The planar cell polarity (PCP) pathway is critical for proper embryonic development of the neural tube and heart. Mutations in these genes have previously been implicated in the pathogenesis of neural tube defects (NTDs), but not in congenital heart defects (CHDs) in humans. We systematically identified the mutation patterns of CELSR1–3, one family of the core PCP genes, in human cohorts composed of 352 individuals with NTDs, 412 with CHDs and matched controls. A total of 72 disease-specific, rare, novel, coding mutations were identified, of which 37 were identified in patients with CHDs and 36 in patients with NTDs. Most of these mutations differed between the two cohorts, because only one novel missense mutation in CELSR1 (c.2609G>A p.P870L) was identified in both NTD and CHD patients. Both in vivo and in vitro assays revealed that CELSR1 P870L is a gain-of-function mutation. It up-regulates not only the PCP pathway, but also canonical WNT signalling in cells, and also induces both NTDs and CHDs in zebrafish embryos. As almost equal numbers of mutations were identified in each cohort, our results provided the first evidence that mutations in CELSR genes are as likely to be associated with CHDs as with NTDs, although the specific mutations differ between the two cohorts. Such differences in mutation panels suggested that CELSRs [cadherin, EGF (epidermal growth factor), LAG (laminin A G-type repeat), seven-pass receptors)] might be regulated differently during the development of these two organ systems.

CLINICAL PERSPECTIVES

  • Our study provided first evidence that mutations in CELSR genes are as likely to be associated with CHDs as with NTDs. Our results accentuated the importance of genetic variants in PCP genes in the aetiology of both NTDs and CHDs.

  • Distinct mutation patterns between two cohorts suggest that different regulations/factors contribute to different phenotypes.

  • The novel missense mutation of CELSR1 (c.2609G>A p.P870L), which was identified in both NTD and CHD patients, is a gain-of-function mutation in vivo and in vitro. It could cause both NTD and CHD in zebrafish embryos and up-regulate not only PCP but also canonical WNT signalling in cells. Such gain-of-function mutation suggested that hyperactive CELSR1 could disrupt the development of both neural tube and heart.

  • Mutations compromise crosstalk of different developmental pathways will provide some insight on those severe birth defects.

INTRODUCTION

Birth defects are a major cause of morbidity and mortality in newborn infants worldwide [1]. Neural tube defects (NTDs) and congenital heart defects (CHDs) are among the most common human structural congenital malformations. NTDs, including craniorachischisis, anencephaly and spina bifida, are severe defects of the brain and spinal cord resulting from a failure of neural tube closure (NTC), and occur with a prevalence of around 1–2 infants per 1000 births [24]. CHDs are the most common congenital anomaly in newborn infants with a prevalence of approximately 8–9 per 1000 live births worldwide [5,6]. CHDs are complicated disorders including a large set of structural and functional deficits, the most obvious defects being ventricular septal defects (VSDs) and conotruncal defects [7]. Although NTDs and CHDs are different disorders, they share several genetic and developmental features. First, both classes of defects are severe disorders that occur mostly sporadically [8,9]. Second, both the neural tube and the heart tube arise during the first trimester, resulting in an embryologically critical period that partially overlaps for the development of NTDs and CHDs [7]. Third, folic acid supplementation has been shown to effectively prevent both defects [7]. Last, but not least, many key morphogenetic signalling pathways have been shown to be involved in both birth defects, including: Wnt, planar cell polarity (PCP), sonic hedgehog (SHH) and Notch pathways [3,10,11,12]. Most of the evidence for the involvement of these morphogenetic pathway genes came from animal models with knockout or mutated genes, e.g. a large forward screen using echocardiography on C57BL/6J fetal mice that had undergone chemical mutagenesis identified 34 cilia-related genes contributing to CHDs [13]. However, it is unclear how mutations in these pathways contribute differently to the development of these two congenital defects in human populations.

One of those key morphogenetic pathways, the PCP pathway, is important for the organization of cell sheets in the tissue plane, so it is essential for multiple developmental processes including: cellular orientation, directional migration and convergent extension during NTC and heart morphogenesis [1416]. The ‘core’ PCP genes include: VANGL1, VANGL2 (Van Gogh homologues), CELSR1, CELSR2, CELSR3 (Starry night homologues), DVL1, DVL2, DVL3 (Dishevelled homologues), PRICKLE1, PRICKLE2 (Prickle homologues) and ANKRD6 (Diego homologue), which are highly conserved from invertebrates to vertebrates [15,16]. Genetic modification of PCP genes, including Celsr1, Vangl2, Dvl1/2, Fz3/6 and Ptk7, in mouse models resulted in severe NTDs, such as craniorachischisis, implicating the essential role of core PCP genes in NTC [15,17]. The mouse models also revealed that PCP signalling is critical for heart development, because several PCP gene-knockout mouse models also expressed CHD phenotypes, including Vangl2, Dvl2/3, Ptk7, Fz1/2, Fz2/7 and Fuz [1821]. Those models provide evidence that PCP signalling is involved in establishing the early heart tube, polarizing cardiomyocytes, remodelling the cardiac outflow tract and playing crucial roles in the directed migration of cardiac crest cells [20,22,23].

Largely based on mouse model studies, we hypothesized that mutations in PCP genes could also be associated with human NTDs and CHDs. Our own work and that of others clearly demonstrated that alterations in PCP genes contribute to the aetiology of human NTDs. Mutations in PCP genes, including VANGL2 [24,25], SEC24B [26] and CELSR1 [2729], have been identified in patients with NTDs. However, there is no report of an association between PCP genes and an increased risk for CHDs in humans to date [20].

In the present study we focused on one set of the core PCP genes, the CELSRs [cadherin, epidermal growth factor (EGF), laminin A G-type repeats (LAG), seven-pass G-type receptors]. CELSRs are a special subgroup of adhesion G-protein-coupled receptors that function in contact-mediated homophilic communication. They were first discovered in Drosophila sp. as flamingo (fmi) for regulating PCP signalling [30], and three homologues named CELSR13 were identified in mammals. For the most part, CELSRs form haemophilic dimers on adjacent cells to mediate PCP signalling, but they also interact with other Wnt receptors such as Frizzled (FZ) and Vang(l)s [31,32].

Recent studies have supported rare and new mutations making a significant contribution to the population prevalence of severe diseases such as NTDs and CHDs, which have an impact on mortality and reproductive fitness [5,33,34], e.g. a large exome-sequencing study of 1213 CHD trios also showed that new protein-damaging mutations in genes, highly expressed in the developing heart and brain, are associated with both CHDs and neural developmental disorders [35]. Another whole-exome-sequencing study on syndromic and non-syndromic CHD (S-CHD and NS-CHD) patients not only strongly supported the finding that new protein-truncating variants (PTVs) in known CHD-associated genes are enriched in S-CHDs, but also showed that rare inherited PTVs from unaffected parents in CHD-associated genes contribute significantly to NS-CHD [36]. For a better understanding of the association between CELSR1–3 gene variants and the risk of NTDs and CHDs in human cohorts, we sequenced all exons of CELSR family genes (CELSR1, CELSR2, CELSR3) for an NTD cohort with 352 cases and 224 matched controls, and a CHD cohort with 412 cases and 213 matched controls. As we do not have biological samples from the parents in our cohorts to identify new mutations in the patients, we focused on those coding mutations that are rare, especially novel mutations not previously reported in 1000 Genomes Project, dbSNP or Exome Aggregation Consortium (ExAC) databases (east Asian). As a result, a total of 72 disease-specific, rare, novel, coding mutations were identified among the three CELSR genes in our cohorts, of which 42 are non-synonymous mutations that cause amino acid changes. It is interesting that almost equal numbers of novel coding mutations were identified in the CHD and NTD cohorts, with only one overlapping mutation, CELSR1 c.2609G>A p.P870L. Our in vitro and in vivo functional analyses showed that CELSR1 c.2609G>A p.P870L is a gain-of-function mutation, with the ability to increase not only PCP but also canonical Wnt signalling. Our results provided the first evidence that mutations of CELSRs are probably associated with CHDs, but with a different panel of mutations from those in NTDs. Such a difference in mutation panels could suggest that CELSRs might be regulated differently or have had different interaction partners during the development of the two organ systems.

MATERIALS AND METHODS

Human cases

Blood samples from 412 CHD patients (mean age 2.9±2.7 years, 55.6% male) were collected from the Cardiovascular Disease Institute of Jinan Military Command between March 2010 and March 2012. Sporadic CHD cases were diagnosed on the basis of echocardiography, with some diagnoses further confirmed surgically. Patients who had clinical features of developmental anomalies were excluded. Patients were also excluded if they had a positive family history of CHD in a first-degree relative, maternal diabetes mellitus, maternal exposure to known teratogens or any therapeutic drugs during gestation. All the CHD cases were classified according to previously described methodology (detailed diagnosis information on the patients is shown in Table 1) [37]. The 213 controls (mean age 7.1±3.7 years, 49.8% male) were ethnically and gender-matched, unrelated healthy volunteers recruited from the same geographical area (Table 1). Controls with any congenital anomalies or cardiac disease were excluded.

