Proper hematopoietic cell fate decisions require co-ordinated functions of transcription factors, their associated co-regulators, and histone-modifying enzymes. Growth factor independence 1 (GFI1) is a zinc finger transcriptional repressor and master regulator of normal and malignant hematopoiesis. While several GFI1-interacting proteins have been described, how GFI1 leverages these relationships to carry out transcriptional repression remains unclear. Here, we describe a functional axis involving GFI1, SMYD2, and LSD1 that is a critical contributor to GFI1-mediated transcriptional repression. SMYD2 methylates lysine-8 (K8) within a -8KSKK11- motif embedded in the GFI1 SNAG domain. Methylation-defective GFI1 SNAG domain lacks repressor function due to failure of LSD1 recruitment and persistence of promoter H3K4 di-methyl marks. Methylation-defective GFI1 also fails to complement GFI1 depletion phenotypes in developing zebrafish and lacks pro-growth and survival functions in lymphoid leukemia cells. Our data show a discrete methylation event in the GFI1 SNAG domain that facilitates recruitment of LSD1 to enable transcriptional repression and co-ordinate control of hematopoietic cell fate in both normal and malignant settings.
Growth factor independence 1 (GFI1) is a transcriptional repressor and master regulator of normal and malignant hematopoiesis. GFI1 is comprised of a 20 amino acid N-terminal Snail/GFI1 (SNAG) domain, a C-terminal concatemer of six C2H2-type zinc fingers, and a linker region that separates them . GFI1-mediated transcriptional repression requires DNA binding via zinc fingers 3, 4, and 5 . The SNAG domain is the dominant contributor to transcriptional repression by GFI1 . GFI1 and its paralog, GFI1B, share 95% identity within their SNAG domains and have a high degree of conservation within their zinc fingers . This conservation is reflected by their recognition of a common response element [TAAATCAC(A/T)GCA; response element in bold face] and shared-binding partners, yet knockin experiments in mice show GFI1 and GFI1B only partially substitute for one another in hematopoietic development .
GFI1 contributes to hematopoietic stem cell (HSC) quiescence and self-renewal [5,6]. GFI1 is also an obligate transcription factor in granulopoiesis and is equally critical to establish and maintain lymphoid fate in development of the adaptive immune system [3,7]. Mutations (N382S, K403R) within GFI1 zinc fingers act in a dominant negative fashion to cause severe congenital neutropenia [8,9]. In the malignant setting, GFI1 is required to establish and maintain lymphoid leukemia/lymphoma, at least in part by restricting pro-apoptotic functions of p53 [10–12]. These reports highlight the critical roles filled by GFI1 in both normal and malignant hematopoiesis.
GFI1 functions through discrete interactions with partner DNA-binding proteins, transcriptional co-regulators and histone-modifying enzymes, including ETS1, MTG8, MTG16, G9a, HDAC1, and LSD1 [13–16]. Factors influencing GFI1 interactions with these partners to govern transcriptional repression are incompletely understood. By virtue of its demethylase activity toward mono- and/or di-methylated histone H3, lysine 4 (H3K4me1/me2), LSD1 is the dominant effector for GFI1-mediated transcriptional repression and is the only protein known to bind the SNAG domain in either GFI1 or GFI1B . A single amino acid substitution from proline to alanine at residue 2 (P2A) within the SNAG domain profoundly impairs LSD1 binding and GFI1 repressor function . Likewise, disruption of LSD1 binding to the GFI1B SNAG domain impairs GFI1B-mediated gene repression and disrupts proper erythroid and megakaryocytic differentiation . Notably, substitutions at nearby residues (R3A, S4A, K10A, and K11A) have minimal impact on GFI1 repressor function . These data suggest a residue-specific impact upon LSD1 binding and GFI1 function via the SNAG domain, yet leaves as an open question how this impact is established and whether GFI1–LSD1 binding and repressor functions might be regulated by modifications at these residues.
