Page 136 - Read Online

Page 4 of 14 Broadwin et al. Vessel Plus 2023;7:25 https://dx.doi.org/10.20517/2574-1209.2023.103

at various sites using a total injection volume of 50 ug of EVs suspended in 2 mL of sterile saline (3 NDC
and 3 HFD) or vehicle control (3 NDC and 3 HFD) were performed in the area supplied by the LCx artery.
[13]
At 19 weeks of age, all pigs were humanely euthanized and cardiac tissue was harvested .

Myocardial perfusion
In a protocol previously reported by our group at the time of terminal harvest, 5 mL of Lutetium and
Europium labeled microsphere were injected into the left atrium. Simultaneously, 10 mL of blood was
collected from the femoral artery. This procedure was then repeated under paced conditions at 150 bpm
using Samarium-labeled microspheres. Post-harvest left ventricular tissues was weighed dried and sent to
Biophysics Assay Laboratory along with the corresponding blood samples. Quantification of microsphere
[13]
content in relation to blood flow could then be calculated for all myocardial segments .
DNA isolation
DNA was isolated from a 30 mg sample of homogenized ventricular tissue using a standard Qiagen DNeasy
procedure per manufacturer’s instructions, and the quality and quantity of DNA was determined
[20]
spectrophotometrically (Implen, CA, USA) .

Methylome analysis
DNA methylation profiling was performed via reduced-representation bisulfite sequencing (RRBS)
(Epigentek, Farmingdale, NY, USA).

Bioinformatic analysis was performed at the Computational Biology Core, Center for Computation and
[21]
Visualization of Brown University. Initial quality control of raw reads was performed using FASTQC .
Reads were then trimmed using Trim Galore and Trimmomatic [22,23] . Then, using Bismark, the reference
[24]
genome was prepared and trimmed reads aligned . Bismark methylation extraction could then be used to
extract methylation information, at which point the edgeR package could be used to find differentially
[25]
methylated loci (DML) .

Loci were filtered to include chr1-18, X, and Y chromosomes, exclude loci that were never methylated or
always methylated (as these provide no information about differential methylation) and include only CpGs
with at least 1 count (methylated or unmethylated) across all samples. The resulting sample size after
filtering was 68,998 loci.

For each lotus left after filtering, we calculated the distance to the nearest Ensembl gene using the
distanceToNearest function in the GenomicRanges package. Methylated and unmethylated counts for loci
that were within 2 kilobases of a gene were aggregated (summed), resulting in methylated and unmethylated
counts across 7,770 gene promoters. The summed counts per gene promoter were used for downstream
statistical analysis.

The glmFit function from edgeR was used to fit a negative binomial generalized log-linear model to test for
differential methylation in gene promoters. The experimental design matrix was constructed using
modelMatrixMeth from edgeR with a factorial experimental design (~0 + group), where group was a factor
variable with levels comprised of each combination of treatment, diet, and tissue. We dropped the intercept
from our model to parameterize it as a means model. This approach allowed us to build contrast vectors to
find differentially methylated loci for specific comparisons of interest, which we performed using the
glmLRT function from edgeR. Promoters were considered differentially methylated if the nominal P-value
was < 0.05. Results were filtered based on this nominal P-value threshold.