Simultaneous measurement of chromatin accessibility, DNA methylation, and nucleosome phasing in single cells

  1. Sebastian Pott  Is a corresponding author
  1. University of Chicago, United States
6 figures and 2 additional files

Figures

Figure 1 with 3 supplements
Overview of scNOMe-seq procedure.

(a) Schematic of GpC methyltransferase-based mapping of chromatin accessibility and simultaneous detection of endogenous DNA methylation. (b) Schematic of scNOMe-seq procedure introduced in this …

https://doi.org/10.7554/eLife.23203.003
Figure 1—figure supplement 1
FACS profile from Hoechst stained nuclei to assess DNA content.

Nuclei were stained with Hoechst 33342 DNA dye and nuclei with DNA content corresponding to the G1-phase of the cell cycle were sorted into individual wells in a 96 well plate. Aggregates and debris …

https://doi.org/10.7554/eLife.23203.004
Figure 1—figure supplement 2
Schematic of experimental set up.

A total of 19 individual cells from GM12878 were profiled in this study, 12 of these cells were exposed to GpC MTase and seven were subjected to the same process without exposure to MTase. For K562 …

https://doi.org/10.7554/eLife.23203.005
Figure 1—figure supplement 3
Number of covered GpC and CpG dinucleotides is proportional to the number of total bases covered.

Number of covered cytosines in GpC and CpG dinucleotides plotted against the total number of nucleotides covered per sample. This comparison suggests that there is no strong bias towards or against …

https://doi.org/10.7554/eLife.23203.006
Figure 2 with 15 supplements
scNOMe-seq data reveal how accessibility in single cells underlies observed DNaseI hypersensitivity in a population of cells.

(a) Average GpC methylation level (blue) and CpG methylation level (orange) at DHSs in GM12878 cells. Regions are centered on the middle of DNase-seq peak locations. Shown is the average methylation …

https://doi.org/10.7554/eLife.23203.007
Figure 2—figure supplement 1
Average CpG and GpC methylation levels in single cells.

Boxplots representing the methylation level at CpG and GpC dinucleotides for groups of cells (GM12878 w/ and w/o MTase,K562 w/ MTase). GM12878 and K562 cells show different levels of CpG …

https://doi.org/10.7554/eLife.23203.008
Figure 2—figure supplement 2
Heatmaps of average GpC and CpG methylation across DHS regions in GM12878 cells.

Each row represents data from an individual cell, both treated and control samples are plotted together. Cells were grouped using hierarchical clustering based on GpC methylation (left) and CpG …

https://doi.org/10.7554/eLife.23203.009
Figure 2—figure supplement 3
Average GpC and CpG methylation across DHS regions in K562 cells.

Average GpC methylation level (blue) and CpG methylation level (orange) at DNase Hypersensitive sites (DHSs) in K562 cells. Regions are centered on the middle of DNase-seq peak locations. Shown is …

https://doi.org/10.7554/eLife.23203.010
Figure 2—figure supplement 4
Heatmaps of average GpC and CpG methylation across DHS regions in K562 cells.

Each row represents data from an individual cell. Cells were grouped using hierarchical clustering based on GpC methylation (left) and CpG methylation (right) within 2 kb regions around DHSs.

https://doi.org/10.7554/eLife.23203.011
Figure 2—figure supplement 5
Distribution of counts of GpCs within DHSs in GM12878 and K562 cells.

Histogram shows the number of GpCs per DHS in GM12878 cells (left) and K562 cells (right). While each GpC dinucleotide can be measured on both strands and would therefore yield a count of two …

https://doi.org/10.7554/eLife.23203.012
Figure 2—figure supplement 6
Proportion of DHSs at different cutoffs for GpCs and CpGs.

Bar graphs display proportion of DHSs that contain at least the number of GpCs (left) and CpGs (right) indicated. The proportions are given in relation to the total number of DHSs in each cell line. …

https://doi.org/10.7554/eLife.23203.013
Figure 2—figure supplement 7
Relationship between DNase-seq peak score and GpC and CpG methylation in GM12878 and K562 cells.