Table 1
Demographic characteristics in CHD cases and controls
VariableCase (%)Controls (%)
Sequencing group 412 213 
Region  Shandong Province 
Age: years (mean±S.D.) 2.9±2.7 7.1±3.7 
Gender   
Male [(no. (%)] 229 (55.6) 106 (49.8) 
Female [(no. (%)] 183 (44.4) 107 (50.2) 
CHD classification   
Septation defects [(no. (%)] 136 (33.0)  
Conotruncal defects [(no. (%)] 139 (33.7)  
RVOTO [(no. (%)] 44 (10.7)  
PDA defect [(no. (%)] 17 (4.1)  
LVOTO defects [(no. (%)] 19 (4.6)  
AVSD [(no. (%)] 20 (5.3)  
APVR defects [(no. (%)] 11 (2.7)  
Others [(no. (%)] 26 (6.3)  
VariableCase (%)Controls (%)
Sequencing group 412 213 
Region  Shandong Province 
Age: years (mean±S.D.) 2.9±2.7 7.1±3.7 
Gender   
Male [(no. (%)] 229 (55.6) 106 (49.8) 
Female [(no. (%)] 183 (44.4) 107 (50.2) 
CHD classification   
Septation defects [(no. (%)] 136 (33.0)  
Conotruncal defects [(no. (%)] 139 (33.7)  
RVOTO [(no. (%)] 44 (10.7)  
PDA defect [(no. (%)] 17 (4.1)  
LVOTO defects [(no. (%)] 19 (4.6)  
AVSD [(no. (%)] 20 (5.3)  
APVR defects [(no. (%)] 11 (2.7)  
Others [(no. (%)] 26 (6.3)  

APVR, anomalous pulmonary venous return; AVSD, atrioventricular septal defect; LVOTO, left ventricular outflow tract obstruction; PDA, patent ductus arteriosus; RVOTO, right ventricular outflow tract obstruction.

‘Others’ includes CHDs with other defects such as heterotaxy.

NTD samples were collected from either aborted fetuses or children with spina bifida aged <10 years from the 1990s to the 2010s. A total of 352 NTD samples (38.1% male, 51.4% female, 10.5% unknown) were collected from Shanxi Province (270), Liaoning (28), Heilongjiang (15), Suzhou (14) and Tianjin (25) (Table 2). Of the 352 cases 309 were tissue samples from aborted fetuses with severe rostral NTDs, such as craniorachischisis or anencephaly. The remaining 43 cases represented children aged <10 years with spina bifida, from whom blood samples were collected. The 224 controls (40.6% male, 58.9% female, 0.5% unknown) were unrelated healthy volunteers recruited from Shanxi Province (192 tissue samples of aborted fetuses from unrelated healthy volunteers) and Shanghai (32 blood samples from healthy college freshman students) who were ethnically and gender matched (Table 2).

Table 2
Demographic characteristics in NTD cohort
VariableCase (%)Controls (%)
Sequencing Group 352 224 
Region   
Tianjin 25  
Suzhou 14  
Liaoning 28  
Heilongjiang 15  
Shanxi 270 192 
Shanghai  32 
Age: weeks/years (mean ± SD) 
Tianjin 3.13 ± 3.4 y  
Jiangsu N.D.  
Liaoning 26.7 ± 8.1 w  
 3.2 ± 4.5 y  
Heilongjiang N.D.  
Shanxi 23.4 ± 6.4 w 20.0 ± 3.2 w 
Shanghai  18.0 y 
Gender 
Male 134(38.1%) 91(40.6%) 
Female 181(51.4%) 132(58.9%) 
Unknown 37(10.5%) 1(0.5%) 
CRS 26 (7.4%)  
EC 73 (20.7%)  
AE 74 (21.0%)  
EX 3 (0.85%)  
SB 255 (72.4%)  
NTD 1 (0.2%)  
VariableCase (%)Controls (%)
Sequencing Group 352 224 
Region   
Tianjin 25  
Suzhou 14  
Liaoning 28  
Heilongjiang 15  
Shanxi 270 192 
Shanghai  32 
Age: weeks/years (mean ± SD) 
Tianjin 3.13 ± 3.4 y  
Jiangsu N.D.  
Liaoning 26.7 ± 8.1 w  
 3.2 ± 4.5 y  
Heilongjiang N.D.  
Shanxi 23.4 ± 6.4 w 20.0 ± 3.2 w 
Shanghai  18.0 y 
Gender 
Male 134(38.1%) 91(40.6%) 
Female 181(51.4%) 132(58.9%) 
Unknown 37(10.5%) 1(0.5%) 
CRS 26 (7.4%)  
EC 73 (20.7%)  
AE 74 (21.0%)  
EX 3 (0.85%)  
SB 255 (72.4%)  
NTD 1 (0.2%)  

CRS, craniorachischisis; EC, encephalocele; AE, anencephaly; EX, exencephaly; SB, spina bifida

The studies described in the present paper were conducted in accordance with the Declaration of Helsinki. Protocols were reviewed and approved by the Ethics Committee of the School of Life Sciences, Fudan University and local ethics committees before the start of the present study. Written informed consent from the parents or guardians of the children was obtained.

Approximately 2 ml of peripheral blood or 50 μg of tissue samples was collected from each test case. Genomic DNA of each test case was isolated from peripheral blood or tissue samples, using conventional reagents, and quantified using a NanoDrop2000 (Thermo Scientific).

DNA sequencing, genotyping and data analysis

The genomic structures of human CELSR1–3 genes were determined using NCBI GenBank (NM_014246, NM_001408 and NM_001407). The 5′-UTR, 3′-UTR and coding regions in CELSR1–3 were detected by next-generation sequencing. Genomic DNA-fragment libraries and target enrichment using a probe specific for CELSR1–3 were performed using the Agilent SureSelect XT Custom enrichment system with some modifications, which was carried out at GBP Biotechnology. In brief, 1 μg of genomic DNA was subjected to ultrasonic energy fragmentation with the aid of a Covaris S2 instrument. The average size of genomic DNA fragments was approximately 250 bp. Each fragmented genomic DNA library was ligated with an indexed adapter, as previously described [38]. The ligated libraries were gel purified and, in turn, PCR amplified (Phusion, Thermo Scientific). The quality and quantity of the prepared libraries were assessed using an Agilent 2100 Bioanalyzer. Each of the 48 individual libraries was equivalently pooled and then hybridized to RNA library baits, and the targeted sequences were purified and amplified (Herculase II fusion, Stratagene). Sequencing was performed on an Illumina HiSeq2000 DNA sequencer (version 3). The fastq files that were obtained were mapped to a hg19 reference sequence using Burrows–Wheeler Aligner's Smith–Waterman Alignment (BWA-SW) [39], and the resulting SAM files were converted to BAM files, followed by sorting and removal of duplicates using SAMtools [40]. Combined variant calling was conducted on the resulting BAM files using SAMtools pileup, followed by VarScan [41]. Variants [single nucleotide variants (SNVs) and Indels] were annotated with information from Ensembl release 79 using the Variant Effect Predictor [42], based on the hg19/GRCh37 database.

To confirm the genotyping results from next-generation sequencing, 21 randomly selected, missense mutations in 28 CHD cases, and all exons of CELSR1 within 176 randomly selected NTD samples, were amplified by PCR and investigated using Sanger DNA sequencing. When compared with the next-generation data, the results were 98.04% concordant. The primers used in the PCR and DNA-sequencing reactions are listed in Supplementary Tables 3 and 4.

All confirmed, case-specific, coding, single nucleotide polymorphisms (SNPs) were further annotated as ‘known’ or ‘novel’, depending on whether they had been previously reported in dbSNP (version 137) (http://www.ncbi.nlm.nih.gov/projects/SNP/index.html), the 1000 Genomes Project [43] or ExAC (Cambridge, MA: http://exac.broadinstitute.org) [44]. All missense mutations were also evaluated using SIFT (sorting intolerant from tolerant: http://sift.jcvi.org/) [45,46] and PolyPhen-2 (Polymorphism Phenotyping v2: http://genetics.bwh.harvard.edu/pph/) for potentially damaging mutations [47].

Plasmids

Myc-tagged human CELSR1 was purchased from Origene Technologies (RC211652). We then performed the site-directed mutation CELSR1 P870L with primers (forward: 5′-GACAACGGCATCCTGCAGAAATCAGAC-3′; reverse: 5′-GTCTGATTTCTGCAGGATGCCGTTGTC-3′). The CELSR1 cDNA and CELSR1 P870L cDNA were subcloned into the pCMV6-AC-GFP vector between the SgfI and HindIII restriction sites. The mutation in the CELSR1 P870L plasmid was validated by DNA sequencing with primers 5′-ACAGGCCTGTC TTTCAGAGC-3′ and 5′-TGTACACGGCCACATTCTCC-3′. The pFR-luciferase reporter construct was obtained from Stratagene. The Topflash-luciferase reporter construct was a gift from Dr Tao P. Zhong at Fudan University.

