Successful ATAC-Seq From Snap-Frozen Equine Tissues

Abstract

An assay for transposase-accessible chromatin with high-throughput sequencing (ATAC-seq) has become an increasingly popular method to assess genome-wide chromatin accessibility in isolated nuclei from fresh tissues. However, many biobanks contain only snap-frozen tissue samples. While ATAC-seq has been applied to frozen brain tissues in human, its applicability in a wide variety of tissues in horse remains unclear. The Functional Annotation of Animal Genome (FAANG) project is an international collaboration aimed to provide high quality functional annotation of animal genomes. The equine FAANG initiative has generated a biobank of over 80 tissues from two reference female animals and experiments to begin to characterize tissue specificity of genome function for prioritized tissues have been performed. Due to the logistics of tissue collection and storage, extracting nuclei from a large number of tissues for ATAC-seq at the time of collection is not always practical. To assess the feasibility of using stored frozen tissues for ATAC-seq and to provide a guideline for the equine FAANG project, we compared ATAC-seq results from nuclei isolated from frozen tissue to cryopreserved nuclei (CN) isolated at the time of tissue harvest in liver, a highly cellular homogenous tissue, and lamina, a relatively acellular tissue unique to the horse. We identified 20,000-33,000 accessible chromatin regions in lamina and 22-61,000 in liver, with consistently more peaks identified using CN isolated at time of tissue collection. Our results suggest that frozen tissues are an acceptable substitute when CN are not available. For more challenging tissues such as lamina, nuclei extraction at the time of tissue collection is still preferred for optimal results. Therefore, tissue type and accessibility to intact nuclei should be considered when designing ATAC-seq experiments.

Keywords: FAANG; chromatin; cryopreserved; epigenetics; horse.

FIGURE 1

A schematic of the experimental design. All samples were prepared at UC Davis prior to shipment to the core laboratories. Samples used were obtained from an equine biobank of two horses (AH1 and AH2), as previously described (Burns et al., 2018).

FIGURE 2

Read coverage correlation between libraries. Read depth was normalized across all libraries. (A) Principal component analysis of genome coverage, showing the first two principal components. (B) Pearson correlation of genome coverage in liver (left) and lamina (right) libraries. Linkage was calculated using Farthest Point Algorithm. (C) Fingerprint plot of genome coverage in liver (left) and lamina (right) libraries. (D) Enrichment as measured by FRiP in each library.

FIGURE 3

HMMRATAC peak calling statistics. (A) Number of peaks, (B) peak length distribution, (C) peak score distribution, and (D) percent of genome covered by peaks for each library. (E,F) Peak metrics assessed using ChIP-seq dataset in liver (E) and lamina (F) libraries.

FIGURE 4

Filtered ATAC-seq peaks. (A) Intersection plot of quality filtered peaks from each library. Bottom left panel shows filtered peak count in each library; bottom right panel shows different intersections (BedTools, 1 bp minimum) of peaks where filled dots indicate presence of peaks in corresponding library; Top panel shows peak count in each intersection. (B) Relationship between promoter accessibility and gene expression (mean vst transformed count) in liver (top left) and lamina (top right). Green cell in ATAC peaks indicate presence of ATAC peaks and black cells indicate absence. Bottom panel shows bigwig tracks of RNA-seq and ATAC-seq read abundance (normalized using RPKM)near APO genes (left, liver specific) and F2RL1 (right, lamina specific) transcription start sites.