Parallel recovery of chromatin accessibility and gene expression dynamics from frozen human regulatory T cells

Abstract

Epigenetic features such as DNA accessibility dictate transcriptional regulation in a cell type- and cell state- specific manner, and mapping this in health vs. disease in clinically relevant material is opening the door to new mechanistic insights and new targets for therapy. Assay for Transposase Accessible Chromatin Sequencing (ATAC-seq) allows chromatin accessibility profiling from low cell input, making it tractable on rare cell populations, such as regulatory T (Treg) cells. However, little is known about the compatibility of the assay with cryopreserved rare cell populations. Here we demonstrate the robustness of an ATAC-seq protocol comparing primary Treg cells recovered from fresh or cryopreserved PBMC samples, in the steady state and in response to stimulation. We extend this method to explore the feasibility of conducting simultaneous quantitation of chromatin accessibility and transcriptome from a single aliquot of 50,000 cryopreserved Treg cells. Profiling of chromatin accessibility and gene expression in parallel within the same pool of cells controls for cellular heterogeneity and is particularly beneficial when constrained by limited input material. Overall, we observed a high correlation of accessibility patterns and transcription factor dynamics between fresh and cryopreserved samples. Furthermore, highly similar transcriptomic profiles were obtained from whole cells and from the supernatants recovered from ATAC-seq reactions. We highlight the feasibility of applying these techniques to profile the epigenomic landscape of cells recovered from cryopreservation biorepositories.

Figure 1

The complexity of frozen samples recapitulates that of the fresh samples. Viability (a) and absolute cell count (b) of the fresh and thawed PBMC samples. (c) Recovery (%) of live PBMC recovery after thawing. The cell viability and recovery of PBMC samples were determined using trypan blue dye exclusion test. (d) Flow cytometric profile from a representative fresh and thawed PBMCs sample and the gating strategy to isolate Conventional T (Tconv) and Regulatory T (Treg) cells. (e–f) Colour dot plots showing events from fresh (e) and thawed (f) CD4 positive cells contributing to t-SNE (t-distributed stochastic neighbour embedding) distribution with manual clustering.

Figure 2

The quality of ATAC-seq from thawed Treg cells closely recapitulates that of the fresh cells. (a) The insert size distribution of fresh and thawed Treg ATAC-seq libraries. (b) Full library complexity and extrapolated yield curve of fresh and thawed ATAC-seq libraries. Complexity measurements are plotted against highest complexity (100% uniquely mapped reads). (c) Distribution of ATAC-seq signal at ± 1.5 kb transcriptional start sites (TSS). Signal coverage is calculated from reads per million mapped reads for each sample. (d) Percentage of reads mapping to mitochondrial genome. Deeper colour is used to depict the most desirable value of the statistic and range following a linear scale starting at 0 (black) and ending at the maximum value (yellow). All values were determined from the full depth of aligned reads. Statistical significance of the difference between the fresh and thawed samples is computed by Paired Student’s t-Test. The data shown in a-c are representative of sequencing reads pooled from three donors.

Figure 3

Chromatin accessibility is maintained through cryopreservation. (a) The fraction of reads in peaks for fresh and thawed Treg ATAC-seq libraries. Statistical significance of the difference between the fresh and thawed samples is computed by Paired Student’s t-Test. (b) Distribution of ATAC-seq peaks across distinct genomic feature expressed in proportion of annotated peaks. Promoters are defined by -1 kb to + 100 bp TSS region. Other, peaks annotated to 3’ UTR, 5’UTR, miRNA, non-coding RNA and TTS (transcription termination site). (c, d) Scatter plot of the read count per million (CPM) reads in ATAC-seq peaks identified from fresh and thawed samples during resting (c) and in response to stimulation (d). Pearson’s correlation coefficient value is indicated. (e) Correlogram showing the association of CPM reads for all the ATAC-seq samples generated for this study. (f, g) Chromatin accessibility profiles of fresh and thawed Treg cells during resting and stimulated state at the IL2RA (f) and FOXP3 (g) loci. ATAC-seq signal is intersected with T cell super enhancers, Treg chromatin states from Roadmap Epigenomics Project and FOXP3 binding sites. Each ATAC-seq track represents signal pooled from three donors. Browser view was generated using UCSC genome browser. The calculation for a-g was performed on pooled data representative of three donors.

Figure 4

Thawed Treg cells demonstrate comparable TF activity to fresh cells. ATAC-seq was performed on sorted resting and stimulated fresh and thawed Treg cells. (a) Correlation of stimulation-responsive chromatin accessibility changes in fresh and thawed Treg cells (expressed as fold changes in reads in peaks between resting and stimulated states). Differentially accessible genomic loci between resting and stimulated states were defined as genes having an FDR of less than 0.05 and log-fold change that is significantly greater (red) or lower (blue) than 1.2 (equivalent to a 2.3-fold difference between conditions). ATAC-seq peaks were annotated to the nearest TSS and common differentially accessible peaks with a log-fold change that is significantly greater/lower than 4 are annotated by gene symbol. Top 20 common differentially accessible Treg signature genes (Ferraro et al. ) are highlighted in green. (b–d) Histogram comparing fragments from fresh and thawed samples at FOXP3 (b), CTCF, RUNX1, CREB1, GATA3, IRF1 and YY1 binding motifs during resting (c) and stimulated (d) states. TF occupancy signal is computed on the genome-wide ATAC-seq footprints matching the corresponding motifs obtained from JASPAR database. Histograms are generated from Treg chromatin accessibility signals computed from pooled sequencing data representative of three donors.

Figure 5

Parallel measurement of chromatin accessibility and transcriptome dynamics in human Treg cells. RNA-seq was performed using RNA isolated from whole cells or supernatant (cytoplasmic) fraction recovered from ATAC-seq reaction prepared from thawed Treg cells. (a) Correlation of stimulation-responsive chromatin accessibility and gene expression changes in thawed, whole Treg cells (expressed as fold changes between resting and stimulated states). Differentially accessible promoter/expressed genes between resting and stimulated states were defined as genes having an FDR of less than 0.05 and log-fold change that is significantly greater or lower than 1.5. Promoters having differential chromatin accessibility and expression changes upon stimulation are highlighted. Blue line indicates loess fit to the distribution. (b) Differential analysis of transcriptomic profiles generated from ATAC-seq supernatant (cytoplasmic) fraction and whole Treg cells. Differentially expressed genes were defined as genes having an FDR of less than 0.05 and log-fold change that is significantly greater (red) or lower (blue) than 1.5. Top 20 differentially accessible genes are annotated by gene symbols. (c) Correlation of transcriptomic profiles generated from ATAC-seq supernatant fraction and whole cells. The expression for each gene is represented by log count per million (CPM) of sequencing reads. Differentially expressed genes are highlighted in red. Results shown are representative of four donors.