Zebrafish injection

Studies were undertaken using AB strain zebrafish (Danio rerio) under standard conditions at 28°C. All the CELSR1 plasmids were diluted in RNA-free water. A mixture (2.3 nl) of 0.3 mM zcelsr1a-MO (MO: morpholino or antisense morpholino-modified oligonucleotides) and 0.3 mM zcelsr1b-MO (antisense to zebrafish celsr1a and celsr1b start-codon region) were injected into each one- to two-cell stage zebrafish embryo. Zebrafish celsr1 morpholinos were purchased from Gene Tools and the sequences are: zcelsr1a: 5′-ATGGTGTAAAACTCCGCAAACAGGC-3′; zcelsr1b: 5′-ATGGCTCACGCGCATCTCATTGGGA-3′. In the rescue (add-back) study, zcelsr1-MOs were co-injected with either CELSR1-WT or CELSR1-MUT, or empty vector plasmids into zebrafish embryos at the one- to two-cell stage. From each plasmid 2.3 nl was injected at a concentration of 25 ng/μl, 50 ng/μl or 100 ng/μl, resulting in injection of 57.5 pg, 115 pg or 230 pg at each concentration. In the over-expression study, each plasmid was individually injected into each one- to two-cell zebrafish embryo at the same dosage. Then 48 h post-injection, photos were taken of the harvested embryos and their morphology was grossly analysed. All statistical tests were two tailed with P<0.05 as the significance level.

Cell culture, transfection and dual-luciferase reporter assay

Wnt3a mouse fibroblast cell line (ATCC CRL-2647) was kindly provided by Dr Xiliang Zha at Fudan University. MDCK (Madin–Darby canine kidney) cells were seeded in 24-well cell culture plates and incubated until they reached 80% confluence. The cells from each well were transfected with 300 ng of Firefly-luciferase reporter (pFR or Topflash), 300 ng of CELSR1-WT/CELSR1-MUT/empty vector plasmids and 10 ng of Renilla-luciferase plasmid (pRLCMV) as an internal control, using Lipofectamine 3000 (Invitrogen). The transfected cells were cultured in high-glucose Dulbecco's modified Eagle medium (Gibco) with 5% FBS. For the pFR-luciferase assay, the cells were lysed and assayed for luciferase activity 36 h post-transfection. For the Topflash-luciferase assay, 24 h after transfection the cells were treated with either Wnt3a conditional medium or control medium for another 48 h, then they were lysed and assayed for luciferase activity. Both Renilla- and Firefly-luciferase activities were analysed using the dual-luciferase reporter assay system (Promega), as described previously [48]. Luciferase activity was corrected for transfection efficiency (pFR/pRLCMV or Topflash/pRLCMV) and extraction yield (via total protein assay). Four independent repeats were performed and all statistical tests were two tailed with P<0.05 as the significance level.

RESULTS

Identification of CELSR1–3 variants in two cohorts

A Chinese CHD and a Chinese NTD cohort were used in this study. Due to temporal and geographical differences, each cohort contained its own matched control group. The detailed information on our cohorts is provided in Tables 1 and 2 for the CHD and NTD cohorts, respectively. The 5′-UTR, 3′-UTR and coding regions of CELSR1–3 genes were sequenced in all samples and controls. In our cohorts, there are more male patients affected in the CHD cohort and more female patients in the NTD cohort (see Tables 1 and 2). Such significant gender-based differences in the prevalence of the malformations was not unexpected, because it has been observed in several previous reports in different populations [4951], suggesting that the gender effect in these two birth defects is not affected by population differences.

As both NTDs and CHDs are severe disorders that have an impact on mortality and reproductive fitness, a very large negative selection eliminates mutations from the human population. Therefore, it is most likely that rare and new mutations make a significant contribution to the population prevalence of these defects, especially for sporadic cases [5,33,34]. Based on this hypothesis, we filtered out the SNVs that have a major allele frequency (MAF) <1% as rare SNVs. There is a total of 184 and 181 rare SNVs in the CELSR1–3 genes identified in our CHD and NTD cohorts, respectively (Figure 1A). Within them, 103 and 95 are case-specific rare SNVs in the CHD and NTD cohorts, respectively (Figure 1A). Within these case-specific rare SNVs, there are 79 and 87 case-specific rare, coding mutations in CHD and NTD cohorts, respectively (Figure 1A).

Comparison of mutations identified in NTD and CHD cohorts

Figure 1
Comparison of mutations identified in NTD and CHD cohorts

(A) Four variant categories were identified for detailed analyses. These are SNVs with MAF <1%, novel SNVs, coding SNVs with MAF <1% and novel coding SNVs. The numbers of SNVs for each category that is identified in CHD or NTD cohorts are listed in the left column. Within each category, the numbers of case-specific SNVs are listed on the right column. (B) The non-synonymous and synonymous SNVs within novel/coding/case-specific SNVs in two cohorts. (C) Illustration of CELSR1–3 protein structure with all novel/case-specific/(missense+frameshift) SNVs identified in NTD and CHD cohorts. Missense SNVs identified in NTD or CHD patients are labelled in black and red letters, respectively.

Figure 1
Comparison of mutations identified in NTD and CHD cohorts

(A) Four variant categories were identified for detailed analyses. These are SNVs with MAF <1%, novel SNVs, coding SNVs with MAF <1% and novel coding SNVs. The numbers of SNVs for each category that is identified in CHD or NTD cohorts are listed in the left column. Within each category, the numbers of case-specific SNVs are listed on the right column. (B) The non-synonymous and synonymous SNVs within novel/coding/case-specific SNVs in two cohorts. (C) Illustration of CELSR1–3 protein structure with all novel/case-specific/(missense+frameshift) SNVs identified in NTD and CHD cohorts. Missense SNVs identified in NTD or CHD patients are labelled in black and red letters, respectively.

We further filtered out all the novel coding SNVs that did not exist in the dbSNP137, 1000 Genomes or ExAC (eastern Asian only) databases, leaving 57 and 59 variants that were identified in the CHD and NTD cohorts, respectively. Of these, 37 in CHDs and 36 in NTDs are case-specific novel, coding variants (Figure 1A). Within these case-specific novel, coding variants, 21 missense mutations and 16 synonymous mutations were identified in the CHD cohort (Figures 1B and 1C, Table 3 and see Supplementary Table S1), although 22 missense mutations and 14 synonymous mutations were identified in the NTD cohort (Figures 1B and 1C, Table 4 and see Supplementary Table S2). Through all the filtering processes, there are comparable numbers of SNVs identified in the CELSR1–3 genes in each category between our CHD and NTD cohorts. It is interesting that most of those novel mutations differ between CHD and NTD cohorts. Only one case-specific novel SNV was shared by the two cohorts, and this is a missense mutation in CELSR1 (CELSR1 c.2609G>A p.P870L) (Figure 1B).

Table 3
Novel non-synonymous mutations of CELSR1-3 detected in CHDs but not in controls

aMAF from ExAC database. bShared variants between CHD and NTD cohorts.

