Single-nuclei isoform RNA sequencing reveals combination patterns of transcript elements across human brain cell types

Single-nuclei RNA-Seq is widely employed to investigate cell types, especially of human brain and frozen samples. In contrast to single-cell approaches, many single-nuclei reads are purely intronic. Here, using microfluidics, PCR-based artifact removal, target enrichment, and long-read sequencing, we developed single-nuclei isoform RNA-sequencing (‘SnISOr-Seq’), and applied it to human adult frontal cortex. SnISOr-Seq dramatically increased the fraction of informative reads. We found that exons associated with autism exhibit coordinated and highly cell-type specific inclusion. We discovered two distinct combination patterns: first, those distinguishing neural cell types, enriched in TSS-exon, exon-polyA-site, and non-adjacent exon pairs. Second, those with multiple configurations within one cell type, enriched in adjacent exon pairs. Furthermore, adjacent exons are predominantly mutually-associated, while distant exons are frequently mutually-exclusive. Finally, we observed that human-specific exons are almost as tightly coordinated as conserved exons. SnISOr-Seq enables single-nuclei long-read isoform analysis in human brain, and in any frozen or hard-to-dissociate sample. 10X single-nuclear RNA sequencing done on two replicates of human Frontal Cortex samples. PacBio done on the same cDNA, after linear PCR. 10X single-cell RNA sequencing followed by Oxford Nanopore long read sequencing as a control sample. FACS sorted nuclei obtained from Human prefrontal cortex samples for bulk RNA sequencing on Illumina, also used as control.