Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Apr 1.
Published in final edited form as: Hypertension. 2022 Jan 7;79(4):761–772. doi: 10.1161/HYPERTENSIONAHA.121.18584

Unique Associations of DNA Methylation Regions With 24-Hour Blood Pressure Phenotypes in Blacks

(DNA Methylation and 24-hour Blood Pressures)

Michelle L Roberts 1,#, Theodore A Kotchen 2,#, Xiaoqing Pan 1,3,#, Yingchuan Li 1,4, Chun Yang 1, Pengyuan Liu 1,5, Tao Wang 6, Purushottam W Laud 6, Thomas H Chelius 7, Yannick Munyura 2, David L Mattson 1,8, Yong Liu 1, Allen W Cowley Jr 1, Srividya Kidambi 2,*, Mingyu Liang 1,*
PMCID: PMC8917053  NIHMSID: NIHMS1767324  PMID: 34994206

Abstract

Background:

Epigenetic marks (e.g., DNA methylation) may capture the effect of gene-environment interactions. DNA methylation is involved in blood pressure (BP) regulation and hypertension development; however, no studies have evaluated its relationship with 24-hour BP phenotypes (daytime, nighttime, and 24-hour average BPs).

Methods:

We examined the association of whole blood DNA methylation with 24-hour BP phenotypes and clinic BPs in a discovery cohort of 281 African Americans using reduced representation bisulfite sequencing (RRBS). We developed a deep and region-specific methylation sequencing method, Bisulfite ULtrapLEx Targeted Sequencing (BULLET-Seq), and utilized it to validate our findings in a separate validation cohort (n = 117).

Results:

Analysis of 38,215 DNA methylation regions (MRs), derived from 1,549,368 CpG sites across the genome, identified up to seventy-two regions that were significantly associated with 24-hour BP phenotypes. No MR was significantly associated with clinic BP. Two to three MRs were significantly associated with various 24-hour BP phenotypes after adjustment for age, sex, and BMI. Together, these MRs explained up to 16.5% of the variance of 24-hour average BP, while age, sex, and BMI explained up to 11.0% of the variance. Analysis of one of the MRs in an independent cohort using BULLET-Seq confirmed its association with 24-hour average BP phenotype.

Conclusions:

We identified several MRs that explain a substantial portion of variances in 24-hour BP phenotypes, which might be excellent markers of cumulative effect of factors influencing 24-hour BP levels. The BULLET-Seq workflow has potential to be suitable for clinical testing and population screenings on a large scale.

Keywords: epigenetics, blood pressure, hypertension, 24-hour blood pressure, DNA methylation

INTRODUCTION

Cardiovascular disease (CVD) is the leading cause of death, not only in the US, but globally as well1. Hypertension, a major CVD risk factor, affects over one billion individuals worldwide and has an estimated yearly expense of over $73 billion in the United States alone2. Based on family studies, heritability of blood pressure (BP) is in the range of 30–60%3–6. This indicates that environmental and lifestyle factors play a major role in the determination of BP, perhaps by influencing physiology through epigenetic modifications that alter regulation of genes, without changes in the DNA sequence7. In addition, genetic approaches such as linkage studies, genome-wide association studies (GWAS), and candidate gene analyses have identified gene variants that account for only a small fraction of the observed heritability in BP phenotypes as well as other predictors of cardiovascular disease risk8. One explanation for this “missing heritability” may be related to epigenetic modifications9, 10.

Epigenetic mechanisms may capture the biological influence of environmental and lifestyle factors in a quantifiable and analyzable molecular form9. DNA methylation is a naturally occurring epigenetic modification of genomic DNA which occurs primarily in cytosine-phospho-guanine dinucleotide (CpG) sites and involves addition of a methyl group on to the carbon number 5 (C5 position) of the cytosine to form 5-methylcytosine (5mC). DNA methylation is involved in numerous cellular functions including chromatin remodeling, regulation of gene expression, cell differentiation, and regulation of alternative splicing.

Increasing evidence indicates that DNA methylation is associated with, and might contribute mechanistically to, the development of hypertension7. In a multi-cohort analysis involving over 17,000 individuals of European, African American, and Hispanic ancestry, Richard et al. (2017) analyzed associations of clinic BP with genome-wide methylation in whole blood cells or a small number with peripheral blood CD4+ T lymphocytes. Of 31 individual and unrelated CpGs discovered (p < 1.03×10−7), 13 independently replicated CpG sites (p < 1.63×10−3) were identified and accounted for an additional 1.4% (systolic blood pressure (SBP)) and 2.0% (diastolic blood pressure (DBP)) variance over models with traditional covariates including age, sex, and body mass index (BMI)11.

Previous studies of the associations of epigenetic modifications with hypertension and BP have been based on one-time office/research visit measurements of BP11. Monitoring BP for a whole 24-h period captures BP variability and diurnal variation and enables detection of masked, white coat, or nocturnal hypertension which would otherwise lead to unnecessary therapies and poor outcomes. Association of epigenetic marks such as DNA methylation with 24-hour BP phenotypes (daytime, nighttime, and 24-hour average BPs) has not been examined. Due to the robustness of 24-h BP compared to clinic BP as a phenotype, analysis of associations of DNA methylation with 24-hour BP phenotypes may yield unique insights.

In the current study, we evaluated the associations of DNA methylation patterns in whole blood with 24-hour average, daytime, and nighttime BPs, as well as clinic BPs (parameters studied within each phenotype included SBP, DBP, pulse pressure (PP), and mean arterial pressure (MAP)) in an established African American cohort. In addition, unlike several previous epigenome-wide studies, we analyzed methylation of DNA regions instead of individual CpG sites. Due to strong correlation among CpG sites within a genomic region of about a few hundred base pairs, analyzing methylated regions (MR) can dramatically reduce the dimensionality of methylation data and thus increase the power and robustness of identifying significant BP-associated MRs, especially for low-coverage regions. Moreover, we developed a method for quantifying methylation levels in a specific DNA region that was suitable for clinical testing and population screenings and applied the method in a validation study.

METHODS

The authors declare that all supporting data are available within the article [and its online supplemental files].

Methods describing recruitment of study participants, clinical assessments, DNA extraction, and RRBS are described in Supplemental Materials12–18.

Bioinformatic analysis of identified methylation regions

The genomic coordinates of transcription start sites (TSS) based on GRCh37 were downloaded from the UCSC Genome Browser. TSS regions were defined as 1,000 bp upstream and downstream of a TSS. Similar to our previous study19, databases including the CCCTC-binding factor (CTCF) database20 and enhancer mark peak regions by Encyclopedia of DNA Elements (ENCODE)21 were applied to identify CTCF-binding regions and enhancer regions overlapping with 24-hour BP phenotype-associated MRs.

