Functional rare variant in a C/EBP beta binding site in NINJ2 gene increases the risk of coronary artery disease

Characteristics of study subjects

The samples involved in the current research were selected from the GeneID cohort [16–23]. To avoid geographical confounding, our genetic association studies between polymorphisms and disease were carried out in two independent populations. The GeneID-Central population, which contained 1,440 cases with CAD and 2,660 CAD-free controls, was collected from central China (Hubei province). The GeneID-North population consists of 1, 354 CAD patients and 1, 471 CAD-free controls, and are selected from the patients in hospitals in the northern area of China.

Basic clinical and demographic characteristics of the study populations were shown in Table 1. Initially, the association analysis was conducted in the GeneID-Central population that included 1,440 CAD cases and 2,660 non-CAD controls as a discovery population. The mean ages for cases and controls were 60.42 ± 12.14 and 61.25 ± 9.83 years, respectively in the GeneID-Central population. The proportions of females in cases and controls was 38.75% and 38.38%, respectively (Table 1). Association from the discovery population was then verified in another independent replication case-control population (GeneID-North population) and included 1,354 cases and 1,471 controls. The GeneID-North population included 44.60% females in cases and 46.20% in controls. The mean age for the cases was 63.67 ± 12.69 years versus 62.17 ± 10.13 years for controls (Table 1).

Variant rs34166160 in NINJ2 associates with CAD in two independent populations

The genotype of rs34166160 did not deviate from the Hardy-Weinberg equilibrium in controls in both the discovery GeneID-Central population (p = 0.99) and the replication GeneID-North population (p = 0.99). The allelic frequencies of rs34166160 in CAD cases were significantly different from those in controls in the discovery GeneID-Central population (Table 2). The frequency of minor A allele of rs34166160 was 0.44% in CAD cases and 0.13% in controls. A significant association was identified with an unadjusted p value (p_-obs) of 6.39 × 10⁻³ and an OR of 3.39. The association was also significant (an adjusted p or p_-adj = 8.12 × 10⁻³ with and OR of 3.10) after adjusting for covariates (diabetes mellitus, age, hypertension, gender, and plasma lipid concentrations) (Table 2).

To further validate the association between rs34166160 and CAD, we genotype for the rs34166160 variant in the replication GeneID-North population that contained 1,354 CAD cases and 1,471 controls. Interestingly, we also found that the A allele of the rs34166160 variant confers risk to CAD (p_-obs = 3.33 × 10⁻³ with an OR of 3.14; p_-adj = 0.01 with an OR of 2.74) in the replication population (Table 2).

After performing a meta-analysis, two populations were combined to yield one large population that included 2,794 cases and 4,131 controls. In the combined population, SNP rs34166160 was found to confer a highly significant risk to CAD (p_-obs = 1.57 × 10⁻⁵, OR = 3.58). Moreover, after adjustment for covariates including sex and age, this association remained significant (p_-adj = 3.41 × 10⁻⁴, OR = 2.80) in the combined population. These results showed that A allele of rs34166160 in the NINJ2 gene was significantly associated with CAD in the studied Chinese population.

Real-time RT-PCR analysis identified that the A allele of rs34166160 is associated with the increased expression level of NINJ2 mRNA

Rs34166160 (Chr12:623,339, hg38 version of human genome data) is located in the first intron of the NINJ2 gene. Bioinformatical analysis predicts that rs34166160 is located in an enhancer region according to the Ensembl database (Regulatory Feature: ENSR00000047562, Chr12:621,401-623,800, hg38, http://asia.ensembl.org/), and in a DNase I hypersensitive cluster (chr12:623,241-623,470) in ENCODE data of UCSC genome browser (https://genome.ucsc.edu). ENCODE Candidate Cis-Regulatory Elements (cCREs) also showed that rs34166160 was predicted to be located in an enhancer of NINJ2 (Chr12:623,197-623,527, ENCODE Accession: EH38E1585530).