NucleotideAmino acidCaseMAF inMAF in east
Genechangechangeno.DomainSIFTPolyPhenSexPhenotypeCHDAsiana
CELSR1 c.1795G>A p.Arg599Trp Cadherin_repeat 0.01 0.826 VSD, ASD, PDA, PFO 0.000084 Not reported 
CELSR1 c.2609G>A p.Pro870Leub 1 Cadherin_repeat 0 1 M PMVSD, PFO 0.00089 
CELSR1 c.2644T>C p.Ile882Val Cadherin_repeat 0.032 ASD 0.00161 Not reported 
       ASD, PDA   
CELSR1 c.4834C>A p.Val1612Leu Laminin G 0.01 0.959 DORV, PS, MPVSD 0.000828 Not reported 
CELSR1 c.5363C>T p.Ser1788Asn Laminin G 0.43 0.02 TOF 0.000816 Not reported 
CELSR1 c.6337T>C p.Arg2113Gly HormR 0.06 VSD, PFO, PH 0.000808 Not reported 
CELSR1 c.6902G>A p.Pro2301Leu 0.38 0.01 VSD, ASD 0.00289 
       TOF, ASD, PDA   
CELSR1 c.7801C>T p.Asp2601Asn 7TM 0.33 TOF, PA 0.000821 
CELSR1 c.8252A>G p.Leu2751Pro Cytotail 0.06 0.967 TECD 0.003632 Not reported 
       VSD, ASD   
       DORV, PS   
CELSR2 c.1637G>A p.Gly546Glu Cadherin_repeat 0.25 0.003 PDA, MI 0.000805 Not reported 
CELSR2 c.1850T>C p.Val617Ala Cadherin_repeat 0.19 0.003 PMVSD, PFO 0.000808 Not reported 
CELSR2 c.3145G>A p.Val1049Ile Cadherin_repeat 0.8 0.008 TECD 0.000806 Not reported 
CELSR2 c.3650C>T p.Thr1217Met 0.08 0.073 AS 0.001658 
       TA, SV, PH   
CELSR2 c.5275C>T p.Arg1759Trp Laminin G 0.01 PS, ASD 0.000885 
CELSR2 c.6064C>T p.Leu2022Phe HormR 0.02 Ebstein's anomaly, TI 0.001613 
       VSD, SASD, PDA   
CELSR3 c.2222C>A p.Ser741Ile Cadherin_repeat 0.05 0.998 TOF 0.000809 Not reported 
CELSR3 c.5132G>T p.Ala1711Glu Laminin G 0.11 0.991 VSD 0.000821 Not reported 
CELSR3 c.5800G>T p.Pro1934Thr Laminin G 0.31 0.23 TOF 0.001385 Not reported 
CELSR3 c.6269G>A p.Ala2090Val EGF_Lam 0.03 0.627 TAPVC, ASD 0.000911 Not reported 
CELSR3 c.8512G>A p.Arg2838Trp Cytotail 0.02 0.999 VSD, ASD, PDA, PFO 0.00098 Not reported 
CELSR3 c.8537C>T p.Arg2846Gln Cytotail 0.13 0.453 PMVSD  
NucleotideAmino acidCaseMAF inMAF in east
Genechangechangeno.DomainSIFTPolyPhenSexPhenotypeCHDAsiana
CELSR1 c.1795G>A p.Arg599Trp Cadherin_repeat 0.01 0.826 VSD, ASD, PDA, PFO 0.000084 Not reported 
CELSR1 c.2609G>A p.Pro870Leub 1 Cadherin_repeat 0 1 M PMVSD, PFO 0.00089 
CELSR1 c.2644T>C p.Ile882Val Cadherin_repeat 0.032 ASD 0.00161 Not reported 
       ASD, PDA   
CELSR1 c.4834C>A p.Val1612Leu Laminin G 0.01 0.959 DORV, PS, MPVSD 0.000828 Not reported 
CELSR1 c.5363C>T p.Ser1788Asn Laminin G 0.43 0.02 TOF 0.000816 Not reported 
CELSR1 c.6337T>C p.Arg2113Gly HormR 0.06 VSD, PFO, PH 0.000808 Not reported 
CELSR1 c.6902G>A p.Pro2301Leu 0.38 0.01 VSD, ASD 0.00289 
       TOF, ASD, PDA   
CELSR1 c.7801C>T p.Asp2601Asn 7TM 0.33 TOF, PA 0.000821 
CELSR1 c.8252A>G p.Leu2751Pro Cytotail 0.06 0.967 TECD 0.003632 Not reported 
       VSD, ASD   
       DORV, PS   
CELSR2 c.1637G>A p.Gly546Glu Cadherin_repeat 0.25 0.003 PDA, MI 0.000805 Not reported 
CELSR2 c.1850T>C p.Val617Ala Cadherin_repeat 0.19 0.003 PMVSD, PFO 0.000808 Not reported 
CELSR2 c.3145G>A p.Val1049Ile Cadherin_repeat 0.8 0.008 TECD 0.000806 Not reported 
CELSR2 c.3650C>T p.Thr1217Met 0.08 0.073 AS 0.001658 
       TA, SV, PH   
CELSR2 c.5275C>T p.Arg1759Trp Laminin G 0.01 PS, ASD 0.000885 
CELSR2 c.6064C>T p.Leu2022Phe HormR 0.02 Ebstein's anomaly, TI 0.001613 
       VSD, SASD, PDA   
CELSR3 c.2222C>A p.Ser741Ile Cadherin_repeat 0.05 0.998 TOF 0.000809 Not reported 
CELSR3 c.5132G>T p.Ala1711Glu Laminin G 0.11 0.991 VSD 0.000821 Not reported 
CELSR3 c.5800G>T p.Pro1934Thr Laminin G 0.31 0.23 TOF 0.001385 Not reported 
CELSR3 c.6269G>A p.Ala2090Val EGF_Lam 0.03 0.627 TAPVC, ASD 0.000911 Not reported 
CELSR3 c.8512G>A p.Arg2838Trp Cytotail 0.02 0.999 VSD, ASD, PDA, PFO 0.00098 Not reported 
CELSR3 c.8537C>T p.Arg2846Gln Cytotail 0.13 0.453 PMVSD  

AS, aortic stenosis; ASD, atrial septal defect; DORV, double outlet right ventricle; MI, mitral insufficiency; PA, pulmonary atresia; PDA, patent ductus arteriosus; PMVSD, perimembranous VSD; PAPVD, partial anomalous pulmonary venous drainage; PECD, partial endocardial defect; PH, pulmonary hypertension; PS, pulmonary stenosis; MPVSD, muscular portion VSD; SASD, secundum atrial septal defect; SV, single ventricle; TA, tricuspid atresia; TAPVC, total anomalous pulmonary venous connection; TECD, transitional endocardial defect; TI, tricuspid incompetence; TOF, tetralogy of Fallot.

The Bold is the mutant identified in both cohorts.

CELSR1 P870L is a gain-of-function mutation in zebrafish

The only case-specific novel, coding SNV identified in both diseases is CELSR1 c.2609G>A p.P870L, which is located in the sixth cadherin tandem repeat domain. It was identified in a 17-week-old female fetus, with both anencephaly and myelomeningocele, and a 3-year-old CHD patient who presented with a patent foramen ovale (PFO) and a VSD (Tables 3 and 4).

Table 4
Novel non-synonymous mutations of CELSR1–3 detected in NTDs but not in controls
NucleotideAmino acidCaseMAF inMAF in east
Genechangechangeno.DomainSIFTPolyPhenSexPhenotypeNTDsAsiana
CELSR1 c.2609G>A p.Pro870Leub 1 Cadherin_repeat 0 1 F CRS 0.00089 0 
CELSR1 c.3562T>A p.Thr1188Ser Cadherin_repeat 0.41 0.036 SB 0.00098 Not reported 
CELSR1 c.4222C>T p.Ala1408Thr EGF_CA 0.65 0.101 SB 0.00134 Not reported 
CELSR1 c.4969G>A p.Arg1657Trp EGF_CA 0.09 0.002 ND SB 0.00019 
CELSR1 c.5504C>T p.Arg1835His 0.34 0.002 AE, SB 0.00092 
CELSR2 c.980G>A p.Arg327His Cadherin_repeat 0.31 0.057 SB 0.00089 Not reported 
CELSR2 c.1882A>G p.Ser628Gly Cadherin_repeat 0.32 0.999 AE, SB 0.00089 Not reported 
CELSR2 c.2593A>C p.Ile865Leu Cadherin_repeat 0.62 0.012 AE, SB 0.00090 Not reported 
CELSR2 c.4741A>G p.Asn1581Asp EGF_CA 0.18 0.01 SB 0.0009 Not reported 
CELSR2 c.6077C>T p.Thr2026Met HormR SB 0.0009 
CELSR2 c.6457C>G p.Arg2153Gly GAIN 0.12 0.981 EC 0.00095 Not reported 
CELSR2 c.7045G>A p.Val2349Ile GPS 0.39 0.094 AE, SB 0.00090 
CELSR2 c.7438C>T p.Arg2480Cys 7TM SB 0.0009 Not reported 
CELSR2 c.6043_6044 p.Arg2015Gly fs*22 HormR – – CRS 0.0009 Not reported 
CELSR2 c.7182_7184 p.Phe2397Leu fs*584 7TM – – SB 0.0009 Not reported 
CELSR3 c.581C>A p.Gly194Val Signal peptide 0.02 0.002 AE, SB 0.0009 Not reported 
CELSR3 c.598C>T p.Gly200Ser 0.66 SB 0.0009 
CELSR3 c.1060C>T p.Ala354Thr Cadherin_repeat 0.57 0.003 SB 0.0013 Not reported 
CELSR3 c.1336C>T p.Val446Met Cadherin_repeat 0.05 0.786 CRS 0.0036 Not reported 
CELSR3 c.5261C>T p.Gly1754Asp EGF_CA CRS 0.0009 Not reported 
CELSR3 c.8587C>T p.Ala2863Thr Cytotail 0.43 0.555 N.D. SB 0.00092 
NucleotideAmino acidCaseMAF inMAF in east
Genechangechangeno.DomainSIFTPolyPhenSexPhenotypeNTDsAsiana
CELSR1 c.2609G>A p.Pro870Leub 1 Cadherin_repeat 0 1 F CRS 0.00089 0 
CELSR1 c.3562T>A p.Thr1188Ser Cadherin_repeat 0.41 0.036 SB 0.00098 Not reported 
CELSR1 c.4222C>T p.Ala1408Thr EGF_CA 0.65 0.101 SB 0.00134 Not reported 
CELSR1 c.4969G>A p.Arg1657Trp EGF_CA 0.09 0.002 ND SB 0.00019 
CELSR1 c.5504C>T p.Arg1835His 0.34 0.002 AE, SB 0.00092 
CELSR2 c.980G>A p.Arg327His Cadherin_repeat 0.31 0.057 SB 0.00089 Not reported 
CELSR2 c.1882A>G p.Ser628Gly Cadherin_repeat 0.32 0.999 AE, SB 0.00089 Not reported 
CELSR2 c.2593A>C p.Ile865Leu Cadherin_repeat 0.62 0.012 AE, SB 0.00090 Not reported 
CELSR2 c.4741A>G p.Asn1581Asp EGF_CA 0.18 0.01 SB 0.0009 Not reported 
CELSR2 c.6077C>T p.Thr2026Met HormR SB 0.0009 
CELSR2 c.6457C>G p.Arg2153Gly GAIN 0.12 0.981 EC 0.00095 Not reported 
CELSR2 c.7045G>A p.Val2349Ile GPS 0.39 0.094 AE, SB 0.00090 
CELSR2 c.7438C>T p.Arg2480Cys 7TM SB 0.0009 Not reported 
CELSR2 c.6043_6044 p.Arg2015Gly fs*22 HormR – – CRS 0.0009 Not reported 
CELSR2 c.7182_7184 p.Phe2397Leu fs*584 7TM – – SB 0.0009 Not reported 
CELSR3 c.581C>A p.Gly194Val Signal peptide 0.02 0.002 AE, SB 0.0009 Not reported 
CELSR3 c.598C>T p.Gly200Ser 0.66 SB 0.0009 
CELSR3 c.1060C>T p.Ala354Thr Cadherin_repeat 0.57 0.003 SB 0.0013 Not reported 
CELSR3 c.1336C>T p.Val446Met Cadherin_repeat 0.05 0.786 CRS 0.0036 Not reported 
CELSR3 c.5261C>T p.Gly1754Asp EGF_CA CRS 0.0009 Not reported 
CELSR3 c.8587C>T p.Ala2863Thr Cytotail 0.43 0.555 N.D. SB 0.00092 