The SNAG domain in GFI1 contains a family- and species-conserved lysine-serine-lysine-lysine (KSKK) motif occurring at residues 8 through 11 . Over 300 UniProt genes encode a KSKK motif, including one occurring from residues 370 to 373 in p53. Residue-specific lysine methylation events in the p53 -370KSKK373- motif, catalyzed by protein lysine methyltransferases SMYD2, SETD7, G9a, and Glp, dynamically regulate p53 function [17–20]. We hypothesized the GFI1 -8KSKK11- motif may be similarly regulated by methylation, and given functional connections between GFI1 and p53, that the same methyltransferases may govern GFI1 transcriptional repression function attributable to SNAG–LSD1 binding. We show that SMYD2 binds and methylates K8 within the GFI1 SNAG domain to promote its interaction with LSD1. Leucine substitution at K8 (K8L) abolishes LSD1 binding required for transcriptional repression by the SNAG domain in isolation. Likewise, GFI1-K8L fails to complement the defect in zebrafish primitive erythropoiesis brought on by gfi1aa depletion and is unable to support T-ALL cell survival after depletion of endogenously expressed GFI1. These data indicate that SNAG domain methylation governs GFI1–LSD1 axis functions and when considered with other GFI1 posttranslational modifications, may be one of several regulatory inputs to this critical determinant of hematopoietic cell fate decisions.
Materials and methods
Reagents and antibodies
Mouse monoclonal α-FLAG (M2) and α-tubulin (B-5-1-2) antibodies were obtained from Sigma-Aldrich. Rabbit polyclonal α-myc (A-14) and goat polyclonal α-GFI1 (sc-8558, N-20) antibodies were obtained from Santa Cruz Biotechnology. Rabbit polyclonal α-H3 (ab1791) and rabbit monoclonal α-H3K4me2 (ab32356) antibodies were obtained from Abcam. Rabbit monoclonal α-LSD1 (C69G12, #2184) antibody was obtained from Cell Signaling Technology. Horseradish peroxidase (HRP)-conjugated α-mouse, α-rabbit, and α-goat IgG were obtained from Jackson Immunoresearch. LSD1 inhibitor, HCI-2509, has been described previously . Restriction endonucleases, polymerases, and ligases were purchased from New England Biosciences. All other materials were of reagent grade.
Plasmids and subcloning
GFI1:3X-FLAG constructs were a generous gift from Dr Tarik Moroy . P2A and K8L substitutions were generated by two-stage PCR and splicing by overlap extension. GFI1 SNAG domain derivative fusion proteins with the Gal4 DNA-binding domain were generated by first annealing complementary single-stranded DNA oligonucleotides encoding SNAG domain derivative sense and antisense strands. The sense strand incorporated an EcoRI site at its 5′ end, whereas the antisense strand contained a MluI site at its 5′ end. Annealed oligonucleotides encoding SNAG domain derivatives were restricted with EcoRI/MluI and subcloned into EcoRI/MluI restricted pCMV5. The Gal4 DNA-binding domain sequence (encoding amino acid residues 1–147) was inserted in-frame and 3′ of the respective SNAG domain derivative using XbaI/BamHI restriction sites. LSD1 constructs have been recently described . Complete open-reading frame sequences for all constructs were confirmed by automated di-deoxy sequencing in the University of Utah DNA Sequencing HSC Core Facility. Primer sequences used to generate constructs are provided in Supplementary Material.
Cell lines and culture
Cos7L cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% FBS, GlutaMAX, and pen-strep. CCRF-CEM cells were maintained in RPMI 1640 media supplemented with 10% FBS, GlutaMAX, and pen-strep. Gal4-UAS-TK-Luc HEK 293T-Rex cells (referred to as HEK293 reporter cells throughout the text) were generously provided by Raphael Margueron of the Institut Curie- Unité de Génétique et de Biologie du Développement, Paris, France and maintained in DMEM supplemented with 10% FBS, GlutaMAX, and pen-strep.
Luciferase assays were performed using a dual-luciferase assay kit according to the manufacturer's protocols (Promega). Briefly, 3.25 × 105 Gal4-UAS-TK-Luc HEK 293T-Rex cells were seeded per well in six-well plates. Transfection of plasmids expressing SNAG:Gal4 derivatives and additional plasmids shown in figures was performed in triplicate using Lipofectamine 2000 according to the manufacturer's protocols. Reporter activity was measured in cell lysates collected between 24 and 48 h post-transfection. Firefly luciferase activity was normalized to a constitutive, co-transfected Renilla luciferase control. Statistical significance was determined by two-sided unpaired t-tests (P-values, * <0.05, ** <0.005, *** <0.0005, **** <0.00005).
