Introduction

Congenital heart defect (CHD) is the most common class of birth defects worldwide. It is estimated that every 8–10 of 1000 newborns can have CHD, and 10 out of 100 aborted fetus have heart anomalies.1 The etiology of CHD is complicated, and both environmental and genetic causes can perturb heart development and lead to heart malformations.2 The whole-exome sequencing has identified coding-region mutations in about 400 genes contributing to CHD risk, which only explains ~10% of sporadic CHD cases,3 leaving a large group of sporadic CHD cases with no explicitly revealed causes.

Tbx2 is a member of T-box transcription factors, which plays an active role in cardiogenesis.4 It is finely regulated in a complex regulation net, including upstream transcription factors and signaling molecules, e.g., Tbx20, retinoid acid, and Bmp2,5,6,7,8 and downstream targeting genes such as Nkx2-5, Tbx5, Msx1, and Msx2.9,10,11 Studies in mouse model showed homozygous mice fetus with Tbx2−/− null mutation had defects of the atrioventricular canal and septum of the outflow tract.12

Considering the dose-dependent function manner of Tbx2,5 we presumed that any factor altering gene expression level, for example, changes in its noncoding region such as 3′ untranslated region (3′UTR) or promoter region, may interfere cardiogenesis and induce CHD. Our previous study demonstrated that noncoding variants in TBX5 3′UTR increased CHD risk by regulating gene expression levels.13 Few studies have identified mutations in TBX2 coding region to induce human CHD till now, while variants of TBX2 in the promoter region14 were indicated to cause ventricular septal defect. However, the contribution of polymorphisms in TBX2 3′UTR, which is critically important for microRNAs (miRNAs) to bind mRNA and regulate gene expression, remains unknown. In the present study, we analyzed the contribution of TBX2 3′UTR to CHD risk in a Han Chinese cohort composed of 1241 CHD patients and 1285 healthy controls, filtered out the most probable CHD subtypes, and further studied the mechanism involved primarily.

Methods

Study subjects

Cases in the present study comprised complex CHD subjects that necessitated surgical intervention due to critical, life-threatening symptoms or failure of congestive cardiac dysfunction to respond to medications. Besides, the inclusion criteria for the simple CHD cases, mainly ventricular septal defect (VSD) and atrial septal defect (ASD), were clinical and/or echocardiographic evidence of a significant shunt through the lesions.15 From January 2014 to May 2016, we recruited 1241 consecutive Han Chinese CHD children from two different cohorts and 1285 healthy controls who were matched to cases in terms of age and sex from the same geographical area over the same period. Among them, 648 controls and 693 CHD cases were enrolled from the Institute of Cardiovascular Research, Jinan Military General Hospital (Jinan, Shandong Province, China), as well as 637 controls and 548 cases from Children’s Hospital of Fudan University (Shanghai, China). Cases of isolated patent ductus arteriosus, patent foramen ovale, small-size septal defects, congenitally corrected transposition of the great arteries, bicuspid aortic valve, coronary anomalies, as well as CHD that relate mainly to the vascular system were excluded from the present study. All patients with genetic syndromes (e.g., Down’s syndrome, Holt–Oram syndrome, Alagille syndrome, DiGeorge syndrome, William syndrome, and Noonan syndrome) or family history of CHD were excluded. Also, cases combined with other non-cardiovascular malformations, tumor or systematic diseases were not recruited.

According to the commonly used criteria,16 the 1241 CHD cases were classified into nine broad categories, including anomalous pulmonary venous return (APVR), atrioventricular septal defects (AVSD), complex CHD, conotruncal defects, heterotaxy (including isomerism and mirror-imagery), left ventricular outflow tract obstruction (LVOTO), right ventricular outflow tract obstruction (RVOTO), septal defects, and CHD with other associations.

The study protocol was approved by the medical ethics committee of Children’s Hospital of Fudan University. Written informed consents were obtained from every participants or their legal guardians.

Variant identification and genotyping

Blood of CHD children was collected at the same time when they received other necessary blood sampling for disease diagnosis, while in the cohort of healthy control, we used the residual samples from their blood test. Genomic DNA was isolated from the peripheral venous blood using conventional reagents. The single-nucleotide polymorphisms (SNPs) in a cohort of 288 cases and 288 controls firstly collected were identified using Sanger Sequencing. Genotyping of rs59382073 in samples collected later was performed using SNaPshot assay (ABI, Waltham, MA) according to the manufacturer’s instructions. Briefly, the TBX2 3′UTR sequences were amplified through PCR using forward primer GAGGCGGCCAATGAACTGCA and reverse primer CTGTGCGCTCTGTGGGTC. After removing excessive primers and deoxynucleotides using ExolI and SAP (Takara bio, Tokyo, Japan), specific SNaPshot primer (tttttAACACAGTCACTTGGGAGCT) was annealed to the PCR products, separated by capillary electrophoresis (ABI, Waltham, MA), and analyzed by Peak Scanner.

