RNA isoform diversity, splicing variants and switching in single cells of the Alzheimer's disease brain - PubMed (original) (raw)
RNA isoform diversity, splicing variants and switching in single cells of the Alzheimer's disease brain
Anis Shahnaee et al. Commun Biol. 2026.
Abstract
Alzheimer's disease (AD) is the most common cause of dementia, yet its molecular causes remain incompletely understood. RNA diversity in part arising from dysregulated splicing may contribute to AD pathogenesis, however the ability to interrogate the resulting full-length isoforms in single brain cells has been technologically limited. Here we report the use of PacBio Kinnex long-read sequencing combined with 10X Genomics single-cell preparations to identify both annotated and unannotated RNA isoforms. Eight AD and seven non-diseased post-mortem human brains yield ~70,000 single nuclei showing diverse, differentially expressed and switched transcripts in multiple genes. Cell-type-specific isoform expression and variants with intra-exonic junctions associated with reverse transcriptase-mediated somatic gene recombination are also detected. Novel isoforms are altered in AD, with examples of CHI3L1 and SEPTIN4. Kinnex detection of RNA isoforms from single nuclei highlights vast isoform diversity amongst brain cell types, representing an under-explored element in AD and other brain disorders.
© 2026. The Author(s).
Conflict of interest statement
Competing interests: J.C. has an employment relationship with Neurocrine Biosciences, a company that may potentially benefit from the research results. J.C.’s relationship with Neurocrine Biosciences has been reviewed and approved by Sanford Burnham Prebys Medical Discovery Institute in accordance with its conflict of interest policies. All other authors declare no competing interests.
Figures
Fig. 1. Single-nucleus short-read RNA-sequencing and Kinnex long-read sequencing workflow and cell type-specific analyses.
a Schematic representation of experimental workflow for single-nucleus short-read RNA-sequencing and Kinnex long-read RNA-sequencing. Figure 1a. was partially created with BioRender.
. b UMAP plot colored by cell type assignments. c Bar plot showing the proportions of each cell type in AD and ND samples (ns: not significant; Two-sided t-test). d Differentially expressed genes (DEGs) identified across different cell types in AD and ND samples (absolute log2 fold change > 0.25 and adjusted _p_-value < 0.05, Wilcoxon Rank Sum test with Bonferroni correction). e–h Circos plots of five selected Gene Ontology (GO) terms for excitatory neurons, microglia, oligodendrocytes, and astrocytes with associated genes (false discovery rate (FDR) of <0.05 with the Benjamini-Hochberg (BH) test). Conserved GO terms are shown in the same color.
Fig. 2. Overview of Kinnex long-read sequencing quality control and isoform filtering criteria.
a Schematic representation of SQANTI3 categories used for isoform classification. b Scatter plot showing the number of isoforms with CPM ⩾1 in different numbers of samples. c Distribution of isoforms by highest CPM cutoff that would meet the four-sample cutoff. d Distribution of isoform lengths in the confident Kinnex dataset. e Bar plot showing the distribution of the number of exons in each confident isoform. f Bar plot showing the number of isoforms in each SQANTI3 category that met or did not meet the confidence cut-off. g The number of confident isoforms identified per gene. h The number of isoforms per sample group that had CPM ⩾1. Zero samples means that isoforms were detected with CPM < 1. Gray boxes indicate isoforms that did not meet the confidence cut-off. i Bar plot showing the proportion of reads assigned to each SQANTI3 category in AD and ND samples without applying the confidence cut-off. j Bar plot showing the proportion of reads assigned to each SQANTI3 category in AD and ND samples after the confidence cut-off. These isoforms were used for further analyses.
Fig. 3. Isoforms with intra-exonic junctions.
a Novel APP transcripts containing intra-exonic junctions (IEJs, boxes). b Novel TARDBP transcripts containing intra-exonic junctions (IEJs, boxes).
