Fig. 4. Focus on TSPAN14 locus.
a, Splicing QTL (sQTL)-GWAS integration results. Known TSPAN14 transcripts (GENCODE v38; green, coding sequences; gray, noncoding) plotted with −log10(P) for (1) EADB GWAS stage I (n = 487,511) signal (black), (2) sQTL signal for chr10:80509471–80510106 junction (supporting cryptic exon 1) in the EADB Belgian LCL sQTL catalog (n = 70 individuals, blue) and (3) sQTL signal for chr10:80512269–80512719 junction in the MayoRNAseq TCX sQTL catalog (n = 259 individuals, red); hg38 genomic position is shown above. LCL and brain-based sQTL coloc and sTWAS analyses associate ADD risk with these junctions that suggest cryptic splicing within ADAM10-interacting domain of TSPAN14 (magenta), which was predicted to result in two cryptic exons. b, Long-read sequencing validation of TSPAN14 cryptic exons. Nanopore sequencing results (Supplementary Note) in the zoomed-in region of chr10:80506973–80516400 (cumulative coverage in log10 scale). Pooled LCL cDNA sample sequenced for cDNA Amplicon2 shown in blue. cDNA Amplicon1 was sequenced on biologically independent hippocampal (HPC; n = 16, red), frontal cortex (FC; n = 18, pink) and LCL (n = 59, orange) cDNA samples. Green, canonical exons (8–12); dotted black lines, canonical splicing; blue, cryptic exon 1 (>45 bp); red, cryptic exon 2 (118 bp). All annotated junctions use canonical splice donor (GT) and acceptor (AG) sites. c,d, sQTL-GWAS colocalization plots for chr10:80509471–80510106 (supporting cryptic exon 1) in the EADB Belgian LCL sQTL catalog (n = 70 individuals) (c) and chr10:80512269–80512719 (supporting cryptic exon 2) in the MayoRNAseq TCX sQTL catalog (n = 259 individuals) (d). sQTL signals for the two junctions colocalize with ADD signal (PP4s of 98.8% and 97.4%, respectively), and sTWAS associates with increased preference for the cryptic splicing with decreased ADD risk (sTWAS P = 6.28 × 10−12 and 1.6 × 10−13, sTWAS Z = −6.9 and −7.4, respectively). y axis, sQTL −log10(P); x axis, EADB GWAS stage I −log10(P). LD r2 values calculated within EADB-TOPMed dataset (n = 42,140) based on the lead variant rs6586028 (purple) are indicated on a color scale.