Targeted DNA methylation sequencing and data analysis

Bisulfite/Bisulfite-Specific PCR ULtrapLEx Targeted Sequencing (BULLET-Seq) (Figure 1) was developed in our laboratory for validating specific MRs for a high number of indexed and pooled subject samples for simultaneous and deep sequencing using the reagents, kits, and services listed in Supplemental Table S1. Experimental samples underwent bisulfite conversion using 1 ug of DNA with the EpiTect Fast Bisulfite Conversion kit. The sample was eluted in 20 uL elution buffer, including an additional elution and centrifugation step with the same eluate, and the resulting samples was referred to as bisulfite-converted DNA. Bisulfite-specific PCR (BSP) primers were designed, optimized, and the reactions were performed using enzymes and cycling parameters outlined in Supplemental Tables S2 and S3. BSP products were pooled after visualization of band intensity on a 0.8% Tris-acetate-EDTA (TAE) gel with UV imaging. Remaining steps of the protocol were largely identical to the RRBS method previously described and involves A-tailing, ligation of an unmethylated adapter, PCR amplifications, and magnetic bead purifications in between steps and for the final libraries which were quantitatively and qualitatively assessed17.

FIGURE 1: Overview of the BULLET-Seq method for both reference controls and experimental Samples.

FIGURE 1:

Open in a new tab

This flowchart describes the steps of the BULLET-Seq workflow pertaining to the reference standard DNA methylation dilution series, the experimental sample preparation, and the subsequent NGS library preparation method. For the technical assessment of the BULLET-Seq method, the Std PCR before bisulfite conversion does not carry over DNA methylation and can be utilized with endogenously methylated standards for pooling into the desired ratios (10:0, 9:1, 8:2, 7:3, 6:4, 5:5) of methylated to unmethylated DNA. Asterisk colors representing extra steps: orange: gel purification; maroon: target confirmation by Blunt-TOPO cloning and Sanger sequencing; green: assess sample by gel quality and quantity; blue: magnetic beads purification. Yellow box: 10:0 dilution for methylation rate of reference NA10847 for a chr19 MR, as derived from RRBS run average. BULLET-Seq: Bisulfite ULtrapLEx Targeted Sequencing, gDNA: genomic DNA, Std: Standard, BSP: bisulfite-specific PCR, HS: high-sensitivity, dsDNA: double-stranded DNA, Me: methylated, UnMe: unmethylated, MRs: methylation regions, NGS: next generation sequencing.

Samples were diluted 1:10 and 1:100 with molecular-grade water for use in 10 uL quantitative-PCR (qPCR) reactions for estimating pooling volumes with a standard dilution series. Primers bound to p7 and p5 sequences of the libraries for amplification. Triplicate cycle threshold values (CTs) were averaged, and the median CT was used to determine the degree of difference relative to each subject sample CT for pooling indexed subject libraries. The resulting single library pool concentration and molarity was assessed for appropriate dilutions and sequencing on the MiSeq platform using the paired end 2×300 bp v3 sequencing kit. BULLET-Seq reads were processed in silico and average methylation rates of MRs were calculated in the same manner as RRBS.

Technical assessment of the BULLET-Seq method

To assess the sensitivity and reliability of BULLET-Seq, we created a series of standard dilutions (10:0, 9:1, 8:2, 7:3, 6:4, 5:5) in ratios of methylated DNA (BSP product that captures the methylation status after bisulfite conversion) to unmethylated DNA (an initial outer PCR and subsequent nested BSP that does not carry over methylation status). The dilution curves were created using two genomic DNA samples, NA07357 and NA10847 (provided by the CEPH Biobank, Paris, France), which served as reference controls and were sequenced using RRBS alongside experimental samples. The methylation rate for each MR of each reference was calculated and averaged from the RRBS analysis (NA07357 n = 16 sequencing lanes and NA10847 n = 14). BSP/nested BSP primers and outer PCR primers were designed and optimized using enzymes and cycling parameters as outlined in Supplemental Tables S2 and S3. Outer PCR products were gel-excised, purified, and subjected to bisulfite conversion. BSPs and nested BSPs were performed on the bisulfite-converted DNA and bisulfite-converted outer PCR products, respectively. All BSP products were gel-excised, purified, and analyzed by blunt-end TOPO cloning and sequencing to confirm error-free amplification.

Pure reference BSP and nested BSP sample concentrations and molarities were assessed, and a correction factor was applied as described in Sanders et al., (2017)22 to account for a higher specificity of one of two assays. Samples were pooled in equimolar fashion based on the respective ratio of methylated DNA to unmethylated DNA for each MR of each reference sample. For each dilution ratio (10:0, 9:1, 8:2, 7:3, 6:4, 5:5), all MRs for the reference were pooled, yielding a total of 6 methylated DNA to unmethylated DNA sample standards per reference control ranging from 50–100% of the reference RRBS methylation rates in 10% increments. Library preparation and data analysis were the same as described for experimental samples. Observed versus expected methylation rates were plotted in Microsoft Excel with best fit line and correlation (R2).

Statistical analysis

Age, BMI, and BP measurements were reported as mean ± SD. For the discovery and validation cohorts, linear regression analyses were performed to investigate the association between methylation rate of each MR and BP, with or without adjustment for covariates age, sex, and BMI. In the linear regression model, BP is a dependent variable and methylation rate and/or other covariates are independent variables. The significance of association was determined with Benjamini–Hochberg adjusted FDR < 0.05. In univariate regression analysis, BP variance was partitioned into two components contributed by one MR and residual errors. In multiple regression analysis, BP variance was partitioned into three components contributed by either one MR or the linear combination of MRs, the linear combination of covariates, and residual errors, allowing for the calculation of the percentage of BP variance explained by MRs or covariates. All statistical analyses were performed using the R statistical package Version 3.5.1.

RESULTS

The discovery cohort and RRBS analysis

Of the 281 discovery cohort subjects (49.8% women) studied, 50.5% were hypertensive. At the time of study, mean age of the participants was 44 ± 7 years and mean BMI was 28 ± 5 kg/m2. Twenty-four percent of the subjects were taking antihypertensive medications at the time of the research study visit. Levels of 24-hour average, daytime and nighttime BPs, and clinic BP parameters (SBP, DBP, MAP, and PP) of the study subjects are shown in Supplemental Table S4. The correlation between 24-hour average BP and clinic BP measurements was modest with Pearson correlation coefficient being 0.77 for SBP and 0.75 for DBP.

DNA methylation from RRBS profiles showed an average bisulfite conversion rate of 99.9% (Supplemental Table S5). Approximately 19.9 million reads from each sample were uniquely mapped to the reference genome (GRCh37/hg19). We merged the data from all subjects based on CpG coordinates, yielding 1,549,367 CpG sites (>5X) that were detected in at least 95% of the subjects. These CpG sites formed 38,215 MRs as defined by metilene and binomial testing (Supplemental Figure S1).

Methylation regions associated with BP without adjustment for covariates

Regression analysis was performed separately for each of the 38,215 MRs. Several MRs were significantly (FDR < 0.05) associated with 24-hour average, daytime, or nighttime BP phenotype parameters SBP, DBP, PP, or MAP, ranging from 2 MRs for nighttime PP to 72 MRs for 24-hour average SBP (Supplemental Tables S6, S7, S8). Many of these MRs were also associated with clinic BPs (unadjusted p < 0.05), however there were no significant associations once the Benjamini–Hochberg adjustment was applied (FDR > 0.05).