Based on its position and the prediction results, we hypothesized that rs34166160 may be associated with the expression level of NINJ2 and enrolled 89 healthy study subjects undergoing annual physical examinations into GeneID to perform eQTL analysis. We measured the mRNA expression levels of NINJ2 using blood samples from 9 people with heterozygous genotype of rs34166160 (AC genotype) and 80 people with wildtype of rs34166160 (CC genotype). The results demonstrated that the mean relative expression level of NINJ2 in the 9 AC genotype carriers was significantly higher than that in the 80 CC genotype carriers (p = 0.02, by a Kruskal-Wallis test) (Figure 1). These data suggest that the minor and risk A allele of rs34166160 is associated with an increased expression level of NINJ2 mRNA.

Rs34166160 exhibits allelic differences in transcriptional activity via CCAAT-box/enhancer-binding protein beta (C/EBP beta)

Through transcript factor binding prediction analysis, we found rs34166160 located in the C/EBP beta binding region in both the PROMO (http://alggen.lsi.upc.es/) and JASPAR (http://jaspar.genereg.net/) database (Figure 2A). First, we tested whether C/EBP beta can affect the expression level of NINJ2 in Human Umbilical Vein Endothelial Cells (HUVEC). The siRNA approach was used to knockdown the expression of C/EBP beta in HUVEC, and the results showed that compared to controls, expression of NINJ2 was decreased by knocking down the expression of C/EBP beta in HUVEC (p < 0.05) (Figure 2B).

To validate whether the genomic region overlapping rs34166160 has regulatory activity and effect as a potential enhancer, we cloned a 1020 bp length genomic fragment overlapping rs34166160 harboring each allele into the PGL-3-promoter luciferase vector which contains a minimal SV-40 promoter. We first performed Dual-luciferase assays in Hela cells which did not express C/EBP beta. The results seemed to be no significant difference in transcriptional activity between fragments containing the C allele and A allele of rs34166160 without C/EBP beta expression in Hela cells (Figure 3A). However, when the luciferase transcriptional reporter assays were performed under the condition of exogenous overexpression of C/EBP beta in Hela cells, the DNA segment containing A allele of rs34166160 was about 1.3-fold than the DNA segment contained C allele of rs34166160 (p < 0.05) (Figure 3A). To further investigate the allelic differences in transcriptional activity of rs34166160, we cloned a 30 bp length core genomic segment overlapping rs34166160 harboring each allele into the PGL-3-promoter luciferase vectors containing a minimal SV-40 promoter and performed the transcriptional reporter assays same as above. The data showed the same results as the 1020 bp length segment, and the A allele of rs34166160 showed more luciferase activity than the DNA segment containing the C allele of rs34166160 (P < 0.05) (Figure 3B). The data suggest that variant rs34166160 impacts as an enhancer, and thereby may be a functional variant at the 12p13 locus for CAD.

We also performed EMSA assay using HUVEC nuclear extracts and probes flanking the genomic region of rs34166160 which is located in chr12:622,485-622,524 (hg38). Incubation of biotin-labeled probes with different allele of rs34166160 will combined with protein of nuclear extracts from HUVEC, and the probe with A allele showed more binding activity than C allele (Figure 3C). An addition of a 200-fold excess of unlabelled sequence was used to see whether the DNA-protein interaction is specific. The results demonstrated that rs34166160 exhibits allelic differences in transcriptional activity, and may be induced via C/EBP beta.

All these data suggest that the A allele of rs34166160, which confers more risk to CAD, have a higher transcriptional activity compared with the non-risk C allele under the condition of overexpression of C/EBP beta.

Risk allele A of rs34166160 has a higher transcriptional activity under the condition of endothelial activation by LPS stimuli

Previous studies have shown that NINJ2 can regulate LPS-induced endothelial activation and inflammation by interacting with TLR4, and the expression of ninjurin2 in HUVEC could be induced by LPS [19]. Vascular inflammation is known to be the critical and initial step in atherosclerosis. In addition, C/EBP beta was reported to mediate LPS-induced endothelial activation and inflammation. Therefore, we hypothesized that the enhancer containing rs34166160 might mediate the LPS-induced increase in NINJ2 expression in HUVECs.

