Hypomethylation in MTNR1B: a novel epigenetic marker for atherosclerosis profiling using stenosis radiophenotype and blood inflammatory cells

Patients and evaluation of traditional cardiovascular risk factors

The first patient set was selected for DREAM sequencing from patients with ischemic stroke over the age of 65 years. Buffy coats from patient samples were prospectively deposited in the Human Bio-Resource Bank of the Chungnam National University Hospital from June 2013 to February 2015 (IRB review number: CNUH-2013-01-019). From this patient set, we selected eight patients with the severe-stenosis radiophenotype who had more than three vessels with > 50% stenosis in 11 intracranial vessels (middle cerebral arteries, anterior cerebral arteries, posterior cerebral arteries, intracranial internal carotid arteries, vertebral arteries of both sides, and basilar artery) on time-of-flight MRA [28] and more than two vessels with > 50% stenosis in the bilateral CCA and proximal intracranial arteries via CDU [29]. Eight patients with the no-stenosis radiophenotype who had no-stenosis of the intracranial and extracranial vessels in the two imaging studies were also selected (Additional file 1: Table S1).

A second set of patients was selected to validate the significance of the identified gene-specific promoter methylation markers in patients with ischemic stroke whose buffy coats were prospectively deposited in the Stroke Registry of the Chungnam National University Hospital from May 2017 to May 2020 (IRB review number: CNUH-2017-04-054). We selected 334 patients with the stenosis radiophenotype with > 1 stenosis and > 50% stenosis in the 11 intracranial vessels or four extracranial vessels via imaging studies, and 50 patients with no-stenosis radiophenotype in intracranial or extracranial vessels (Table 1).

Patients with ischemic stroke due to cardioembolic causes were excluded from the study. The first and second sets of patients were evaluated based on clinical cardiovascular risk factors including age, sex, body mass index, and previous history of hypertension, diabetes, and smoking, and lastly, laboratory tests including hemoglobin A1c, fasting blood glucose, total cholesterol, triglyceride, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, lipoprotein (a), apolipoprotein A and B, high-sensitivity C-reactive protein, homocysteine, white blood cell count, hemoglobin, and platelet count measured during the fasting state within 24 h after admission.

Isolation of buffy coats and individual inflammatory cells from blood

Buffy coats were separated from blood samples (3 mL) of 18 patients in the first set after centrifugation for 15 min at 1000 rpm in EDTA-coated tubes. For 384 patients in the second set, buffy coats were separated using density gradient centrifugation (Ficoll-Paque PLUS, cat. no. 17-1440-02, Merck KGaA, Darmstadt, Germany) from blood samples (3 mL) collected in citrate-coated tubes during the fasting state within 24 h after admission. In 170 of 384 patients in the second set, T cells (Dynabeads® CD3; cat. no. 11151D, Invitrogen, USA), B cells (CD19 Pan B; cat. no. 11143D, Invitrogen), and monocytes (CD14; cat. no. 11149D, Invitrogen) were sorted from the buffy coats. All buffy coats and individual inflammatory cells from the first and second patient sets were stored at − 80 ℃ before DNA extraction. DNA from the buffy coats and individual inflammatory cell types was extracted using a kit (DNeasy Blood & Tissue Kit, Qiagen, Germany) and stored at − 20 °C until further use.

Vascular endothelial cell culture

For wild-type endothelial cells, we cultured five human umbilical vein endothelial cell (HUVEC) lines (cat. no. C-005-5C, lot no. 1774129, Gibco, USA; cat. no. C-015-5C, lot no. 1391153, Gibco; cat. no. PCS-100-010, lot no. 70006858, American Type Culture Collection [ATCC]; cat. no. PCS-100-010, lot no. 70008844, ATCC; cat. no. PCS-100-013, lot no. 80616172, ATCC) and two human aortic endothelial cell (HAEC) lines (cat. no. PCS-100-011, lot. no. 64389694 and 70008309, ATCC). HUVECs and HAECs were cultured in M199 growth media (cat. no. 31100035, Gibco) containing less than 20% fetal bovine serum (cat. no. 12483-020, Gibco), 2% human serum, 2 mmol/L L-glutamine (cat. no. 25030-081, Gibco), and 50 μg/mL endothelial cell growth supplement (cat. no. 356006, BD Biosciences, USA). All HUVEC and HAEC lines were subcultured for up to five passages, and DNA was then extracted using a kit (DNeasy Blood & Tissue Kit, Qiagen) and stored at − 20 °C until further use.

Collection of atherosclerotic plaque and non-plaque intima from the CCA of cadavers

We collected atherosclerotic plaque and non-plaque intima dissected from the CCA of 20 cadavers (male-to-female ratio = 14:6, age = 63.1 ± 18.8 years, mean ± standard deviation). Elevated atherosclerotic plaques in the intimal layer were distinguished from the surrounding non-plaque intima via naked-eye observation. The plaque and non-plaque regions were clearly dissected along the plaque margins, and then harvested by peeling them from the media layer of the individual CCAs and stored at − 80 °C until DNA extraction. The plaque and non-plaque intima were examined using histological hematoxylin and eosin staining. DNA from 100 mg of the formalin-fixed plaque and non-plaque intima was extracted as previously described [30], suspended in 100 μL of 1X Tris–EDTA buffer (pH 8.0), and then stored at − 20 °C until further use. The vascular tissues from cadavers were provided by the Human Resources Center in the Department of Anatomy, College of Medicine at the Chungnam National University.

