NEW YORK (GenomeWeb) – A team from the US and Netherlands has demonstrated that it can get a more refined look at cell subtypes, find new transcript isoforms, and bolster genome annotation with a fluorescence-activated cell sorting-free method for separating cell types and sequencing RNA isoforms in individual cells.
For a paper published in Nature Biotechnology today, the researchers applied this microfluidic amplification-based method — called "single-cell isoform RNA-Seq," or ScISOr-Seq — to thousands of cells from the cerebellum region of the mouse brain, including neurons, astrocytes, microglia, and Purkinje or Granule cell subtypes. In the process, they were able to find cell type-specific expression patterns for tens of thousands of new or known transcript isoforms.
"ScISOr-Seq enables long-read, full-length RNA-seq in single cells that can be clustered into cell types with a very low identification error rate," corresponding author Hagen Tilgner, a researcher with Weill Cornell Medicine's Brain and Mind Research Institute and Center for Neurogenetics, and his colleagues wrote.
Starting with cerebellum samples dissected from mice at post-natal day one, the researchers funneled disassociated cells from bulk tissue samples into the 10X Genomics Chromium controller system before amplifying complementary DNA coinciding to the RNA and using it to prepare libraries for their subsequent Illumina, Pacific Biosystems, and Oxford Nanopore MinIon sequencing experiments.
"Short-read 3' sequencing provided molecular counts for each gene and cell, which enabled clustering of cells and cell types assignment using cell type-specific markers," the authors explained. They also used the PacBio circular consensus sequences and other PacBio or Oxford Nanopore reads to profile full-length RNA isoforms.
Based on nearly 3,900 molecular identifiers and information at more than 1,400 genes per cell, for example, the team identified 17 clusters of cell types and subtypes for 6,627 individual mouse cerebellar cells subjected to ScISOr-Seq. The PacBio or MinIon long reads made it possible to find RNA transcript isoforms expressed in one or more of these cell clusters.
The researchers went on to replicate these findings in additional cells, and compare their ScISOr-Seq results to those found by sparse isoform sequencing. Using cell type-specific expression profiles for 16,872 new and 18,173 known transcript isoforms, meanwhile, they also demonstrated that they could improve version 10 of the Gencode mouse genome.
Based these and other results, the authors suggested that the ScISOr-Seq strategy should prove useful for profiling cell type-specific isoform splicing in other situations as well, including analyses of human brain samples from individuals with neurological disease-related genetic risk variants.
"In the future, we envisage that for genetic risk factors for brain disorders, for example, Alzheimer's-disease-related genes, such as MAPT, BIN1, and APOE, the effects of disease-associated single-nucleotide polymorphisms can be explained by analyzing cell-type-specific isoform expression using ScISOr-Seq," the authors concluded.