A TFIID-SAGA Perturbation that Targets MYB and Suppresses Acute Myeloid Leukemia

Abstract

Targeting of general coactivators is an emerging strategy to interfere with oncogenic transcription factors (TFs). However, coactivator perturbations often lead to pleiotropic effects by influencing numerous TFs. Here we identify TAF12, a subunit of TFIID and SAGA coactivator complexes, as a selective requirement for acute myeloid leukemia (AML) progression. We trace this dependency to a direct interaction between the TAF12/TAF4 histone-fold heterodimer and the transactivation domain of MYB, a TF with established roles in leukemogenesis. Ectopic expression of the TAF4 histone-fold fragment can efficiently squelch TAF12 in cells, suppress MYB, and regress AML in mice. Our study reveals a strategy for potent MYB inhibition in AML and highlights how an oncogenic TF can be selectively neutralized by targeting a general coactivator complex.

Keywords: MYB; SAGA; TAF12; TFIID; acute myeloid leukemia; coactivator; epigenetics.

Figure 1. A TFIID-focused shRNA screen reveals that acute myeloid leukemia cells are hypersensitive to TAF12 knockdown

(A) Negative selection shRNA screening strategy. 86 shRNAs targeting 16 TFIID subunits were cloned into the MLS-E retroviral vector. The effect of individual shRNAs on cell proliferation was evaluated by measuring GFP depletion in a competition-based assay. (B) Summary of shRNA screening results. The GFP depletion of shRNAs targeting TFIID subunits, a negative control shRNA targeting Renilla luciferase (shREN.713) and a positive control shRNA targeting RPA3 (shRPA3.457) is displayed. GFP depletion is plotted as a day 2 to day 12 ratio of GFP% for RN2 cells and a day 2 to day 8 ratio of GFP% for normal myeloid progenitors. (C) Western blot analysis in RN2 cells and normal myeloid progenitor cultures on day 3 post infection with control or TAF12 shRNA (#364). (D) Western blot analysis in RN2 cells transduced with FLAG-TAF12 cDNA or empty retroviral vector. (E) TAF12 shRNA/cDNA rescue assay in RN2 cells. Retrovirally (MSCV) expressed FLAG-TAF12 cDNA is linked to GFP and retrovirally expressed shTAF12.364 is linked with mCherry. GFP/mCherry double positive cell depletion was measured to evaluate rescue. n=3. (F) Bright-field images of methylcellulose-based colony formation assays in RN2 cells and normal myeloid progenitors on day 7 post viral transduction with control or TAF12 (#364) shRNAs. Scale bar represents 1 mm. (G) Quantification of immature/blast colonies shown in (F). n=3. (H) May-Grünwald Giemsa staining of RN2 cells on day 4 post infection with control or TAF12 shRNA (#364). Scale bar represents 10 μm. (I) Flow cytometry analysis of RN2 cells stained with antibodies against c-Kit (surface marker of leukemia stem cells) and Mac-1 (surface marker of macrophage cells) on day 4 post infection with control or TAF12 (#364) shRNAs. (J–Q) GFP depletion assays and western blot analysis in murine MLL-AF9/Nras^G12D AML, i.e. RN2 (J), murine DNMT3A^R882H/NRAS^G12D AML (K), murine NUP98-JARID1A AML (L), murine NIH 3T3 fibroblast (M), murine B16-F10 melanoma cells (N), murine Kras^G12D/p53^−/− lung cancer cells (O), murine 4T1 breast cancer cells (P) and murine MycCaP prostate cancer cells (M) after retroviral transduction with control or TAF12 (#364) shRNAs. n=3. All bar graphs represent the mean ±3SEM. See also Figure S1.

