An “all-access” pass to a concert might get you behind the scenes with the band, but as a scientist wouldn’t you rather go behind the scenes in a cell, and get the inside scoop on the regulation of gene expression? (No? Just us?) Fortunately, new research from the labs of Cole Trapnell and Jay Shendure at the University of Washington School of Medicine gives us an all-access pass into mouse chromatin, and the unique architecture therein. This massive undertaking to generate a single-cell atlas of mouse chromatin accessibility is more impressive than any concert you’ve ever been to, guaranteed.
This groundbreaking work was accomplished using single cell combinatorial indexing (sci-) combined with ATAC-seq (Assay for Transposase-Accessible Chromatin using sequencing). This method generates genome-wide chromatin accessibility data (for thousands of cells at a time) which can be traced to a single cell via unique “barcodes” integrated during transposase activity and PCR amplification. This team isolated 13 different mouse tissue types, yielding over 100,000 cells for sequencing. From this, they identified more than 400,000 accessible sites in the mouse chromatin. This enormous accessibility dataset underwent two iterations of algorithmic grouping to yield 85 clusters of cells—each cluster representing a distinct chromatin accessibility pattern.
To determine what cell types comprised each cluster, they turned to Cicero—an algorithm pioneered in the Trapnell lab. Cicero uses single-cell-ATAC-seq data to connect distal regulatory elements with target genes by identifying and scoring co-accessible DNA elements. Cicero, in combination with cell type-specific markers, allowed researchers to assign cell types to 69 of the 85 clusters.
Here’s a glimpse of what chromatin accessibility data at the single cell level can reveal:
- By applying machine learning via a “neural network”, they could identify cell-specific influential motifs within differentially accessible chromatin, and even speculate novel motifs that may influence gene activity
- The chromatin accessibility patterns of similar cells types from different tissues could be compared
- Endothelial cells (from lung, liver, brain, kidney or heart) appear to have unique, tissue-specific chromatin patterns, whereas monocytes and macrophages have more similar chromatin patterns regardless of tissue type
- Temporal ordering of accessibility states can be used to investigate “trajectories”
- Accessibility data from mouse bone marrow can be sequentially ordered, revealing a timeline of changing chromatin accessibility during blood cell development
- The mouse chromatin accessibility dataset can be overlapped with human genome-wide association studies to infer enrichment of heritability for human diseases
- Many diseases show predictable enrichments: autoimmune diseases are enriched in leukocyte clusters, neurological traits are enriched in neuronal cells, etc.
- There are also unpredictable enrichments—for example, bipolar disorder was strongly enriched specifically for excitatory neurons—which could lead to increased understanding of disease states
Even at the level of chromatin accessibility, this paper demonstrates what a powerful tool a single-cell atlas can be and reiterates the need to move forward toward a human cell atlas.
Click over to Cell, August 2018 to read the full paper, no ticket purchase required.
If you’d like to read more about the ATAC-Seq method, please visit this great blog article from our friends at Active Motif – Complete Guide to Understanding and Using ATAC-Seq.