AE, anencephaly, CRS, craniorachischisis; EX, exencephaly; EC, encephalocele; HD, hydranencephaly; ND, not determined; SB, spina bifida.

aMAF from ExAC database.

bShared variants between CHD and NTD cohorts.

The Bold is the mutant identified in both cohorts.

Given the fact that this CELSR1 mutation was the only one detected in both cohorts, we thought that this specific variant deserved further investigation. To investigate the function of CELSR1 p.P870L (CELSR1-MUT), we initially explored its potential in the in vivo pathogenetic effect of CELSR1-MUT in zebrafish. It has previously been shown that celsr1a and celsr1b are highly expressed during neural tube convergence/extension stages in zebrafish [5254]. Antisense morpholino-modified oligonucleotides (MOs) for both zebrafish celsr1 isoforms (zcelsr1a and zcelsr1b) were injected together into one- to two-cell zebrafish embryos. Then, 48 h post-injection, both NTD and CHD phenotypes were observed in these zcelsr1 knock-down fish (Figures 2A and 2B). Of the MO-injected fish, 70.6% showed shortening of the anterior–posterior axis, which is probably due to convergence extension defects of the neural tube (Figures 2A and 2C). Approximately 80% of the embryos exhibiting an NTD phenotype also had an enlarged pericardium (a CHD phenotype), which accounted for 56.6% of the total MO-injected fish (Figures 2B and 2D).

Human CELSR1-MUT (p.P870L) caused severe NTD and CHD phenotypes in zebrafish

Figure 2
Human CELSR1-MUT (p.P870L) caused severe NTD and CHD phenotypes in zebrafish

Two types of abnormalities, (A) NTDs and (B) pericardial abnormalities, were observed in zebrafish embryos after MO (0.3 mM zcelsr1a-MO and 0.3 mM zcelsr1b-MO) injection. The severity of the NTD was determined based on the length and curve of the neural tube, as shown in (A). Both CELSR1-WT and CELSR1-MUT could rescue the NTDs and pericardial defects in zebrafish embryos when co-injected with antisense MOs. Approximately 2.3 nl of 25 ng/μl, 50 ng/μl and 100 ng/μl of each of the CELSR1 plasmids was used. (C, D) The response frequencies of NTDs and pericardial defects. Over-expression of CELSR1-WT and -MUT (p.P870L) alone also induced (E) NTDs and (F) pericardial abnormalities. Plasmids (115 pg) were injected into each zebrafish embryo. The number above each bar is the total number of embryos examined under each experimental condition. P was calculated by χ2 analysis. ***P<0.001, **P<0.01, *P<0.05. Red arrows (B) indicate the paricardial cavity of zebrafish embryoes.

Figure 2
Human CELSR1-MUT (p.P870L) caused severe NTD and CHD phenotypes in zebrafish

Two types of abnormalities, (A) NTDs and (B) pericardial abnormalities, were observed in zebrafish embryos after MO (0.3 mM zcelsr1a-MO and 0.3 mM zcelsr1b-MO) injection. The severity of the NTD was determined based on the length and curve of the neural tube, as shown in (A). Both CELSR1-WT and CELSR1-MUT could rescue the NTDs and pericardial defects in zebrafish embryos when co-injected with antisense MOs. Approximately 2.3 nl of 25 ng/μl, 50 ng/μl and 100 ng/μl of each of the CELSR1 plasmids was used. (C, D) The response frequencies of NTDs and pericardial defects. Over-expression of CELSR1-WT and -MUT (p.P870L) alone also induced (E) NTDs and (F) pericardial abnormalities. Plasmids (115 pg) were injected into each zebrafish embryo. The number above each bar is the total number of embryos examined under each experimental condition. P was calculated by χ2 analysis. ***P<0.001, **P<0.01, *P<0.05. Red arrows (B) indicate the paricardial cavity of zebrafish embryoes.

We also tested whether human CELSR1 protein could rescue these defects when added back to those knock-down embryos. We co-injected embryos with a mixture of antisense MOs and with either human CELSR1-WT or CELSR1-MUT plasmid. The response frequency of both NTDs and CHDs caused by MO knock-down was significantly reduced by both CELSR1-WT and -MUT in a dosage-dependent manner (Figures 2C and 2D). It is interesting that CELSR1-MUT demonstrated a significantly more robust ability to rescue the normal phenotype than CELSR-WT for both phenotypes at the same concentrations (Figures 2C and 2D). For example, 115 pg of CELSR1-MUT co-injected with MOs could reduce the NTD prevalence from 70.6% to 37.1%, whereas CELSR1-WT co-injected with MOs could reduce the NTD response frequency only from 70.6% to 53.9% (Figure 2C). Similar rescue effects were observed for the pericardial defects, because a co-injection of 115 pg of CELSR1-MUT could reduce the CHD prevalence from 56.6% to 23.1%, whereas a CELSR1-WT co-injection could reduce the CHD response frequency only from 56.6% to 33.3% (Figure 2D). These results suggest that CELSR1-MUT is a gain-of-function mutation in vivo.

A similar gain-of-function effect was also observed in an over-expression experiment when CELSR1-WT and -MUT were injected alone into zebrafish embryos. A significantly higher response frequency of NTDs and CHDs was observed in zebrafish embryos injected with CELSR1-MUT compared with those injected with CELSR1-WT at the same concentration (79.7% vs 55.5% NTDs and 68.4% vs 46.6% CHDs) (Figures 2E and 2F). Our results showed that hyperactivity of CELSR1, by either mutation or over-expression, could lead to both NTDs and CHDs in zebrafish.

Luciferase reporter assay showed that CELSR1 P870L increases not only PCP but also Wnt signalling

To explore the functional effect of the CELSR1 P870L mutation, we examined the effect of CELSR1 on the PCP pathway using a JNK-responsive, reporter pFR-luciferase co-transfected with either CELSR1-WT or CELSR1-MUT into MDCK cells. The results showed that, although CELSR1-WT activated the PCP pathway in MDCK cells, the CELSR1-MUT actually resulted in significantly higher activation of the PCP pathway (Figure 3A).

CELSR1-MUT (p.P870L) could increase both PCP and WNT signalling

Figure 3
CELSR1-MUT (p.P870L) could increase both PCP and WNT signalling

CELSR-WT, CELSR-MUT and an empty vector (as control) were co-transfected with either (A) PCP signalling reporter, JNK-responsive, pFR-luciferase reporter or (B) Wnt signalling reporter, β-catenin-responsive, Topflash-luciferase reporter into MDCK cells, respectively. After 48 h, cells were collected and lysed for luciferase assay. In (B) the Wnt signalling assay, the cells were treated with either control medium or Wnt3a conditional medium 36 h after transfection and incubated for another 12 h before harvesting the cells for a luciferase assay. ***P<0.001, **P<0.01, *P<0.05.

Figure 3
CELSR1-MUT (p.P870L) could increase both PCP and WNT signalling

CELSR-WT, CELSR-MUT and an empty vector (as control) were co-transfected with either (A) PCP signalling reporter, JNK-responsive, pFR-luciferase reporter or (B) Wnt signalling reporter, β-catenin-responsive, Topflash-luciferase reporter into MDCK cells, respectively. After 48 h, cells were collected and lysed for luciferase assay. In (B) the Wnt signalling assay, the cells were treated with either control medium or Wnt3a conditional medium 36 h after transfection and incubated for another 12 h before harvesting the cells for a luciferase assay. ***P<0.001, **P<0.01, *P<0.05.