Chromatin immune precipitation and quantitative PCR
Briefly, transfection of Gal4-UAS-TK-Luc HEK293 T-Rex cells was performed using Lipofectamine 2000 according to the manufacturer's protocols. Forty-eight hours post-transfection, cells were cross-linked in 1% formaldehyde for 15 min at room temperature. Cross-linking was terminated with 125 mM glycine for 5 min at room temperature. Cells were washed with ice cold PBS and lysed in ice cold Farnham lysis buffer [24,25]. Cell lysates were sonicated at 5 W for 5 second-on/25 second-off intervals for a total of 1 min and 30 s. Antibody-coated Dynabeads (Thermo Fisher Scientific) were added to clarified lysates and incubated at 4°C overnight. Beads were washed three times with LiCl buffer (100 mM Tris, 500 mM LiCl, 1% NP-40, 1% sodium deoxycholate, and pH 7.5) followed by a single wash in TE buffer (10 mM Tris, 1 mM EDTA and pH 8.0). Cross-links were reversed in IP elution buffer (100 mM NaHCO3 and 1% SDS) at 65°C overnight. Immunoprecipitated DNA was recovered using ChIP DNA Clean and Concentrator (Zymo Research) and quantified using the PicoGreen system (Thermo Fisher Scientific). PCR was performed using the Luc-TSS-FWD and Luc-TSS-REV primers (see Supplementary Material). Fold enrichment was determined by the ΔΔCt method and normalized to input DNA. All qPCRs were performed in triplicate using SsoFast EvaGreen Supermix reagent (Bio-Rad).
Zebrafish morpholino and mRNA injections
Zebrafish were maintained in the aquatics shared resource at the Huntsman Cancer Institute. Morpholinos were purchased from Gene Tools, LLC. mRNAs were generated using mMessage mMachine T7 transcription reactions (Ambion, Thermo Fischer). Embryos were microinjected at the one-cell stage with 1 nl of 20 µM splice-blocking gfi1aa and 200 µM kdm1a morpholino with 10% phenol red and sterile water. mRNAs were coinjected with morpholino at a final concentration of 100 ng/µl. Unspliced gfi1aa mRNA was detected in gfi1aa morpholino-injected embryos by PCR amplification of the morpholino-targeting intron 1/exon 2 boundary within cDNA preparations as described recently . Morpholino and primer sequences are provided in Supplementary Materials. Dechorionated, 48 h post-fertilization (hpf) embryos were stained in a solution of 40% ethanol, 0.65% H2O2, 0.01 M sodium acetate, and 0.6 mg/ml o-dianisidine in the dark on a rolling rotisserie for 2 h. Embryos were thoroughly washed with PBS-T and visualized on a dissecting microscope. Primitive erythropoiesis was quantified using a scoring system from 1 to 4, indicating minimal (1), modest (2), moderate (3), and complete (4) hemoglobinization as revealed by o-dianisidine staining. More than 60 embryos were scored for each experimental condition. Statistical significance was determined by Wilcoxon–Mann–Whitney testing in GraphPad Prism 6 (P-values, * <0.05, ** <0.005, *** <0.0005, **** <0.00005).
Transfection, immunoprecipitation and drug treatments
Cos7L cells were transfected using Lipofectamine 2000 (Life Technologies) as recently described . Cell lysates were harvested in phospho-protecting lysis buffer 48 h post-transfection and immunoprecipitation was performed from clarified cell lysates using α-FLAG (M2) antibody and Protein G Sepharose beads . Immune complexes and clarified lysates from transfections were probed using antibodies indicated in the figures and figure legends. SMYD2 inhibitor, LLY-507, was prepared in DMSO and dosed as shown in figure 2 . The LSD1 inhibitor, HCI-2509, was prepared in DMSO and dosed as shown in figure 4 .
In vitro peptide binding
GFI1 SNAG domain peptides were synthesized by GenScript to >95% purity and conjugated to biotin. SNAG domain peptides were incubated at 4°C with purified human recombinant LSD1 or extracts prepared from CCRF-CEM cells under non-denaturing conditions. Biotinylated peptides and bound LSD1 were isolated using streptavidin agarose beads (Thermo Fisher Scientific), washed extensively in incubation buffer (50 mM HEPES, pH 7.5), and subjected to α-LSD1 western blotting as described above.
