Layer-specific chromatin accessibility landscapes reveal regulatory networks in adult mouse visual cortex
- Allen Institute for Brain Science, United States
Figures
Figure 1 with 1 supplement
Overview of 500 cell ATAC-seq.
(a) Mouse visual cortex was isolated from transgenic mice by brain sectioning and microdissection, and dissociated into single-cell suspension. 500 fluorescently labeled cells were isolated from the …
-
Figure 1—source data 1
Cre-line cell type composition table, as plotted in Figure 1C.
- https://doi.org/10.7554/eLife.21883.003
-
Figure 1—source data 2
Fragment size frequencies for single replicates of each cell class.
- https://doi.org/10.7554/eLife.21883.004
Figure 1—figure supplement 1
Quality control plots for ATAC-seq libraries.
Each library is composed of DNA from 500 cells. For each library, we plotted the complexity curve derived from preseq output, the insert sizes derived using Picard Tools, and ATF2 footprinting from …
Figure 2 with 1 supplement
Peak locations relative to TSS and histone modifications.
(a) Histogram of peak positions relative to the nearest TSS location. Distance to nearest TSS was used to group peaks into three upstream categories (−3, –2, and −1) and three downstream categories …
-
Figure 2—source data 1
Distributions of peak locations relative to TSS, used for Figure 2A.
- https://doi.org/10.7554/eLife.21883.007
-
Figure 2—source data 2
Histone modification frequencies for peaks by cell class and distance bin, used for Figure 2B.
- https://doi.org/10.7554/eLife.21883.008
Figure 2—figure supplement 1
Comparisons of ATAC-seq peaks to TSS locations and Camk2a-Cre-derived histone ChIP-seq data.
(a) Absolute distances from peaks to the nearest TSS. Peaks separate into two clear populations when plotted on a log10 distance scale: p, proximal (<2 kb from a TSS); d, distal (>2 kb from TSS). (b)…
Figure 3 with 5 supplements
ATAC-seq samples cluster by cell class and reveal class-specific chromatin accessibility.
(a) Correlation between each sample pair was based on the number of overlapping HotSpot regions weighted by normalized accessibility scores for each sample. The pairwise correlation scores were then …
Figure 3—figure supplement 1
Hierarchical clustering of samples based on TSS-proximal or TSS-distal HotSpot results.
Top and bottom row, clustering of samples based on their HotSpot regions and peaks, respectively. HotSpot regions vary in length from 10 bp to 7.7 kb (median = 312 bp), while peaks are uniformly 150 …
Figure 3—figure supplement 2
Intersections among peak sets derived from different cell classes.
We first derived a merged peak set by Diffbind for all peaks identified in all cell classes. We counted how often each merged peak overlapped peaks called for all 3 replicates of each cell class, the…
Figure 3—figure supplement 3
Correlation of HotSpot peak sets from this study and ENCODE tissue DNase-seq.
To test if our datasets agree with previously published brain-derived chromatin accessibility datasets, we obtained HotSpot peaks derived from DNase-seq of 15 tissues from the Mouse ENCODE database. …
Figure 3—figure supplement 4
Correlation of brain-derived DNase-seq and ATAC-seq datasets.
(a) To test if our data showed similar patterns of chromatin accessibility as previously published mouse cortical ATAC-seq datasets, we performed DiffBind analysis of our datasets (underlined), …
Figure 3—figure supplement 5
Fraction of Reads in Peaks.
(a) Fraction of reads in peaks (FRiP) scores were calculated for each sample by counting the number of downsampled BAM reads (3.2M per sample) that overlapped the corresponding HotSpot peaks in that …
Figure 4 with 2 supplements
Chromatin accessibility corresponds to cell class-specific transcription.
(a) For each pair of cell classes, we identified differentially expressed genes (adjusted p-value < 0.05 and fold change > 2), then separated genes into two groups based on the class with higher …
-
Figure 4—source data 1
Peak accessibility scores (TMM) for peaks associated with gene sets in Figure 4A.
- https://doi.org/10.7554/eLife.21883.017
-
Figure 4—source data 2
Mann-Whitney test results for each comparison in Figure 4A.
- https://doi.org/10.7554/eLife.21883.018
-
Figure 4—source data 3
Gene expression data for the heatmap at the bottom of Figure 4B.
- https://doi.org/10.7554/eLife.21883.019
-
Figure 4—source data 4
Differential accessibility and –log10(pvalue) scores used to generate the volcano plot in Figure 4B.
- https://doi.org/10.7554/eLife.21883.020
-
Figure 4—source data 5
Gene expression data for the heatmap at the bottom of Figure 4C.
- https://doi.org/10.7554/eLife.21883.021
-
Figure 4—source data 6
Differential accessibility and –log10(pvalue) scores used to generate the volcano plot in Figure 4C.
- https://doi.org/10.7554/eLife.21883.022
Figure 4—figure supplement 1
Pairwise comparisons of peak accessibility for peaks associated with differentially-expressed genes.
(a) To assess the overall correspondence between differentially expressed genes and differentially accessible chromatin, we plotted the statistical significance for each peak that was positionally …
Figure 4—figure supplement 2
Permutation of peak-gene associations.
To test the correlation between gene expression and nearby peak accessibility, we calculated mean peak accessibility scores for each peak and mean gene expression scores for each gene across …
Figure 5 with 2 supplements
Clustering of peaks and genes reveals common patterns of chromatin accessibility and gene expression.
(a) Scaled module profiles derived from k-means clustering of peaks and genes (Materials and methods). Points represent median values, and shaded areas represent percentiles as shown in the legend. …
-
Figure 5—source data 1
Fisher’s exact test result values presented in Figure 5B.