Fig. 4. Discovery of novel transcripts of AD-related genes via Kinnex sequencing with cross-dataset comparisons.
a Confident FSM BIN1 transcripts (navy blue) and novel transcript PB.94751 (yellow). Arrows indicate the locations of the forward and reverse primers used for amplification. Dashed arrow indicates a primer that spans the exon-exon junction. b RT-PCR gel validation of the novel isoform, PB.94751. The band at 107 bp confirms that PB.94751 was detected in all 16 samples. Nuclease-free water (NFW) serves as the negative control. c BIN1 isoform expression in individual samples from Kinnex dataset. AD samples with CPM ⩾ 1 are represented as red circles; ND samples with CPM ⩾ 1 are represented as blue circles; and AD and ND samples with CPM < 1 are represented as gray diamonds. d Heatmap showing normalized gene expression of BIN1 gene across different cell types from short-read sequencing data. Expression was not significantly changed between AD and ND in any cell type (Wilcoxon rank-sum test with Bonferroni correction). e Confident FSM APOE transcript (navy blue) and novel transcript PB.82051 (yellow). Arrows indicate the locations of the forward and reverse primers used for amplification. Dashed arrow indicates a primer that spans the exon-exon junction. f RT-PCR gel validation of the novel isoform, showing a band at 133 bp confirming the presence of the isoform in all 16 samples. Nuclease-free water (NFW) serves as the negative control. g APOE isoform expression in individual samples from Kinnex dataset. AD samples with CPM ⩾ 1 are represented as red circles; ND samples with CPM ⩾ 1 are represented as blue circles; and AD and ND samples with CPM < 1 are represented as gray diamonds. h Normalized gene expression of APOE gene across different cell types from short-read sequencing data. Asterisks (*) indicate DEGs (Wilcoxon rank-sum test with Bonferonni correction). UpSet plots showing FSM (i), NIC (j), and NNC (k) isoforms from our confident set detected in other datasets.
Fig. 5. Disease-associated novel isoforms expression and transcript switching in AD.
a Volcano plot of differentially expressed transcripts (DETs). Red dots represent up-regulated DETs in AD (L2FC > 1 and adjusted _p_-value < 0.05); blue dots represent down-regulated DETs in AD (L2FC < −1 and adjusted _p_-value < 0.05); and black dots indicate non-significant transcripts (adjusted _p_-value > 0.05). Transcripts marked with (‡) sign represent the same isoform that is both differentially expressed and switched. Isoform names are colored by SQANTI3 category: FSM (black), NIC (green), NNC (yellow). b Proportion of the reads for each transcript of SEPTIN4 gene in AD and ND samples. ENST00000317256.10 (light blue) is differentially expressed and significantly switched between ND and AD brains. Numbers above each bar indicate total read count. c Proportion of reads for each transcript of GTPBP6 gene in AD and ND samples. ENST00000711232.1 transcript (light blue) is differentially expressed and significantly switched between ND and AD brains. Numbers above each bar indicate total read count. d Proportion of the reads for each transcript of LUC7L2 gene in AD and ND samples. ENST00000482860 (teal) differentially expressed and significantly switched between ND and AD brains. Numbers above each bar indicate total read count. e Proportion of the reads for each transcript of STX7 gene in AD and ND samples. ENST00000475879.1 (tan) is switched between ND and AD brains. Numbers above each bar indicate total read count. Normalized gene expression across cell types from short-read sequencing data for SEPTIN4 (f), GTPBP6 (g), LUC7L2 (h), and STX7 (i).
Fig. 6. DET isoforms expression.
