3.1 Characteristics of the subjects and variants
This study adopted the method of case-control study. This study included 949 CAD cases with a mean age of 66.2±10.7 years and 892 controls with a mean age of 65.6±12.7 years. Cases and controls were frequency matched according to age and gender. The basic clinical data of the study subjects were shown in Table 1. The pairwise LD of the five SNPs was calculated for the 949 case patients and 892 age- and sex-frequency-matched controls. The haplotype block was generated from the Haploview software (supplementary 2).
Table 1. General characteristics of CAD cases and controls
Variants
|
Case(N=949)
|
Control(N=892)
|
p-value
|
Male/female, %
|
49.3/50.7
|
49.8/50.1
|
0.82
|
Age(years)
|
66.2±10.7
|
65.6±12.7
|
0.30
|
BMI (kg/m2)
|
23.4±3.0
|
23.6±3.2
|
0.50
|
Smoking, (yes/no)
|
25.2/74.8
|
16.0/84.0
|
<0.01
|
Drinking, (yes/no)
|
11.1/88.9
|
17.9/82.1
|
<0.01
|
FPG (mmol/L)
|
6.7±2.8
|
6.0±2.5
|
<0.01
|
SBP
|
136.0±19.4
|
131.3±18.4
|
<0.01
|
DBP
|
76.6±12.4
|
80.0±10.8
|
<0.01
|
TC (mmol/L)
|
4.3±1.2
|
4.8±1.1
|
<0.01
|
TG (mmol/L)
|
1.6±1.2
|
1.5±1.1
|
0.10
|
HDL (mmol/L)
|
1.2±0.4
|
1.3±0.5
|
<0.01
|
LDL (mmol/L)
|
2.5±1.0
|
2.9±1.0
|
<0.01
|
Hypertension, (yes/no)
|
31.9/68.1
|
56.1/43.9
|
<0.01
|
Diabetes, (yes/no)
|
70.0/30.0
|
86.8/13.2
|
<0.01
|
Coronary angiography
|
378
|
4
|
…
|
0-vessel disease, %
|
143
|
4
|
…
|
1-vessel disease, %
|
115
|
…
|
…
|
2-vessel disease, %
|
64
|
…
|
…
|
3-vessel disease, %
|
56
|
…
|
…
|
LM, %
|
19
|
…
|
…
|
LAD, %
|
217
|
…
|
…
|
RAD, %
|
112
|
…
|
…
|
LCX, %
|
109
|
…
|
…
|
Abbreviations: FPG, fasting plasma glucose; SBP, Systolic blood pressure; DBP, Diastolic blood pressure; TC, Total cholesterol; TG, triglycerides; HDL, High density lipoprotein; LDL, Low-density lipoprotein; LM, left main coronary artery disease; LAD, left anterior descending artery disease; LCX, left circumflex artery disease; RAD, right coronary artery disease.
3.2 Association between SNPs and CAD risk
The distribution of genotype and allele frequency in the case group and the control group were shown in Table 2. The results of univariate analysis showed that there was no statistical difference in the frequency distribution of the five SNPs between the case-control groups. Considering that confounding factors may bias the results of the study, we further adopted a binary unconditional logistic regression model. As represented in the table, SNP rs2327433 in TARID was associated with CAD after adjusting for sex, age, smoking, drinking, diabetes and five SNPs. The rs2327433 GG genotype was found as a risk factor for CAD in adults (OR=2.74, 95%CI: 1.10-6.83, P=0.03). As shown in supplementary 3, stratified analysis indicated a significant association between SNPs (rs2327433 and rs1966248) and risk of CAD. Our results showed that genotype GG of the SNP rs2327433 was associated with risk of CAD in the male’s group and the drinking group (OR =3.37, 95%CI: 1.22-9.32, P=0.02 and OR=9.72, 95%CI: 1.14-82.83, P=0.04, respectively). The results indicated that genotypes AA carriers of rs1966248 had significant association with CAD risk in non-diabetes subjects (OR=1.45, 95%CI: 1.04-2.03, P=0.03).
