RNA splicing programs define tissue compartments and cell types at single-cell resolution

  1. Julia Eve Olivieri
  2. Roozbeh Dehghannasiri
  3. Peter L Wang
  4. SoRi Jang
  5. Antoine de Morree
  6. Serena Y Tan
  7. Jingsi Ming
  8. Angela Ruohao Wu
  9. Tabula Sapiens Consortium
  10. Stephen R Quake
  11. Mark A Krasnow
  12. Julia Salzman  Is a corresponding author
  1. Institute for Computational and Mathematical Engineering, Stanford University, United States
  2. Department of Biomedical Data Science, Stanford University, United States
  3. Department of Biochemistry, Stanford University, United States
  4. Department of Neurology and Neurological Sciences, Stanford University School of Medicine, United States
  5. Department of Pathology, Stanford University Medical Center, United States
  6. Academy for Statistics and Interdisciplinary Sciences, Faculty of Economics and Management,East China Normal University, China
  7. Department of Mathematics, The Hong Kong University of Science and Technology, China
  8. Department of Chemical and Biological Engineering, The Hong Kong University of Science and Technology, China
  9. Chan Zuckerberg Biohub, United States
  10. Department of Bioengineering, Stanford University, United States
8 figures, 1 table and 6 additional files

Figures

Analysis of alternative splicing in single-cell RNA-seq.

(A) Human, mouse lemur, and mouse single-cell RNA-seq from 10X and SS2 were run through the SpliZ pipeline for differential splicing discovery. (B) 10X data from the first human individual contains …

Figure 2 with 3 supplements
Compartment-specific alternative splicing revealed by applying the SpliZ to scRNA-seq data.

(A) Dot and sashimi plots showing LIMCH1 compartment-specific exon skipping (involving 5′ splice site [5′ SS] 41,619,440 and two 3′ splice sites [3′ SS] 41,638,932 and 41,644,500) impacting a …

Figure 2—figure supplement 1
Gene expression and the SpliZ value of MYL6 across single cells in the Tabula Sapiens dataset.

Coloring Tabula Sapiens cells by both gene expression and SpliZ value shows that MYL6 is ubiquitously expressed and that the SpliZ is independent of gene expression for these cases. Plots are …

Figure 2—figure supplement 2
Gene expression and the SpliZ value of RPS24 across single cells in the Tabula Sapiens dataset.

Coloring Tabula Sapiens cells by both gene expression and SpliZ value shows that RPS24 is ubiquitously expressed and that the SpliZ is independent of gene expression for these cases. Plots are …

Figure 2—figure supplement 3
Gene expression and the SpliZ value of ATP5F1C across single cells in the Tabula Sapiens dataset.

Coloring Tabula Sapiens cells by both gene expression and SpliZ value shows that ATP5F1C is ubiquitously expressed and that the SpliZ is independent of gene expression for these cases. Plots are …

Figure 3 with 2 supplements
Compartment-specific alternative splicing in MYL6.

Differential alternative splicing between compartments for MYL6 is driven by an exon skipping event with orthologous splice sites (SS) in (A) human (5′ SS: 56,160,320 and two 3′ SSs: 56,161,387 and …

Figure 3—figure supplement 1
Uncropped plots of MYL6 expression.

(A) Human, (B) lemur, and (C) mouse for the splice sites shown in Figure 3.

Figure 3—figure supplement 2
FISH validation for MYL6 alternative splicing in cells isolated from human muscle.

Indicated cell types were isolated from human muscle and stained by RNA FISH. Example images are shown on the left, with the exon 6 isoform shown in red, the 5–7 isoform shown in gray in the DIC …

Figure 4 with 2 supplements
RPS24 has striking compartment-specific alternative splicing.

(A) Each colored circle in the plot represents one RPS24 junction that uniquely identifies an isoform (junction with green circle represents two isoforms). For each cell type (y axis), the median of …

Figure 4—figure supplement 1
FISH validation for RPS24 alternative splicing in cells isolated from human muscle.

Indicated cell types were isolated from human muscle and stained by RNA FISH. Example images are shown on the left, with the -a-b+c isoform shown in red, the -a-b-c isoform shown in gray in the DIC …

Figure 4—figure supplement 2
RPS24 isoforms quantified by the Bowtie2 aligner validate STAR-based discoveries.

Bar plots for the (A) endothelial, (B) epithelial, (C) immune, and (D) stromal compartments show proportions of each isoform for each species, with error bars corresponding to 95% binomial …

Figure 5 with 1 supplement
Correlations show high concordance of the SpliZ values for significant genes across biological replicates.

(A) When subsetted to only shared junctions and shared cell types, the SpliZ values for significant genes for both 10X datasets are highly concordant (Pearson correlation of 0.776). (B) Comparing …

Figure 5—figure supplement 1
Choosing effect size filters.

Effect size filters were chosen based on correlation analysis of (A) Both human 10X datasets and (B) separately for SpliZVD by correlation of both human SS2 datasets.

Figure 6 with 1 supplement
Cell-type-specific and conserved alternative splicing in TPM1.

Conserved splicing in TPM1 is recovered in (A) human, (B) mouse lemur, and (C) mouse. TPM1 has a pattern of differential splicing involving two cassette exons and an alternate 5′ end (as shown by …

Figure 6—figure supplement 1
Cell-type-specific splicing in other members of the tropomyosin family.

Both TPM2 (A) and TPM3 (B) exhibit cell-type-specific splicing patterns in human.

Figure 7 with 1 supplement
Cell-type-specific alternative splicing is prevalent in human genes and can reveal novel cell subpopulations.

