A functional polymorphism in the promoter of TUG1 is associated with an increased risk of ischaemic stroke

Abstract Taurine‐upregulated gene 1 (TUG1), a kind of long non‐coding RNAs (lncRNAs), was up‐regulated in ischaemic stroke (IS) with the function of promoting neuron apoptosis. In this study, we aimed to investigate the association of TUG1 polymorphisms with IS risk. The TUG1 polymorphisms were genotyped using a custom‐by‐design 48‐Plex SNPscan kit. The promoter activity was measured using the dual luciferase reporter assay. Relative expression of TUG1 in IS patients was analysed using quantitative PCR and the binding of TUG1 rs2240183 polymorphism to transcription factor was analysed using chromatin immunoprecipitation (ChIP) assay. The rs2240183 CT/CC genotypes and C allele in the promoter of TUG1 were associated with an increased risk of IS (CT/CC vs. TT: adjusted OR = 1.70, 95% CI, 1.16‐2.49, P = 0.006; C vs. T: adjusted OR = 1.47, 95% CI, 1.12‐1.93, P = 0.005). Logistic regression analysis showed that the rs2240183 was a risk factor of IS besides TC, TG, HDL‐C, LDL‐C, VLDL‐C, Apo‐A1, Apo‐B and NEFA. Further functional analysis revealed that the TUG1 rs2240183 C allele exhibited higher transcriptional activity and TUG1 expression levels (P < 0.01). The ChIP assay showed that the rs2240183 C allele binds to transcriptional factor GATA‐1. These findings indicate that the rs2240183 C allele was associated with a higher risk of IS possibly by binding to GATA‐1 and elevating TUG1 levels.

RNAs can be divided into two major categories: small (<200 nt) and long non-coding RNAs (≥200 nt, lncRNAs). [15][16][17] Recently, the expression, function and mechanisms of lncRNAs had been widely studied in various diseases, including IS. For example, taurine-upregulated gene 1 (TUG1) was found to be up-regulated in cerebral ischaemic animals and oxygen-glucose deprived (OGD) cells, with the function of promoting neuron apoptosis by sponging microRNA (miR)-9 and increasing the expression of pro-apoptosis gene Bcl-2-like 11 (Bcl2l11). 15,18 These findings highlighted that TUG1 may be involved in the pathological process of IS.
Previous work has reported that single nucleotide polymorphisms (SNPs) related to lncRNAs may affect individual's susceptibility to IS. [19][20][21][22] The risk variants included rs217727 TT genotype in lncRNA H19, rs145204276 del/del genotype in the promoter of lncRNA growth arrest-specific 5 (GAS5) and rs1537378 GG, rs2184061 AA, rs7044859 AA and rs7865618AA in the antisense non-coding RNA in the INK4 locus (ANRIL). [19][20][21][22] To date, no study has reported the association between the TUG1 polymorphisms and risk of IS. In this case-control study, we investigated whether the TUG1 polymorphisms were associated with the occurrence of IS in a Chinese Han population.

| Ethics, consent and permissions
The study protocol was approved by the Ethics Committee of the Affiliated Hospital of Youjiang Medical University for Nationalities.
All patients agreed to participate in the study and provided informed consent.

| Consent to publish
The participants signed the consent to publish the data.

| Study population
The study population consisted of 571 controls and 556 patients with IS. The patients were consecutively enrolled from the Affiliated Hospital of Youjiang Medical University for Nationalities, Guangxi, China, between October 2010 and September 2015. Detailed information of sample collection has been described in our previous work. 14,23 Briefly, the diagnosis of IS was determined according to clinical symptoms, physical examinations and cranial magnetic resonance imaging and/or cranial computed tomography. We excluded those patients with haemorrhagic, autoimmune or chronic inflammatory diseases and family history of stroke. Controls were healthy volunteers after physical examination in the hospital. We excluded those patients who had brain tumours, autoimmune diseases, haematological disorder and family history of stroke. The following clinical data were collected: age, sex, smoking status, total cholesterol (TC), triglyceride (TG), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), very low-density lipoprotein cholesterol (VLDL-C), apolipoprotein A1 (Apo-A1), apolipoprotein B (Apo-B), homocysteic acid (Hcy) and non-esterified fatty acid (NEFA). The interval time between IS onset and biochemical test was within 1 day. All patients were unrelated Han Chinese living in Guangxi province.

| SNPs selection
We selected SNPs using the following criteria: (a) tagSNPs in lncRNA TUG1; (b) in silico prediction revealed potentially functional SNPs in the promoter region of TUG1; (c) minor allele frequency is more than 5% in Chinese Han population. Finally, five SNPs (ie, rs224 0183, rs5749201, rs5753409, rs7284767 and rs8139350) were selected for further analysis.