There was considerable overlap of MRs associated with the various BP parameters (SBP, DBP, PP, and MAP) (Figure 2 & Supplemental Figure S3) within each phenotype. For example, of the 72 MRs associated with 24-hour average SBP, 30 were also associated with 24-hour average MAP and 10 were further associated with 24-hour average DBP (Supplemental Figure S3A). Similarly, several MRs associated with SBP, DBP, PP or MAP measured during daytime or nighttime overlapped (Supplemental Figures S3B, S3C). MRs associated with 24-hour average, daytime or nighttime BPs also overlapped partially (Supplemental Figures S3A-S3D). Another example is that of the 72 MRs associated with 24-hour average SBP, 37 were associated with daytime SBP and 16 with nighttime SBP. MRs associated with nighttime DBP were also a subset of the MRs associated with daytime DBP, which in turn was a subset of MRs associated with 24-hour average DBP (Figure 1B). Partial overlap was also observed for the genes that harbor intragenic or TSS MRs significantly associated with SBP, DBP, PP or MAP for 24-hour average, daytime, or nighttime BPs (Supplemental Figure S4).

FIGURE 2. Overlap of MRs associated with 24-hour average, daytime, or nighttime blood pressures without covariate adjustment.

FIGURE 2.

Open in a new tab

Venn diagrams of MRs significantly associated with and overlapping with 24-hour average, daytime, or nighttime BP phenotypes as represented numerically and as a percent of total MRs for A) SBP, B) DBP, C) MAP, and D) PP. SBP: systolic blood pressure; DBP: diastolic blood pressure; PP: pulse pressure; MAP: mean arterial pressure.

On average, each of the 72 MRs associated with 24-hour average SBP without adjustment for covariates explained 6.4% ± 0.8% of the 24-hour average SBP variance (Supplemental Table S6). Similar fractions of 24-hour average DBP, PP and MAP as well as daytime or nighttime SBP, DBP, PP and MAP were explained by each of the associated MRs (Supplemental Tables S6, S7, S8). When considered jointly, the MRs significantly associated with each of the BP phenotypes explained substantial fractions of BP variance, ranging from 15.2% of nighttime PP variance explained by 2 significantly associated MRs to 66.6% of the 24-hour average SBP variance explained by 72 significantly associated MRs (Table 1).

Table 1.

Combined effect of newly identified MRs on 24-hour average, daytime, or nighttime BP variances.

BP Phenotype No covariate adjustment Adjusted for age, sex, and BMI
# of MRs MRs Residuals # of MRs Covariates MRs Residuals
SBP.24h avg 72 66.6% 31.2% 2 11.0% 14.2% 75.7%
DBP.24h avg 12 34.8% 64.8% 2 5.9% 12.9% 78.8%
PP.24h avg 22 46.8% 52.2% 3 8.8% 16.5% 76.2%
MAP.24h avg 34 51.9% 47.1% 2 6.8% 12.9% 77.7%
SBP.day 38 53.3% 44.3% 0 NA NA NA
DBP.day 8 27.5% 72.2% 0 NA NA NA
PP.day 28 49.0% 49.2% 10 4.0% 33.9% 59.5%
MAP.day 7 23.6% 76.1% 0 NA NA NA
SBP.night 20 43.4% 54.0% 0 NA NA NA
DBP.night 4 26.9% 73.0% 2 3.6% 14.6% 79.6%
PP.night 2 15.3% 84.2% 2 4.2% 14.3% 80.3%
MAP.night 5 26.4% 72.9% 2 4.4% 14.1% 79.1%
Open in a new tab

The columns titled MRs, Covariates, and Residuals show percentages of BP variance explained by the identified MRs considered jointly, the adjusted covariates considered jointly, and not explained by the covariates and MRs, respectively.

BP: blood pressure, MR: methylation region, BMI: body mass index, SBP: systolic blood pressure, DBP: diastolic blood pressure, PP: pulse pressure, MAP: mean arterial pressure, avg: average.

The MR locations (intergenic, intragenic, and TSS regions) associated with 24-hour BP phenotypes are shown in Supplemental Figure S2. Regarding nuclear chromatin organization and looping and transcriptional regulation, several MRs associated with 24-hour BP phenotypes overlapped with CTCF-binding sites and enhancers (Supplemental Tables S6, S7, and S8). CTCF is a zinc-finger protein that works collectively with other molecules to form loops that organize the nuclear chromatin, ultimately affecting gene regulation. There were no overlaps found between the identified MRs and GWAS-nominated single nucleotide polymorphisms (SNPs) associated with BPs19.

The intragenic and TSS MRs associated with 24-hour average SBP, DBP, PP or MAP were located in 57 genes and the intragenic and TSS MRs associated with 24-hour average, daytime, or nighttime BP phenotypes were positioned in 67 genes. A Database for Annotation, Visualization, and Integrated Discovery (DAVID) analysis24 indicated that “renal” and “developmental” were the most enriched “disease classes” in both sets of genes, although the enrichment did not reach statistical significance based on FDR. A Metascape analysis25 indicated that these genes were enriched for genes related to several Gene Ontology terms, such as epithelial cell differentiation, calcium ion regulated exocytosis and cell junction organization (Supplemental Table S9).

Methylation regions associated with BP after adjustment for covariates

Several MRs, ranging from 2 to 10, were significantly associated with 24-hour average, daytime, or nighttime BP phenotype parameters of SBP, DBP, PP, or MAP after adjustment for covariates including age, sex, and BMI (Table 2, Supplemental Table S10). Although clinic BPs were associated with covariate-adjusted MRs significantly (p < 0.05) prior to FDR adjustments, none of the 38,215 MRs were associated with clinic BP parameters after adjustment for covariates and FDR, (Supplemental Table S10). CTCF sites did not overlap with covariate-adjusted MRs of significance for clinic BP parameters, although there were many overlapping with enhancers (Table 2, Supplemental Table S10).

Table 2.

DNA methylation regions associated 24-hour average, daytime, or nighttime blood pressures following adjustment for age, sex, and BMI.

MR Information 24h BP
BP chr start end α p adjusted p Prop.cov Prop.MR
SBP 24h avg 13 25,745,246 25,745,301 −4.1 1.04E-06 0.026 12.7% 8.1%
19 47,569,729 47,569,789 19.5 1.34E-06 0.026 8.3% 7.7%
DBP 24h avg 19 47,569,729 47,569,789 12.7 1.80E-06 0.034 6.9% 7.6%
5 145,725,340 145,725,376 −5.4 1.55E-06 0.034 6.7% 8.0%
PP 24h avg 9 139,317,613 139,317,697 1.9 4.92E-07 0.019 7.0% 8.3%
11 44,330,609 44,330,643 −5.6 1.86E-06 0.035 7.3% 8.4%
13 25,745,246 25,745,301 −1.9 3.39E-06 0.043 9.4% 7.6%
MAP 24h avg 19 47,569,729 47,569,789 15.0 8.63E-07 0.033 7.8% 8.0%
5 145,725,340 145,725,376 −6.1 2.36E-06 0.045 7.7% 7.7%
PP day 9 139,317,613 139,317,697 1.9 3.53E-07 0.013 7.2% 8.5%
11 44,330,609 44,330,643 −5.6 1.11E-06 0.021 7.4% 8.8%
13 65,723,859 65,723,946 9.3 1.97E-06 0.025 5.6% 7.6%
16 88,512,223 88,512,278 3.0 3.31E-06 0.032 6.7% 7.1%
13 25,745,246 25,745,301 −1.9 5.06E-06 0.036 9.5% 7.4%
2 133,031,534 133,031,592 5.0 5.65E-06 0.036 5.7% 7.1%
16 2,229,956 2,230,045 11.1 6.97E-06 0.038 4.1% 7.6%
10 88,295,236 88,295,267 −6.6 1.21E-05 0.049 6.6% 6.9%
10 134,276,268 134,276,333 6.5 1.04E-05 0.049 6.4% 7.0%
9 139,404,808 139,404,851 7.1 1.29E-05 0.049 6.7% 6.5%
DBP night 5 145,725,340 145,725,376 −6.2 5.84E-07 0.022 4.3% 9.2%
19 47,569,729 47,569,789 14.1 2.11E-06 0.040 4.7% 8.2%
PP night 7 158,789,546 158,789,867 4.9 3.88E-07 0.015 4.4% 9.4%
11 44,330,609 44,330,643 −6.2 2.61E-06 0.050 5.0% 8.8%
MAP night 19 47,569,729 47,569,789 16.1 1.70E-06 0.032 5.7% 8.3%
5 145,725,340 145,725,376 −6.8 1.30E-06 0.032 5.1% 8.6%
Open in a new tab