GFI1 SNAG domain peptides were incubated with human recombinant SMYD2 (Cayman Chemical) and 3H-S-adenosylmethionine (SAM) at 30°C. Biotinylated peptide substrates were purified from reaction mixtures using streptavidin agarose beads (Thermo Fisher Scientific). Beads were washed extensively with reaction buffer, resuspended in scintillation cocktail, and 3H incorporation measured using a Beckman Coulter LS 6500 liquid scintillation counter. Statistical significance was determined by two-sided unpaired t-tests (P-values, * <0.05, ** <0.005, *** <0.0005, **** <0.00005).
shRNA and flow cytometry
GFI1 depletion was achieved in cell lines via an shRNA targeting the 3′-UTR of GFI1 subcloned into the pMKO1 vector (a generous gift of Dr Stephen Lessnick), relative to content-matched scramble control. Selection was performed in growth medium containing 3 µg/ml puromycin for 3 days. Apoptosis in stable cells was determined by Annexin V/propidium iodide (PI) staining on a FACSCanto Analyzer (BD).
Transcriptome RNA sequencing was performed as follows. Total RNA from four biological replicates for each experimental condition was isolated using the Qiagen RNeasy Mini Kit with on-column DNAse treatment. RNA quality was confirmed by RNA integrity number scoring on the Agilent RNA ScreenTape Assay. The Illumina TruSeq Stranded Total RNA Sample Prep Kit with Ribo-Zero Gold was used for sequencing library preparation. An Illumina HiSeq 50 Cycle Single-Read Sequencing version 4 sequencing protocol was performed and reads were aligned to the hg38 human genome assembly. Differentially expressed genes were identified at a >2-fold change threshold and an adjusted P-value of 0.01 relative to vector control CCRF-CEM cells.
SNAG K8 methylation by SMYD2 promotes LSD1 binding and GFI1-mediated transcriptional repression
Transcriptional repression by GFI1 can be predominantly attributed to LSD1 recruitment by the SNAG domain . To gain additional insights into determinants of LSD1 binding, we examined the primary structure of GFI family proteins to find near complete conservation of a -8KSKK11- motif embedded within the SNAG domains of GFI1 and GFI1B among diverse species (Figure 1A). An analogous motif in p53 is subject to methylation by specific methyltransferases to regulate dynamically p53 function. We hypothesized that methylation on -8KSKK11- lysine residues may similarly affect functions of the GFI1 SNAG domain.
Lysine 8 contributes to transcriptional repression by the GFI1 SNAG domain.
To test the contribution of -8KSKK11- motif residues to GFI1-mediated transcriptional repression, we utilized a cell-based, heterologous transcriptional reporter stably integrated into the host cell genome and regulated by a concatemer of five Gal4–UAS elements juxtaposed to a HSV thymidine kinase (TK) minimal promoter (Figure 1B) . This system enables transcriptional output to be attributed specifically to motifs in GFI1 incorporated into fusion proteins with Gal4, and free from partnerships GFI1 might have with other DNA-binding proteins on an endogenous promoter. Likewise, the system permits transcriptional output to be measured in a chromatinized context and for assessment of promoter occupancy and histone modifications to be made in a unified fashion. When fused to the Gal4 DNA-binding domain, wild-type (WT) SNAG or derivatives with discrete mutations can be examined for impact upon transcription, LSD1 recruitment, and H3K4 methylation status. We generated WT, K8L, K10L, and K11L SNAG:Gal4 fusion proteins and Δ1C GFI1:Gal4, comprised of GFI1 residues 1–340 fused to Gal4 (Figure 1B), then confirmed their expression by western blot (Figure 1C), and tested their impact on transcriptional output from the integrated reporter. The SNAG domain alone, fused to Gal4, was sufficient to repress reporter output comparable to Δ1C-WT:Gal4 (Figure 1D), confirming a dominant role for the SNAG domain in GFI1-mediated transcriptional repression. Leucine substitution at K8 impaired repressor function, quantitatively similar to that observed with P2A substitution (Figure 1D) . Leucine substitution at either K10 or K11 had little impact on repressor function (Figure 1D).