Table 2
Association of 5 SNP in TARID gene with CAD Cases and Controls in the Chinese Han population
SNPs | Genetype | Case (n, %) | Control (n, %) | Crude OR (95%CI) | p-value | Adjusted OR (95%CI) | p-value* |
rs2327429 | TT | 235(25.4) | 206(23.9) | 1.00 | | | |
TC | 466(50.4) | 457(53.0) | 0.89(0.71-1.12) | 0.33 | 0.63(0.33-1.19) | 0.15 |
CC | 223(24.1) | 199(23.1) | 0.98(0.75-1.28) | 0.90 | 0.44(0.16-1.17) | 0.10 |
Allele | | | | | | |
T | 912(49.4) | 855(49.6) | 1.00 | | | |
C | 936(50.6) | 869(50.4) | 0.99(0.87-1.13) | 0.88 | | |
rs2327433 | AA | 594(66.3) | 579(69.5) | 1.00 | | | |
AG | 271(30.2) | 236(28.3) | 1.12(0.91-1.38) | 0.29 | 1.19(0.85-1.67) | 0.31 |
GG | 31(3.5) | 18(2.2) | 1.68(0.93-3.03) | 0.09 | 2.74(1.10-6.83) | 0.03 |
Allele | | | | | | |
A | 1459(81.4) | 1394(83.7) | 1.00 | | | |
G | 333(18.4) | 272(16.3) | 1.17(0.98-1.40) | 0.08 | | |
rs12190287 | CC | 315(36.5) | 285(34.7) | 1.00 | | | |
CG | 411(47.6) | 419(51.0) | 0.89(0.72-1.10) | 0.27 | 1.21(0.75-1.97) | 0.44 |
GG | 137(15.9) | 118(14.4) | 1.05(0.78-1.41) | 0.74 | 1.48(0.65-3.41) | 0.35 |
Allele | | | | | | |
C | 1041(60.3) | 989(60.2) | 1.00 | | | |
G | 685(39.7) | 655(39.8) | 0.99(0.87-1.14) | 0.93 | | |
AA | 170(18.8) | 139(16.6) | 1.00 | | | |
AT | 424(46.9) | 391(46.8) | 1.07(0.87-1.32) | 0.44 | 1.17(0.83-1.65) | 0.38 |
TT | 310(34.3) | 306(16.6) | 1.21(0.92-1.59) | 0.49 | 1.29(0.74-2.24) | 0.37 |
rs1966248 | Allele | | | | | | |
T | 1044(57.7) | 1003(60.0) | 1.00 | | | |
A | 764(42.3) | 669(40.0) | 1.10(0.96-1.26) | 0.18 | | |
rs6569912 | CC | 273(30.5) | 255(30.6) | 1.00 | | | |
CT | 434(48.5) | 416(49.9) | 0.97(0.78-1.21) | 0.82 | 1.20(0.72-1.90) | 0.48 |
TT | 187(20.9) | 163(19.5) | 1.07(0.82-1.41) | 0.62 | 1.37(0.61-3.08) | 0.45 |
Allele | | | | | | |
C | 980(54.8) | 926(55.5) | 1.00 | | | |
T | 808(45.2) | 742(44.5) | 1.03(0.9-1.18) | 0.68 | | |
* The P values and OR (95% CI) were calculated from logistic regression analyses adjusting for sex, age, smoking, drinking, diabetes and five SNPs. |
The results of GMDR model analysis for the entire case-control data were shown in Table 3, which showed that the best model was the four-factor model. This model was adjusted for smoke. There were significant interactions with polymorphisms in the four genes in TARID.
Table 3
The models to predict risk of CAD by GMDR.