To determine whether rs34166160 is involved in endothelial cell activation and inflammation induced by LPS stimulation, the PGL-3-promoter luciferase vectors containing the 30 bp length genomic fragment overlapping rs34166160 harboring each allele were transfected into HUVEC, and the transcriptional activity was measured using luciferase assay with or without LPS stimuli. The results showed that there were no significant differences in luciferase activity between A allele and C allele without stimulation of LPS (Figure 4). However, under the stimulation of LPS, the A allele of rs34166160 showed more luciferase activity than the DNA segment containing C allele of rs34166160 (Figure 4). These results indicated that the A allele of rs34166160, which is shown to confer more risk to CAD, have a higher transcriptional activity compared with non-risk C allele under inflammatory stimuli, and may have allelic differences in the process of inflammation.

We also investigated the expression level of C/EBP beta and ninjurin2 in HUVEC in the condition of LPS stimuli (1 ug/mL), and the results showed that both C/EBP beta and ninjurin2 expression were increased after LPS stimulation (Figure 5A). What is more, the increased expression of ninjurin2 induced by LPS stimulation can be blocked by knockdown of C/EBP beta (Figure 5B). These results indicated that rs34166160 plays an important role in LPS inducing endothelial cell activation through the binding of C/EBP beta transcription factor.

Study subjects

The samples involved in the current research were selected from the GeneID cohort [17, 19, 21]. All participants are of Han nationality according to their self-description. All the research about human subjects was approved by the ethics committee of the university (Huazhong University of Science and Technology, HUST). The study confirms the guidelines set forth by the Declaration of Helsinki, and all study subjects give written informed consent.

Diagnosis of CAD and myocardial infarction (MI) was carried out by two independent expert cardiologists based on the results of coronary angiography. The criteria of myocardial infarction (MI) and CAD were according to the standard ACC/AHA guidelines [49], and as previous studies [50–53]. Patients with greater than 70% of luminal stenosis in at least one main vessel by coronary angiography, coronary artery bypass graft, percutaneous coronary intervention, and/or MI were diagnosed as CAD patients. The diagnosis of MI was based on typical chest pain of ≥30 min duration, characteristic electrocardiographical patterns of acute MI, and significant elevation of cardiac enzymes (such as CK-MB, lactate dehydrogenase) and troponin I or T. Patients that were identified by angiography with myocardial spasm and myocardial bridge were excluded from the study. Subjects with congenital heart disease, childhood hypertension, and type I diabetes mellitus were also excluded. Control subjects were general population controls and did not show any evidence of CAD or MI as indicated by their medical records or by electrocardiographic analysis.

Basic data such as gender, age, diabetes mellitus, hypertension and lipid concentrations (total cholesterol, triglyceride, LDL cholesterol and HDL cholesterol) were collected from medical records.

SNP genotyping

Genotyping for SNP rs34166160 was carried out using the TaqMan allelic discrimination assay as previously described [54]. For each round of genotyping, 10 ng of genomic DNA was used in a total volume of 5 μl and containing 1× TaqMan Genotyping Master Mix (Life Technologies) and 1× TaqMan SNP genotyping probe (Life Technologies, CA, USA). The PCR reaction and post-PCR endpoint plate reading was carried out using the Applied Biosystems 7900HT Fast Real-Time PCR System (Life Technologies, CA, USA) according to the manufacturer’s instructions. To ensure the quality of the experiment, appropriate negative controls were included in each round of genotyping. The genotyping results of TaqMan assay were verified by direct Sanger DNA sequencing of 48 subjects randomly selected.