Plasmid constructs

We constructed plasmids for validation of the function of either allele of the SNP rs59382073 using the psiCHECKTM-2 vector (Promega, Madison, WI). A 1081-bp fragment in TBX2 3′UTR was cloned from genomic DNA which contains the G allele. Later we designed primers with point mutation (Forward: TGGGAGCTTCTGGGCTAGGTCCAGATCCGCTCCAG; Reverse: CTGGAGCGGATCTGGACCTAGCCCAGAAGCTCCCA) and used the KOD-Plus (Takara bio, Tokyo, Japan) DNA polymerase to construct vectors containing the corresponding T allele plasmid. Then PCR products were digested by XhoI and NotI (Takara bio, Tokyo, Japan), and then cloned into the 3′UTR of the Renilla luciferase gene of the psiCHECKTM-2 vector, respectively. The recombined plasmids sequences were confirmed by Sanger sequencing.

Transfection and luciferase reporter assays

Human embryonic kidney 293 (HEK-293T) cells and rat cardiomyocyte H9c2 cells were grown using Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum (Thermo Fisher Scientific, Waltham, MA), added 1% penicillin/streptomycin solution (10,000 U/mL and 10 mg/mL, respectively). Mycoplasma contamination was regularly checked. Cells were cultured in 24-well plates and transfected with 500 ng psiCHECK2-G, psiCHECK2-T, or psiCHECK2-basic vector using 1.5 μL Lipofectamine 2000 (Life Technologies, Waltham, MA) at 70% confluence. Cotransfection with miRNAs were performed with extra 15 pM miRNA mimics or inhibitors (GenePharm, Shanghai, China) mixing with the constructed vectors in both HEK-293T cells and H9c2 cells. After 24 h of transfection, the transfected cells were lysed in 100 μL per well passive lysis buffer (Promega, Madison, WI), gently rocked at room temperature for 15 min, centrifuged at 6000 rpm for 3 min, and transferred to black 96-well plate. The lysate was then assayed for luciferase activity with the Dual-Luciferase Reporter Assay System (Promega, Madison, WI) using GloMax 96 Microplate Lumiometer (Promega, Madison, WI). Transfection experiments were performed in three replicates, and each luciferase assay was carried out in triplicate.

Statistical analyses

Independent sample t-tests (two-tail) were used to compare continuous demographic and phenotypic variables between controls and cases. Allele frequencies were confirmed by gene counting. Hardy–Weinberg equilibrium, genotype and allele frequency comparisons between groups were performed using the Chi-square tests. To evaluate the associations between the genotypes and CHD risk, odds ratios (ORs) and 95% confidence intervals (CIs) were calculated by using unconditional logistic regression analysis with adjustment for age and sex. The statistical packages SPSS 20.0 (SPSS, Chicago, IL) and GraphPad Prism (San Diego, CA) were used for the statistical analyses and graphing. Statistical differences were considered significant if P < 0.05.

Results

TBX2 3′UTR polymorphism rs59382073 significantly increased CHD susceptibility in the Han Chinese population

Five polymorphisms with minor allele frequency (MAF) over 0.05 were found in the 288 cases and 288 controls firstly collected, including rs59382073, rs1058004, rs729782, rs3198587, and rs729781. Among them the frequencies of rs59382073 alleles were significantly different between control and case cohort (P = 0.01) (Table 1). Therefore, the variant rs59382073 was further studied in two independent case–control cohorts from Shanghai and Shandong.

Table 1 Genotyping of TBX2 3′UTR SNPs identified in 288 CHD patients and 288 controls
Full size table

Analyses showed control population in both groups were in Hardy–Weinberg equilibrium, as well as in combination (HWE-p > 0.05). T allele frequencies in cases were significantly higher compared to controls in combination (7.86% vs 4.55%, P = 1.00 × 10−6), Shanghai (6.93% vs 4.87%, P = 0.03), and Shandong (8.59% vs 4.24%, P = 5.00 × 10−6) cohorts, respectively (Table 2).

Table 2 TBX2 3′UTR SNP rs59382073 allele frequencies comparison
Full size table

In consideration of rare proportion of TT genotype, we applied the dominant model of inheritance in the association studies to enhance the statistical power. Logistic regression analyses revealed that T allele carriers (homozygous TT and heterozygous GT individuals) had a significantly higher risk of CHD compared to wild-type GG subjects in both Shanghai (OR = 1.61, 95% CI = 1.12–2.31, P = 0.01) and Shandong group (OR = 2.17, 95% CI = 1.54–3.05, P = 1.00 × 10−5). The combination showed a 1.89-fold increased CHD risk (OR = 1.89, 95% CI = 1.48–2.46, P = 4.48 × 10−7) for the GT and TT genotypes compared to GG subjects  (Table 3). Hence, TBX2 3′UTR variant rs59382073 was a significant contributor to CHD risk in two cohorts both separately and together.