BP: blood pressure, chr: chromosome, pos: position, MR: methylation region, BMI: body mass index, SBP: systolic blood pressure, DBP: diastolic blood pressure, PP: pulse pressure, MAP: mean arterial pressure, avg: average, α: directionality and effect size estimate, Prop.cov: percentage of BP variance explained by the adjusted covariates; Prop.MR: percentage of BP variance explained by the identified MR.

Each of the MRs associated with 24-hour average SBP after adjustment for age, sex, and BMI explained on average 7.9% of the 24-hour average SBP variance, while the covariates explained 10.5% of the variance (Table 2, Supplemental Table S10). Comparable fractions of 24-hour average DBP, PP and MAP as well as day- or nighttime SBP, DBP, PP and MAP were explained by each of the associated MRs and covariates when significantly associated MRs were identified after adjustment for covariates (Table 2). Considered jointly, the 2 MRs significantly associated with 24-hour average SBP after adjustment for age, sex and BMI explained 14.2% of the 24-hour SBP variance, while the covariates explained 11.0% (Table 1). Similarly, substantial fractions of the variance of other BP phenotypes were explained by the associated MRs considered jointly when significantly associated MRs were identified (Table 1).

Previously published BP-associated DNA methylation sites

Another method for examination of DNA methylation is by the HumanMethylation 450K BeadChip Array that contains approximately ~480,000 probes representing specifically selected CpGs. Of the 1,549,368 CpG sites detected in RRBS, there were 91,057 (5.85%) CpGs that overlapped from the BeadChip array. In the study by Richard et al., there were 9 CpGs of significance from the datasets reported that were also detected in our RRBS data11. Before covariate and FDR adjustments, several of these individual CpGs (cg07797660, cg11938080, cg06340364, and cg00533891) were significantly associated with multiple 24-hour average, daytime, nighttime, and/or clinic BP phenotypes in our study (p < 0.05) (Supplemental Table S11). One CpG site, cg06340364 (chr7:6,655,973), showed significant negative associations with nearly all the studied BP phenotypes (α = −0.375 to −3.966; p < 0.05) with the exception of nighttime DBP and clinic DBP (p > 0.05). After FDR adjustments for 9 CpG sites, cg06340364 was significantly associated with 24-hour average SBP and clinic SBP (FDR < 0.05) (Supplemental Table S11). After covariate adjustments, cg06340364 showed associations with 24-hour average SBP, MAP, and PP, daytime SBP, MAP, and PP, and nighttime SBP and PP in our study (p < 0.05) while a chr10 CpG (cg00533891) showed a significant association with daytime SBP (p < 0.05) (Supplemental Table S11). None of the 9 CpG sites were significantly associated with any of the BP phenotypes after both covariate and FDR adjustments. The chr7 CpG site, cg06340364, was the closest to significance with clinic SBP (FDR = 0.051) (Supplemental Table S11).

Development of a method for analyzing specific methylation regions that is suitable for clinical testing and population screenings

We developed a method called BULLET-Seq to quantify methylation levels at specific DNA regions as described in Methods (Figure 1). Three MRs of interest that were significantly associated with various 24-hour BP phenotype measurements, including an intergenic region of chr5, an area in the TSS of AMER2 on chr13, and an intragenic region of ZC3H4 on chr19 (Table 2, Supplemental Table S10), were chosen to assess the reliability of BULLET-Seq. The three MRs were chosen because they represented significant associations in many BP phenotype categories and allowed for the testing of BULLET-Seq for different levels of methylation. The average methylation rates for the two reference samples, NA07357 and NA10847, are 3.67% and 11.96% (chr5 region), 1.01% and 0.56% (chr13 region), and 83.75% and 76.18% (chr19 region), respectively.

A 6-point dilution series was generated for each of the two reference samples as described in Methods. The dilutions series were analyzed in parallel with other samples using BULLET-Seq in two separate runs of sample preparation and sequencing. The quality control data showed similar high base-calling scores and mapping rates for both runs whether the total reads yielded was in the millions per sample (Run 1; low n) or a few thousand reads per sample (Run 2; high n) (Supplemental Table S12). The chr19 MR, which had methylation rates ranging from 38.1% to 83.7% in the dilution series, showed strong correlation between methylation rates obtained from BULLET-Seq and expected methylation rates (R2 = 0.95 – 0.97; Figure 3). Reference NA10847 showed some variation between runs but had a steeper slope closer to y = x (Figure 3B).

FIGURE 3. BULLET-Seq reference sample dilution series for a chr19 MR show high accuracy and precision.

FIGURE 3.

Open in a new tab

Expected versus observed (BULLET-Seq; n = 2) standard dilution curves for reference samples NA07357 (A) and NA10847 (B) with SD error bars. Equation of line and correlation (R2) of expected versus observed in the inset. SD: standard deviation, MR: methylation region.

Dilution series of the chr5 and chr13 MRs had methylation rates ranging from 1.83% to 3.67% and from 0.51% to 1.01%, respectively, for NA07357, and from 5.98% to 11.96% and from 0.28% to 0.56% for NA10847, respectively. The correlation in the chr5 MR dilution curve was strong for NA10847 (R2 =0.96) but modest for NA07357 (R2 = 0.82) (Supplemental Figures S5A and S5B). The lack of correlation in chr13 MR dilution curves is indicated by R2 = 0.27 for NA07357 and R2 = 0.03 for NA10847 (Supplemental Figures S5C and S5D). These data indicate that BULLET-Seq is able to accurately quantify the 10% changes in the dilution series when methylation rates ranged from ~6% to 90% (e.g., the chr5 MR in the NA10847 dilution series and the chr19 MR in both reference samples), can modestly measure these changes when rates range from ~2% to 4% (e.g., the chr5 region in NA07357), and is questionable when methylation rates are below 2% (e.g., the chr13 region).

