GFI1 Snag Domain Methylation Directs LSD1 Recruitment to Control Transcription and Cell Fate Determination in Hematopoiesis

GFI1 is a zinc-finger transcriptional repressor and master regulator of growth, differentiation and survival. Elevated GFI1 expression is observed in many cancer types, and targeted depletion of GFI1 in experimental models of some cancers leads to tumor regression. GFI1 confers growth and survival advantages to cells in which it is expressed. Its impact can be attributed at least in part to repression of cell cycle checkpoint proteins such as p21 and to inhibition of the p53-induced DNA damage response. Understanding determinants of GFI1-mediated transcriptional repression may allow control over normal cell and tissue homeostasis and enable rationally conceived opportunities for cancer therapy.

GFI1 is comprised of an N-terminal, 20 amino acid SNAG domain, a C-terminal concatemer of six zinc fingers and a linker that separates them. The GFI1 SNAG domain is responsible for recruitment of LSD1 to GFI1-regulated genes while its zinc fingers recognize and bind a response element found within their promoters. LSD1 recruitment by GFI1 is essential for GFI1 function as a transcriptional repressor, yet the molecular features that enable regulation of the GFI1-LSD1 relationship to control cell fate are not known.

We used primitive erythropoiesis in Zebrafish and a chromatinized luciferase reporter system in human cells to interrogate details of the GFI1-LSD1 relationship in hematopoiesis and transcriptional control. We show that GFI1-LSD1 binding via the SNAG domain is required for primitive erythropoiesis and that LSD1 depletion phenocopies Gfi1 loss. In parallel, we show the SNAG domain is the dominant transcriptional repression motif in GFI1, that transcriptional repression attributable to the GFI1 SNAG domain requires LSD1 recruitment and is accompanied by demethylation of histone H3K4 at the reporter locus. SNAG domain derivatives devoid of LSD1 binding fail to repress gene expression. To gain additional mechanistic insights, we carefully examined the SNAG domain primary structure. We show that the GFI1 SNAG domain harbors a primary structural motif, -⁸KSKK¹¹-, similar to that observed in p53 (-³⁷⁰KSKK³⁷³-) where it is known to alter p53 transcriptional activity through site-specific methylation. Lysine (K) 370 monomethylation (K370me) deactivates p53, while dimethylation (K370me2) activates it. Notably, LSD1 can demethylate both K370me and K370me2. This suggests an activation-inactivation cycle for p53 controlled by LSD1 and p53-K370 protein lysine methyltransferases (PKMTs). Like p53, we find GFI1's SNAG domain is methylated, and when dimethylated (me2) on K8, dramatically enhances LSD1 binding. However, unlike K370me1/2 in p53, K8me2 in GFI1 is not a substrate for LSD1, and K8me2-SNAG peptide is a potent competitive inhibitor of LSD1-mediated lysine demethylation in vitro. Mutation of K8 in GFI1 to leucine (GFI1-K8L) abolishes LSD1 binding and GFI1-mediated transcriptional repression. Identical results are observed when analyzing transcriptional repression by the SNAG domain in isolation. In Zebrafish primitive erythropoiesis, GFI1-K8L expression fails to complement the Gfi1 depletion phenotype, while injection of K8me2-SNAG peptide phenocopies GFI1 or LSD1 depletion in this same assay. We then screened p53 -³⁷⁰KSKK³⁷³- PKMTs, SMYD2, SETD7 and G9a for their specificity toward lysine residues in the GFI1 -⁸KSKK¹¹- motif. We find PKMT specificity toward GFI1's -⁸KSKK¹¹- motif analogous to that reported for p53. Our findings indicate that SNAG domain methylation modulates transcriptional repression and cell fate determination functions of GFI1, and suggest methylation-dependent integration of GFI1 and p53 actions in cell growth and survival through a shared structural motif.