Table 3 TBX2 3′UTR SNP rs59382073 association with CHD risk in two independent cohorts
Full size table

TBX2 rs59382073 T allele was significantly associated with susceptibility of RVOTO, conotruncal, and septal defect

To confirm which CHD subtype is closely associated with rs59382073 polymorphism, a stratified analysis was performed according to standard CHD classifications as previously described,16 including APVR (34, 2.7%), AVSD (30, 2.4%), conotruncal defect (192, 15.5%), complex (49, 3.9%), heterotaxy (19, 1.5%), LVOTO (52, 4.2%), RVOTO (71, 5.7%), septal defect (769, 62.0%), and other CHDs (25, 2.0%). The results showed that T allele carriers had a significantly increased risk of RVOTO (OR = 2.57, 95% CI = 1.39–4.76, P = 2.64 × 10−3), conotruncal (OR = 2.25, 95% CI = 1.48–3.42, P = 1.40 × 10−4), and septal defect (OR = 1.98, 95% CI = 1.50–2.60, P = 1.00 × 10−6) (Table 4).

Table 4 Stratified analysis of CHD subtypes risk and rs59382073 variant
Full size table

MiR-3940 and miR-708 repressed luciferase expression in rs59382073 T allele construct

Luciferase assays were conducted in both HEK-293T and H9c2 cells to determine the functional effect of rs59382073 polymorphism. We first transfected cells with psiCHECK2-G, psiCHECK2-T and psiCHECK2-basic (as a negative control) vectors respectively. We found that relative Renilla luciferase activity decreased in psiCHECK2-T and psiCHECK2-G vector transfected cells compared to psiCHECK2-Basic cells in both cell lines and the T construct showed the lowest expression level (Fig. 1). This result indicated that rs59382073 T allele was associated with decreased gene expression compared to the G allele.

Fig. 1
figure 1

Transfection of vector containing rs59382073 T allele decreased luciferase activities. a In HEK-293T cells, compared to psiCHECK2-Basic transfected cells, psiCHECK2-G luciferase activity decreased by 29.01% (P = 4.8 × 10−8) and psiCHECK2-T by 44.88% (P = 2.8 × 10−14). b In H9c2 cells, psiCHECK2-G luciferase activity decreased by 9.9% (P = 0.02) and psiCHECK2-T by 29.61% (P = 1.2 × 10−3). Bars, SE. *P < 0.05, **P < 0.001

Full size image

Further, to explore whether the rs59382073 polymorphism changes CHD risk by affecting miRNAs biding affinities, we used RegRNA (http://regrna2.mbc.nctu.edu.tw/) to predict potential miRNAs binding sites. This prediction showed that the G to T variation created additional biding sites for five miRNAs, namely miR-127-5p, miR-28-5p, miR-3139, miR-3940, and miR-708 (Fig. 2). We then cotransfeced HEK-293T and H9c2 cells with psiCHECK2-G, psiCHECK2-T, or psiCHECK2-basic construct and miRNA mimics or negative control oligos. MiR-3940 and miR-708 exaggerated the discrepancy of luciferase activity level between psiCHECK2-G and psiCHECK2-T transfected HEK-293T cells (Fig. 3a), and so did miR-708 in H9c2 cells (Fig. 3b). Then we cotransfected cells with vector constructs and inhibitors of miR-3940, miR-708, miR-3139 or negative control oligos. Inhibitors of miR-3940 and miR-708 in HEK-293T cells and inhibitor of miR-3940 and miR-708 in H9c2 cells alleviated the discrepancy.

Fig. 2
figure 2

Five additional miRNAs binding to rs59382073 T variant predicted by RegRNA

Full size image
Fig. 3
figure 3

Rs59382073 G to T variation created additional miRNAs binding sites and caused suppressed reporter gene expression. MiR-3940 and miR-708 in HEK-293T cells (a) and miR-708 in H9c2 cells (b) exaggerated the discrepancy of luciferase activity level between psiCHECK2-G and psiCHECK2-T transfected cells. c In HEK-293T cells, inhibitors of miR-3940 and miR-708 alleviated the discrepancy. d In H9c2 cells, inhibitors of miR-3940 and miR-708 alleviated the discrepancy. N.C. negative control miRNAs. Bars, SE. *P < 0.05. **P < 0.001