LSD1 is the dominant effector of GFI1-mediated transcriptional repression and directly binds the GFI1 SNAG domain . To test whether the transcriptional repression defect of the K8L derivative reflected impaired LSD1 binding, we expressed FLAG-tagged WT, K8L, P2A, and ΔSNAG derivatives of GFI1 in Cos7L cells and determined their ability to engage endogenously expressed LSD1 in α-FLAG co-precipitation assays. K8L substitution abolished the interaction between GFI1 and LSD1, comparable to SNAG domain deletion or P2A substitution (Figure 1E). These results indicate that the GFI1 SNAG domain alone is sufficient for transcriptional repression, and by supporting LSD1 recruitment, K8 is essential to this function of the SNAG domain.
We then tested whether methyltransferases, including SMYD2, SETD7, and G9a, influence SNAG domain-mediated transcriptional repression. Methyltransferases were coexpressed with SNAG-WT:Gal4 and reporter activity was quantified. SMYD2 coexpression enhanced repression by SNAG-WT:Gal4, whereas SETD7 and G9a co-expression had minimal impact (Figure 2A). Given the functional relationship between SMYD2 and transcriptional repression, we asked whether SMYD2 bound the SNAG domain. FLAG-tagged SMYD2 was transiently expressed alongside Gal4-tagged WT and K8L SNAG domains. Gal4 western blot of FLAG-immune complexes revealed an equivalent interaction between SMYD2 and both WT and K8L SNAG domains (Figure 2B). We then tested SMYD2 methyltransferase activity toward GFI1 SNAG peptide either without or with the K8L substitution (Figure 2C). WT or K8L SNAG peptides were incubated with recombinant SMYD2 and 3H-SAM. SNAG peptides were purified via their biotin tag and methylation was measured by scintillation counting. K8 substitution significantly reduced activity of SMYD2 toward the SNAG domain, suggesting that K8 is a SMYD2 methyl-acceptor residue in GFI1. To test whether K8 is required for enhanced repressor function conferred by SMYD2, we expressed WT or K8L SNAG:Gal4 alongside SMYD2 using the integrated reporter system. SNAG-K8L:Gal4 was insensitive to additional repression by SMYD2 (Figure 2D). Likewise, an inhibitor of SMYD2, LLY-507, caused concentration-dependent reversal of repression only for SNAG-WT but not for its -K8L variant (Figure 2E) . These data suggest that SMYD2 contributes to GFI1-mediated transcriptional repression via methylation of K8 within the SNAG domain.
SMYD2-mediated methylation at K8 of the SNAG domain contributes to repressor function.
SNAG domain K8 methylation recruits LSD1 for H3K4 demethylation and gene repression
SMYD2 binds and methylates the GFI1 SNAG domain at K8 and contributes to GFI1 repressor function. To address directly the role of K8 methylation for SNAG–LSD1 binding, we synthesized biotinylated SNAG peptide di-methylated on K8 (SNAG-K8me2), then compared its interaction with LSD1 to that of unmethylated SNAG peptide. Recombinant human LSD1 (hLSD1) and CCRF-CEM cell lysates containing LSD1 were incubated with biotinylated SNAG or SNAG-K8me2 peptides, then co-purified on streptavidin-conjugated beads. SNAG-K8me2 peptide extracts hLSD1 from solution quantitatively while unmethylated SNAG peptide bound only a small minority of the hLSD1 available, leaving >90% in solution (Figure 3A). Similarly, the SNAG–LSD1 interaction was nearly undetectable using unmethylated SNAG peptide and CCRF-CEM extracts, while LSD1 binding increased dramatically using SNAG-K8me2 peptide (Figure 3B). These findings indicate that K8 methylation strongly favors SNAG–LSD1 binding.
K8 methylation favors SNAG–LSD1 binding.