- https://doi.org/10.7554/eLife.21883.026
-
Figure 5—source data 2
Quantile values for gene clusters presented in Figure 5A.
- https://doi.org/10.7554/eLife.21883.027
-
Figure 5—source data 3
Quantile values for peak clusters presented in Figure 5A.
- https://doi.org/10.7554/eLife.21883.028
Figure 5—figure supplement 1
Example instance of initial k-means clustering of differentially accessible peaks and differentially expressed genes.
We first performed k-means clustering on peaks (left) and genes (right) with k = 15. This resulted in overclustering, as many clusters are extremely similar to each other. Top heatmaps show Pearson …
Figure 5—figure supplement 2
Log-odds ratios for tests of Peak-Gene module association.
Log-odds ratios for Fisher's exact tests of enrichment between each pair of peak and gene modules, corresponding to the p-values reported in (Figure 5).
Figure 6 with 2 supplements
Peak module analysis reveals layer-specific enrichment of transcription factor motifs.
(a) Select TF motif families are significantly enriched or depleted in specific peak modules. Enrichment and depletion were calculated relative to unrelated modules (Figure 6—figure supplement 1) usi…
-
Figure 6—source data 1
AME result p-values, as plotted in Figure 6A.
- https://doi.org/10.7554/eLife.21883.032
-
Figure 6—source data 2
Gene expression values used for Figure 6B.
- https://doi.org/10.7554/eLife.21883.033
-
Figure 6—source data 3
FOXP motif Tn5 insertion frequency data.
- https://doi.org/10.7554/eLife.21883.034
-
Figure 6—source data 4
NEUROD motif Tn5 insertion frequency data.
- https://doi.org/10.7554/eLife.21883.035
-
Figure 6—source data 5
RFX motif Tn5 insertion frequency data.
- https://doi.org/10.7554/eLife.21883.036
Figure 6—figure supplement 1
Background set selection for AME and top significant AME results for each peak module.
(a) For each peak module (rows), the background modules are represented as colored boxes. These sets of foreground and background peaks were used for AME analysis. (b) Representative motif LOGOS for …
Figure 6—figure supplement 2
Transcription factor gene expression in individual cells arranged by cell class.
https://doi.org/10.7554/eLife.21883.038
Figure 7 with 1 supplement
Putative regulatory interactions that govern layer-specific chromatin and transcriptomic state in glutamatergic cell classes.
(a) Putative regulatory interactions between key TFs. (b) Putative regulatory interactions between key TFs and other differentially expressed TFs in glutamatergic cell classes. Key TFs have bold …
-
Figure 7—source data 1
Data used to build the network presented in Figure 7B and Figure 8.
- https://doi.org/10.7554/eLife.21883.040
Figure 7—figure supplement 1
Flowcharts for selecting TFs and TF targets to construct a network of regulatory interactions.
(a) Flowchart for selecting key TFs that act as activators or repressors by binding differentially enriched TF motifs. These genes were used as nodes in our network diagram. (b) Flowchart for …
Figure 8
Gene expression patterns of layer-specific transcription factors.
The location and identity of each node is the same as in the regulatory network presented in Figure 7. The color of each node represents the normalized, average gene expression across all cells in …
Figure 9
The Layer-specific regulatory landscape of Nfia.
(a) Nfia gene expression distributions based on scRNA-seq in each neuronal cell class. (b) Volcano plot showing all peaks that are positionally associated with Nfia in L6 and L4. Significantly …
-
Figure 9—source data 1
Nfia expression values used to generate the plot in Figure 9A.
- https://doi.org/10.7554/eLife.21883.044
-
Figure 9—source data 2
Peak statistics for peaks positionally associated with Nfia, used to generate Figure 9B.
- https://doi.org/10.7554/eLife.21883.045
Figure 10
Cell class-specific regulatory domains downstream of the Cux1 TSS.
(a) Cux1 gene expression distributions based on scRNA-seq in each neuronal cell class. (b) Volcano plot shoing all peaks that are positionally associated with Cux1 in a pairwise comparison between …
-
Figure 10—source data 1
Cux1 expression values used to generate the plot in Figure 10A.
- https://doi.org/10.7554/eLife.21883.047
-
Figure 10—source data 2
Peak statistics for peaks positionally associated with Cux1, used to generate Figure 10B.
- https://doi.org/10.7554/eLife.21883.048
Figure 11
Previously published TF-binding sites near Pou3f2 and Foxp2 observed during development are not accessible in adult mouse cortex.
(a) Previously described RORB-binding sites (BS1 and BS2) near the Pou3f2 TSS are not accessible in any of the adult cell classes we examined. (b) Same as in (a), but for a POU3F2 binding site in …
Figure 12
FOXP motif accessibility within or near the Mef2c gene.
(a) Chromatin accessibility upstream of the Mef2c TSS. Orange boxes, putative FOXP binding sites that are significantly less accessible in L6 than in upper layers; Gray boxes, …
Additional files
-
Supplementary file 1
Libraries, Peaks, and associated statistics.
(a) Library and alignment statistics. (b) Merged DiffBind peak locations, mean TMM scores per cell class, and pairwise DiffBind p-values. (c) Associations between peaks and nearest genes, average expression scores for associated genes, and pairwise DESeq2 differential expression p-values.
- https://doi.org/10.7554/eLife.21883.051
-
Supplementary file 2
Motif analysis results.
(a) Peak module AME results. (b) FIMO motif search results. (c) Network interaction table.
- https://doi.org/10.7554/eLife.21883.052
-
Supplementary file 3
Downsampled RNA-seq data.
- https://doi.org/10.7554/eLife.21883.053