Figure 2. Evaluation of TAF12 knockdown in AML and normal tissues in vivo

(A) Bioluminescent imaging of mice transplanted with RN2 cells transduced with doxycycline (dox) regulated shRNAs. Mice were injected with 500,000 RN2 cells and treated with dox on day 0 and imaged on day 7, 9 and 11 post transplantation. 4 or 5 mice were used for each cohort and representative images are shown. (B) Quantification of bioluminescent imaging from (A). Values represent photons per second (p/s) of bioluminescent signal detection. p value was calculated using unpaired Student’s t-test. n=4–5. (C) Survival curves of mice transplanted with RN2 cells transduced with the indicated dox-regulated shRNAs. p value was calculated using Log-rank (Mantel-Cox) test. n=4–5. (D) Bioluminescent imaging of mice transplanted with DNMT3A^R882H/NRAS^G12D AML cells transduced with CRISPR guide RNAs targeting a control region (sgROSA) or TAF12 (sgTAF12 #4.1). Mice were injected with 15,000 cells and imaged on day 8, 10 and 12 post transplantation. 4 or 5 mice were used for each cohort and representative images are shown. (E) Quantification of bioluminescent imaging from (D). Values represent photons per second (p/s) of bioluminescent signal detection. Data are presented as mean ± SEM, n=4–5. p value was calculated using unpaired Student’s t-test. (F) Survival curves of mice transplanted with DNMT3A^R882H/NRAS^G12D AML cells transduced with indicated CRISPR guide RNAs (sgRNAs). p value was calculated using Log-rank (Mantel-Cox) test. 9–10 mice were used for each cohort. (G) Transgenic shRNA expression strategy. TRE: Tetracycline response element; rtTA-M2: reverse tetracycline-controlled transactivator M2 variant. All mice evaluated were heterozygous for the shRNA cassette and homozygous for the rtTA-M2 transgene. (H) Representative western blot analysis of whole tissue lysates prepared from mice treated with dox in the drinking water and chow for 4 weeks. (I) Hematoxylin and eosin staining of tissues from mice treated with dox for 4 weeks. Scale bar represents 100 μm. (J–K) Flow cytometry analysis of bone marrow (J) and spleen (K) stained with antibodies against the indicated cell surface markers from transgenic mice treated with dox in the drinking water and chow for 4 weeks. Gating was first performed on GFP⁺/shRNA⁺ cells, prior to measurement of marker staining. p value was calculated by unpaired Student’s t test, *p<0.05. n=3. All bar graphs represent the mean ±3SEM. See also Figure S2.

Figure 3. TAF12 is required to sustain the transcriptional output of MYB

(A) RT-PCR analysis of Myb mRNA level in murine AML and non-AML cell lines. Measurements were normalized to Gapdh mRNA level. n=3. (B–I) GFP depletion assays in the murine MLL-AF9/Nras^G12D AML, i.e. RN2 (B), murine NUP98-JARID1A AML (C), murine DNMT3A^R882H/NRAS^G12D AML (D), murine Kras^G12D/p53^−/− lung cancer cells (E), murine MycCap prostate cancer cells (F), murine 4T1 breast cancer cells (G), murine B16-F10 melanoma cells (H) and immortalized fibroblast iMEF (I) transduced with control (shREN) or MYB shRNAs. n=3. (J–K) RNA-seq comparison of TAF12 knockdown to MYB (J) or MLL-AF9 (K) knockdown in RN2 cells. Scatter plot depicts the fold-change in RPKM values of 8,044 expressed genes (RPKM =5 in shREN samples) comparing two independent shRNAs targeting TAF12, MYB or MLL-AF9 to shREN. (L) Unbiased Gene Set Enrichment Analysis (GSEA) of shTAF12 versus shREN in RN2 cells. Normalized enrichment score (NES) and family-wise error rate (FWER) p value were ranked and plotted for 13,321 genes sets, which include all of the Molecular Signature Database v5.1 (MSigDB) and a MYB gene signature, defined as the top 200 down-regulated genes upon MYB knockdown in RN2 cells. Each gene set is depicted as a single dot. (M) GSEA plot of the murine MYB target gene signature upon TAF12 knockdown in RN2 cells. (N) Unbiased GSEA analysis of sgTAF12 versus sgROSA in the human AML cell line MOLM-13, as described in (L), for evaluating the effects in the MSigDB gene sets and a MYB gene signature, defined as the top 200 down-regulated genes in MOLM-13 cells following sgRNA-mediated targeting of MYB. (O) Heat map showing the relative MYB mRNA level of 173 AML patients samples in the TCGA database. Patients with relative high and low MYB level were defined by z score >0.5 and <−0.5, respectively. (P–Q) GSEA analysis of patients with high versus low MYB level for evaluating the effects in the MYB-dependent (P) or TAF12-dependent (Q) gene signatures, defined as the top 200 down-regulated genes upon sgRNA-mediated targeting of MYB and TAF12 in MOLM-13 cells, respectively. All bar graphs represent the mean ±3SEM. See also Figure S3.