To clarify LSD1's contribution to GFI1 transcriptional repression in the integrated reporter system, LSD1 was co-expressed alongside SNAG-WT:Gal4 and reporter activity was quantified. Enforced LSD1 expression enhances transcriptional repression by SNAG-WT:Gal4 with no discernible impact on SNAG-K8L:Gal4 (Figure 4A). Expression of a catalytically inactive LSD1 derivative (K661A) reverses the enhanced repression conferred by WT LSD1 expression (Figure 4B), while an LSD1 inhibitor, HCI-2509, causes concentration-dependent reversal of repression by SNAG-WT:Gal4 but not by SNAG-K8L:Gal4 (Figure 4C) . We then used chromatin immunoprecipitation and quantitative PCR (ChIP-qPCR) to investigate how loss of K8 methylation might affect LSD1 recruitment and H3K4me2 status, and to correlate with transcriptional output from the integrated reporter system. ChIP-qPCR revealed that WT and K8L SNAG:Gal4 equivalently occupy the reporter promoter (Figure 4D). However, K8L substitution profoundly reduced LSD1 occupancy compared with SNAG-WT:Gal4 (Figure 4E) and resulted in enrichment of H3K4me2 marks at the promoter locus (Figure 4F). Combined, these data indicate that K8 methylation strongly favors LSD1 recruitment by the SNAG domain to direct H3K4 demethylation and to bring about transcriptional repression.
K8 is required for LSD1 recruitment and H3K4 demethylation at a GFI1 target gene.
K8 methylation contributes to GFI1 function during developmental erythropoiesis
GFI1 contributes to early myelo- and lymphopoiesis in mice and humans . In the developing zebrafish, the GFI1 homolog, gfi1aa, is required for primitive erythropoiesis . We leveraged this functional requirement to test the importance of GFI1 K8 methylation in vivo. Hemoglobinized primitive erythrocytes in zebrafish were visualized by whole-mount o-dianisidine staining at 48 h post-fertilization (Figure 5A). Embryo staining was scored across a numeric continuum representing minimal to complete erythropoiesis . Consistent with previous reports, morpholino-mediated depletion of gfi1aa impairs primitive erythropoiesis and this defect could be complemented by co-injection of morpholino-resistant mRNA encoding WT rat Gfi1 (Figure 5A) [26,30]. However, co-injection of Gfi1 K8L mRNA failed to complement the erythropoiesis defect brought on by gfi1aa depletion, just as was observed using mRNA for the LSD1-binding deficient mutant, Gfi1 P2A. In contrast, co-injection of mRNA encoding either GFI1 K10L or K11L was able to complement the gfi1aa depletion phenotype (Figure 5B). Notably, morpholino-mediated depletion of the zebrafish LSD1 homolog, kdm1a, phenocopies the defect in primitive erythropoiesis shown for gfi1aa depletion. This phenotype was reversed by co-injection of WT human LSD1 mRNA (Figure 5C). These data highlight the critical importance of K8-dependent GFI1–LSD1 binding to a GFI1-driven outcome in developmental hematopoiesis and suggest that K8 methylation is a critical determinant of GFI1–LSD1 axis function.
K8 methylation is required for zebrafish primitive erythropoiesis.
K8 methylation is required for GFI1 pro-growth and survival functions in lymphoid leukemia cells
GFI1 is critical for the initiation and maintenance of lymphoblastic leukemia . GFI1 conditional deletion impedes lymphoma development in animal models and causes regression of established lymphomas [12,31]. We tested the importance of K8 for pro-growth and survival functions of GFI1 in lymphoid leukemia cells. Using a GFI1-targeted shRNA, we depleted GFI1 in the lymphoblastic leukemia cell line, CCRF-CEM (Figure 6A), and determined apoptosis relative to a content-matched vector control (Figure 6B,C). Apoptosis was determined by flow cytometry using concurrent PI and Annexin V staining. Control cells display size and complexity features consistent with lymphocytes. A limited subset of cells underwent apoptosis (27% Annexin V+/PI−). Following GFI1 depletion, the lymphocyte population displayed increased apoptosis (62% Annexin V+/PI−) and a population with small size and widely divergent side scatter emerged. This emergent population, which increased with time following GFI1 depletion, is consistent with secondary necrosis (98% Annexin V+/PI+; Figure 6B). Quantifying both apoptotic and secondarily necrotic populations reveals that GFI1 is required for CCRF-CEM cell survival (Figure 6C).
GFI1 is required for lymphoid leukemia cell survival.
