Genetic and epigenetic associations of ANRIL with coronary artery disease and risk factors

Background Both DNA genotype and methylation of antisense non-coding RNA in the INK4 locus (ANRIL) have been robustly associated with coronary artery disease (CAD), but the interdependent mechanisms of genotype and methylation remain unclear. Methods Eighteen tag single nucleotide polymorphisms (SNPs) of ANRIL were genotyped in a matched case–control study (cases 503 and controls 503). DNA methylation of ANRIL and the INK4/ARF locus (p14ARF, p15INK4b and p16INK4a) was measured using pyrosequencing in the same set of samples (cases 100 and controls 100). Results Polymorphisms of ANRIL (rs1004638, rs1333048 and rs1333050) were significantly associated with CAD (p < 0.05). The incidence of CAD, multi-vessel disease, and modified Gensini scores demonstrated a strong, direct association with ANRIL gene dosage (p < 0.05). There was no significant association between ANRIL polymorphisms and myocardial infarction/acute coronary syndrome (MI/ACS) (p > 0.05). Methylation levels of ANRIL were similar between the two studied groups (p > 0.05), but were different in the rs1004638 genotype, with AA and AT genotype having a higher level of ANRIL methylation (pos4, p = 0.006; pos8, p = 0.019). Further Spearman analyses indicated that methylation levels of ANRIL were positively associated with systolic blood pressure (pos6, r = 0.248, p = 0.013), diastolic blood pressure (pos3, r = 0.213, p = 0.034; pos6, r = 0.220, p = 0.028), and triglyceride (pos4, r = 0.253, p = 0.013), and negatively associated with high-density lipoprotein cholesterol (pos2, r = − 0.243, p = 0.017). Additionally, we identified 12 transcription factor binding sites (TFBS) within the methylated ANRIL region, and functional annotation indicated these TFBS were associated with basal transcription. Methylation at the INK4/ARF locus was not associated with ANRIL genotype. Conclusions These results indicate that ANRIL genotype (tag SNPs rs1004638, rs1333048 and rs1333050) mainly affects coronary atherosclerosis, but not MI/ACS. There may be allele-related DNA methylation and allele-related binding of transcription factors within the ANRIL promoter. Supplementary Information The online version contains supplementary material available at 10.1186/s12920-021-01094-8.

Most of the 9p21.3 CAD-associated genomic variants are located within ANRIL [7], also, expression of ANRIL has been shown to associated with CAD and MI [8,9], which makes ANRIL the most robust genetic marker of CAD today. However, inconsistent results have been reported [10]. Whether ANRIL polymorphisms are associated with the likelihood of CAD and its main effect (coronary atherosclerosis or plaque instability) on CAD remains controversial [11]. As an effector gene in 9p21. 3, ANRIL can regulate its adjacent INK4/ARF locus in cis, as well as the distant loci in trans [9,10,12]. Causal variants at ANRIL can disrupt predicted transcription factor binding sites (TFBS) [13] and modulate ANRIL expression and/or structure [9,14,15], playing a pivotal role in mediating the 9p21.3 susceptibility for CVD. On the other hand, epigenetic modification, such as DNA methylation, at the ANRIL and INK4/ARF loci has been implicated in the genetic cause of CVD [16][17][18][19]. In cancer cells, transcription of ANRIL and p16 INK4a is regulated by the methylation status of p16 INK4a [20]. However, the relationship between polymorphism of ANRIL and DNA methylation of ANRIL and the INK4/ARF locus in CAD patients has not been examined. In this study, we systematically examined the polymorphisms of ANRIL using haplotype tag SNPs and detected their associations with CAD risk and the severity of coronary atherosclerosis, and then explored the potential relationship between the polymorphism of ANRIL and DNA methylation of ANRIL and the INK4/ARF locus in a Chinese population.