Figure 4. Physical interaction between MYB and the TAF12/TAF4 heterodimer

(A) IP-western blot analysis evaluating the interaction between FLAG-MYB and endogenous TFIID subunits. FLAG IP was performed in nuclear lysates from RN2 cells stably expressing FLAG tagged MYB^FL(full-length), MYB^1-327 fragment or empty vector followed by western blotting for the indicated TFIID subunits. (B) Western blot analysis of TFIID subunits pulled down by recombinant GST or GST tagged transactivation domain (TAD) of MYB from RN2 nuclear lysates. (C) Illustration of MYB domain architecture and fragments used for shRNA/cDNA rescue experiments. DBD: DNA-binding domain, TAD: transactivation domain, NRD: negative regulatory domain. (D) GFP depletion assay evaluating the FLAG-MYB fragments shown in (C) to rescue the growth-arrest in RN2 cells caused by GFP-linked MYB shRNA (#2652). Bar graph represents the mean ±3SEM, n=3. (E) Illustration of murine TAF12 and TAF4 domain architectures. Crystal structure represents the human TAF12/TAF4 histone-fold domain (HFD) heterodimer (PDB:1H3O) (Werten et al., 2002). (F) GFP depletion assay evaluating the ability of full-length (FL) TAF12 versus the TAF12-HFD fragment (amino acids 50–130) to rescue the growth-arrest in RN2 cells caused by GFP-linked TAF12 shRNA (#364). Bar graph represents the mean ±3SEM, n=3. (G) Illustration of MYB truncation fragments used for co-IP experiments. (H) Co-IP western blot analysis evaluating the interaction between the indicated MYB fragments and the TAF4/TAF12 HFD heterodimer. HEK 293T cells were co-transfected with HA tagged TAF12, untagged TAF4-HFD, and one of the FLAG-MYB fragments shown in (G) or empty vector. FLAG IP was performed in nuclear lysates 48 hr post transfection. (I) Silver staining analysis of the recombinant GST, GST tagged TAF12/TAF4 HFD heterodimer (purified from E. coli) and FLAG-MYB^1-327 (purified from HEK 293T). (J) Binding assay using the proteins shown in (I). GST proteins were immobilized on glutathione beads, followed by incubation with FLAG-MYB^1-327. The bead-associated complexes were eluted and western blotted with anti-FLAG antibodies. (K) Co-IP western blot analysis evaluating the interaction between MYB and TADA1/TAF12 or TAF4/TAF12 HFD heterodimer. HEK 293T cells were co-transfected with HA tagged TAF12-HFD, untagged TAF4-HFD or TADA1-HFD, and FLAG-MYB (FL or 1-327 fragment) or empty vector. FLAG IP was performed in nuclear lysates 48 hr post transfection. (L) Mass spectrometry analysis of TFIID subunits following FLAG-TAF10 IP in RN2 cells expressing a dox-regulated TAF12 shRNA (#364) or control shRNA. Total peptide counts for the indicated subunits is shown, normalized to the control shRNA samples. (M) Western blot analysis of the indicated TFIID subunits in RN2 cells on day 3 post infection of control or TAF12 (#364) shRNAs. See also Figure S4.

Figure 5. MYB recruits TAF12 to select cis elements in AML cells to support transcriptional activation

(A) ChIP-seq density plots of MYB, FLAG-TAF12, TAF2, TAF10, H3K27ac and H3K4me3 enrichment at MYB-occupied promoter and distal regions. MYB peaks located within −1 kb to +100 bp of transcription start site were defined as promoter regions, and others were defined as distal regions. Each row represents a 2 kb interval centered on the summit of each MYB peak and were ranked based on MYB ChIP-seq tag counts. FLAG-TAF12 ChIP-seq was performed in RN2 cells stably expressing FLAG tagged TAF12 following 24 hr dox-induced expression of control or MYB (#2652) shRNAs. TAF2 and TAF10 ChIP-seq were performed in parental RN2 cells. MYB, H3K27ac and H3K4me3 ChIP-seq data were from previous studies generated in parental RN2 cells (Roe et al., 2015; Shi et al., 2013). (B) Global changes in FLAG-TAF12 enrichment following MYB knockdown. Fold-change in TAF12 tag counts for 19,780 TAF12 peaks in shMYB (#2652) versus shREN transduced RN2 cells are ranked and plotted. Examples of MYB dependent and independent peaks are indicated. (C–D) ChIP-seq profiles of H3K27ac, MYB, TAF2 and FLAG-TAF12 at Ctsg (C), a genomic locus with MYB-dependent TAF12, and Slpi (D), a site with MYB-independent TAF12, in RN2 cells. (E–F) ChIP-qPCR analysis of TAF12 (E) and TAF2 (F) at indicated MYB dependent and independent TAF12 loci following MYB knockdown. ChIP-qPCR was performed in RN2 cells following 24 hr dox-induced expression of control or MYB (#2652) shRNAs. Bar graphs represent the mean ±3SEM, n=3. (G) GSEA analysis evaluating the expression of 326 expressed genes (RPMI =5) located nearest to the top 500 MYB-dependent TAF12 peaks following MYB or TAF12 knockdown. (H) Transcription factor affinity prediction of the top 500 MYB-dependent TAF12 peaks versus 500 random TAF12 peaks. DNA sequences flanking 200 bp of each peak summit were used for motif analysis. (I) Ontology analysis of genes located nearest to the top 500 MYB-dependent TAF12 peaks versus genes located nearest to 500 random TAF12 peaks using the Genomic Regions Enrichment of Annotations Tool (GREAT). In parentheses is the ontology identifier of each indicated pathway in the GREAT database. See also Figure S5.