Real-time quantitative reverse transcription PCR analysis

Real-Time Quantitative Reverse Transcription PCR (qRT-PCR) was used to determine the mRNA expression level of the NINJ2 among subjects with different genotypes of rs34166160. RNA was obtained from blood sample by Trizol (Life Technologies, CA, USA). Real-Time Quantitative Reverse Transcription PCR Analysis was used to determine the mRNA expression level of the NINJ2 gene among subjects with different genotypes of rs34166160. Quantitative real-time PCR analysis was carried out according to the MIQE guidelines. Total RNA samples were extracted from human peripheral blood leukocytes using Trizol reagent (Life Technologies, CA, USA). Quantification of RNA samples was performed using a spectrophotometer (NanoDrop, Thermo Scientific, NH, USA). 2 μg of total RNA was used for reverse transcription with Superscript II reverse transcriptase (Life Technologies, CA, USA) and oligo (dT)₁₈. A standard two steps real-time PCR assay was performed using an ABI 7500fast Genetic Analyzer (Life Technologies, CA, USA). The primers for NINJ2 were 5′-ATGCGGCTGAAGGCGGTGCTG-3′ (forward) and 5′-TGGCTGCGTTGTTGAGCTGGTTG-3′ (reverse). The primers for the β-actin reference gene (GeneBank ID: BC014861) are 5′-GGACTTCGAGCAGGAGATGG-3′ (forward) and 5′-GCACCGTGTTGGCGTAGAGG-3′ (reverse). Each PCR reaction was performed in a final volume of 10 μL reaction mixture containing 5 μL of 2X PCR master mixture with ROX (Roche Applied Science, IN), 2 μL of cDNA, 0.4 μL of 10 pM primers, and 2.6 μL of ddH2O. Each reaction was performed in triplicate. The cycling conditions were 95°C for 10 minutes and 40 cycles of 95°C for 15 seconds and 60°C elongation for 45 seconds. After the PCR reaction, Cq values (threshold cycle) of a target gene (NINJ2) (Cq T) or reference gene β-actin (GeneBank ID: BC014861, Cq E) were computed using the RQ Manager program (version 1.3) and SDS (version 2.3). Reaction with a Cq of ≥40 or with the difference between Cq and mean Cq greater than 0.5 were excluded for further analysis. For each individual, the relative expression level ΔCq (Cq T-Cq E) of a target gene was normalized with the reference gene and then transformed into relative quantity using RQ formula (RQ = 2^−ΔΔCq, ΔΔCq= individual’s ΔCq-calibrator’s ΔCq). The calibrator was a mixed cDNA sample pooled from 10 randomly selected individuals. The RQ for the calibrator was normalized to 1. After outliers were excluded, a Kruskal-Wallis test (nonparametric analysis of variance) was used to compare the differences for mean RQ values of NINJ2 between different genotypes of SNP rs34166160.

Cell culture and treatment

HUVEC were obtained from Wuhan Pricells (Wuhan, Hubei, China). Primary Umbilical Vein Endothelial Cells (HUVECs) were obtained from Wuhan Pricells (Wuhan, Hubei, China). The research was performed using 3 distinct cell batches. HUVECs were maintained in endothelial cell medium (ECM, ScienCell, CA, USA) containing Endothelial Cell Growth Supplement (ECGS, ScienCell, CA, USA), 20% FBS (Gibco, MD, USA), 100 mg/L heparin (Invitrogen, CA, USA), 1 mM Sodium Pyruvate and 1% penicillin/streptomycin (Invitrogen, CA, USA) and used at passage 3–6. Hela cells were cultured in DMEM (Gibco, MD, USA) containing 10% FBS, L-glutamine (2 mM), penicillin G (100 units/ml), streptomycin (10 mg/ml). All cells were grown in an incubator with a humidified atmosphere of 5% CO₂ at 37°C. Cell were plated in 24-well plates for 24 h and accompanied by incubation with or without LPS (1 ug/mL) for another 24 h.

