The Transcription Factor PU.1 Controls Alternative Splicing of Cflip By Binding to SRSF2/10 Splice Factors in AML Therapy Responses

cFLIP (CFLAR) is an inactive homolog of caspase 8 that functions downstream of death-receptor mediated apoptosis. CFLAR gene is alternatively spliced and exists in three isoforms: a long isoform (cFLIP_L), which partially inhibits CASP8 activity, and two short forms (cFLIP_R and cFLIP_S) that entirely block caspase 8 activity. Moreover, cFLIP_L is crucial for macrophage differentiation. Therefore, the ratio between cFLIP_L and cFLIP_S/R may regulate myeloid differentiation. The transcription factor PU.1 (SPI1) is necessary for myeloid differentiation as well as myeloid cell survival. PU.1 is expressed at low levels in hematopoietic progenitors and increasing levels of PU.1 promote macrophage and granulocyte differentiation. Additionally, aberrant low PU.1 expression contributes to an immature myeloid phenotype, e.g. seen in acute myeloid leukemia (AML). Interestingly, two studies indicated that high PU.1 protein levels were associated with alternative splicing in a murine erythroleukemia model by either direct binding to splice factors or RNA. Based on these observations we hypothesized that PU.1 controls alternative splicing of cFLIP during myeloid differentiation.

In line with above hypothesis, differentiating primary CD34⁺ hematopoietic progenitors into granulocytes and monocytes or AML cells into granulocytes shifted cFLIP splicing in favor of cFLIP_L paralleled by increased PU.1 mRNA and protein levels. Next, we assessed the direct impact of PU.1 protein on splicing during granulocytic differentiation by knocking down PU.1 in NB4 acute promyelocytic leukemia (APL) and MOLM13 AML cells. Silencing PU.1 in NB4 and MOLM13 cells decreased cell death and differentiation during all-trans retinoic acid (ATRA)-mediated granulocytic differentiation. Moreover, knocking down PU.1 reduced cFLIP_L/S protein and mRNA ratios during ATRA treatment. Interestingly, increasing the cFLIP_L/S ratio by knocking down specifically cFLIP_S or overexpressing cFLIP_L reduced MOLM13 cell viability during ATRA treatment.

Since the canonical function of CFLAR isoforms is the inhibition of extrinsic apoptosis, we next tested if increased cFLIP_S levels found in AML cells depleted for PU.1 would show reduced sensitivity to TRAIL treatment. Indeed, TRAIL-induced extrinsic apoptosis via TRAIL receptor was significantly reduced in PU.1 depleted MOLM-13 AML cells.

To investigate if PU.1 is directly involved in CFLAR splicing, we analyzed the CFLAR pre-mRNA sequence encompassing the alternative splicing site for PU.1 and splice factor binding motifs. We identified one putative PU.1 and SRSF2 as well as SRSF10 splice factor binding sites in the CFLAR pre-mRNA sequence. Using RNA immunoprecipitation, PU.1 pull-down and proximity ligation assays, we showed that all three proteins bind the CFLAR mRNA and physically interact with each other.

Lastly, we asked if PU.1 additionally influences cFLIP alternative splicing via regulating splice factor transcription. Knocking down PU.1 in MOLM13 cells resulted in significantly reduced SRSF2 and SRSF10 mRNA and protein levels. We confirmed PU.1 binding to their promoter using ChIP assays. Accordingly, these splice factors showed altered mRNA expression when comparing mature granulocytes to primary AML cells. Since silencing SRSF2 and SRSF10 switched the CFLAR_L/S isoform ratio towards the long isoform, we suggest that PU.1-mediated induction of these splice factores represents a negative feedback loop.

In conclusion, we provide evidence that PU.1 directly regulates alternative splicing of CFLAR during ATRA-mediated neutrophil differentiation and TRAIL-induced cell death in human AML cells. We propose a model where PU.1 binds to CFLAR pre-mRNA and blocks SRSF2 and SRSF10-mediated splicing ultimatively leading to exon 7 skipping and predominant expression of cFLIP_L.

No relevant conflicts of interest to declare.