Factor | Training Bal. Acc | Testing Bal. Acc | Sign Test(P)* | CV consistency |
[rs1966248] | 0.5199 | 0.4875 | 3(0.9453) | 6/10 |
[rs1966248 rs6569912] | 0.5329 | 0.4662 | 0(1.0000) | 5/10 |
[rs1966248rs12190287 rs6569912] | 0.5559 | 0.4922 | 4(0.8281) | 6/10 |
[ rs1966248 rs2327433 rs12190287 rs6569912] | 0.5876 | 0.5189 | 9(0.0107) | 10/10 |
[ rs1966248 rs2327429 rs2327433 rs12190287 rs6569912] | 0.6018 | 0.5036 | 7(0.1719) | 10/10 |
*Adjusted for smoking. GMDR: Versatile software for detecting gene-gene and gene-environment interactions underlying complex traits. |
We used the eQTL research platform provided by the GTEx (version 8) database to analyze the relationship between SNP rs2327433 and the expression levels of host genes and adjacent CAD-related genes. The eQTL analysis can confirm SNP-related gene expression differences[20]. We have listed the existing organizational data in supplementary 4. The results showed that the different genotypes of rs2327433 was significant correlated with the expression level of TCF21 gene. The results reveal that in the esophageal mucosa (enriched endothelial cells) the expression of TCF21 is lower in the GG genotype of SNP rs2327433. TCF21 is a protein-coding gene adjacent to the downstream of TARID. By performing eQTL analysis in aortic tissue from the STARNET database25 on this larger set of SNPs, studies found that the CAD risk imparted by rs2327433 was also correlated with lower TCF21 expression from that allele[21]. Reduced TCF21 expression is associated with increased coronary disease risk[21].
3.3 Association between CAD-related variant rs2327433 and the severity of CAD
The binary logistic regression analyses revealed that the genotype of rs2327433 was related to the proportion of CAD patients with left anterior descending artery disease (LAD) (Figure 1a) and left circumflex artery disease (LCX) (Figure 1b) after adjusting for traditional risk factors such as age, smoking, gender, drinking (P=0.025 and P=0.025, respectively). We further assessed the effect of SNPs on the development of CAD. The results showed that the minor allele frequency of rs2327433 was significantly correlated with the severity of the disease (P = 0.029) (Figure 1c).
3.4 TARID silencing affected cell cycle progression and the expression of cell cycle-related genes
LncRNA TARID activates the expression of TCF21 by inducing promoter methylation[17]. The SNP site rs2327433 of lncRNA TARID is related to the expression of TCF21. On the esophageal mucosa, the expression of TCF21 is lower in the GG genotype of SNP rs2327433. We speculate that the GG genotype of SNP rs2327433 affects the expression of lncRNA TARID, preventing it from activating TCF21. Then the expression of lncRNA TARID in patients with coronary heart disease is not clear, we use human peripheral blood mononuclear cells to detect the expression level of lncRNA TARID. The results showed that after normalization with housekeeping gene (18s), compared with healthy subjects, the expression level of TARID in peripheral blood mononuclear cells of CAD cases was significantly reduced (P <0.01) (Figure 2a).
Studies have shown that TCF21 affects cell cycle progression of cells[22]. However, it is not clear whether the silencing of lncRNA TARID can affect the cell cycle through the TCF21 pathway. As displayed in Figure 2b, the expression of TARID RNA in THP-1 or HUVEC cells transfected with siTARID-1 were significantly lower than in NC group, respectively. To assess how TARID silencing affects cell cycle progression, we conducted flow cytometry to analyze the distribution of cells across the major phases of the cell cycle in THP-1 cells transfected with TARID siRNA. The results revealed that TARID silencing resulted in an increase of THP-1 cells in the S-phase from 36.88% to 42.97 %, and decreased accumulation in the G2 phase from 4.85–2.93%, relative to the control group (Figure 2c).
To further study the effect of TARID in cell cycle regulation, qRT-PCR was performed to determine cell cycle-related genes. As shown in Figure 2d, TARID silencing up-regulated CDK1 and PCNA in THP-1 cells or HUVECs.