Validation of a DNA methylation region associated with BP phenotypes using BULLET-Seq

We applied BULLET-Seq to analyze the chr19 MR in a separate cohort of participants. Of the 117 independent validation cohort subjects (46.2% women) studied, 59.0% were hypertensive, and 37.6% of the subjects were taking antihypertensive medications at the time of the clinic visit. The mean age of the participants was 42 ± 7 years and mean BMI was 29 ± 5 kg/m2 (overweight). The 24-hour BP phenotypes and clinic BPs were shown in Supplemental Table 13. Mean BPs were alike between validation and discovery cohorts despite a higher percentage of hypertensives in the validation cohort. The range of BP was narrower in the validation cohort than the discovery cohort. For example, interquartile ranges of 24-h average SBP and DBP were 26 and 15 mmHg, respectively, for the validation cohort compared with 31 and 19 mmHg for the discovery cohort. Pearson correlation coefficients for the correlation between 24-hour average BP and clinic BP in the validation cohort were 0.62 for SBP and 0.58 for DBP.

Indexed experimental samples from validation cohort participants (n = 117) were multiplexed and analyzed using BULLET-Seq in a single sequencing run. Base-calling had high Q30 scores (mean > 90%) and mapping rates were approximately 41%, on average (Supplemental Table S12). Regression analyses were performed after DNA methylation rates were Logit-transformed. The chr19 MR was significantly associated with 24-h average BPs (SBP, DBP, and MAP) after adjustment for covariates (FDR < 0.05) (Table 3), confirming findings from the discovery cohort. The association with nighttime BPs (DBP and MAP) was not significant. The proportion of BP variance in the validation cohort explained by covariates was 11.8% - 17.2% and the proportion of the BP variance explained by the chr19 MR was 1.08 – 1.75% (Table 3).

Table 3.

Validation of the chr19 MR significantly associated with 24-hour average and nighttime BP phenotypes after adjustment for covariates age, sex, and BMI.

BP Phenotype Discovery (n = 281) Validation (n = 117)
α p FDR-adj p Prop.cov Prop.MR α p Prop.cov Prop.MR
SBP 24h avg 19.5 1.34E-06 0.0257 8.3% 7.7% 4.7 0.0492 11.8% 1.08%
DBP 24h avg 12.7 1.80E-06 0.0344 6.9% 7.6% 3.2 0.0257 17.2% 1.75%
MAP 24h avg 15.0 8.63E-07 0.0330 7.8% 8.0% 3.5 0.0354 16.3% 1.33%
DBP night 14.1 2.11E-06 0.0403 4.7% 8.2% 1.5 0.3924 16.4% 0.02%
MAP night 16.1 1.70E-06 0.0325 5.7% 8.3% 2.0 0.3090 5.7% 0.05%
Open in a new tab

BP: blood pressure, chr: chromosome, MR: methylation region, SBP: systolic blood pressure, DBP: diastolic blood pressure, MAP: mean arterial pressure, avg: average, Prop.cov: percentage of BP variance explained by the adjusted covariates; Prop.MR: percentage of BP variance explained by the identified MR, FDR: false discovery rate.

DISCUSSION

In this study, we identified DNA MRs specifically associated with 24-hour BP phenotypes (24-hour average, daytime, and nighttime) for the first time. These MRs explained a remarkably large proportion of BP variance for 24-hour BP phenotype parameters (SBP, DBP, PP, and MAP) above and beyond traditional risk factors such as age, sex, and BMI explained. None of these MRs were associated with clinic BP parameters indicating uniqueness of 24 h BP measurement as a phenotype. In addition, we developed a method that is suitable for analyzing methylation levels at specific DNA regions in clinic- and population- level screenings.

We used RRBS and metilene analytical method to detect MRs of interest. Some prior population studies of blood pressure have used HumanMethylation BeadChip assays for interrogation of specific, individual CpG sites. Both methods have their own advantages and disadvantages in comparison23–27. One of the advantages of RRBS is the ability to search for MRs spanning multiple consecutive CpG sites, which may capture functionally relevant methylation events28. Because of strong correlations among CpG sites within a genomic region, analysis performed with MRs compared to individual CpG sites reduces the dimensionality of methylation data and increases the robustness of identifying important methylation events, especially for low-coverage regions such as intergenic regions or non-CpG island regions.

Importantly, we were able to identify several MRs that were associated with 24-hour BP phenotypes after adjustment for age, sex, and BMI at the significance level of FDR<0.05. The 24-hour average, daytime, and nighttime BP phenotypes used in this study are likely more stable and less susceptible to transient or random changes than clinic BP taken at a single sitting. Previous studies have shown that while RNA abundance in blood cells is vulnerable to short-term stressors, DNA methylation in the same cells is less susceptible and better correlated with long-term phenotypes29. Our finding supports the notion that DNA methylation reflects stable, rather than transient, change in the genome function. In addition, our analysis indicates that the overlaps between the MRs that we discovered and individual CpG sites represented on methylation arrays are limited. In other words, we would not have discovered many of these MRs if we had used the arrays.

Several MRs remained significantly associated with 24-hour BP phenotypes even after adjustment for age, sex, and BMI. These MRs individually and collectively explained several percent of BP variance (~14%). This is remarkably high proportion of BP variance that can be explained by these newly detected MRs, beyond what was explained by traditional risk factors of age, sex, and BMI. In comparison, age, sex, and BMI jointly explained no more than 11% of the 24-h average BP variance in our study. Further, hundreds of SNPs identified by GWAS often individually explain a fraction of one percent of clinic BP variance and jointly a few percent. More than a dozen previously identified CpG sites jointly account for an additional 1–2% of clinic BP variance over models with age, sex, and BMI11. It is possible that MRs performed better than age and BMI due to modest age range in our cohort and poor association of BMI with BP in African Americans30, 31. In the validation cohort, the chr19 MR explains 1–2% of 24-hour average BP variance, which is less than the 7–8% explained in the discovery cohort. This may be in part because the validation cohort has narrower ranges of BP than the discovery cohort. Nevertheless, it is still remarkable for a single MR to explain 1–2% of BP variance.

Several factors may contribute to the MRs explaining such large proportions of 24-hour BP phenotype variances. It is possible MRs may be more powerful than individual CpG sites, as explained above, in detecting associations with a phenotype. Moreover, DNA methylation at these genomic regions might be influenced by environmental or genetic factors that collectively have a large effect on 24-hour average, daytime, or nighttime BP variance. In other words, the findings of this study appear to demonstrate the power of epigenomic or functional genomic analysis for capturing the cumulative effects of environmental and genetic factors, including factors that may not have been measured9, 32, 33. The MRs that we identified may be excellent indicators and biomarkers of the cumulative effect of environmental or genetic factors that influence 24-hour average, daytime, or nighttime BP phenotypes.

Even though our study failed to show significant associations of clinic BPs with specific MRs after correction for FDR <0.05, previous studies have reported associations of DNA methylation at several genomic loci in whole blood with clinic BP levels in thousands of subjects7, 11, 34. Several of these CpG sites were associated with various BP phenotypes in our study with marginal significance, demonstrating consistency with prior studies. However, despite the strengths of our methods of MR analysis, the sample size may have limited the statistical power for identifying significant associations between DNA methylation and clinic BP in the current study. In addition, 24-hour BP phenotypes were measured after the subjects had been removed from antihypertensive drugs for at least 1 week. In contrast, clinic BP was measured with 24% of the subjects being on BP-lowering medications in the discovery cohort and 37.6% subjects in the validation cohort. It is unknown the extent to which this may have contributed to the more significant associations between DNA methylation and 24-hour average BP measurements than clinic BP.