a Confident FSM isoform (navy blue) and NIC isoform (green) of SEPTIN4. Arrows indicate forward and reverse primers. Dashed arrows indicate primers spanning an exon-exon junction. Gel electrophoresis validation of RT-PCR for SEPTIN4 PB.73144 NIC isoform (b) and SEPTIN4 ENST00000317256.10 isoform (c). Nuclease-free water (NFW) was used as a negative control. d Confident FSM isoform (navy blue), NIC isoform (green), and NNC isoform (yellow) of CHI3L1. Arrows indicate forward and reverse primers. Dashed arrows indicate primers spanning an exon-exon junction. e CHI3L1 isoform expression in individual samples from Kinnex dataset. AD samples with CPM ⩾ 1 are represented as red circles; ND samples with CPM ⩾ 1 are represented as blue circles; and AD and ND samples with CPM < 1 are represented as gray diamonds. f Heatmap showing normalized gene expression of CHI3L1 gene across different cell types from short-read sequencing data. Asterisk (*) indicates DEGs (Wilcoxon rank-sum test with Bonferroni correction). g Gel electrophoresis validation of RT-PCR for CHI3L1 PB.178972 NIC isoform and CHI3L1 ENST00000255409.8 isoform. Nuclease-free water (NFW) was used as a negative control. h RT-qPCR validation of the FSM and NIC transcript. NIC was significantly upregulated in AD brains (_p_-value: 0.0289; Mann-Whitney U test, (two-tailed)), whereas FSM isoform showed no difference (not significant: ns = 0.0721). Error bars indicate SEM.
Fig. 7. Isoform expression pattern across cell types in AD and ND brains.
a–f Proportions of reads for different isoform categories (FSM, ISM, NIC, NNC, genic, antisense, and fusion) across different cell types in AD and ND. g Number of isoforms detected in each cell type, separated by neuronal (excitatory and inhibitory neurons) and non-neuronal (astrocytes, microglia, oligodendrocytes, and OPC) and shared across the neuronal and non-neuronal. h–l Number of unique isoforms expressed in only one cell type, suggesting transcript specificity across cell types. m Heatmap showing cell-type proportion in which each isoform was detected from neurological genes across neurons and glia.
Fig. 8. Transcripts of AD-related genes.
a Confident FSM CLU transcripts (navy blue), PB.149034 (green) NIC isoform, and PB.149064 (yellow) NNC isoform. b CLU isoform expression in individual samples from Kinnex dataset. AD samples with CPM ⩾ 1 are represented as red circles; ND samples with CPM ⩾ 1 are represented as blue circles; and AD and ND samples with CPM < 1 are represented as gray diamonds. c Heatmap showing normalized gene expression of CLU gene across different cell types from short-read sequencing data. Expression was not significantly changed between AD and ND in any cell type. d Confident FSM APP transcripts (navy blue). e APP isoform expression in individual samples from Kinnex dataset. AD samples with CPM ⩾ 1 are represented as red circles; ND samples with CPM ⩾ 1 are represented as blue circles; and AD and ND samples with CPM < 1 are represented as gray diamonds. f Heatmap showing normalized gene expression of APP gene across different cell types from short-read sequencing data. Expression was not significantly changed between AD and ND in any cell type (Wilcoxon rank-sum test with Bonferroni correction). g Confident FSM MAPT transcripts (navy blue), PB.4087 NIC isoform (green), and PB.4075 NNC isoform (yellow). h Heatmap showing normalized gene expression of MAPT gene across different cell types from short-read sequencing data. Expression was not significantly changed between AD and ND in any cell type (Wilcoxon rank-sum test with Bonferroni correction). i MAPT isoform expression in individual samples from Kinnex dataset. AD samples with CPM ⩾ 1 are represented as red circles; ND samples with CPM ⩾ 1 are represented as blue circles; and AD and ND samples with CPM < 1 are represented as gray diamonds.
Fig. 9. DET Isoform expression pattern across cell-types in AD and ND brains.
a Heatmap showing proportion of cells in which each isoform was detected from DET genes across neurons and glia.
Update of
- RNA isoform diversity, splicing variants, and switching in single cells of the Alzheimer's disease brain.
Shahnaee A, Liu CS, Ngo T, Palmer CR, Ziomek D, Park C, Tan VP, Jimenez NL, Chun J. Shahnaee A, et al. Res Sq [Preprint]. 2025 Sep 29:rs.3.rs-7436496. doi: 10.21203/rs.3.rs-7436496/v1. Res Sq. 2025. PMID: 41255979 Free PMC article. Updated. Preprint.
References
MeSH terms
Substances
LinkOut - more resources
Full Text Sources
Medical