As CELSR1 interacts not only with PCP receptors but also with Wnt receptors, we additionally examined the effect of CELSR1-MUT on canonical Wnt signalling using the Topflash-luciferase reporter, which is a β-catenin-responsive reporter. When CELSR1-WT or CELSR1-MUT was over-expressed in MDCK cells, both CELSR1-WT and CELSR1-MUT could activate the Wnt pathway under Wnt3a stimulation (Figure 3B). It is interesting that CELSR1-MUT had a much stronger effect in promoting canonical Wnt signalling, because this mutant protein could significantly activate the Wnt pathway even without the Wnt3a ligand (Figure 3B). Consistent with the in vivo data, these in vitro data strongly suggested that CELSR1-MUT is a gain-of-function mutation.

DISCUSSION

Previous studies have shown that CELSR1 is more strongly associated with NTDs than CELSR2 and CELSR3, given that Celsr2 and Celsr3 mutant mice lacked an NTD phenotype apart from an occasional hydrocephalus [31,55]. There are several previous reports linking CELSR1 non-synonymous mutations to human NTDs [2729]. Although one previous study reported that rs646776 near CELSR2 is associated with early onset myocardial infarction [56], no association between mutations in CELSR1–3 and CHDs has been reported so far. In the present study, we evaluated CELSR1–3 mutations in two separate CHD and NTD cohorts, with comparable numbers of cases and controls. We identified almost equal numbers of variants between the two cohorts throughout all the filtration processes involving both rare SNVs and novel coding SNVs. In the end, a total of 72 case-specific novel mutations in CELSR1–3 were identified (37 in CHDs and 36 in NTDs). These results indicate that CELSR1–3 probably make comparable contributions to the aetiology of NTDs and CHDs. Therefore, our data provide evidence that mutations in CELSRs may be associated with increased risk not only for NTDs but also for CHDs in humans. To our knowledge, this is the first report of CELSR mutations associated with human CHDs.

Most of the mutations that we identified in the present study differ between the two malformation cohorts. It is possible that factors, including differences in location and age groups, may contribute to the observed differences in distinct CELSR mutations. Alternatively, different selective pressures were present for the two developmental systems, and CELSRs might interact with different molecules during the development of these individual organ systems in humans. Future study of the molecular control of CELSR gene functions and whole-genome sequencing studies, with larger sample sizes and parent samples, will certainly provide more information about the relationship between mutations in the CELSR genes and the risk for NTDs and CHDs in humans.

The single novel CELSR1 mutation (c.2609G>A p.P870L) observed in both NTD and CHD patients led to both defects in the zebrafish model. Therefore, it is very likely that the gain-of-function mutation CELSR1 P870L also affects both neural tube and heart development in human embryos. The different phenotypes in the two patients again suggested that molecular regulation of CELSR1 gene function may differ in humans. There could have been different genetic, epigenetic or environmental factors that interacted with the same CELSR1 during the development of these two early embryonic systems in humans. CELSR1 P870L mutation leads to hyperactivity of CELSR1. The interaction of genetic elements or environmental factors with this hyperactive CELSR1 may determine the outcome. An interesting observation is that CELSR1 P870L not only increased PCP signalling, but also induced canonical Wnt signalling. This has not been reported before because previous studies have shown only that CELSR1 acts homophilically to transmit intercellular PCP signals [31,57]. However, as CELSR1 interacts not only with other core PCP proteins such as VANGL2 but also with the Wnt receptor FRIZZLED [32], it is possible that CELSR1 P870L could increase its interaction with FRIZZLED to activate a canonical Wnt pathway even without the ligand Wnt3a. Therefore, such cross-talk with other signalling pathways might explain such a ‘syndromic’ effect of this mutant. Other variants in the PCP pathway that compromise cross-talk with canonical Wnt signalling or other morphogenetic pathways might be involved in determining the risk for NTDs and CHDs in humans. It is interesting that a similar phenomenon was observed in a mouse model of another PCP gene, the PCP effector Fuz, which is also involved in cilia, and the SHH pathway, in which the knockout mice presented with both NTD and CHD phenotypes [18]. It is highly likely that FUZ plays an important role in the aetiology of human NTDs and CHDs. Besides Fuz, it is possible that those PCP mutants, such as Vangl2 and Ptk7, that had both NTD and CHD phenotypes in knockout mice models [1820] would be involved in the aetiology of both NTDs and CHDs in humans. Future investigations should focus on mutations that can compromise cross-talk of different developmental pathways, which could provide some insight into these severe birth defects.

In conclusion, we systematically identified a set of rare novel mutations in CELSR1–3 genes in human CHD and NTD patients that have the potential to be used diagnostically. Our data support our hypothesis that core PCP gene CELSR1–3 mutations are related to both NTDs and CHDs in humans. The shared mutation, CELSR1 P870L, has the potential to help us have a better understanding of the developmental overlap between NTDs and CHDs. Our data suggest that mutations in PCP genes, especially hyperactive mutations such as CELSR1 P870L, are likely to be associated with both NTDs and CHDs in humans. Further understanding of the regulations and cross-talk between PCP and other signalling pathways, such as the canonical Wnt pathway, during early embryonic development, is critical for identifying how best to prevent these preventable birth defects.

AUTHOR CONTRIBUTION

H. Wang and Y. Zheng designed the study and Y. Zheng, H. Wang and R. Finnell prepared the manuscript. X. Qiao and Y. Liu performed lab experiments. X. Qiao, P. Li, H. Li, B. Qiao, T. Zhang and X. Yang participated in the recruitment of patients and sequencing confirmation. P. Li, H. Li, X. Yang and Z. Chen performed the statistical analysis. All authors reviewed the manuscript and contributed to its scientific content. X. Qiao and Y. Liu contributed equally to this paper.

Thanks to Dr Tao P. Zhong and Dr Xiliang Zha at Fudan University, for making a gift to us of the Topflash-luciferase reporter construct, and kindly providing the Wnt3a mouse fibroblast cell line (ATCC CRL-2647), respectively.

FUNDING

This work was supported by grants from the 973 Program [2013CB945403 to H. Wang, 2013CB945404 to Y. Zheng], the National Natural Science Foundation of China [81430005, 81270232], the National Key Basic Research Program of China [2016YFC1000502 to H. Wang] and the Commission for Science and Technology of Shanghai Municipality [13JC1407600 to H. Wang and 14JC1401000 to Y. Zheng].

Abbreviations

     
  • CHD

    congenital heart defect

  •  
  • CELSR

    cadherin, EGF, LAG, seven-pass G-type receptor

  •  
  • EGF

    epidermal growth factor

  •  
  • ExAC

    Exome Aggregation Consortium

  •  
  • LAG

    laminin A G-type repeat

  •  
  • MAF

    major allele frequency

  •  
  • MDCK

    Madin–Darby canine kidney

  •  
  • MO

    morpholino-modified oligonucleotide

  •  
  • NTC

    neural tube closure

  •  
  • NTD

    neural tube defect

  •  
  • NS-CHD

    non-syndromic CHD

  •  
  • PCP

    planar cell polarity

  •  
  • PFO

    patent foramen ovale

  •  
  • pRLCMV

    Renilla-luciferase plasmid

  •  
  • PTV

    protein-truncating variant

  •  
  • S-CHD

    syndromic CHD

  •  
  • SHH

    sonic hedgehog

  •  
  • SIFT

    sorting intolerant from tolerant

  •  
  • SNP

    single nucleotide polymorphism

  •  
  • SNV

    single nucleotide variant

References

References
1
Webber
 
D.M.
MacLeod
 
S.L.
Bamshad
 
M.J.
Shaw
 
G.M.
Finnell
 
R.H.
Shete
 
S.S.
Witte
 
J.S.
Erickson
 
S.W.
Murphy
 
L.D.
Hobbs
 
C.
 
Developments in our understanding of the genetic basis of birth defects
Birth Defects Res. A Clin. Mol. Teratol.
2015
, vol. 
103
 (pg. 
680
-
691
)
[PubMed]
2
Detrait
 
E.R.
George
 
T.M.
Etchevers
 
H.C.
Gilbert
 
J.R.
Vekemans
 
M.
Speer
 
M.C.
 
Human neural tube defects: developmental biology, epidemiology, and genetics
Neurotoxicol. Teratol.
2005
, vol. 
27
 (pg. 
515
-
524
)
[PubMed]
3
Greene
 
N.D.
Copp
 
A.J.
 
Neural tube defects
Annu. Rev. Neurosci.
2014
, vol. 
37
 (pg. 
221
-
242
)
[PubMed]
4
Greene
 
N.D.
Stanier
 
P.
Copp
 
A.J.
 
Genetics of human neural tube defects
Hum. Mol. Genet.
2009
, vol. 
18
 
R2
(pg. 
R113
-
R129
)
[PubMed]
5
Fahed
 
A.C.
Gelb
 
B.D.
Seidman
 
J.G.
Seidman
 
C.E.
 
Genetics of congenital heart disease: the glass half empty
Circ. Res.
2013
, vol. 
112
 (pg. 
707
-
720
)
[PubMed]
6
Hoffman
 
J.I.
Kaplan
 
S.
 
The incidence of congenital heart disease
J. Am. Coll. Cardiol.
2002
, vol. 
39
 (pg. 
1890
-
1900
)
[PubMed]
7
Czeizel
 
A.E.
Dudas
 
I.
Vereczkey
 
A.
Banhidy
 
F.
 