While the primary importance of identifying various 24-hour BP phenotypes lies in their ability to assist in the diagnosis of white-coat hypertension, masked hypertension, and/or nocturnal hypertension as well as to help assess the cardiovascular and renal disease risk35, 36, multiple measurements during a specific timeframe also provides stability to BP values which can vary moment to moment due to transient ambient and physiologic environmental influences37. While this variability is essential for body to respond to numerous physiological demands, it may sometimes obscure the prevailing BP value and multiple daily measurements as shown in this study are a way to alleviate that risk38.

Several factors have been known to affect the BP at various times during a 24-hour period39. The day-night differences in BP, characterized by morning surge and nocturnal dips40, are due to our internal body clock, differences in mental and physical activities based on the time of the day, and changes in sympathetic nervous system activity41–43. The contribution of cyclic changes in the environment - temperature, humidity, noise, and light - and behavior - food, liquid, salt, and stimulant consumption, emotional/mental stress, posture, and physical activity intensity - to the 24 h BP pattern is well established42. In addition, age, sex, and baseline blood pressure affect these patterns as well. These physiologic day-night differences in BP seem to track with changes in heart rate as well as plasma and urinary catecholamine levels41. Pathologic variability in BP, for example, non-dipping of BP at night seen in hypertensives as well as some normotensive African Americans, has been attributed to stress, sleep quality, and obesity with resulting hormonal changes such as changes in cortisol levels and RAAS activation44. Given these various factors affecting these phenotypes, it is not surprising there are some differences in the MR that are associated with each of these phenotypes.

When interrogating regions of DNA methylation for validation purposes, many opt for using quantitative methylation-specific PCR or Sanger sequencing of BSP or methylation-specific PCR products. These methods generate limited data for usually a small number of samples45–47. The targeted BULLET-Seq method that we have developed is extremely deep in sequence coverage and has the sensitivity of reliably detecting 10% changes in methylation rates in specific DNA regions with 6% or greater methylation rates. The sensitivity of BULLET-Seq for any given methylation region can be assessed using dilution curves as we did in the current study. In addition, BULLET-Seq is inexpensive and amenable to ultraplexing and can be scaled up to analyze hundreds or more samples simultaneously. Others have reported methods that share some features of BULLET-Seq but none have assessed these methods or applied them at the scale that we present here45, 47–49. The main limitation of BULLET-Seq is the difficulty in the design of bisulfite-specific primers. In addition, BSP is a type of PCR that requires fine-tuned optimization. Nonetheless, the sensitive, inexpensive, and highly scalable BULLET-Seq is a versatile method for analyzing methylation in specific DNA regions that is suitable for development as a clinical test and a population screening tool.

Identification of methylation marks associated with substantial proportion of variance in ambulatory BP measurements opens a new frontier in the investigation of effects of gene-environment interactions determining BP levels.

PERSPECTIVES

Changes in DNA methylation may contribute mechanistically to changes in BP7, 16. Several of the MRs identified in the current study are located in regulatory DNA regions including enhancers and CTCF binding sites. The gene and pathway analysis indicated that the identified MRs may influence genes involved in pathways with known roles in BP regulation or the development of hypertension. However, it remains to be determined whether the changes in these DNA methylation regions, which were identified in whole blood, reflect changes of gene regulation in cell types directly involved in physiological mechanisms for BP regulation. Until proven otherwise, the DNA methylation regions that we have identified should be considered markers, not mechanistic determinants of BP. Further studies should investigate these markers in more diverse populations and with larger numbers of participants who have either ambulatory or 24-hour BP data available.

Supplementary Material

Supplemental Publication Material
Supp3
Supp1

PATHOPHYSIOLOGIC NOVELTY AND RELEVANCE.

What is new?

  • Relationship between methylation modifications detected using RRBS technique and 24-hour blood pressure (BP) measurements was studied.

  • Discovery cohort (281 African American) with 24-hour BP measurements was evaluated.

  • Deep and region-specific methylation sequencing method, Bisulfite ULtrapLEx Targeted Sequencing (BULLET-Seq) was developed to validate the findings in a separate cohort (n=17).

What is relevant?

  • Several methylation regions explaining a substantial portion of variances in 24-hour BP phenotypes were identified, and one of them was confirmed of its association using BULLET-Seq.

What are the Pathophysiological Implications?

  • Epigenetic marks (e.g., DNA methylation changes) may capture the effect of gene-environment interactions.

  • Methylation changes may be involved in BP regulation and hypertension development.

SOURCES OF FUNDING

This work was supported by the American Heart Association (15SFRN23910002), the National Institutes of Health (HL149620, HL121233), and the Advancing a Healthier Wisconsin Endowment.

Footnotes

DISCLOSURES

None.