Expression vector construction

The predicted 1020 bp length C/EBP beta binding region flanking rs34166160 in the first intron of NINJ2 (Chr12:622,453-623,472, hg38) was generated by PCR and using human genomic DNA as a template. The primer sequences used for amplification are 5′-GAGCTCGGGCTACCTAAAGAGAGGAAGA-3′ contained a Sac I restriction enzyme cutting site and 5′-CTCGAGCTTCCCTGATTTTGGCTGGTAC-3′ contained a Xho I site. The 1020 bp length PCR products were digested with Sac I and Xho I, and sub-cloned into the pGL3-Promoter luciferase plasmid (Promega, Madison, WI, USA), resulting in the pGL3-promoter-1020 bp-A or pGL3-promoter-1020 bp-C plasmid. The 30 bp length fragments flanking rs34166160 was synthesized by Genewiz ltd (Suzhou, China). The DNA fragment (5′-CGGGGTACCTCTGTCCCCCTCCCCCACTGCTACCCGAGCCTCGAGCGG-3′ for C allele of rs34166160 and 5′-CGGGGTACCTCTGTCCCCCTCCCCAACTGCTACCCGAGCCTCGAGCGG-3′ for A allele of rs34166160) and their corresponding complement chains were annealed and then sub-cloned into the pGL3-Promoter luciferase plasmid after digesting by Kpn I and Xho I. Human C/EBP beta ORF (NM_005194) was purchased from Origene (Rockville, MD, USA) and sub-cloned into pENTR vector (Thermo fisher, Pittsburgh, PA, USA).

Luciferase reporter assay

Hela cells co-transfected with 150 ng of either the pENTER control vector or pENTER- C/EBP beta and 50 ng pGL3-promoter luciferase reporter plasmids contained C allele or A allele of rs34166160 in a 96-well plate [55]. The transfection efficiency was normalized by co-transfecting of 4 ng pRL-TK plasmid. Cells were lysed using a passive lysis buffer after 48 hours of transfection (Promega, Madison, WI, USA). The dual-luciferase assay was analyzed using the dual-luciferase assay kit (Promega, Madison, WI, USA) using GloMax illuminometer (Promega, Madison, WI, USA).

Western blot

Western blot was performed as standard protocol [56]. The antibodies used for western blot analysis include an anti- C/EBP beta antibody (1:1500, Proteintech, Wuhan, China), an anti-alpha tubulin (1:5000, Merck Millipore, Darmstadt, Germany) and a goat polyclonal anti-NINJ2 antibody (1:1200, R&D Systems, Minneapolis, MN, USA).

Electrophoretic Mobility Shift Assay (EMSA)

pENTER- C/EBP beta plasmid was transfected into HUVEC for 40 hours using Viafect™ reagent. EMSA probes were designed according to the predicted C/EBP beta binding site on SNP rs34166160. The sequence of wild type C allele probe (5′-AGCCTCTGTCCCCCTCCCCACTGCTACCCGA-3′) and the minor A allele probe (5′-AGCCTCTGTCCCCCTCCCAACTGCTACCCGA-3′) was synthesized in Sangong Biotech and with a 5′-end-labeled biotin. NE-PER Nuclear and Cytoplasmic Extract kit was used to extract nuclear proteins (Thermo fisher, Pittsburgh, PA, USA). 2 μg of extracted nuclear protein was hybridized with WT and Mutation DNA probes, respectively. Protein-DNA complexes were analyzed on a 6% nondenaturing PAGE gel. Excessive unlabeled probes were used as competition for EMSA experiments. LightShift™ Chemiluminescent EMSA Kit was used for the EMSA assay (Thermo fisher, Pittsburgh, PA, USA).

Statistical analysis

For genetic association studies, statistical analysis was carried out as previously reported [17, 21]. Pearson’s 2 × 2 contingency tables and Chi-square tests as implemented in PLINK version 1.08 were used to analyze the SNP allelic association study [57]. Risk factors including lipid concentrations, age, gender, hypertension and diabetes mellitus and lipids profiles were used to adjust the association using multivariate logistic regression analysis. PS:Power and Simple Size Calculation software was used to analyze the statistical power [58]. For western blot and other functional studies, results from at least three independent experiments are shown and presented as mean ± SD. Student’s t tests were used for data analysis between the two groups using GraphPad Prism 6 software. P < 0.05 means that there is a significant difference between the data and is statistically significant (^*means P < 0.05).