Improve Single Cell Transcriptional Profiling with CRISPRclean
Transcriptional profiling has been revolutionized with high throughput, massively parallel, single cell RNA sequencing. Microfluidic/microdroplet technologies recently introduced to the market produce complex data sets that allow investigators to deconvolute complex cell mixtures, identify cell types present in healthy and diseased tissues, and create cell type specific transcriptional signatures. These technologies are dramatically enhancing our ability to identify transcriptional and cellular perturbations driving disease at the individual cell level.
CRISPRclean is a post-library depletion method that utilizes CRISPR-Cas technology to deplete uninformative molecules from a sequencing library.
Overexpressed and housekeeping transcripts dominate sequencing reads
Microfluidic/microdroplet technologies are expensive and enable only sparse sampling of RNAs from each cell, with many genes represented by only 1 or 2 sequencing reads. This limit is due partially to the fact that libraries are dominated by an abundance of housekeeping RNAs, which dominate sequencing reads and limit detection of even moderately expressed transcripts that often drive biological differences between cell types.
CRISPRclean removes 100 overexpressed, housekeeping transcripts so you can see all transcripts, even those genes that were represented by only 1 or 2 sequencing reads.
Our goal was to target genes of little biological interest that are consistently at high abundance in single cell sequence data. We apply CRISPRclean to remove 100 overexpressed, housekeeping transcripts that are abundantly expressed in all cell types. The genes consist of ribosomal and mitochondrial protein coding transcripts (mRNA with poly A tails).
The SAMPLES & Method
Matched diseased and normal atherosclerotic carotid artery tissue collected during carotid endarterectomy from three donors
The normal tissue samples contain ~3,500 cells and the diseased tissue samples have ~11,000 cells. Each 10x Genomics library was sequenced on an Illumina NextSeq instrument with ~250 million 2 x 150 PE reads corresponding to ~60K reads per cell for the normal and ~24K reads per cell for the diseased tissue samples (sequencing saturation of ~85% and ~65% for normal and diseased sample types, respectively). The prepared libraries were subsequently depleted with the CRISPRclean scRNA depletion product, as described in Figure 1. The samples were sequenced on the same NextSeq instrument again, for a similar read length and depth as the original (non-depleted) samples. All sequence data was processed through Cell Ranger, subject to standard mitochondrial and doublet detection algorithms, and partitioned into cell-types using Monocle3. Differential expression was performed using a generalized linear model to control for patient-specific transcriptomic effects.
Effect of abundant gene depletion on single cell gene expression
Differential gene expression results demonstrate that the vast majority of genes targeted for depletion display the greatest reduction in expression across all genes analyzed in both health and diseased samples. Figure 2B shows that mitochondrial gene UMI counts are dramatically lower after depletion. As a result of mitochondrial and ribosomal gene depletion, total UMI count per cell are lower as well (Figure 2A).
Depletion results in an increase of 2,747 UMIs (12.6% of all genes in the library) derived from healthy tissue and 4,260 (18.4% of all genes in the library) derived from diseased tissue (Table 2) indicating that depletion has improved the ability to detect certain genes.
Since depleted libraries are derived from undepleted ones, 10x Genomics cell barcodes are identical between depleted and undepleted libraries of the same type. Thus we can compare these libraries for alterations in UMI numbers per cell in the form of linear regression plots for selected individual genes in Figure 3.
The plots show a dramatic reduction in UMIs per cell for the genes selected for depletion (Figure 3 Tissue Gene RRS12). The plots also show an increase in UMIs per cell for certain (non-targeted) genes as a consequence of depletion (Figure 3 Tissue Gene CCL14).
Effect of abundant gene depletion on cell type identification and resolution
Figure 4A shows a UMAP plot with an overlay of the original and depleted cell data demonstrating that cell type identification is highly preserved.
Figure 4B shows the same plot adjusted to account for the depletion of the 100 targeted genes (i.e., depletion effect regressed out). The results indicate that depletion does not affect resolution of cell types within the cell population.
CRISPRclean increases the ability to detect rare transcripts and adds confidence to differential gene expression data when applied to 10x Genomics single cell RNA sequencing libraries
CRISPRclean may allow the user to glean new biologically relevant information from scRNA-seq data and, thus, gain novel insight into the genetic and cellular basis for disease.
Early access sign up for CRISPRclean Single Cell RNA Seq kit is now open!
Fill out the contact form below and ask to be an early access user. We will reach out to you with the latest details.