Subjects
Angiographic CAD was determined by blinded coronary angiographic analysis and defined as having > 50% diameter stenosis in at least 1 major epicardial coronary artery. Modified Gensini coronary scores were used to assess the severity of CAD [21,22]. For a vessel to be scored, stenosis > 50% had to be noted in an epicardial coronary vessel of its major branches [23].
Subjects with other cardiac diseases (congenital heart disease, cardiomyopathy, or rheumatic heart disease), cerebrovascular or neurological diseases, cancer, severe liver or kidney disease were excluded from the study. Furthermore, patients who had received angioplasty, intravenous thrombolysis, coronary artery stents, or coronary artery bypass graft surgery before the enrollment were also excluded.
Finally, 503 angiographic CAD cases and 503 age-(3-year bands) and sex-matched controls were selected from consecutive patients undergoing diagnostic or interventional coronary angiography within the First Affiliated Hospital of Shantou University Medical College from April 20, 2015 to August 20, 2016. One hundred eighty-eight MI/ACS and 188 age-(3-year bands) and sex-matched angiographic CAD controls (CAD patients without MI/ACS) were selected according to the third universal definition of myocardial infarction [24] from the same hospital during the same period. The controls were free of MI/ACS by questionnaires, history-taking, detection of troponin-T and myocardial enzymes, electrocardiography, chest X-ray, and Doppler echocardiography.
Demographic and clinical data required for this study were obtained from physician and hospital records, and included age, sex, health history (CAD, MI/ACS, hypertension, type 2 diabetes, hyperlipidemia), vital signs at entry, medication use, personal hobbies (smoking, alcohol use), total cholesterol (TC), low-density lipoprotein cholesterol (LDL), high-density lipoprotein cholesterol (HDL), triglycerides (TG), glycosylated hemoglobin (HbAlc), creatinine, uric acid, and systolic and diastolic blood pressure (SBP and DBP).
All subjects were from the Chinese Han population and gave informed consent prior to the study. Ethics approval was obtained from the Ethics Committee of the First Affiliated Hospital of Shantou University Medical College.

SNP selection and genotyping
An arterial blood sample was taken at the time of the catheterization for deoxyribonucleic acid extraction and subsequent genotyping. Genomic deoxyribonucleic acid was isolated with a FlexGen Blood DNA Kit according to the manufacturer's protocol (CoWin Biosciences). Eighteen tag SNPs of ANRIL were selected by Haploview software (Version 4.1). The minor allele frequency of each SNP was > 5% in the HapMap of the Chinese Han Beijing (CHB) population (see Additional file 1: Table S1).
Genotyping was performed with the Access Array micro-fluidics PCR platform (Fluidigm Corporation, South San Francisco, California, USA) according to the standard instructions [25]. Fluidigm SNP Genotyping Analysis (Version 4.3.2) software was used for data management and analyses.

Methylation analysis of ANRIL, p14 ARF , p15 INK4b and p16 INK4a
Genomic DNA was isolated as mentioned above. Bisulfite treatment of DNA was done by using the EZ DNA Meth-ylationTM kit (ZYMO Research, Orange, CA) according to the manufacturer's protocol. Then the resulting bisulfite-treated DNA was purified and eluted in 20 μl of M-elution buffer, and 4 μl of this was used in the methylation-specific PCR (MSP) amplification. Primers for MSP amplification were designed with the use of the bioinformatics program (http:// www. uroge ne. org/ methp rimer/ index1. html) and are shown in Additional file 1: Table S2. After amplification, the MSP products were analyzed by bisulfite pyrosequencing (PyroMark Q96 System version 2.0.6, Qiagen).

Functional annotation of the ANRIL methylation region
PROMO (version 8.3 of TRANSFAC) was used to predict the putative TFBS within the ANRIL methylation region. The Database for Annotation, Visualization and Integrated Discovery (DAVID) v6.8 was used to annotate the function of the predicted TFBS. Pathways that were potentially affected by gene DNA methylation were generated by Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis. SPSS 19.0 (IBM Corp© 2010) and SNPassoc for R statistical package were used for statistical analyses. Haploview software (version 4.1) was used for analyses of the pairwise linkage disequilibrium, haplotype structure and selecting tagging SNPs. All p-values were two sided, and p < 0.05 was considered statistically significant.