| Genotyping
About 2-3 mL of ethylene diamine tetraacetic acid-anticoagulated peripheral blood was taken from each participant before treatment.
Genomic DNA was extracted using a salting-out method. 24 The SNPs were genotyped using a custom-by-design 48-Plex SNPscan kit on an ABI3730XL sequencer (Genesky Biotechnologies Inc, Shanghai, China). 25 This technique was based on double ligation and multiplex fluorescence PCR. For quality control, about 5% of all samples were randomly selected for Sanger sequencing and the results were 100% consistent.

| Plasmid construction and dual-luciferase reporter assay
TUG1 promoter sequence containing the rs224 0183 CC or rs224 0183 TT genotype was amplified using the following primers:

| Chromatin immunoprecipitation (ChIP) assay
ChIP assay was performed using a commercial kit (ThermoFisher Scientific, Waltham, MA, USA) following the manufacturer's protocol. Briefly, HEK293 cells were cross-linked in 1% formaldehyde and sonicated to produce fragments of 200-1000 bp. Antibodies against GATA-1 and rabbit IgG (Abcam) were used to treat the samples to introduce immunoprecipitation. After incubation overnight at 4°C, the immune complexes were washed, reversely cross-linked and finally re-suspended in 20 μL of double-distilled H 2 O. The diluted DNA and input DNA were amplified by PCR using the primers: 5'-TCCTTATCCCAAAGGCTTCC-3' (forward) and 5'-TGTTCCAGCTTCACCAAAGA-3' (reverse). The PCR products were analysed using Sanger sequencing. bio-x.cn/myAna lysis.php). 27 Logistic regression was used to identify risk factors of IS P values less than 0.05 were considered statistically significant.

| Characteristics of the study population
The characteristics of the study population are presented in Table 1. The distributions of age, gender, smoking status and TC level were not significantly different between cases and controls.
The levels of TG, LDL-C, VLDL-C and Apo-B were higher whereas the levels of HDL-C, Apo-A1, Hcy and NEFA were lower in IS patients (P < 0.05).

| Main effect of TUG1 polymorphisms on IS risk
The genotype and allelic frequencies of the five SNPs between cases and controls are summarized in Table 2 Similarly, increased risk of IS was also observed in allele comparison with adjusted OR of 1.47 (95% CI, 1.12-1.93; P = 0.005). No significant association between rs5749201, rs5753409, rs7284767 and rs8139350 and IS risk was found.
LD analysis showed that the rs5749201, rs5753409, rs7284767 and rs8139350 were in moderate LD and haplotype analysis was then performed. As shown in Table 3, eight common haplotypes were detected. The AGGC haplotype tended but did not reach the significance to decrease the risk of IS compared to the TGAC haplotype (OR = 0.78, 95% CI, 0.61-1.00, P = 0.05).

| Stratification analysis
Stratification analysis was performed between the TUG1 polymorphisms and clinical characteristics of IS. As shown in Table 4, patients carrying rs5749201 AT/AA genotypes had lower levels of TC, HDL-C and Apo-A1 compared to the rs5749201 TT carriers (P = 0.03, 0.003, and 0.007 respectively). However, no significant association was observed between rs224 0183, rs5753409, rs7284767 and rs8139350 and clinical characteristics of IS.

| Multivariate logistic regression analysis
Logistic regression was carried out to determine the risk factors of IS. As shown in Table 5

| The rs224 0183 C allele increased the transcriptional activity
To determine whether the rs224 0183 in the promoter of TUG1 affected the luciferase activity, we constructed plasmids containing the rs224 0183 C or T allele and measured the reporter activity. The schematic representation of the plasmid construction is shown in Figure 1A. As shown in Figure 1B, both rs224 0183 C and T allele exhibited a higher luciferase activity compared to the empty vector (**P < 0.01). Importantly, the rs224 0183 C allele had a higher luciferase activity compared to the rs224 0183 T allele (**P < 0.01). TA B L E 2 Association between TUG1 polymorphisms and risk of IS

| The rs224 0183 CC genotype was associated with increased levels of TUG1
To determine whether the rs224 0183 influenced TUG1 expression, we analysed the levels of TUG1 in 81 IS patients using qPCR.
As shown in Figure 2A, the rs224 0183 CC carriers had increased levels of TUG1 compared to the rs224 0183 TT carriers (**P < 0.01).
These findings were consistent with results from the expression Quantitative Trait Loci (eQTL) ( Figure 2B). The rs224 0183 CC genotype was associated with higher gene expression in several single tissues, such as whole blood, cerebellar hemisphere, cortex and hippocampus ( Figure 2C-F) (P < 0.001).

|
The rs224 0183 C allele binds to transcription factor GATA-1 In silico analysis predicted that rs224 0183 C but not rs224 0183 T binds to transcription factor GATA-1. ChIP assay was then used to validate the allele-specific transcription factor binding. PCR-electrophoresis revealed that the DNA fragments immunoprecipitated specifically with the anti-GATA-1 antibody rather than non-specific IgG ( Figure 3). Further sequencing showed that the GATA-1 binding region contained the rs224 0183 CC genotype of TUG1 promoter.