Full size image

Discussion

TBX2 gene localizes in chromosome 17q23, containing seven exons,17,18,19 but no mutation in the coding region was clarified in previous studies. Previous studies reported genomic deletions and duplication at 17q23.1–q23.2, containing TBX2 gene, in patients with multiple developmental abnormalities including CHDs. A patient carrying a 131 kb duplication, encompassing the TBX2 gene, at 17q23.2 was manifested with mild mental retardation, prenatal onset growth retardation, cerebellar hypoplasia and complex heart defect.20 Another series of seven patients having 2.2 or 2.8 Mb deletion at 17q23.1q23.2, flanked by segmental duplications, were manifested with mild to moderate developmental delay, hand and foot anomalies, heart defects, in most cases either patent ductus arteriosus or atrial septal defect, and other anomalies. Within the 2.2 Mb region of overlap, TBX2 and TBX4 were strong candidates contributing to microdeletion manifestations.14 Variants in the promoter region of TBX2 (g.59477201C>T, g.59477347G>A, g.59477353delG, and g.59477371G>A) were also reported to be associated with VSD, may acting by decreasing gene expression.14 Taken together, alteration in TBX2 levels contributed to both syndromic and isolated CHD. Thus, we presumed that genetic variations in TBX2 3′UTR, which might also modulate gene expression levels, have impact on CHD development. In the present study, we examined the association between the SNPs within TBX2 3′UTR and CHD risk in a large sample set including 1241 cases and 1285 controls of Han Chinese from Shanghai and Shandong. Case–control studies showed that the polymorphism rs59382073 is significantly associated with the increased susceptibility of CHD (OR = 1.89, 95% CI = 1.48–2.46, P = 4.48 × 10−7) in the Han Chinese population, which has not been reported before.

During heart development, Tbx2 expression is restricted to the atrioventricular channel and outflow tract, and the downregulation of Tbx2 gene is essential for the right ventricle wall formation.21 The lineage tracing studies demonstrated that the Tbx2 expressing cells finally become the ventricular septum.22,23,24 Tbx2 can inhibit Tbx5-induced chamber differentiation and endocardial cushion formation,9,25 thus keeping myocytes in the primary phenotype, allowing these regions to develop another fate.4 Our study detected that, compared to GG genotypes, GT and TT genotype had a significantly increased risk of RVOTO (OR = 2.57, 95% CI = 1.39–4.76, P = 2.64 × 10−3), conotruncal (OR = 2.25, 95% CI = 1.48–3.42, P = 1.40 × 10−4), and septal defect (OR = 1.98, 95% CI = 1.50–2.60, P = 1.00 × 10−6). These affected subtypes were consistent with former molecular studies that the precise regulation of Tbx2 expression is essential to entire right ventricle and septum development.22,23,24

MiRNAs can negatively regulate gene expression by interacting with 3′ UTR of the target mRNA, cleaving or degrading the mRNA and/or inhibiting its translation and transcription.26,27,28 Studies have shown that genetic variations within 3′UTR can affect the binding affinity of miRNAs, leading to heart malformations.13,29,30,31 Using the dual-luciferase reporter assays, our study found rs59382073 G to T variation decreased reporter gene expression in both HEK-293T and H9c2 cells. Furthermore, we found five additional miRNA binding sites created by G to T variation predicted by RegRNA. Then we identified two miRNAs, miR-3940 and miR-708, that significantly exaggerated the reporter gene expression discrepancy, while their corresponding inhibitors alleviated it. Therefore, we presumed that T allele carriers have increased risks of CHD due to the reduction of TBX2 level by giving rise to biding sites for the two miRNAs and facilitating their negative regulation.

Our study discovered the association between CHD risks and rs59382073 in TBX2 3′ UTR, and revealed that the G to T variation created miRNAs binding sites for miR-3940, miR-708 and miR-3139to downregulate gene expression. Due to the insufficient cases of CHD subtypes such as APVR, AVSD, heterotaxy, and other associations, future work is needed to expand the sample size of these lesions to guarantee enough statistical power for subanalyses. Another limitation of our work is the lack of heart tissues to verify TBX2 levels in different genotypes, due to the difficulties of getting prompt cardiac samples in CHD subjects. Future studies are needed to verify the in vivo role of miR-3940 and miR-708 in decreasing TBX2 level and to explore the possible value of the two miRNAs in early CHD diagnosis.

In conclusion, the T allele of the SNP rs59382073 in TBX2 3′ UTR is associated with increased CHD susceptibility in Han Chinese population, and carriers of T allele had significantly greater risk of CHD, especially of RVOTO, conotruncal defect, and septal defect. Functional studies demonstrated that the G to T alteration created additional binding sites for miR-3940 and miR-708 to repress gene expression. Our study sheds light on genetic variants in TBX2 3′UTR in sporadic CHD, highlighting CHD risk assessment for the Han Chinese population.