Page 36 - Read Online

Page 6 Qu et al. J Transl Genet Genom 2023;7:3-16 https://dx.doi.org/10.20517/jtgg.2022.16

many kinds of high-throughput count data, including ChIP-Seq, was used to identify regions with peaks
that have significant differential enrichment of H3k27ac activity (Benjamini-Hochberg corrected P ≤ 0.05)
[20]
relative between the four different groups of mice: WT-LFD, WT HFD,Apc Min/+ -LFD,Apc Min/+ -HFD . The
effect size estimate for the differential enrichment between the two groups at a specific H3K27ac region was
measured as log (fold-change). These regions of differential enrichment were termed VEL .
[7]
2
RNA-Seq
RNA was isolated from the intestinal epithelia samples using 1 mL of TRIzol (Life Technologies; Carlsbad,
California; 15596-026) according to the manufacturer’s protocol by the CWRU Translational Resource
Core. RNA-seq libraries were prepared and sequenced on an Illumina NextSeq 550 platform by the CWRU
Genomics Core. Data were assessed for quality and trimmed for adapter sequences using Trim Galore!
v0.4.2 (Babraham Bioinformatics; Cambridge, UK), a wrapper script for FastQC and Cutadapt. Reads that
passed quality control were then aligned to the mouse reference genome (mm9) using the STAR aligner
[21]
v2.5.3 . Alignment for the sequences was guided using the GENCODE annotation for mm9. Reads were
analyzed for differential expression using Cufflinks v2.2.1, an RNA-Seq analysis package that reports the
fragments per kilobase of exon per million fragments mapped for each gene . A differential analysis report
[22]
was generated using the cuffdiff command performed pairwise for each mouse group to identify
differentially expressed genes (DEGs). The effect size estimate of each gene’s differential expression between
the two groups was measured as log (fold-change). Significant differential expression was identified using a
2
cutoff of q-value < 0.05.

Functional enrichment analysis
Using the default regulatory domain definition (basal promoter 5 + 1 kb and extension up to 1 Mb),
Genomic Regions Enrichment of Annotations Tool (GREAT) v4.04 was used to identify genes predicted to
[23]
be associated with ChIP-Seq VELs . Search Tool for the Retrieval of Interacting Genes/Proteins (STRING)
v11.5 was then used to identify biological pathways from the KEGG pathway database found to be
significantly associated with these predicted genes (Benjamini-Hochberg corrected P ≤ 0.05) . STRING
[24]
was also used to identify pathways significantly associated with DEGs identified from RNA-Seq.

Correlation of VELs and target gene expression
DEGs predicted to be associated with VELs were identified according to the analysis of both ChIP-Seq and
RNA-Seq data. The list of DEGs found from RNA-Seq was compared to the list of genes predicted by
GREAT to be associated with the VELs found from ChIP-Seq. For each identical gene found in both lists,
the RNA fold-change of the DEG was compared to the average H3K27ac fold-change of all of the associated
VELs. VEL-gene associations in which both fold-changes were concordant (increase in RNA fold-change
and increase in H3K27ac fold-change OR decrease in RNA fold-change and decrease in H3K27ac fold-
change) were identified.

Hypergeometric optimization of enrichment analysis
Hypergeometric optimization of enrichment (HOMER) software was used to identify de novo motifs among
[25]
the VELs (Available from: http://homer.ucsd.edu/homer/) . These de novo motifs were then compared to
known motifs to find the closest match to identify putative transcription factors that may be involved in
interaction with the VELs.

RESULTS
High-fat diet-induced epigenomic and transcriptomic changes
H3K27ac ChIP-Seq profiles were obtained from the small intestinal epithelia samples of WT andApc Min/+
mice fed with either HFD or LFD for three days. More than 40,000 H3K27ac peaks were identified for each