| D ISCUSS I ON
In this study, we presented two major findings. One is that TUG1 TUG1, a 7.2 kb lncRNA, was initially discovered in taurine-treated mouse retinal cells and played key roles in retinal development. 28 It is also observed to be up-regulated in nervous system diseases. 18 In atherosclerosis, the up-regulation of TUG1 enhances endothelial cell apoptosis by regulating miR-26a, 29 promotes vascular smooth muscle cell proliferation by regulating miR-21/PTEN axis 30 and stimulates proliferation and migration of endothelial cells via the Wnt pathway. 31 Knockdown of TUG1 ameliorates atherosclerosis by modulating fibroblast growth factor 1 via miR-133a. 32 Under ischemia, TUG1 silencing promotes cell survival and decreases cell apoptosis by sponging miR-9 and decreasing Bcl2l11 protein. 18 All the above mentioned miRNAs are key mediators in the pathology of IS. [33][34][35][36] miR-26a promotes angiogenesis in a rat model of IS via the PI3K/ AKT and MAPK/ERK pathway. 33  Based on this background, we suggested that SNPs in TUG1 may be related to the risk of IS. Our results confirmed this hypothesis and we found that the rs224 0183 CT/CC genotypes in the promoter of TUG1 were associated with an increased risk of IS in Table 2.
However, the rs224 0183 was not associated with clinical characteristics of IS in stratification analysis as shown in Table 4. One possible reason may be that the reference is different. In Table 2 reference is controls, whereas in Table 4, all the volunteers were IS patients and the reference is part of patients. Also we cannot exclude the possibility that the results may occur by chance. A similar phenomenon was also observed in overall analysis of rs5749201.
Although no significant association of rs5749201 with IS risk was observed compared to controls (Table 2), rs5749201 was associated with TC, HDL-C and Apo-A in IS patients ( Table 4). The reason for the discrepancy may be explained by the possibility mentioned above.
rs224 0183 being a risk factor was also identified using subsequent   Next, we explored the possible mechanism for rs224 0183 C increasing the risk of IS. We firstly used in silico analysis to predict the binding of rs224 0183 and transcriptional factor and found that rs224 0183 C but not T allele can bind to transcription factor GATA-1. In this study, CHIP assay was performed and the allelespecific GATA-1 binding was validated. These findings suggest that rs224 0183 C promoted the expression of TUG1 by binding to GATA-1, which may be the possible mechanism for rs224 0183 C increasing the risk of IS. It is well known that transcription factor binding to the promoter sequence may promote the transcriptional activity. We then evaluated whether rs224 0183 influenced the transcriptional activity and TUG1 expression using the dual-luciferase reporter gene assay and qPCR technique. We found that the rs224 0183 C allele exhibited a higher reporter activity and patients carrying the rs224 0183 CC genotype had higher levels of TUG1. These findings were consistent with the results from eQTL F I G U R E 2 The rs224 0183 CC genotype was associated with increased levels of TUG1. A, quantitative PCR was used to examine the expression levels of TUG1 in ischaemic stroke patients. Compared to rs224 0183 TT carriers, rs224 0183 CC carriers had increased levels of TUG1 (**P < 0.01). Expression Quantitative Trait Loci analysis of rs224 0183 with gene expression in single tissue (B), whole blood (C), cerebellar hemisphere (D), cortex (E) and hippocampus (F) in several tissues, such as whole blood, brain, spinal cord, spleen, testis, liver and stomach, further supporting that our findings were reasonable and reliable.
There are some limitations in this study. We collected hospital-based controls and thus the selection bias cannot be removed completely. Because of lack of data of alcohol consumption, geneenvironment interaction analysis could not be performed. Because of these limitations, further investigations are necessary to confirm the significance of rs224 0183 as a risk factor for IS. The interaction of TUG1 polymorphisms with environmental factors is of great value.
In conclusion, we demonstrated that the rs224 0183 CT/CC genotypes in the promoter of lncRNA TUG1 were risk factors for the development of IS. We also observed that rs224 0183 C was associated with higher promoter activity and TUG1 expression levels.
Furthermore, results from ChIP may provide an underlying mechanism for the observed association with the susceptibility of IS.
Because IS is a disease with high mortality and disability rate, it is of great importance to understand the risk factors for the development and progression of IS, which will benefit for the early prevention and intervention of IS.

CO N FLI C T O F I NTE R E S T S
The authors confirm that there are no conflicts of interest.

R E FE R E N C E S
F I G U R E 3 GATA-1 binds to the promoter region of TUG1 containing the rs224 0183 CC genotype. ChIP was performed using anti-GATA-1 antibody and anti-IgG antibody in HEK293 cells. The promoter region of TUG1 containing the rs224 0183 was analysed using PCR-electrophoresis