Targeting FIP1L1 reverses APA trends in t(8;21) AML. (A) Extent of 3'UTR shortening, in all patient samples and primary HSPC controls, of the 235 genes with shortened 3'UTRs in both t(8;21) and non-t(8;21) patients. For each gene, the average and standard deviation (SD) of dPAS usage were determined. Z scores were then calculated and assigned to each sample in the cohort for the given gene. Each bar of the graph represents the average z score of all 235 genes in the indicated patient or healthy control, a measure of the overall degree of shortening. Data are mean ± SD. (B) Negative correlation between FIP1L1 expression, calculated by RNA-sequencing, and the extent of 3'UTR shortening per patient. Among all APA regulators, FIP1L1 expression is most correlated to 3'UTR shortening. Correlation was determined by linear regression. (C) Reverse transcription quantitative polymerase chain reaction (RT-qPCR) analysis of BAALC 3'UTR length in Kasumi-1 cells transduced with shRNAs targeting FIP1L1 [shRNA (1), (2), or (3)] or a control shRNA. Usage of either the middle or distal PAS was measured relative to total BAALC mRNA using primer pairs upstream of the middle poly(A) site (mPAS) and most proximal poly(A) site (pPAS), respectively. Relative 3'UTR length in FIP1L1 knockdown Kasumi-1 cells was normalized to 3'UTR length in cells transduced with the control shRNA. Data are mean ± SD of 3 independent experiments. *P < .05, ***P < .001, 1-way analysis of variance (ANOVA) with a post hoc Tukey test. (D) RT-qPCR analysis of total BAALC mRNA normalized to β-Actin mRNA in Kasumi-1 cells upon shRNA knockdown of FIP1L1. mRNA levels in knockdown cells were normalized to cells transduced with the control shRNA. Data are mean ± SD of 3 independent experiments. ***P < .001, one-way ANOVA with a post hoc Tukey test. (E) Western blot showing FIP1L1, BAALC, and actin (loading control) protein in Kasumi-1 cells after transduction with control shRNAs or shRNAs targeting FIP1L1. BAALC protein was quantified by normalizing BAALC signal intensity to actin signal intensity using Li-Cor Image Studio software. Normalized protein quantifications are shown in the bar graph to the right. n = 5 for shControl and shFIP1L1 (2); n = 3 for shFIP1L1 (1) and shFIP1L1 (3). Data are mean ± SD. ***P < .001, 1-way ANOVA with a post hoc Tukey test. (F) Scatter plot showing the change in expression of pPAS and dPAS isoforms, per gene, in Kasumi-1 cells after FIP1L1 knockdown. Significant differences in PAS usage were calculated by using the Fisher’s exact test, comparing PAS usage in Kasumi-1 cells transduced with both shRNAs targeting FIP1L1 vs the control shRNA. APA events are classified and divided according to type: 3'UTR-APA (top) and CDS-APA (bottom). Each dot corresponds to a single gene. Blue dots indicate significantly more pPAS usage; red dots indicate significantly more dPAS usage. (G) Venn diagrams showing: (left) the overlap of genes with 3'UTR shortening in t(8;21) AML blasts that were significantly lengthened upon FIP1L1 knockdown in Kasumi-1 cells and (right) the overlap of genes with CDS lengthening in t(8;21) AML blasts that were significantly shortened upon FIP1L1 knockdown in Kasumi-1 cells. (H) Genome browser tracks depicting normalized sequencing reads (reads per million [RPM]) in the BAALC gene obtained from 3'READS of Kasumi-1 cells transduced with a control shRNA or shRNAs targeting FIP1L1. The full BAALC genomic structure is shown (top) with the purple, boxed region expanded (below). The percent usage of each indicated PAS was calculated by using 3'READS and is shown in the bar graph to the right. n = 3, shControl and shFIP1L1 (2); n = 4 shFIP1L1 (1).