Folate deficiency and folic acid supplementation: the prevention of neural-tube defects and congenital heart defects
Nutrients
2013
, vol. 
5
 (pg. 
4760
-
4775
)
[PubMed]
8
Copp
 
A.J.
Stanier
 
P.
Greene
 
N.D.
 
Neural tube defects: recent advances, unsolved questions, and controversies
Lancet Neurol.
2013
, vol. 
12
 (pg. 
799
-
810
)
[PubMed]
9
Vecoli
 
C.
Pulignani
 
S.
Foffa
 
I.
Andreassi
 
M.G.
 
Congenital heart disease: the crossroads of genetics, epigenetics and environment
Curr. Genomics
2014
, vol. 
15
 (pg. 
390
-
399
)
[PubMed]
10
Copp
 
A.J.
Greene
 
N.D.
 
Neural tube defects–disorders of neurulation and related embryonic processes
Wiley Interdiscip. Rev. Dev. Biol.
2013
, vol. 
2
 (pg. 
213
-
227
)
[PubMed]
11
Koefoed
 
K.
Veland
 
I.R.
Pedersen
 
L.B.
Larsen
 
L.A.
Christensen
 
S.T.
 
Cilia and coordination of signaling networks during heart development
Organogenesis
2014
, vol. 
10
 (pg. 
108
-
125
)
[PubMed]
12
Yuan
 
S.
Zaidi
 
S.
Brueckner
 
M.
 
Congenital heart disease: emerging themes linking genetics and development
Curr. Opin. Genet. Dev.
2013
, vol. 
23
 (pg. 
352
-
359
)
[PubMed]
13
Li
 
Y.
Klena
 
N.T.
Gabriel
 
G.C.
Liu
 
X.
Kim
 
A.J.
Lemke
 
K.
Chen
 
Y.
Chatterjee
 
B.
Devine
 
W.
Damerla
 
R.R.
, et al 
Global genetic analysis in mice unveils central role for cilia in congenital heart disease
Nature
2015
, vol. 
521
 (pg. 
520
-
524
)
[PubMed]
14
Munoz-Soriano
 
V.
Belacortu
 
Y.
Paricio
 
N.
 
Planar cell polarity signaling in collective cell movements during morphogenesis and disease
Curr. Genomics
2012
, vol. 
13
 (pg. 
609
-
622
)
[PubMed]
15
Simons
 
M.
Mlodzik
 
M.
 
Planar cell polarity signaling: from fly development to human disease
Annu. Rev. Genet.
2008
, vol. 
42
 (pg. 
517
-
540
)
[PubMed]
16
Wansleeben
 
C.
Meijlink
 
F.
 
The planar cell polarity pathway in vertebrate development
Dev. Dyn.
2011
, vol. 
240
 (pg. 
616
-
626
)
[PubMed]
17
Murdoch
 
J.N.
Damrau
 
C.
Paudyal
 
A.
Bogani
 
D.
Wells
 
S.
Greene
 
N.D.
Stanier
 
P.
Copp
 
A.J.
 
Genetic interactions between planar cell polarity genes cause diverse neural tube defects in mice
Dis. Model Mech.
2014
, vol. 
7
 (pg. 
1153
-
1163
)
[PubMed]
18
Gray
 
R.S.
Abitua
 
P.B.
Wlodarczyk
 
B.J.
Szabo-Rogers
 
H.L.
Blanchard
 
O.
Lee
 
I.
Weiss
 
G.S.
Liu
 
K.J.
Marcotte
 
E.M.
Wallingford
 
J.B.
, et al 
The planar cell polarity effector Fuz is essential for targeted membrane trafficking, ciliogenesis and mouse embryonic development
Nat. Cell Biol.
2009
, vol. 
11
 (pg. 
1225
-
1232
)
[PubMed]
19
Juriloff
 
D.M.
Harris
 
M.J.
 
A consideration of the evidence that genetic defects in planar cell polarity contribute to the etiology of human neural tube defects
Birth Defects Res. A Clin. Mol. Teratol.
2012
, vol. 
94
 (pg. 
824
-
840
)
[PubMed]
20
Wu
 
G.
Ge
 
J.
Huang
 
X.
Hua
 
Y.
Mu
 
D.
 
Planar cell polarity signaling pathway in congenital heart diseases
J. Biomed. Biotechnol.
2011
, vol. 
2011
 pg. 
589414
 
[PubMed]
21
Yu
 
H.
Ye
 
X.
Guo
 
N.
Nathans
 
J.
 
Frizzled 2 and frizzled 7 function redundantly in convergent extension and closure of the ventricular septum and palate: evidence for a network of interacting genes
Development
2012
, vol. 
139
 (pg. 
4383
-
4394
)
[PubMed]
22
Henderson
 
D.J.
Chaudhry
 
B.
 
Getting to the heart of planar cell polarity signaling
Birth Defects Res. A Clin. Mol. Teratol.
2011
, vol. 
91
 (pg. 
460
-
467
)
[PubMed]
23
Henderson
 
D.J.
Phillips
 
H.M.
Chaudhry
 
B.
 
Vang-like 2 and noncanonical Wnt signaling in outflow tract development
Trends Cardiovasc. Med.
2006
, vol. 
16
 (pg. 
38
-
45
)
[PubMed]
24
Kibar
 
Z.
Salem
 
S.
Bosoi
 
C.M.
Pauwels
 
E.
De Marco
 
P.
Merello
 
E.
Bassuk
 
A.G.
Capra
 
V.
Gros
 
P.
 
Contribution of VANGL2 mutations to isolated neural tube defects
Clin. Genet.
2011
, vol. 
80
 (pg. 
76
-
82
)
[PubMed]
25
Lei
 
Y.P.
Zhang
 
T.
Li
 
H.
Wu
 
B.L.
Jin
 
L.
Wang
 
H.Y.
 
VANGL2 mutations in human cranial neural-tube defects
N. Engl. J. Med.
2010
, vol. 
362
 (pg. 
2232
-
2235
)
[PubMed]
26
Yang
 
X.Y.
Zhou
 
X.Y.
Wang
 
Q.Q.
Li
 
H.
Chen
 
Y.
Lei
 
Y.P.
Ma
 
X.H.
Kong
 
P.
Shi
 
Y.
Jin
 
L.
, et al 
Mutations in the COPII vesicle component gene SEC24B are associated with human neural tube defects
Hum. Mutat.
2013
, vol. 
34
 (pg. 
1094
-
1101
)
[PubMed]
27
Allache
 
R.
De Marco
 
P.
Merello
 
E.
Capra
 
V.
Kibar
 
Z.
 
Role of the planar cell polarity gene CELSR1 in neural tube defects and caudal agenesis
Birth Defects Res. A Clin. Mol. Teratol.
2012
, vol. 
94
 (pg. 
176
-
181
)
[PubMed]
28
Lei
 
Y.
Zhu
 
H.
Yang
 
W.
Ross
 
M.E.
Shaw
 
G.M.
Finnell
 
R.H.
 
Identification of novel CELSR1 mutations in spina bifida
PLoS One
2014
, vol. 
9
 pg. 
e92207
 
[PubMed]
29
Robinson
 
A.
Escuin
 
S.
Doudney
 
K.
Vekemans
 
M.
Stevenson
 
R.E.
Greene
 
N.D.
Copp
 
A.J.
Stanier
 
P.
 
Mutations in the planar cell polarity genes CELSR1 and SCRIB are associated with the severe neural tube defect craniorachischisis
Hum. Mutat.
2012
, vol. 
33
 (pg. 
440
-
447
)
[PubMed]
30
Usui
 
T.
Shima
 
Y.
Shimada
 
Y.
Hirano
 
S.
Burgess
 
R.W.
Schwarz
 
T.L.
Takeichi
 
M.
Uemura
 
T.
 
Flamingo, a seven-pass transmembrane cadherin, regulates planar cell polarity under the control of Frizzled
Cell
1999
, vol. 
98
 (pg. 
585
-
595
)
[PubMed]
31
Langenhan
 
T.
Aust
 
G.
Hamann
 
J.
 
Sticky signaling–adhesion class G protein-coupled receptors take the stage
Sci. Signal.
2013
, vol. 
6
 pg. 
re3
 
[PubMed]
32
Yang
 
Y.
Mlodzik
 
M.
 
Wnt-Frizzled/planar cell polarity signaling: cellular orientation by facing the wind (Wnt)
Annu. Rev. Cell Dev. Biol.
2015
, vol. 
31
 (pg. 
623
-
646
)
[PubMed]
33
Veltman
 
J.A.
Brunner
 
H.G.
 