(a) Average GpC methylation and (b), (c) endogenous CpG methylation at DHSs grouped into quartiles based on associated DNase-seq peak scores from lowest to highest scores. ‘Shuffled’ represents …

https://doi.org/10.7554/eLife.23203.014
Figure 2—figure supplement 8
Correlation between GpC methylation and DHS peak score.

Shown are correlation coefficients for comparisons between single cell and bulk NOMe-seq data with DNase-seq peak score for each covered location for (a) GM12878 and (b) K562. Each dot represents …

https://doi.org/10.7554/eLife.23203.015
Figure 2—figure supplement 9
Cumulative distribution of average GpC methylation in DHSs in GM12878 and K562 cells.

Plot of cumulative distribution of GpC methylation for individual GM12878 and K562 cells at DHSs with at least four covered GpC. GM12878 and K562 cells exposed to GpC MTase show similar …

https://doi.org/10.7554/eLife.23203.016
Figure 2—figure supplement 10
Proportion of accessible DHSs remains stable across range of thresholds for methylation levels and covered GpCs per site.

Different thresholds for GpC methylation and number of covered GpC required per individual DHS were used to test how much the number of resulting ‘accessible’ DHSs depended on these parameters. …

https://doi.org/10.7554/eLife.23203.017
Figure 2—figure supplement 11
Cell-to-cell variability in DHSs accessibility reflects DNaseI hypersensitivity of the region.

Pair-wise jaccard distances between GM12878 (a) and K562 (b) cells, respectively, were calculated based on DHSs accessibility in individual cells. DHSs were grouped by DNase-seq peak scores and DHSs …

https://doi.org/10.7554/eLife.23203.018
Figure 2—figure supplement 12
Comparison of correlations between single cell NOMe-seq and bulk NOMe-seq data sets.

Shown are the correlation coefficients for comparisons between GpC methylation in single cell and bulk NOMe-seq data within DNase-seq peak location for a) GM12878 and b) K562. Each dot represents …

https://doi.org/10.7554/eLife.23203.019
Figure 2—figure supplement 13
GpC methylation correlates with DHS peaks scores in individual cells.

High DHS peak scores are associated with higher GpC methylation in single cells. Scatter plot showing relationship between GpC methylation levels and DHS peaks scores for each covered DHS. Each plot …

https://doi.org/10.7554/eLife.23203.020
Figure 2—figure supplement 14
Endogenous CpG methylation is inversely correlated with DHS peak scores in individual cells.

High DHS peak scores are associated with lower endogenous CpG methylation in single cells. Scatter plot showing relationship between CpG methylation levels and DHS peaks score for each covered DHS. …

https://doi.org/10.7554/eLife.23203.021
Figure 2—figure supplement 15
Comparison of CpG and GpC methylation status at individual DHS in single GM12878 cells.

Smoothened scatterplot illustrates the relationship between endogenous CpG methylation and GpC methylation at DHS loci. Each plot shows data from a single GM12878 cell.

https://doi.org/10.7554/eLife.23203.022
Figure 3 with 2 supplements
Single cell NOMe-seq reveals chromatin features closely linked to gene expression.

(a) Average GpC methylation level at TSS in GM12878 cells. Regions are centered on the TSS locations. Shown is the average methylation across a 2 kb window of 12 GM12878 cells. (b) Same as in a) but …

https://doi.org/10.7554/eLife.23203.023
Figure 3—figure supplement 1
Endogenous methylation in gene bodies of single K562 cells.

(a) Average endogenous CpG methylation at gene loci in individual K562 cells. Shown is the average methylation across gene bodies (represented as meta genes) and 50 kb regions upstream and …

https://doi.org/10.7554/eLife.23203.024
Figure 3—figure supplement 2
Chromatin accessibility in promoters correlates with transcript levels of adjacent genes.

(a) Average GpC methylation level at TSS genes in GM12878 cells. Regions are centered on the TSS locations and genes were grouped into quartiles based on their transcript levels in bulk GM12878 …

https://doi.org/10.7554/eLife.23203.025
Figure 4 with 1 supplement
single cell GpC and CpG methylation signal is sufficient to group GM12878 and K562 cells according to their origin.