References

  • 1.Mozaffarian D, Benjamin EJ, Go AS, Arnett DK, Blaha MJ, Cushman M, de Ferranti S, Despres JP, Fullerton HJ, Howard VJ, Huffman MD, Judd SE, Kissela BM, Lackland DT, Lichtman JH, Lisabeth LD, Liu S, Mackey RH, Matchar DB, McGuire DK, Mohler ER 3rd, Moy CS, Muntner P, Mussolino ME, Nasir K, Neumar RW, Nichol G, Palaniappan L, Pandey DK, Reeves MJ, Rodriguez CJ, Sorlie PD, Stein J, Towfighi A, Turan TN, Virani SS, Willey JZ, Woo D, Yeh RW, Turner MB, American Heart Association Statistics C and Stroke Statistics S. Heart disease and stroke statistics−-2015 update: a report from the American Heart Association. Circulation 2015;131:e29–322. [DOI] [PubMed] [Google Scholar]
  • 2.James PA, Oparil S, Carter BL, Cushman WC, Dennison-Himmelfarb C, Handler J, Lackland DT, LeFevre ML, MacKenzie TD, Ogedegbe O, Smith SC Jr., Svetkey LP, Taler SJ, Townsend RR, Wright JT Jr., Narva AS and Ortiz E. 2014 evidence-based guideline for the management of high blood pressure in adults: report from the panel members appointed to the Eighth Joint National Committee (JNC 8). JAMA 2014;311:507–20. [DOI] [PubMed] [Google Scholar]
  • 3.Levy D, DeStefano AL, Larson MG, O’Donnell CJ, Lifton RP, Gavras H, Cupples LA and Myers RH. Evidence for a gene influencing blood pressure on chromosome 17. Genome scan linkage results for longitudinal blood pressure phenotypes in subjects from the framingham heart study. Hypertension 2000;36:477–83. [DOI] [PubMed] [Google Scholar]
  • 4.Pilia G, Chen WM, Scuteri A, Orru M, Albai G, Dei M, Lai S, Usala G, Lai M, Loi P, Mameli C, Vacca L, Deiana M, Olla N, Masala M, Cao A, Najjar SS, Terracciano A, Nedorezov T, Sharov A, Zonderman AB, Abecasis GR, Costa P, Lakatta E and Schlessinger D. Heritability of cardiovascular and personality traits in 6,148 Sardinians. PLoS Genet 2006;2:e132. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Tobin MD, Raleigh SM, Newhouse S, Braund P, Bodycote C, Ogleby J, Cross D, Gracey J, Hayes S, Smith T, Ridge C, Caulfield M, Sheehan NA, Munroe PB, Burton PR and Samani NJ. Association of WNK1 gene polymorphisms and haplotypes with ambulatory blood pressure in the general population. Circulation 2005;112:3423–9. [DOI] [PubMed] [Google Scholar]
  • 6.van Rijn MJ, Schut AF, Aulchenko YS, Deinum J, Sayed-Tabatabaei FA, Yazdanpanah M, Isaacs A, Axenovich TI, Zorkoltseva IV, Zillikens MC, Pols HA, Witteman JC, Oostra BA and van Duijn CM. Heritability of blood pressure traits and the genetic contribution to blood pressure variance explained by four blood-pressure-related genes. J Hypertens 2007;25:565–70. [DOI] [PubMed] [Google Scholar]
  • 7.Liang M. Epigenetic Mechanisms and Hypertension. Hypertension 2018;72:1244–1254. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Cowley AW Jr., Nadeau JH, Baccarelli A, Berecek K, Fornage M, Gibbons GH, Harrison DG, Liang M, Nathanielsz PW, O’Connor DT, Ordovas J, Peng W, Soares MB, Szyf M, Tolunay HE, Wood KC, Zhao K and Galis ZS. Report of the National Heart, Lung, and Blood Institute Working Group on epigenetics and hypertension. Hypertension 2012;59:899–905. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Kotchen TA, Cowley AW Jr. and Liang M. Ushering Hypertension Into a New Era of Precision Medicine. JAMA 2016;315:343–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Liang M, Cowley AW Jr., Mattson DL, Kotchen TA and Liu Y. Epigenomics of hypertension. Semin Nephrol 2013;33:392–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Richard MA, Huan T, Ligthart S, Gondalia R, Jhun MA, Brody JA, Irvin MR, Marioni R, Shen J, Tsai PC, Montasser ME, Jia Y, Syme C, Salfati EL, Boerwinkle E, Guan W, Mosley TH Jr., Bressler J, Morrison AC, Liu C, Mendelson MM, Uitterlinden AG, van Meurs JB, Consortium B, Franco OH, Zhang G, Li Y, Stewart JD, Bis JC, Psaty BM, Chen YI, Kardia SLR, Zhao W, Turner ST, Absher D, Aslibekyan S, Starr JM, McRae AF, Hou L, Just AC, Schwartz JD, Vokonas PS, Menni C, Spector TD, Shuldiner A, Damcott CM, Rotter JI, Palmas W, Liu Y, Paus T, Horvath S, O’Connell JR, Guo X, Pausova Z, Assimes TL, Sotoodehnia N, Smith JA, Arnett DK, Deary IJ, Baccarelli AA, Bell JT, Whitsel E, Dehghan A, Levy D and Fornage M. DNA Methylation Analysis Identifies Loci for Blood Pressure Regulation. Am J Hum Genet 2017;101:888–902. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Juhling F, Kretzmer H, Bernhart SH, Otto C, Stadler PF and Hoffmann S. metilene: fast and sensitive calling of differentially methylated regions from bisulfite sequencing data. Genome Res 2016;26:256–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Kidambi S, Ghosh S, Kotchen JM, Grim CE, Krishnaswami S, Kaldunski ML, Cowley AW Jr., Patel SB and Kotchen TA. Non-replication study of a genome-wide association study for hypertension and blood pressure in African Americans. BMC Med Genet 2012;13:27. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Krueger F and Andrews SR. Bismark: a flexible aligner and methylation caller for Bisulfite-Seq applications. Bioinformatics 2011;27:1571–2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Li Y, Pan X, Roberts ML, Liu P, Kotchen TA, Cowley AW Jr., Mattson DL, Liu Y, Liang M and Kidambi S. Stability of global methylation profiles of whole blood and extracted DNA under different storage durations and conditions. Epigenomics 2018;10:797–811. [DOI] [PubMed] [Google Scholar]
  • 16.Liu P, Liu Y, Liu H, Pan X, Li Y, Usa K, Mishra MK, Nie J and Liang M. Role of DNA De Novo (De)Methylation in the Kidney in Salt-Induced Hypertension. Hypertension 2018;72:1160–1171. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Liu Y, Kriegel AJ and Liang M. Library Preparation for Multiplexed Reduced Representation Bisulfite Sequencing with a Universal Adapter. Methods Mol Biol 2019;2018:177–194. [DOI] [PubMed] [Google Scholar]
  • 18.Liu Y, Liu P, Yang C, Cowley AW, Jr. and Liang M. Base-resolution maps of 5-methylcytosine and 5-hydroxymethylcytosine in Dahl S rats: effect of salt and genomic sequence. Hypertension 2014;63:827–38. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Mishra MK, Liang EY, Geurts AM, Auer PWL, Liu P, Rao S, Greene AS, Liang M and Liu Y. Comparative and Functional Genomic Resource for Mechanistic Studies of Human Blood Pressure-Associated Single Nucleotide Polymorphisms. Hypertension 2020;75:859–868. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Maurano MT, Wang H, John S, Shafer A, Canfield T, Lee K and Stamatoyannopoulos JA. Role of DNA Methylation in Modulating Transcription Factor Occupancy. Cell Rep 2015;12:1184–95. [DOI] [PubMed] [Google Scholar]
  • 21.Consortium EP. An integrated encyclopedia of DNA elements in the human genome. Nature 2012;489:57–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Sanders AD, Falconer E, Hills M, Spierings DCJ and Lansdorp PM. Single-cell template strand sequencing by Strand-seq enables the characterization of individual homologs. Nat Protoc 2017;12:1151–1176. [DOI] [PubMed] [Google Scholar]
  • 23.Gu H, Smith ZD, Bock C, Boyle P, Gnirke A and Meissner A. Preparation of reduced representation bisulfite sequencing libraries for genome-scale DNA methylation profiling. Nat Protoc 2011;6:468–81. [DOI] [PubMed] [Google Scholar]
