37B.

Northern hybridisation with a RAB11FIP4 probe revealed brain-specific expression of two different sized transcripts in fetal tissues: the smaller isoform may coincide with a testis cDNA clone containing a downstream translational initiation codon; the larger isoform probably coincides with the full-length cDNA (http://www.ncbi.nlm.nih.gov/entrez/). Specific transcripts for EVI2A, EVI2B, C17orf42 and RAB11FIP4 (in all tissues other than brain) could only be detected by RT-PCR, thus indicating very low expression levels in the tissues during the considered periods. Furthermore, EVI2B (which has a specific transcript only in heart and aorta at the 15th week), and C17orf42, RAB11FIP4 and RNF135 (which have specific transcripts in the 18th–21st weeks), seem to be expressed at different times during fetal development and growth.

The RT-PCR expression study of genes known to have mouse orthologues was extended to embryo developmental stages before and during heart development. C17orf40 and Suz12 had specific fragments in total embryo, heart and brain at all stages, whereas Centa2 was not present in 10.5 dpc brain. Rnf135, Evi2b, Evi2a and Rab11Fip4 did not have specific fragments at all of the stages analysed, and were not always found in different RT-PCR replications, thus indicating low levels of specific transcripts.

The expression gene profile obtained in this study shows that human C17orf40, SUZ12 and CENTA2 mRNAs have their highest relative expression in heart, and display specific transcripts in mouse embryos before and during heart development. These findings support their possible role in heart development.

The most functionally characterised of the three genes is SUZ12, the human homologue of the Drosophila Su(z)12 polycomb gene, encoding a protein containing a C2H2-type zinc finger domain (Birve et al. 2001), which interacts with polycomb proteins in Drosophila (Kuzmichev et al. 2002). The polycomb group includes repression proteins controlling the spatial expression of Hox genes (McGinnis & Krumlauf, 1992). Heterozygous flies carrying one Su(z)12 loss of function mutant allele show considerable misexpression of the Hox genes (particularly Ubx and Abd-B) needed for correct cardiogenesis (Lovato et al. 2002; Ponzielli et al. 2002; Lo et al. 2002). Less is known about the role of Hox genes in vertebrate cardiogenesis (probably because of the far more complex scenario with four paralogous Hox gene clusters), but it can be speculated that the human homologous SUZ12 gene may play a role in cardiac morphogenesis, and that its haploinsufficiency causes congenital cardiac malformations.

The CENTA2 gene encodes a phosphatidylinositide-binding protein that is predicted to contain a C4 zinc finger motif and two pleckstrin homology (PH) domains responsible for InsP4 binding; further studies are needed to assess the biological role of the CENTA2 product. The C17orf40 gene product contains multiple HAT repeat domains: it is highly conserved during evolution, but its function is still unknown.

An increasing number of genes have been identified as being required for cardiogenesis, as shown by the severe abnormalities in cardiac development caused by experiments in mouse and other model organisms (Epstein, 2000). About 150 induced mutations are known to lead to aberrant cardiac development in mice (http://research.bmn.com/mkmd), but pathogenetic mutations responsible for human congenital heart defects have so far been reported for only a few genes.

These include JAG1, which encodes a ligand for multiple Notch receptors (McCright et al. 2002). Haploinsufficiency for JAG1 causes Alagille syndrome (Oda et al. 1997), which affects the development of the heart, arteries and other organs, whereas missense mutations are responsible for isolated congenital heart defects (Krantz et al. 1999). It has been reported that the PTPN11 gene, which encodes the non-receptor protein tyrosine phosphatase SHP-2, plays a role in cardiomorphogenesis: missense mutations cause pulmonic stenosis, septal defects and hypertrophic cardiomyopathy (two of the typical diagnostic signs of Noonan syndrome) (Tartaglia et al. 2001). Likewise, point mutations of TBX5 (encoding a T-box transcription factor) have been observed in individuals with Holt-Oram syndrome, characterised by septal defects, conductional abnormalities and limb anomalies (Gruenauer-Kloevekorn & Froster, 2003). All of these genes act in the pathways involved in cardiac morphogenesis, and their mutations cause a syndromic phenotype. Mutations in the ELN gene specifically cause supravalvular aortic stenosis due to the generation of unfunctional truncated elastin, a structural protein, and haploinsufficiency of ELN, and other genes, gives rise to other cardiac defects in Williams syndrome (Morris & Mervis, 2000).

Only two genes causing non-syndromic congenital heart defects have been identified through genetic linkage analysis: GATA4, a zinc finger transcription factor, and NKX2-5, which encodes a homeobox transcriptional factor, are expressed early in heart and are essential for the regulation of septation during cardiac morphogenesis in embryo development (Lints et al. 1993; Brand, 2003). In some cases, the identified mutations predict binding impairment of the protein to target DNA leading to haploinsufficiency; in other cases, binding efficiency increases (Schott et al. 1998).

Two GATA4 mutations are associated with cardiac septal defects, and three NKX2-5 mutations have been found to cause pulmonic stenosis, atrial/ventricular septal defects, valve defects and hypertrophic cardiopathy, all of which are shown by our NF1 patients with microdeletion (Venturin et al. 2004b; Epstein, 2000; Garg et al. 2003). The wide spectrum of heart defects caused by NKX2-5 mutations may be explained by the possibility that this transcription factor regulates several genes. Functional studies of murine systems have demonstrated that Gata6 gene expression is regulated by Nkx2-5 through the binding of a known consensus sequence to the promoter (Molkentin et al. 2000).

Haploinsufficiency of the genes involved in the same pathway as Nkx2-5, or the genes that are directly or indirectly regulated by it, may cause heart defects. We sought and found the Nkx2-5 consensus sequence in the murine genes Suz12, Centa2 and C17orf40, which are all expressed early in the mouse embryo, before and during heart development. The consensus sequence in C17orf40 was found at the same distance from the active translation starting site as that in GATA6 (Molkentin et al. 2000).

It is worth noting that Nkx2-5 expression is maintained by the Phc1 gene which, like Suz12, belongs to the polycomb gene family (Shirai et al. 2002). This, together with the essential role of Su(z)12 (the orthologue of Suz12 in Drosophila) in the regulation of cardiogenesis and the early expression in human fetal and mouse embryonic heart, reinforce the possibility that SUZ12 is a candidate gene for congenital heart disease that deserves further investigation.

The variability of the congenital heart diseases reported in patients with NF1 microdeletion syndrome (Venturin et al. 2004b) or with NKX2-5 mutations (Epstein, 2000) suggests that other unlinked loci or environmental factors may play a role in modifying the disease phenotype, and the relatively low (20%) penetrance of cardiac malformations in NF1 microdeleted patients may depend on functional complementation through putative redundant pathways. A genetic background characterised by mutations/variations at additional loci may allow the disease phenotype to be expressed in NF1 patients with haploinsufficiency of the deleted genes.

In conclusion, this study provides data concerning the expression profiles of nine previously poorly characterised genes located in the NF1 microdeletion interval, and provides evidence that indicates three candidate genes for heart diseases that warrant further expression and functional studies.


The authors thank Silvia Moncini for her technical contribution, Bank of cells, tissues and DNAs (Telethon Project, Napoli) for providing RNAs from human fetal aorta and heart at 15th week of gestation. This work was supported by a 2003 grant of FIRST to PR.