CPSF1-Mediated Acute Myeloid Leukemia Fusion 3'UTR Alternation Is a Potential Therapeutic Target

Alternative polyadenylation (APA) can alter the three prime untranslated region (3'UTR) length of mRNAs, crucial for regulating mRNA metabolism and gene expression. Despite the prevalence of APA post-transcriptional regulation in cancers, changes in 3'UTR length by APA and its contribution to leukemia development have not been thoroughly studied. In this study, we demonstrated the significance of APA of leukemic fusion genes in acute myelogenous leukemia (AML) development.

T(8;21) is the most common chromosomal abnormality in AML and encodes the AML1-ETO (AE) fusion gene. The AE 3'UTR has a full length of 5.2kb that contains 4 canonical polyadenylation sites (PAS), such that APA can result in several mRNA isoforms with varying 3'UTR lengths. However, RNA-seq and absolute quantification qPCR revealed that AE mainly uses 2 PAS: 5.2kb (long 3'UTR) and 3.7kb (short 3'UTR). Moreover, the AE short 3'UTR is the major isoform in both t(8;21) AML primary patients and cell lines. Based on these findings, we next hypothesize that changes in PAS usage and thus AE 3'UTR length, can modulate fusion gene expression.

The single cell based dynamic array revealed that AE⁺ non-leukemic differentiated cells coexisted with AML blasts in the diagnostic bone marrow and these non-leukemic AE⁺ cells expressed much lower AE compared to AML blasts (monocytes, 0.19-fold and granulocytes, 0.11-fold, p<0.001). Among these leukemic and AE⁺ non-leukemic cells, AE expression and long 3'UTR usage showed a significant negative correlation (R²=0.64, p<0.01). Moreover, half-life assay using actinomycin D showed that AE long 3'UTR had much shorter half-life than AE short 3'UTR (0.8 hour and 2.8 hours, p<0.01). A dual reporter assay also presented the predominant repression of AE long 3'UTR (3.7 kb: 0.31 and 5.2 kb: 0.08, p<0.001). These data suggest that AE 3'UTR length affects mRNA stability and is an important indicator of fusion gene expression.

The selective inhibition of AE short 3'UTR PAS by morpholino antisense oligonucleotide caused an increase of AE long 3'UTR usage, decrease of total AE expression, and an increase in apoptosis. These results indicate that the inhibition of AE short 3'UTR PAS increases the long 3'UTR usage and the PAS usage of leukemic fusion genes could be a potential therapeutic target.

To find the key regulator of this 3'UTR change, we comprehensively knocked down all APA machinery members by shRNA in t(8;21) AML cell lines. We discovered that CPSF1 (cleavage and polyadenylation specific factor 1) knockdown increased long 3'UTR usage (37% in total AE) and down-regulated overall AE mRNA (0.47-fold). Importantly, CPSF1 was highly expressed in t(8;21) AML patients' leukemia cells compared to healthy CD34⁺ cells (2.7-fold, p=0.01). CPSF1 knockdown also promoted apoptosis and Gene Set Enrichment Analysis revealed that CPSF1 knockdown showed the similar gene signature to AE knockdown. These data suggest that CPSF1 promotes AE 3'UTR shortening, resulting in AE stability and a growth advantage of AML cells. Importantly, we also found that AE regulates CPSF1 expression, suggesting a positive feedback loop of AML1-ETO and CPSF1 in t(8;21) AML.

Finally, we also checked APA of leukemic fusion gene MLL-AF9, which has 2 canonical PAS (0.1kb short 3'UTR and 0.7kb long 3'UTR). Compared to healthy CD34⁺ cells, THP-1 cells (MLL-AF9⁺ AML cell line) predominantly used MLL-AF9 short 3'UTR. Knockdown of CPSF1, which is also up-regulated in MLL-associated AML, showed increase of MLL-AF9 long 3'UTR usage, decrease of MLL-AF9 expression, and an increase of apoptosis in THP-1 cells.

Collectively, we conclude that the polyadenylation regulator CPSF1 controls 3'UTR length of leukemic fusion transcripts, and extending the fusion gene 3'UTR by knocking down CPSF1 reduces fusion gene mRNA stability and restricts cell growth in both AE⁺ and MLL-AF9⁺ AMLs. These results suggest that 3'UTR shortening of oncogenic fusion gene transcripts contributes to leukemogenesis and controlling CPSF1 and 3'UTR usage may be a useful approach to inhibit leukemia progression.