Statistical analysis
Continuous variables were reported as the mean value ± standard deviation (SD). Data with a normal distribution were compared by Student t test or ANOVA test, and those with unequal variance or without a normal distribution were analyzed by a Mann-Whitney rank sum test or Spearman correlation test. Categorical variables were expressed as frequencies with percentages and were compared by the chi-squared (χ 2 ) test. A trend test was also examined by the χ 2 test (linear-by-linear association). Genotypic frequencies in cases and controls were tested for departure from Hardy-Weinberg equilibrium using a Fisher's exact test. Genetic model analyses (dominant, recessive) were applied to assess the significance of SNPs, and the allelic frequencies were compared between cases and controls by χ 2 /Fisher's exact test. The associations between CAD and genotypes of the SNPs were estimated by computing the odds ratios (ORs) and 95% confidence intervals (CIs) from the multivariable logistic regression, and adjusted for sex, age, smoking, alcohol use, hypertension, type 2 diabetes, blood lipid (TC, TG, HDL, LDL), creatinine and uric acid.

General characteristics of the subjects
The main demographic and clinical characteristics of the CAD cases and controls are summarized in Table 1. Mean age of the cases was 59.67 years old (from 39 to 87), mean age of the controls was 60.42 years old (from 33 to 85), there was no significant difference between the two groups (p = 0.198). As expected, other traditional risk factors, such as hypertension, type 2 diabetes and hyperlipidemia were more prevalent in cases than the controls (p < 0.05). Accordingly, mean levels of SBP, HbAlc, TG and HDL were significantly different between the two groups (p < 0.05). DBP, TC and LDL did not differ between the cases and controls (p > 0.05), which could be the result of antihypertensive and cholesterol-lowering drugs in the patients after diagnosis.

Association of ANRIL genotype with CAD
Of the 18 tag SNPs, rs10965227 and rs10965241 did not conform to the Hardy-Weinberg equilibrium test in controls and cases (see Additional file 1: Table S1), and so were not used for further analysis. Univariate analyses found 5 SNPs (rs1004638, rs1333048, rs1333050, rs4977756, rs9632885) to be significantly associated with CAD (see Additional file 1: Table S3), while the other 11 SNPs had no association with CAD (data not shown).
The linkage disequilibrium was defined as D ≥ 95, and five haplotype blocks were identified in the ANRIL region (see Additional file 1: Fig. S1). Haplotype analysis found blocks 1, 2, 3, and 5 were not significantly associated with CAD. rs1004638 and rs1333048 showed strong linkage in block 4, and the haplotype AC (A, C are the risk alleles of rs1004638 and rs1333048, respectively) was significantly associated with an increased risk of CAD (OR = 1.42, 95% CI 1.22-1.66; p < 0.001).

Effect of ANRIL genotype on CAD severity and MI/ACS
There was a strong positive association between the incidence of CAD and increasing gene dose of the rs1004638 (p < 0.001), rs1333048 (p < 0.001) and rs1333050 (p = 0.034) risk variants. Multi-vessel disease (p = 0.027) increased as increasing gene dosage of rs1333048 risk variant. Gensini scores in the mutant homozygote of rs1333048 (CC, p < 0.001) and rs1333050 (TT, p = 0.001) were significantly higher than that of the heterozygote or wild-type homozygote. Patients with two hazardous alleles of ANRIL were more likely to suffer severe CAD (Table 3).
In addition, we performed a stratified analysis by age. Premature CAD (PCAD) in this study was defined as CAD occurring in males < 55 years of age and females < 65 years of age [26]. Accordingly, late-onset CAD (LCAD) was defined as males ≥ 55 years of age and females ≥ 65 years of age. Multivariable logistic regression analysis showed that there were appreciably higher ORs for the risk alleles in the PCAD set (p < 0.05) when compared with the LCAD set (p ≥ 0.05) ( Table 4).
To explore the effect of ANRIL on plaque instability, another case-control study (188 MI/ACS and 188 age-(3-year bands) and sex-matched CAD controls) was carried out. The results showed that there was no  Table S4).

