GFI1B Splice Variants Differentially Regulate Transcription and Megakaryocyte/Erythrocyte Fate in Hematopoiesis

Alternative splicing is a powerful way to expand functional genetic diversity without increasing the number of genes. Recent deep sequencing efforts reveal that 100% of human genes are alternatively spliced. Alternative splicing is enriched in genes that control cell cycle dynamics, transcription and organelle organization during erythrocyte differentiation.

Growth factor independence (GFI) 1B is comprised of an N-terminal SNAG domain, a C-terminal concatemer of six C2H2 zinc fingers (ZnFs), and a linker that joins them. GFI1B is required for both definitive erythropoiesis and megakaryopoiesis. Elevated GFI1B expression is seen in AML, while GFI1B mutations cause inherited platelet disorders. Understanding GFI1B function is critical to addressing its contributions to diseases of hematopoiesis. GFI1B has two splice variants, GFI1B-full and GFI1B-short . GFI1B-short lacks ZnFs 1 and 2, but is otherwise structurally identical to GFI1B-full. Both splice variants are expressed during erythroid differentiation from megakaryocyte-erythrocyte progenitors (MEPs), but their distinctive functions and relative contibutions to this binary fate decision remain largely unexplored.

We used the bipotent, GFI1B-dependent erythroleukemia cell line, K562, to explore fate specifying and mechanistic differences between GFI1B-full and GFI1B-short. Inducible expression of GFI1B-full in K562 cells promotes spontaneous erythroid differentiation and antagonizes 12-O-Tetradecanoylphorbol-13-acetate (TPA)-induced megakaryocyte differentiation. In contrast, GFI1B-short fails to support spontaneous erythroid differentiation and instead enables megakaryocyte fate in response to TPA. These findings suggest ZnFs 1 and 2 control fate-specifying differences between GFI1B splice variants. We have previously shown the GFI1B paralog, GFI1, is SUMOylated and that a portion of the linker region and ZnFs 1-3 provide binding surfaces for the E3 SUMO ligase, Protein Inhibitor of Activated STAT (PIAS)3. To gain additional insights regarding binding relationships between GFI1B and PIAS family proteins, we mapped the binding interface between GFI1B-full and PIAS3 and investigated binding differences between GFI1B-full vs. GFI1B-short for PIAS family members. We find ZnFs 1-3 of GFI1B are necessary and sufficient for PIAS3 binding and predicted impaired PIAS protein binding by GFI1B-short. While modest impairment was observed between GFI1B-short and PIAS proteins, PIASXa binding was completely abolished. We also find that GFI1B-full is SUMOylated on lysine (K)-61 within a yKXD/E SUMOylation consensus element. Yet, GFI1B-short is not SUMOylated, despite this consensus element being present. This suggests that differential recruitment of the SUMOylation machinery by GFI1 splice variants and assembly of distinct, SUMO-dependent transcriptional regulatory complexes may direct binary cell fate decisions. Using proximity-based in situ biotinylation and mass spectral analysis, we recorded the composition of transcriptional regulatory complexes assembled by GFI1B-full and GFI1B-short. We identifed candidate factors whose recruitment by GFI1B may be SUMOylation-dependent, suggesting that GFI1B splicing variation and differential SUMOylation may direct lineage allocation in hematopoiesis.