De novo mutations in human genetic disease
Nat. Rev. Genet.
2012
, vol. 
13
 (pg. 
565
-
575
)
[PubMed]
34
Zaidi
 
S.
Choi
 
M.
Wakimoto
 
H.
Ma
 
L.
Jiang
 
J.
Overton
 
J.D.
Romano-Adesman
 
A.
Bjornson
 
R.D.
Breitbart
 
R.E.
Brown
 
K.K.
, et al 
De novo mutations in histone-modifying genes in congenital heart disease
Nature
2013
, vol. 
498
 (pg. 
220
-
223
)
[PubMed]
35
Homsy
 
J.
Zaidi
 
S.
Shen
 
Y.
Ware
 
J.S.
Samocha
 
K.E.
Karczewski
 
K.J.
DePalma
 
S.R.
McKean
 
D.
Wakimoto
 
H.
Gorham
 
J.
, et al 
De novo mutations in congenital heart disease with neurodevelopmental and other congenital anomalies
Science
2015
, vol. 
350
 (pg. 
1262
-
1266
)
[PubMed]
36
Sifrim
 
A.
Hitz
 
M.P.
Wilsdon
 
A.
Breckpot
 
J.
Turki
 
S.H.
Thienpont
 
B.
McRae
 
J.
Fitzgerald
 
T.W.
Singh
 
T.
Swaminathan
 
G.J.
, et al 
Distinct genetic architectures for syndromic and nonsyndromic congenital heart defects identified by exome sequencing
Nat. Genet.
2016
, vol. 
48
 (pg. 
1060
-
1065
)
[PubMed]
37
Botto
 
L.D.
Lin
 
A.E.
Riehle-Colarusso
 
T.
Malik
 
S.
Correa
 
A.
 
Seeking causes: Classifying and evaluating congenital heart defects in etiologic studies
Birth Defects Res. A Clin. Mol. Teratol.
2007
, vol. 
79
 (pg. 
714
-
727
)
[PubMed]
38
Craig
 
D.W.
Pearson
 
J.V.
Szelinger
 
S.
Sekar
 
A.
Redman
 
M.
Corneveaux
 
J.J.
Pawlowski
 
T.L.
Laub
 
T.
Nunn
 
G.
Stephan
 
D.A.
, et al 
Identification of genetic variants using bar-coded multiplexed sequencing
Nat. Methods
2008
, vol. 
5
 (pg. 
887
-
893
)
[PubMed]
39
Li
 
H.
Durbin
 
R.
 
Fast and accurate long-read alignment with Burrows–Wheeler transform
Bioinformatics
2010
, vol. 
26
 (pg. 
589
-
595
)
[PubMed]
40
Li
 
H.
Handsaker
 
B.
Wysoker
 
A.
Fennell
 
T.
Ruan
 
J.
Homer
 
N.
Marth
 
G.
Abecasis
 
G.
Durbin
 
R.
 
Genome Project Data Processing Subgroup
The sequence alignment/map format and SAMtools
Bioinformatics
2009
, vol. 
25
 (pg. 
2078
-
2079
)
[PubMed]
41
Koboldt
 
D.C.
Chen
 
K.
Wylie
 
T.
Larson
 
D.E.
McLellan
 
M.D.
Mardis
 
E.R.
Weinstock
 
G.M.
Wilson
 
R.K.
Ding
 
L.
 
VarScan: variant detection in massively parallel sequencing of individual and pooled samples
Bioinformatics
2009
, vol. 
25
 (pg. 
2283
-
2285
)
[PubMed]
42
McLaren
 
W.
Pritchard
 
B.
Rios
 
D.
Chen
 
Y.
Flicek
 
P.
Cunningham
 
F.
 
Deriving the consequences of genomic variants with the Ensemble API and SNP Effect Predictor
Bioinformatics
2010
, vol. 
26
 (pg. 
2069
-
2070
)
[PubMed]
43
Consortium
 
G.P.
 
An integrated map of genetic variation from 1,092 human genomes
Nature
2012
, vol. 
491
 (pg. 
56
-
65
)
[PubMed]
44
Lek
 
M.
Karczewski
 
K.J.
Minikel
 
E.V.
Samocha
 
K.E.
Banks
 
E.
Fennell
 
T.
O'Donnell-Luria
 
A.H.
Ware
 
J.S.
Hill
 
A.J.
Cummings
 
B.B.
, et al 
Analysis of protein-coding genetic variation in 60,706 humans
Nature
2016
, vol. 
536
 (pg. 
285
-
291
)
[PubMed]
45
Kumar
 
P.
Henikoff
 
S.
Ng
 
P.C.
 
Predicting the effects of coding non-synonymous variants on protein function using the SIFT algorithm
Nat. Protoc.
2009
, vol. 
4
 (pg. 
1073
-
1081
)
[PubMed]
46
Sim
 
N.L.
Kumar
 
P.
Hu
 
J.
Henikoff
 
S.
Schneider
 
G.
Ng
 
P.C.
 
SIFT web server: predicting effects of amino acid substitutions on proteins
Nucleic. Acids Res.
2012
, vol. 
40
 
Web Server issue
(pg. 
W452
-
W457
)
[PubMed]
47
Adzhubei
 
I.A.
Schmidt
 
S.
Peshkin
 
L.
Ramensky
 
V.E.
Gerasimova
 
A.
Bork
 
P.
Kondrashov
 
A.S.
Sunyaev
 
S.R.
 
A method and server for predicting damaging missense mutations
Nat. Methods.
2010
, vol. 
7
 (pg. 
248
-
249
)
[PubMed]
48
Shi
 
Y.
Ding
 
Y.
Lei
 
Y.P.
Yang
 
X.Y.
Xie
 
G.M.
Wen
 
J.
Cai
 
C.Q.
Li
 
H.
Chen
 
Y.
Zhang
 
T.
, et al 
Identification of novel rare mutations of DACT1 in human neural tube defects
Hum. Mutat.
2012
, vol. 
33
 (pg. 
1450
-
1455
)
[PubMed]
49
Gilboa
 
S.M.
Devine
 
O.J.
Kucik
 
J.E.
Oster
 
M.E.
Riehle-Colarusso
 
T.
Nembhard
 
W.N.
Xu
 
P.
Correa
 
A.
Jenkins
 
K.
Marelli
 
A.J.
 
Congenital heart defects in the United States: estimating the magnitude of the affected population in 2010
Circulation
2016
, vol. 
134
 (pg. 
101
-
109
)
[PubMed]
50
Sokal
 
R.
Tata
 
L.J.
Fleming
 
K.M.
 
Sex prevalence of major congenital anomalies in the United Kingdom: a national population-based study and international comparison meta-analysis
Birth Defects Res. A Clin. Mol. Teratol.
2014
, vol. 
100
 (pg. 
79
-
91
)
[PubMed]
51
Tennant
 
P.W.
Samarasekera
 
S.D.
Pless-Mulloli
 
T.
Rankin
 
J.
 
Sex differences in the prevalence of congenital anomalies: a population-based study
Birth Defects Res. A Clin. Mol. Teratol.
2011
, vol. 
91
 (pg. 
894
-
901
)
[PubMed]
52
Carreira-Barbosa
 
F.
Kajita
 
M.
Morel
 
V.
Wada
 
H.
Okamoto
 
H.
Martinez Arias
 
A.
Fujita
 
Y.
Wilson
 
S.W.
Tada
 
M.
 
Flamingo regulates epiboly and convergence/extension movements through cell cohesive and signalling functions during zebrafish gastrulation
Development
2009
, vol. 
136
 (pg. 
383
-
392
)
[PubMed]
53
Formstone
 
C.J.
Mason
 
I.
 
Combinatorial activity of Flamingo proteins directs convergence and extension within the early zebrafish embryo via the planar cell polarity pathway
Dev. Biol.
2005
, vol. 
282
 (pg. 
320
-
335
)
[PubMed]
54
Harty
 
B.L.
Krishnan
 
A.
Sanchez
 
N.E.
Schioth
 
H.B.
Monk
 
K.R.
 
Defining the gene repertoire and spatiotemporal expression profiles of adhesion G protein-coupled receptors in zebrafish
BMC Genomics
2015
, vol. 
16
 pg. 
62
 
[PubMed]
55
Wang
 
X.J.
Zhang
 
D.L.
Xu
 
Z.G.
Ma
 
M.L.
Wang
 
W.B.
Li
 
L.L.
Han
 
X.L.
Huo
 
Y.
Yu
 
X.
Sun
 
J.P.
 
Understanding cadherin EGF LAG seven-pass G-type receptors
J. Neurochem.
2014
, vol. 
131
 (pg. 
699
-
711
)
[PubMed]
56
Kathiresan
 
S.
Voight
 
B.F.
Purcell
 
S.
Musunuru
 
K.
Ardissino
 
D.
Mannucci
 
P.M.
Anand
 
S.
Engert
 
J.C.
Samani
 
N.J.
, et al 
Myocardial Infarction Genetics Consortium
Genome-wide association of early-onset myocardial infarction with single nucleotide polymorphisms and copy number variants
Nat. Genet.
2009
, vol. 
41
 (pg. 
334
-
341
)
[PubMed]
57
Nishimura
 
T.
Honda
 
H.
Takeichi
 
M.
 
Planar cell polarity links axes of spatial dynamics in neural-tube closure
Cell
2012
, vol. 
149
 (pg. 
1084
-
1097
)
[PubMed]

Author notes

1

These two authors contributed equally to this work.

Supplementary data