Collection of atherosclerotic tissues and buffy coats from patients who underwent CEA

Atherosclerotic plaques were harvested from 26 patients who received a CEA at Dong-A Medical Center from December 1, 2015 to September 30, 2017. All CEA plaque samples were immersed overnight in five volumes of RNAlater (Thermo Fisher Scientific, USA) immediately after collection from each patient and stored in liquid nitrogen until DNA extraction after removal of the solution. Buffy coats from each patient were simultaneously separated from blood samples (3 mL) in an EDTA tube, which was collected immediately before the CEA operation and separated after centrifugation for 15 min at 1000 rpm. DNA from 50 mg of CEA plaques and 100 μL of buffy coats were extracted using a kit (DNeasy Blood & Tissue Kit, cat. no. 69506, Qiagen) according to the manufacturer’s instructions. All CEA plaque and buffy coat samples were stored in the Human Bio-Resource Bank at the Dong-A Medical Center with informed consent.

Profiling and mapping of gene-specific promoter methylation using DREAM sequencing

We profiled gene-specific promoter methylation related to stenosis radiophenotypes using DREAM sequencing [31] of DNA extracted from the buffy coats of eight patients in the first set each with or without the stenosis radiophenotype (Additional file 1: Table S1). In brief, a sequencing library was generated after digestion of the DNA with the SmaI and then XmaI restriction enzymes. The libraries were sequenced via paired-end 36 nt sequencing on the Illumina Genome Analyzer II or Illumina HiSeq 2000. We mapped the reads of 374,165 SmaI sites to the human genome (NCBI36/hg18) reference and signatures corresponding to methylated and unmethylated CpG genome using the Bowtie [32] and Burrows–Wheeler transform [33] aligners. Next, we excluded repetitive sequences (including LINE and Alu), and applied the following criteria to select tags that distinguish between methylation differences in no- and severe-stenosis groups: methylation difference > 5% and p value < 0.1. Finally, we selected target tags located in CpG islands (GC% and observed/expected ratio > 0.6) within the ± 1000 bp sequence from the transcription start site of the specific genes.

Promoter methylation evaluation

The promoter methylation status of the 11 genes identified after the mapping analysis was evaluated using bisulfite pyrosequencing (Fig. 2). Bisulfite treatment of DNA was performed using a kit (cat. no. D5002, Zymo Research, USA) with 1 μg of DNA and stored at − 20 °C until further use. For bisulfite pyrosequencing, we used primer sets (Additional file 1: Table S3) comprising forward and reverse primer pairs for polymerase chain reaction (PCR) and a sequencing primer that amplified promoter CpG islands in individual target genes (Fig. 2). PCR for pyrosequencing of each gene was performed in a total volume of 20 μL with a premix PCR kit (AccuPower® PyroHotStart Taq PCR PreMix, cat. no. K-2611, Bioneer, South Korea) after adding 1 μL (15 ng) of bisulfite-treated DNA and 0.1 mmol/L of forward and reverse primers for individual genes, and one primer for each gene was biotinylated at the 5′-end. After denaturation for 5 min at 95 °C, the reaction cycles were as follows: 45 cycles (95 °C for 30 s, annealing temperature of individual genes for 30 s, and 72 °C for 30 s), and final annealing and extension at 72 °C for 10 min. Next, bisulfite pyrosequencing was performed using a sequencing primer for individual genes and PyroMark Gold Q96 Reagents (cat. no. 972804, Qiagen) and a pyrosequencing machine (PyroMark Q96 ID, Qiagen). Methylation levels in each gene are represented as mean values from all pyrosequenced CpG sites for individual genes.

Statistical analysis

Differences in sex and clinical risk factors were compared using a chi-square test between no-stenosis and stenosis radiophenotypes. Differences in age, laboratory tests, and target gene-specific promoter methylation were compared between the two radiophenotype groups, vascular tissues, and inflammatory cell types using independent Student’s t tests. We compared the prediction performance of the logistic regression model for the stenosis radiophenotype using traditional clinical and laboratory cardiovascular risk factors. The prediction performance was modeled based on the addition of gene-specific promoter methylation markers of seven genes (EFNA2, ENOSF1, GLS2, KNDC1, MTNR1B, PAX8-AS1, and TLDC1) to the traditional risk factors in the second patient set. To mitigate overfitting of the logistic regression prediction model, we performed fivefold cross-validation analysis for the logistic regression model using the traditional clinical and laboratory risk variables, and the seven gene promoter methylation markers. For the cross-validation analysis, we first divided the second set of patients into two datasets: 70% training and 30% test datasets. Using the 70% training dataset, fivefold cross-validation of the logistic regression model was estimated. Performance of the prediction model from the 70% training dataset was validated using the 30% test dataset. The prediction performance differences before and after including the target gene-specific promoter methylation markers were measured and compared according to accuracy, sensitivity to predict the no-stenosis radiophenotype, specificity to predict the stenosis radiophenotype, and AUC. All statistical analyses were performed using SPSS (ver. 24.0, IBM Corp., USA) and R packages (ver. 4.1.3).

留言 (0)

沒有登入
gif