  • 24.Xu Z and Taylor JA. Reliability of DNA methylation measures using Illumina methylation BeadChip. Epigenetics 2021;16:495–502. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Forest M, O’Donnell KJ, Voisin G, Gaudreau H, MacIsaac JL, McEwen LM, Silveira PP, Steiner M, Kobor MS, Meaney MJ and Greenwood CMT. Agreement in DNA methylation levels from the Illumina 450K array across batches, tissues, and time. Epigenetics 2018;13:19–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Carmona JJ, Accomando WP Jr., Binder AM, Hutchinson JN, Pantano L, Izzi B, Just AC, Lin X, Schwartz J, Vokonas PS, Amr SS, Baccarelli AA and Michels KB. Empirical comparison of reduced representation bisulfite sequencing and Infinium BeadChip reproducibility and coverage of DNA methylation in humans. NPJ Genom Med 2017;2:13. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Bose M, Wu C, Pankow JS, Demerath EW, Bressler J, Fornage M, Grove ML, Mosley TH, Hicks C, North K, Kao WH, Zhang Y, Boerwinkle E and Guan W. Evaluation of microarray-based DNA methylation measurement using technical replicates: the Atherosclerosis Risk In Communities (ARIC) Study. BMC Bioinformatics 2014;15:312. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Teschendorff AE and Relton CL. Statistical and integrative system-level analysis of DNA methylation data. Nat Rev Genet 2018;19:129–147. [DOI] [PubMed] [Google Scholar]
  • 29.Chen R, Xia L, Tu K, Duan M, Kukurba K, Li-Pook-Than J, Xie D and Snyder M. Longitudinal personal DNA methylome dynamics in a human with a chronic condition. Nat Med 2018;24:1930–1939. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Kidambi S, Kotchen JM, Krishnaswami S, Grim CE and Kotchen TA. Hypertension, insulin resistance, and aldosterone: sex-specific relationships. J Clin Hypertens (Greenwich) 2009;11:130–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Stevens J, Juhaeri Cai J and Jones DW. The effect of decision rules on the choice of a body mass index cutoff for obesity: examples from African American and white women. Am J Clin Nutr 2002;75:986–92. [DOI] [PubMed] [Google Scholar]
  • 32.Mattson DL and Liang M. Hypertension: From GWAS to functional genomics-based precision medicine. Nat Rev Nephrol 2017;13:195–196. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Touyz RM, Montezano AC, Rios F, Widlansky ME and Liang M. Redox Stress Defines the Small Artery Vasculopathy of Hypertension: How Do We Bridge the Bench-to-Bedside Gap? Circ Res 2017;120:1721–1723. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Huang Y, Ollikainen M, Muniandy M, Zhang T, van Dongen J, Hao G, van der Most PJ, Pan Y, Pervjakova N, Sun YV, Hui Q, Lahti J, Fraszczyk E, Lu X, Sun D, Richard MA, Willemsen G, Heikkila K, Mateo Leach I, Mononen N, Kahonen M, Hurme MA, Raitakari OT, Drake AJ, Perola M, Nuotio ML, Huang Y, Khulan B, Raikkonen K, Wolffenbuttel BHR, Zhernakova A, Fu J, Zhu H, Dong Y, van Vliet-Ostaptchouk JV, Franke L, Eriksson JG, Fornage M, Milani L, Lehtimaki T, Vaccarino V, Boomsma DI, van der Harst P, de Geus EJC, Salomaa V, Li S, Chen W, Su S, Wilson J, Snieder H, Kaprio J and Wang X. Identification, Heritability, and Relation With Gene Expression of Novel DNA Methylation Loci for Blood Pressure. Hypertension 2020;76:195–205. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Verdecchia P, Porcellati C, Schillaci G, Borgioni C, Ciucci A, Battistelli M, Guerrieri M, Gatteschi C, Zampi I, Santucci A, Santucci C, Reboldi G and et al. Ambulatory blood pressure. An independent predictor of prognosis in essential hypertension. Hypertension 1994;24:793–801. [DOI] [PubMed] [Google Scholar]
  • 36.Mancia G and Verdecchia P. Clinical value of ambulatory blood pressure: evidence and limits. Circ Res 2015;116:1034–45. [DOI] [PubMed] [Google Scholar]
  • 37.Mancia G, Ferrari A, Gregorini L, Parati G, Pomidossi G, Bertinieri G, Grassi G, di Rienzo M, Pedotti A and Zanchetti A. Blood pressure and heart rate variabilities in normotensive and hypertensive human beings. Circ Res 1983;53:96–104. [DOI] [PubMed] [Google Scholar]
  • 38.Piper MA, Evans CV, Burda BU, Margolis KL, O’Connor E and Whitlock EP. Diagnostic and predictive accuracy of blood pressure screening methods with consideration of rescreening intervals: a systematic review for the U.S. Preventive Services Task Force. Ann Intern Med 2015;162:192–204. [DOI] [PubMed] [Google Scholar]
  • 39.Kawano Y. Diurnal blood pressure variation and related behavioral factors. Hypertens Res 2011;34:281–5. [DOI] [PubMed] [Google Scholar]
  • 40.Agarwal R. Regulation of circadian blood pressure: from mice to astronauts. Curr Opin Nephrol Hypertens 2010;19:51–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Tuck ML, Stern N and Sowers JR. Enhanced 24-hour norepinephrine and renin secretion in young patients with essential hypertension: relation with the circadian pattern of arterial blood pressure. Am J Cardiol 1985;55:112–5. [DOI] [PubMed] [Google Scholar]
  • 42.Smolensky MH, Hermida RC and Portaluppi F. Circadian mechanisms of 24-hour blood pressure regulation and patterning. Sleep Med Rev 2017;33:4–16. [DOI] [PubMed] [Google Scholar]
  • 43.Lecarpentier Y, Schussler O, Hebert JL and Vallee A. Molecular Mechanisms Underlying the Circadian Rhythm of Blood Pressure in Normotensive Subjects. Curr Hypertens Rep 2020;22:50. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Bowman MA, Buysse DJ, Foust JE, Oyefusi V and Hall MH. Disturbed Sleep as a Mechanism of Race Differences in Nocturnal Blood Pressure Non-Dipping. Curr Hypertens Rep 2019;21:51. [DOI] [PubMed] [Google Scholar]
  • 45.Han Y, Franzen J, Stiehl T, Gobs M, Kuo CC, Nikolic M, Hapala J, Koop BE, Strathmann K, Ritz-Timme S and Wagner W. New targeted approaches for epigenetic age predictions. BMC Biol 2020;18:71. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Jeon K, Min B, Park JS and Kang YK. Simultaneous Methylation-Level Assessment of Hundreds of CpG Sites by Targeted Bisulfite PCR Sequencing (TBPseq). Front Genet 2017;8:97. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Leitao E, Beygo J, Zeschnigk M, Klein-Hitpass L, Bargull M, Rahmann S and Horsthemke B. Locus-Specific DNA Methylation Analysis by Targeted Deep Bisulfite Sequencing. Methods Mol Biol 2018;1767:351–366. [DOI] [PubMed] [Google Scholar]
  • 48.al-Shareef AH, Buss DC, Allen EM and Routledge PA. The effects of charcoal and sorbitol (alone and in combination) on plasma theophylline concentrations after a sustained-release formulation. Hum Exp Toxicol 1990;9:179–82. [DOI] [PubMed] [Google Scholar]
  • 49.Wozniak A, Heidegger A, Piniewska-Rog D, Pospiech E, Xavier C, Pisarek A, Kartasinska E, Boron M, Freire-Aradas A, Wojtas M, de la Puente M, Niederstatter H, Ploski R, Spolnicka M, Kayser M, Phillips C, Parson W, Branicki W and Consortium V. Development of the VISAGE enhanced tool and statistical models for epigenetic age estimation in blood, buccal cells and bones. Aging (Albany NY) 2021;13:6459–6484. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Publication Material
NIHMS1767324-supplement-Supplemental_Publication_Material.docx (1.7MB, docx)
Supp3
NIHMS1767324-supplement-Supp3.xlsx (115.2KB, xlsx)
Supp1
NIHMS1767324-supplement-Supp1.jpg (246.4KB, jpg)

RESOURCES