(a) Heatmap shows similarity scores (pair-wise Jaccard distances) for accessibility between all GM12878 and K562 cells measured on the union set of DHSs from GM12878 and K562 cells. Cells were …

https://doi.org/10.7554/eLife.23203.026
Figure 4—figure supplement 1
Single GM12878 and K562 cells can be grouped based on GpC methylation and endogenous methylation.

(a) Heatmap shows correlation coefficients for GpC methylation between all GM12878 and K562 cells measured on the union set of DHSs from GM12878 and K562 cells. Cells were grouped based on …

https://doi.org/10.7554/eLife.23203.027
Figure 5 with 5 supplements
scNOMe-seq detected characteristic accessibility patterns at CTCF transcription factor binding sites and measured CTCF footprints at individual loci.

(a) Average GpC methylation level (blue) and CpG methylation level (orange) at CTCF binding sites in GM12878 cells. Regions are centered on motif locations. Shown is the average methylation across a …

https://doi.org/10.7554/eLife.23203.028
Figure 5—figure supplement 1
Average GpC methylation and endogenous CpG methylation at CTCF sites in pooled K562 cells.

Average GpC methylation level (blue) and CpG methylation level (orange) at CTCF binding sites in K562 cells. Regions are centered on motif locations. Shown is the average methylation across a 2 kb …

https://doi.org/10.7554/eLife.23203.029
Figure 5—figure supplement 2
Average GpC methylation level at CTCF binding sites in individual K562 cells.

Heatmap shows the average GpC methylation across a 2 kb window centered on the CTCF motif location. Each row corresponds to an individual K562 cell and rows are grouped by hierarchical clustering.

https://doi.org/10.7554/eLife.23203.030
Figure 5—figure supplement 3
Average GpC methylation and endogenous CpG methylation at additional transcription factor binding sites in pools of GM12878 and K562 cells.

Average GpC methylation level (blue) and CpG methylation level (orange) at (a) PU.1 binding sites in GM12878 cells and b) EBF1 binding sites. Regions are centered on motif locations. Shown is the …

https://doi.org/10.7554/eLife.23203.031
Figure 5—figure supplement 4
Scores at CTCF motifs with footprints are significantly higher than those without.

Boxplot representing the CTCF motif scores in regions with and without CTCF footprint of an individual GM12878 cell (GM_1, the same cell as shown in Figure 3d and e). Within individual cells regions …

https://doi.org/10.7554/eLife.23203.032
Figure 5—figure supplement 5
Loci with CTCF footprint in single cells.

At each locus scNOMe-seq data from this study and DNase hypersensitivity data from ENCODE are shown. scNOMe-seq data tracks show methylation status of individual GpCs. Each row corresponds to data …

https://doi.org/10.7554/eLife.23203.033
Figure 6 with 2 supplements
Nucleosome phasing in single cells.

(a) Average GpC methylation level and (b) CpG methylation level at well-positioned nucleosomes in GM12878 cells. Regions are centered on midpoints of top 5% of positioned nucleosomes. Shown is the …

https://doi.org/10.7554/eLife.23203.034
Figure 6—figure supplement 1
Number of nucleotide pairs used for correlation at each offset distance.

Plotted is the number of nucleotide pairs that are found at each offset distance and used to calculate the correlation coefficient at that distance. The number of comparison declines precipitously. …

https://doi.org/10.7554/eLife.23203.035
Figure 6—figure supplement 2
Offset distances with a high proportion of GpC pairs that share methylation status indicate average phasing distance.

Shown is the proportion of nucleotide pairs at each offset distance in which both cytosines are methylated. These measurements yield curves very similar to the distribution of Pearson correlation …

https://doi.org/10.7554/eLife.23203.036

Additional files

Supplementary file 1

Table 1: scNOMe-seq libraries used in this paper and their technical details and alignment summary statistics. Table 2: Primer sequences of primers used for amplification and barcoding of sequencing library. Table 3: Additional datasets used in this study and their sources.

https://doi.org/10.7554/eLife.23203.037
Source code 1

Script to calculate autocorrelation between GpCs at offset distances from 3–400bp.

https://doi.org/10.7554/eLife.23203.038

Download links