To determine the impact of K8 methylation and LSD1 recruitment on cell survival, we determined cell viability in stable CCRF-CEM cell lines expressing FLAG-tagged WT, P2A, and K8L GFI1 derivatives or vector control and simultaneously depleted of endogenous GFI1. We confirmed shRNA-mediated depletion of endogenous GFI1 and enforced expression of FLAG-tagged GFI1 derivatives by western blot (Figure 7A). Cell lines depleted of endogenous GFI1 then re-expressing WT GFI1 retained growth potential, while those expressing P2A, K8L, or vector control failed to expand (Figure 7B). These data indicate that GFI1 derivatives with defective LSD1 binding cannot functionally compensate for GFI1 loss and suggest that K8 methylation may be a critical determinant of lymphoid leukemia cell viability.
K8 is required for GFI1 pro-growth and survival functions in lymphoid leukemia cells.
Total RNA from these cell lines was subjected to RNA sequencing. Clustering provided a list of genes up-regulated with endogenous GFI1 depletion alone and in GFI1-depleted cells expressing WT, K8L, and P2A GFI1 derivatives (Figure 7C). Genes up-regulated following endogenous GFI1 depletion alone, as well as in GFI1-depleted cells expressing K8L and P2A GFI1 derivatives, but not WT GFI1, were subjected to gene ontology (GO) functional term enrichment (Figure 7C). This list was enriched for genes associated with programmed cell death/regulation of apoptosis and for loss of transcriptional repression (Figure 7C). These results suggest that K8 methylation is a critical determinant of LSD1 recruitment by GFI1 to support GFI1 pro-growth and survival functions in lymphoblastic leukemia cells.
GFI1 is a transcriptional repressor and master regulator of cell fate, differentiation and survival in normal and malignant hematopoiesis . The N-terminal SNAG domain is the dominant repressor element within GFI1. Yet, a detailed understanding of mechanisms governing transcriptional control by GFI1 is lacking [1,32]. We focused on the SNAG domain, and in particular, on a -8KSKK11- motif embedded within the SNAG domain to gain additional mechanistic insights into GFI1-mediated transcriptional repression. We find SMYD2 methylates K8 within the SNAG domain, that K8 methylation favors LSD1 recruitment and that this SNAG–LSD1 relationship is required for GFI1–LSD1 axis functions in transcriptional repression and hematopoietic cell fate determination.
Nonhistone proteins, including transcription factors, are subject to lysine methylation. Among them are p53, RB, E2F1, STAT3, MYOD, RELA/p65, AR, ERα, and CEBPβ [33–35]. These factors play critical regulatory functions in cell growth and survival, and likewise instruct the establishment and maintenance of the differentiated state. Thus, lysine methylation occupies a central nexus for cell fate determination, making its co-ordinate control mechanisms among transcription factors essential for normal homeostasis. A growing roster of methyltransferases is responsible for lysine methylation among nonhistone proteins and transcriptional regulators, including SETD7, G9a, NSD2, SETD6, SMYD2, and SMYD3 [33–35]. Often, the methylation events catalyzed by these enzymes can be reversed by the methyl-lysine specific demethylase, LSD1 [33,35]. The mechanistic and molecular consequences of lysine methylation vary for each transcription factor, and perhaps for each methylation event and stoichiometry. For example, RB is methylated at K860 by SMYD2, promoting its interaction with Polycomb group repressor protein L3MBTL1 and correlating with cell cycle arrest . However, RB can also be methylated at K873 by SETD7, which promotes RB's interaction with the heterochromatin protein, HP1, and supports transcriptional repression, cell cycle arrest, and myogenic differentiation . These examples demonstrate discrete methylation events that alter protein–protein and transcriptional co-regulator interactions to carry out specific transcriptional and biological outcomes. This same principle is reflected in our description of the SMYD2–GFI1–LSD1 functional axis. SMYD2 methylates K8 to favor LSD1 binding. K8L substitution abolishes binding between GFI1 and LSD1, profoundly impairs transcriptional repression by the SNAG domain, and renders GFI1 inactive in hematopoietic complementation and growth control. Yet, neither K10L nor K11L substitutions impact these phenotypes, suggesting a measure of residue selectivity in controlling GFI1 function. Further studies will be needed to determine whether K8 methylation within the -8KSKK11- motif of GFI1's SNAG domain alone governs LSD1 recruitment, or whether instead it is one of many methyl group modifications via distinct methyltransferases that are integrated to modulate transcriptional repression. Given that SMYD2 is a common regulator for GFI1 and p53, and that GFI1 exerts dominion over p53 DNA damage and pro-apoptotic responses, it is attractive to speculate that other p53 -370KSKK373- methyltransferases might modify analogous lysine residues in the GFI1 -8KSKK11- motif to achieve co-ordinate control over their counter-regulatory relationship. Functional grouping of methyltransferases and demethylases toward transcription factor pairs has obvious advantages when minimizing regulatory complexity is a goal. We are actively pursuing this notion of shared regulatory factors for GFI1 and its transcriptional partners.