Methylation of ANRIL, p14 ARF , p15 INK4b and p16 INK4a in the study population
To further explore the possible molecular mechanism of CAD, we assessed DNA methylation of the ANRIL, p14 ARF , p15 INK4b and p16 INK4a promoter region in CAD and control subjects. The one-sample K-S test was used to test the normality of DNA methylation levels, and it was found that the data did not conform to a normal distribution, so the Mann Whitney rank sum test was used to compare the differences between groups. As shown in Fig. 1, only one CpG site within the third CpG island (pos3) located upstream of p15 INK4b was hypermethylated in CAD subjects compared to the matched controls (p = 0.025). No differences in ANRIL or p14 ARF methylation were observed (P > 0.05). Methylation of p16 INK4a was at barely detectable levels in both CAD patients and controls, so the data was not shown.

Association of ANRIL genotype with methylation
Further, we assessed the association between CAD risk genotypes and methylation of ANRIL, p14 ARF and p15 INK4b . CAD-associated rs1004638 was associated with methylation of ANRIL. Compared with carriers of TT genotypes, AA and TT genotype carriers of rs1004638 had markedly elevated levels of methylation [ANRIL pos4 (p = 0.006) and pos8 (p = 0.019)]. rs1333048 and rs1333050 had no effect on the methylation level of ANRIL (Fig. 2). There were no differences between CADassociated SNPs (rs1004638, rs1333048, rs1333050) and methylation of p14 ARF or p15 INK4b (data not shown).

Functional annotation of ANRIL methylation region
We predicted 12 TFBS in DNA sequences within the ANRIL methylation region (Fig. 4). Functional annotation by KEGG pathway enrichment analysis indicated these TFBS-binding sites were associated with basal transcription (Fig. 5). Among them, transcription factor IID (TFIID) was identified as a key transcription factor, which is composed of the TATA-binding protein and binds to the core promoter to position the RNA polymerase II properly, serves as the scaffold for assembly of the remainder of the transcription complex, and acts as a channel for regulatory signals [27][28][29][30].