(A) Cell-type-specific exon skipping in PNRC1 (involving one 3′ SS and two 5′ SS’s) is replicated across the four human datasets. Skeletal muscle satellite stem cells include the exon about 50% of …

Figure 7—figure supplement 1
Differential alternative splicing of FYB1 within the immune compartment.

The pattern is replicated in 10X and SS2 data from both human individuals and technologies, with monocytes and macrophages tending to include an exon (at position 39125998) that is usually skipped …

Figure 8 with 2 supplements
Developmentally regulated alternative splicing in mammalian spermatogenesis.

(A) Regulated alternative splicing of MTFR1 during sperm development. Significant negative correlation (Spearman’s correlation = –0.27, p-value = 1.23e-7) between the SpliZ score for gene MTFR1 and …

Figure 8—figure supplement 1
Regulated alternative splicing of MLF1 during sperm development in human, mouse, and mouse lemur.

The same 5′ splice site drives the alternative splicing in human and mouse. The gene structures for MLF1 according to the RefSeq database for human, mouse, and mouse lemur are shown.

Figure 8—figure supplement 2
Regulated alternative splicing of SPTY2D1OS during sperm development in human and mouse lemur.

Regulated splicing in both human and mouse lemur involves one unannotated 5′ splice site. The gene structures according to the RefSeq database for human and mouse lemur are shown.

Tables

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Software, algorithmSICILIANDehghannasiri et al., 2021https://github.com/salzmanlab/SICILIAN, Roozbeh, 2021
Software, algorithmSpliZ PipelineOlivieri et al., 2021https://github.com/juliaolivieri/SpliZ_pipeline, Julia, 2021b
Software, algorithmSTARDobin et al., 2013https://github.com/alexdobin/STAR, Alexander, 2021
Commercial assay or kitBaseScope Duplex Reagent Kit - HsACD (Bio-Techne)cat. no 323,870
Commercial assay or kitBaseScope Probes for MYL6ACD (Bio-Techne)BA-Hs-MYL6-tv1-1zz-st-C2 and BA-Hs-MYL6-tv2-1zz-st
Commercial assay or kitBaseScope Probes for RPS24ACD (Bio-Techne)BA-Hs-RPS24-tva-1zz-st-C2 and BA-Hs-RPS24-tvc-1zz-st
Sequence-based reagentFL-RPS24ex4F1This paperPCR primer/6FAM/CAATGTTGGTGCTGGCAAAA
Sequence-based reagentRPS24ex6R2This paperPCR primerGCAGCACCTTTACTCCTTCGG

Additional files

Supplementary file 1

Dataset summary.

Brief overview of the datasets used in this paper, including tissues analyzed, number of cells, median number of spliced reads per cell, sex, and age.

https://cdn.elifesciences.org/articles/70692/elife-70692-supp1-v2.tsv
Supplementary file 2

Differential alternative splicing per compartment.

Separate table with the p-value based on the SpliZ and SpliZVD for each gene in each dataset, testing differences between compartments. The gene name, SpliZ p-value, SpliZVD p-value, and largest magnitude median for all compartments for SpliZ and SpliZVD for each gene are given by the geneR1A_uniq, perm_pval_adj_scZ, perm_pval_adj_svd_z0, max_abs_median_scZ, and max_abs_median_svd_z0 columns, respectively. (A) Human individual 1 10X; (B) human individual 2 10X; (C) human individual 1 SS2; (D) human individual 2 SS2; (E) lemur individual 1 10X; (F) lemur individual 2 10X; (G) mouse individual 1 10X; (H) mouse individual 2 10X.

https://cdn.elifesciences.org/articles/70692/elife-70692-supp2-v2.xlsx
Supplementary file 3

Differential alternative splicing per cell type.

Separate table with the p-values based on the SpliZ and SpliZVD for each gene in each dataset, testing differences between cell types. The gene name, SpliZ p-value, SpliZVD p-value, and largest magnitude median for all compartments for SpliZ and SpliZVD for that gene are given by the geneR1A_uniq, perm_pval_adj_scZ, perm_pval_adj_svd_z0, max_abs_median_scZ, and max_abs_median_svd_z0 columns, respectively. (A) human individual 1 10X; (B) human individual 2 10X; (C) human individual 1 SS2; (D) human individual 2 SS2; (E) lemur individual 1 10X; (F) lemur individual 2 10X; (G) mouse individual 1 10X; (H) mouse individual 2 10X.

https://cdn.elifesciences.org/articles/70692/elife-70692-supp3-v2.xlsx
Supplementary file 4

Most variable splice sites (SpliZsites).

The most variable splice sites (SpliZsites) for genes with significant alternative splicing in human individual 1 10X data. Each line reports a SpliZsite and contains the coordinates, whether it is an annotated exon, whether it is an exon with known alternative splicing, whether the splice site is in 5′ or 3′ UTR of the gene, and whether the SpliZsite found to be a SpliZsite in mouse lemur and mouse datasets.

https://cdn.elifesciences.org/articles/70692/elife-70692-supp4-v2.csv
Supplementary file 5

Regulated alternative splicing events in spermatogenesis.

The list of genes with significantly regulated alternative splicing during sperm development. Each line contains information about the number of cells, Spearman’s correlation, and its p-value for the human gene and also the same information (based on the mouse and mouse lemur sperm data) for its orthologous genes in mouse and mouse lemur genomes.

https://cdn.elifesciences.org/articles/70692/elife-70692-supp5-v2.txt
Transparent reporting form
https://cdn.elifesciences.org/articles/70692/elife-70692-transrepform1-v2.docx

Download links