Our findings add to an emerging choreography of posttranslation modifications that regulate GFI1 function and provide an example for how a requirement for coincident regulatory inputs may restrict the activity of an epigenetic effector like LSD1. We recently described a role of GFI1 SUMOylation in recruiting the LSD1/CoREST complex to GFI1-regulated promoters . We envision that LSD1 recruitment via the GFI1 SNAG domain can be stabilized, and its demethylase activity stimulated by interactions between CoREST and SUMOylated GFI1. Thus, SUMOylation provides one regulatory input for LSD1 activity when recruited to GFI1-regulated genes via the SNAG domain. Our finding that K8 methylation strongly favors, and K8L substitution abolishes LSD1 binding, GFI1-mediated transcriptional repression, and cell fate decisions suggests a second regulatory input for the GFI1–LSD1 axis.
Our data are consistent with a working model for transcriptional control via the SMYD2–GFI1–LSD1 axis (Figure 8) where transcriptional repression by GFI1 is favored by concurrent, SMYD2-mediated K8 methylation and SUMOylation within the GFI1 linker, together facilitating LSD1/CoREST recruitment and enabling CoREST-dependent activation of LSD1 demethylase activity at target genes . Our work here provides the basis for future work clarifying how these and other posttranslational modifications regulate GFI1 transcriptional control to govern cell fate and survival decisions in normal and malignant hematopoiesis.
A working model for GFI1-mediated LSD1/CoREST recruitment and transcriptional repression.
ChIP-qPCR, chromatin immunoprecipitation and quantitative PCR; DMEM, Dulbecco's modified Eagle's medium; GFI1, growth factor independence 1; GO, gene ontology; hLSD1, human LSD1; HSC, hematopoietic stem cell; PI, propidium iodide; TK, thymidine kinase; WT, wild type.
M.V. designed and conducted experiments, analyzed the data and prepared the draft manuscript. J.S. and D.B developed, prepared and validated reagents used in experiments, and performed the experiments. J.M., D.M. and H.L. contributed to the experimental design and execution of the experiments. C.M.T, J.M.F and S.F performed the experiments and interpreted the generated data. S.S. synthesized the small molecule inhibitor of LSD1 and provided additional reagents for the experiments. M.E.E. designed, conducted and interpreted the experiments, supervised the research and compiled the manuscript. All authors critically reviewed the manuscript.
This work, including the efforts of Michael Eugene Engel, was funded by HHS/NIH/National Cancer Institute (NCI) [P30CA042014]. This work, including the efforts of Michael Eugene Engel, was funded by the HHS/NIH/National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) [K08DK080190]. This work, including the efforts of Michael Eugene Engel, was funded by the St. Baldrick's Foundation (Career Development Award). This work, including the efforts of Sunil Sharma, was funded by the CureSearch Foundation. The funders had no role in the design of experiments, collection of data, or interpretation of study results. Likewise, they had no role in the decision to submit the work for publication.
The authors thank Thomas O'Hare, Mahesh Chandrasekharan, Trudy Oliver, Tim Formosa, and Brad Cairns for their helpful thoughts and suggestions during preparation of this manuscript. We acknowledge support from the DNA/peptide synthesis, Centralized Zebrafish Animal Resource (CZAR), flow cytometry, and DNA sequencing cores of the University of Utah as well as the high-throughput genomics and bioinformatics analysis cores at the Huntsman Cancer Institute.
The Authors declare that there are no competing interests associated with the manuscript.