Discussion
In the present study, we demonstrated that: (1) tag SNPs (rs1004638, rs1333048, rs1333050) of ANRIL significantly influence the hazard of CAD in a Chinese Han population. Homozygous carriers showed higher coronary atherosclerosis risk, whereas heterozygous carriers showed intermediate risk between that of wild-type and homozygous carriers, indicating a genetic dose effect, especially for premature CAD.
(2) ANRIL polymorphism was significantly associated with CAD severity, but not MI/ACS. (3) DNA methylation levels of ANRIL were not associated with CAD, but were associated with rs1004638 and CAD risk factors (SBP, DBP, TG and HDL). (4) Twelve TFBS were predicted within the ANRIL methylation region. These findings indicate that there may be allele-related DNA methylation and allele-related binding of transcription factors within the ANRIL promoter region.
Our results are consistent with previous studies conducted in China and other populations in which ANRIL polymorphisms were found to be associated with CAD [11,31]. A quantitative assessment of the severity of coronary stenosis is better than a binary phenotype [32], while previous studies found that ANRIL confers risk for CAD vs. controls. So, a validated semi-quantitative angiographic score, Gensini scores, was used to estimate severity of CAD in this study. Our results showed a dose-effect relationship between the Gensini scores and ANRIL risk alleles. Homozygotes of the rs1333048 risk allele are more likely (72.79%) to suffer multi-vessel disease than wild-type carriers (60.58%), thus providing clues for predicting severity of CAD. Considering that ANRIL is related to the degree of coronary stenosis, we wondered whether it is also related to MI/ACS. MI/ACS patients are prone to suffer multiple vessel lesions [23]. Over 50% of STsegment elevation MI patients have multiple vessel lesions, which indicates poor prognosis [33]. However, our results demonstrate that there is no association between ANRIL and MI/ACS when both cases and controls have coronary stenosis. Such an analytical method has been used in previous studies as a means to distinguish the effect of genetic factors on CAD from MI [34]. However, by comparing MI cases with healthy controls, Abdul Azeez et al. [35] and Cheng et al. [36] did not distinguish the effect of ANRIL on CAD and MI, but showed that ANRIL is associated with MI in both the Saudi population and Chinese Han population. Given that MI is a more downstream phenotype of CAD, comparisons between MI and healthy controls means comparing CAD with non-CAD. Hence, it can be considered that ANRIL mainly influences coronary atherosclerosis, instead of MI/ACS, which is pathophysiologically distinct from CAD [37].
Risk factors for CAD are both environmental and genetic. PCAD has a stronger genetic susceptibility [38], and heritability of PCAD is more obvious than that of the LCAD [39]. We observed an appreciably higher OR for the risk alleles of ANRIL in PCAD individuals compared with the LCAD individuals. Previous evidence has reported the associations of 9p21 SNPs with PCAD in different populations [40,41]. Our results may favor prediction of PCAD by providing new SNPs.
Epigenetics has been shown to play an important role in the initiation and progression of atherosclerosis. Atherosclerosis is even regarded as an epigenetic disease [42]. Among different epigenetic mechanisms, DNA methylation is the key epigenetic process for CAD and its risk factors [43]. Our results show that p15 INK4b , but not p14 ARF or p16 INK4a , is hypermethylated in CAD cases, consistent with the study of Zhuang et al. [19]. DNA methylation of ANRIL has been reported as an important epigenetic regulatory factor in mediating CAD [16], adiposity [18] and cardiovascular risk [17]. However, its relation with ANRIL polymorphism has not been reported. In the post-GWAS era, increasingly evidence has suggested a complex relationship between SNP, allele-specific binding of transcription factors, and allele-specific DNA methylation for disease risk [44].
Our results found that methylation levels of ANRIL are similar between the cases and controls (p > 0.05), but are different for the rs1004638 genotype, where AA and AT had a higher level of ANRIL methylation (p < 0.05). Furthermore, correlation analysis found ANRIL methylation levels are significantly associated with risk factors for CAD, and functional annotation indicates that the ANRIL methylated region has binding sites for transcription factors that are associated with basal transcription. These results showed there may be allele-related DNA methylation and allele-related binding of transcription factors within the ANRIL promoter. ANRIL may cause CAD via allele-related binding of transcription factors or allele-related gene expression. As for the role of allelerelated DNA methylation, the mediator or the phenotype of CAD, further prospective studies are needed to explore this. Previous methylation quantitative-trait loci (met QTL) analyses also revealed that DNA methylation is under strong genetic influence [45] and most effective methylation is associated with nearby SNPs [46][47][48].
Methylation of p15 INK4b or p14 ARF is not affected by ANRIL genotype. Zhuang et al. also found that methylation of the INK4/ARF locus is not affected by SNPs in the ANRIL locus (typified by rs10757274) [19]. This led Lillycrop et al. to speculate that both risk genotype of ANRIL (rs10757274) and methylation of the INK4/ARF locus independently affect ANRIL expression to mediate disease risk [18]. We are in favor of this hypothesis.

Strengths and limitations
The strengths of this study are the relatively large number of participants with definite diagnosis and detailed clinical characteristics. On the other hand, the genetic (genotype) and epigenetic (methylation) association analysis are from the same set of samples. There are some potential limitations. First, our study is only a preliminary study with just the results of association analysis. Further cellular and molecular experiments are warranted to validate the specific mechanism. Second, the epigenome is dynamic as the environment changes and due to the heterogeneity of patients and variations in therapies, so it is probably inappropriate to extend the present research results to other cells or tissues. Finally, our study is a case-control and cross-sectional design study. Whether methylation of ANRIL is a driver of CAD or a consequence of CAD requires further evaluation. Dynamic detection of ANRIL methylation before and after the occurrence of CAD would be a better choice.

Conclusions
In conclusion, our findings indicate that polymorphisms of ANRIL (rs1004638, rs1333048 and rs1333050) might serve as a genetic biomarker of CAD, but not MI/ACS. There may be allele-related DNA methylation and allelerelated binding of transcription factors within the ANRIL promoter. Our study provides new mechanistic insight into the regulation of ANRIL. Further studies are warranted to illustrate potential mechanisms for crosstalk of genetic factors, allele-related DNA methylation, and allele-related binding of transcription factors.