Figure 6. TAF12 protects MYB from proteasome-mediated degradation

(A) MYB ChIP-qPCR analysis at indicated regions identified via MYB ChIP-seq in RN2 cells infected with control or TAF12 (#364) shRNAs on day 3 post infection. Bar graph represents the mean ±3SEM, n=3. (B) Quantification of the relative MYB occupancy detected by ChIP-qPCR in (A). Data was normalized to shREN sample. Each dot represents individual locus tested in (A) and the error bar represents SD. (C) Western blot analysis in RN2 cells infected with control or TAF12 (#364) shRNAs on day 3 post infection. Quantification of MYB bands with normalization to ACTIN bands was performed using ImageJ software. (D) Western blot analysis of MYB protein level in RN2 cells following dox-regulated shRNA expression for 48 hr, followed by MG132 treatment for 4 hr. 1/2 shREN indicates a 2-fold dilution of shREN sample. (E) Western blot analysis of MYB protein level in RN2 cells retrovirally co-transduced with the indicated cDNA and shRNA constructs. (F) GFP depletion assay evaluating the effect of MYB overexpression on the growth-arrest caused by the GFP-linked TAF12 shRNA (#364) in RN2 cells. Data are plotted as the mean ±3SEM. * p<0.05 using unpaired Students’ t-test, n=3. See also Figure S6.

Figure 7. Squelching of TAF12 with a TAF4 HFD peptide leads to MYB suppression and anti-AML activity in vivo

(A) Illustration of the TAF4-HFD based strategy for interfering with TAF12 and MYB function in AML. (B) Western blot analysis of FLAG-TAF4-HFD expression in RN2 cells. (C) GFP depletion assay evaluating the effect of retrovirally expressing the TAF4-HFD (linked to GFP) in RN2 cells. Prior to TAF4-HFD expression, RN2 cells were transduced with the TAF12 cDNA or empty vector. (D) Bright-field images of methylcellulose-based colony formation assays in RN2 cells and normal myeloid progenitors expressing TAF4-HFD or empty vector on day 7 post viral transduction and day 5 post plating. Scale bar represents 1 mm. (E) Quantification of immature/blast colonies in (D). Data are plotted as mean ±3SEM, n=3. (F) May-Grünwald Giemsa staining of RN2 cells expressing the TAF4-HFD or empty vector (day 4 post infection). Scale bar represents 10 μm. (G) Flow cytometry analysis of RN2 cells stained with antibodies against c-Kit and Mac-1 on day 4 post infection with TAF4-HFD or empty vector control. (H) GSEA analysis of TAF4-HFD overexpression versus empty vector control in RN2 cells, evaluating effects on a MYB dependent gene signature, defined as the top 200 downregulated genes following MYB knockdown in RN2 cells. (I–J) RNA-seq comparison of TAF4-HFD expression to TAF12 (I) or MYB (J) knockdown in RN2 cells. Scatter plot depicts the fold-change in RPKM values of 8,044 expressed genes (RPKM =5 in shREN) comparing two independent TAF12/MYB shRNAs or TAF4-HFD expression to control (shREN or empty vector, respectively). (K) Bioluminescent imaging of mice transplanted with RN2 cells transduced with dox-regulated GFP-TAF4-HFD or GFP control. Mice were treated with dox in the drinking water and chow starting on day 9 and imaged on day 9, 11 and 13 following RN2 cells injection. 4 or 5 mice were used for each cohort. (L) Quantification of bioluminescent imaging in (K). Values represent photons per second (p/s) of bioluminescent signal detection. p value was calculated using unpaired Student’s t-test. Data are plotted as mean ±3SEM, n=4–5. (M) Survival curves of mice transplanted with RN2 cells transduced with dox-regulated GFP-TAF4-HFD or GFP. Mice were treated with dox in the drinking water and chow starting on day 9 following RN2 cell injection. p value was calculated by Log-rank (Mantel-Cox) test. n=4–5. See also Figure S7.