Fig 2: AKAP8 eCLIP identifies its RNA-binding targets and cis-element motifs.a Distribution of AKAP8 eCLIP-binding sites across different regions of gene bodies. b Bar plot comparing the percentage of AKAP8-binding sites in each gene region between epithelial (EPI) and mesenchymal (MES) cell states. P values were computed by Fisher’s Exact test. c Metagenes plotting AKAP8-binding intensity (average log2FC IP/Input) in epithelial state (top panels) or mesenchymal state (bottom panels) across all introns (left panels) or exons (right panels). Introns (black line flanked by two black rectangles) were scaled to 400 bins. The flanking 50 nucleotides were split into 25 bins. Exons (black rectangle flanked by two black lines) were scaled to 500 bins. The flanking 1000 nucleotides were split into 100 bins. d Boxplots with jitters comparing the binding intensity (log2FC IP/Input) of all AKAP8-binding sites located in proximal introns showing that AKAP8 binds more strongly to proximal introns in the epithelial state compared with the mesenchymal state (p computed by Student’s t test, two-tailed). The line within each box represents the median. Upper and lower edges of each box represent 75th and 25th percentile, respectively. The whiskers extend across the entire range of the data. e Weblogos depicting the most significant AKAP8 eCLIP-binding motif in epithelial and mesenchymal cells identified through de novo motif analysis. The two motifs are similar with guanine stretches of at least three nucleotides flanked by uridine or adenine nucleotides. P values were determined by hypergeometric test. f Integrated genome viewer tracks centered on a cassette exon (exon 11) of CLSTN1, which undergoes exon inclusion upon AKAP8 (AK8) knockdown in the epithelial cells. Top four tracks represent autoscaled RPM-normalized RNA seq reads. Bottom two tracks represent IP/input normalized eCLIP signal. Black bricks indicate hnRNPM (hnM) binding motifs (GU-rich). Bottom cartoon shows CLSTN1 cassette exon in light blue flanked by upstream (dark blue) and downstream (pink) constitutive exons. Yellow bar indicates AKAP8 eCLIP-binding site. Zoomed out sequence indicates nucleotide sequence covered by binding site with AKAP8 motifs underlined. Source data are provided as a Source Data file.

Fig 3: AKAP8 antagonizes hnRNPM’s splicing activity.a qRT-PCR (top) and semi-qPCR (bottom) analysis of CD44v8-splicing minigene assay showing that hnRNPM-mediated (hnM) exon skipping is antagonized by AKAP8 (AK8) in a dose-dependent manner. Incl: Inclusion. PSI: Percent Spliced In. b–c qRT-PCR (top) and semi-qPCR (bottom) analysis of CD44v5-splicing minigene assay showing that AKAP8 promotes v5 inclusion b and that AKAP8 inhibits hnRNPM-mediated exon skipping activity c. d, e Semi-qPCR of CD44v8 and CD44v5 minigene reporter assays showing increased exon skipping activity of hnRNPM in AKAP8 knockdown cells. f, g RNA pull-down assays showing that hnRNPM binds to its RNA cis-elements I-8 f and I-5 g more strongly in AKAP8-silenced epithelial HMLE cells. Twist-induced mesenchymal cells did not show binding differences. The sequences of I-8 and I-5 are shown at the bottom of each panel. Error bars indicate s.d. N = 3. (**) P < 0.01, (***) P < 0.001 between indicated groups. P values were tested by Student’s t test, two-tailed, in a, b, c. Source data are provided as a Source Data file.