PRMT4 is a novel coactivator of c-Myb-dependent transcription in haematopoietic cell lines

Abstract

Protein arginine methyltransferase 4 (PRMT4)-dependent methylation of arginine residues in histones and other chromatin-associated proteins plays an important role in the regulation of gene expression. However, the exact mechanism of how PRMT4 activates transcription remains elusive. Here, we identify the chromatin remodeller Mi2α as a novel interaction partner of PRMT4. PRMT4 binds Mi2α and its close relative Mi2β, but not the other components of the repressive Mi2-containing NuRD complex. In the search for the biological role of this interaction, we find that PRMT4 and Mi2α/β interact with the transcription factor c-Myb and cooperatively coactivate c-Myb target gene expression in haematopoietic cell lines. This coactivation requires the methyltransferase and ATPase activity of PRMT4 and Mi2, respectively. Chromatin immunoprecipitation analysis shows that c-Myb target genes are direct transcriptional targets of PRMT4 and Mi2. Knockdown of PRMT4 or Mi2α/β in haematopoietic cells of the erythroid lineage results in diminished transcriptional induction of c-Myb target genes, attenuated cell growth and survival, and deregulated differentiation resembling the effects caused by c-Myb depletion. These findings reveal an important and so far unknown connection between PRMT4 and the chromatin remodeller Mi2 in c-Myb signalling.

Figure 1. Identification of putative interaction partners of PRMT4.

A: PRMT4 resides within a high molecular weight protein complex. For size fractionation by gel filtration chromatography, whole-cell protein extracts were generated from HEK293, Molt-4 and MCF7 cells and subjected to Benzonase treatment. Protein extracts were applied to a Superose 6 column and 1 ml fractions were collected. Fractions (8–19) were stained by Western Blot analysis using anti-PRMT4 antibodies. The column was calibrated using standard protein markers. Accordingly, the size of the PRMT4-containing fractions and the void volume (V₀) are indicated. B: Biochemical purification of endogenous PRMT4. Schematic representation of the chromatographic steps used to purify PRMT4-containing complexes from HEK293 protein extract. C: High molecular weight complexes of PRMT4 remain stably associated during purification procedure. PRMT4-containing MonoQ fractions were size-fractionated using a Superdex 200 column. After collection of fractions (500 µl each), fraction numbers 14–25 were stained by Western Blot analysis using anti-PRMT4 antibodies. The size of the PRMT4-containing fractions and the void volume (V₀) are indicated. D: Affinity purification of endogenous PRMT4 by immunoprecipitation. PRMT4-containing MonoQ fractions were subjected to IP using polyclonal anti-PRMT4 antibodies compared to bead control. Input (1% = 12 µg) and immunoprecipitates were separated by SDS-PAGE. Silver-stained bands specifically detected in the anti-PRMT4 sample were excised and identified by MALDI-TOF peptide mass fingerprint analysis as PRMT4 (asterisks) and Mi2α (arrowheads). SDS-PAGE size markers (in kDa) are shown on the left. Two bands were identified as PRMT4, which are likely different isoforms (of 63 and 65 kDa), but which could not be distinguished by peptides in the mass spectrometry.

Figure 2. PRMT4 interacts with Mi2α and Mi2β.

A, B: Co-immunoprecipitation of PRMT4 with Mi2α. HEK293 cells were transfected with Flag-Mi2α construct (+) or empty vector (−). After 48 hours, cells were lysed and 1 mg protein extract was subjected to IP of (A) endogenous PRMT4 (α-PRMT4) or (B) overexpressed Flag-Mi2α (α-Flag). IPs using isotype control IgG were performed in parallel (α-ctrl). Input (1%) and immunoprecipitates were stained by Western Blot analysis using anti-Flag and anti-PRMT4 antibodies. C: Among the PRMT family, PRMT4 preferentially interacts with Mi2α. Protein lysates of Flag-Mi2α-overexpressing HEK293 cells (as in A, B) were subjected to GST-pulldown experiments. Equal amounts of recombinant proteins GST alone and GST-PRMT1, -3, -4 and -6 (Figure S1) coupled to glutathione beads were incubated with 250 µg of HEK293 extracts. Input (1%) and bound Flag-Mi2α protein were visualised by Western Blot analysis using anti-Flag antibodies. D: Co-immunoprecipitation of PRMT4 with Mi2β. HEK293 cells were transfected with HA-Mi2β construct (+) or empty vector (−). Protein extracts were subjected to IP using anti-PRMT4 (α-PRMT4) antibodies or isotype control IgG (α-ctrl). Input (1%) and precipitates were stained by Western Blot analysis using anti-HA and anti-PRMT4 antibodies.

Figure 3. PRMT4 and Mi2 interact with the transcription factor c-Myb.

A: Co-immunoprecipitation of overexpressed PRMT4 with c-Myb. HEK293 cells were transfected with untagged PRMT4 and HA-c-Myb construct (alone or in combination) and harvested 48 hours after transfection. Protein extracts were subjected to IP using anti-HA (α-HA), anti-PRMT4 (α-PRMT4) antibodies or isotype control IgG (α-ctrl). Input (1%) and precipitates were stained by Western Blot analysis using anti-HA and anti-PRMT4 antibodies. B: Co-immunoprecipitation of endogenous PRMT4 with c-Myb. Jurkat cell extract was incubated with anti-PRMT4 (α-PRMT4) or isotype control IgG (α-ctrl). Input (5%) and precipitates were stained by Western Blot analysis using anti-c-Myb and anti-PRMT4 antibodies. The anti-c-Myb stainings for input and IP shown in the panels derive from the same blot and exposure time revealing that less than 5% of endogenous c-Myb interacts with endogenous PRMT4. C: PRMT4 and c-Myb are direct interaction partners. GST-pulldown with recombinant His-tagged c-Myb was performed using equal amounts of glutathione beads-bound recombinant GST, GST-PRMT1, -4 and -6 proteins. Input and bound His-c-Myb were visualised by Western Blot analysis using anti-c-Myb antibodies. D: Mi2α and Mi2β interact with c-Myb. HEK293 cells were transfected with HA-c-Myb plasmid together with Flag-Mi2α or HA-Mi2β. Protein extracts were incubated with anti-c-Myb antibodies or isotype control IgG (α-ctrl). Input and precipitates were stained by Western Blot analysis using anti-HA and anti-Flag antibodies. Asterisks indicate Flag-Mi2α and arrowheads indicate HA-Mi2β. E: Co-immunoprecipitation of endogenous Mi2α with PRMT4 and c-Myb. Jurkat cell extract was incubated with anti-Mi2α (α-Mi2α) or control IgG (α-ctrl). Input (5%) and precipitates were stained by Western Blot analysis using anti-PRMT4, anti-c-Myb and anti-Mi2α antibodies.

Figure 4. PRMT4 and Mi2 are cooperating transcriptional activators of c-Myb-dependent gene expression in the chicken macrophage cell line HD11.

A: PRMT4 and Mi2α are synergistic coactivators of c-Myb target genes. HD11 cells were transfected with the indicated constructs. After 48 hours, cells were harvested and total RNA was isolated. RT-qPCR was performed for detection of transcript levels of Mim-1. Each mRNA expression was normalised to GAPDH mRNA expression. Transcript levels in empty vector-transfected cells (−) were set to 1. B: Coactivation of c-Myb-dependent gene expression is specific for PRMT4. HD11 cells were transfected with the indicated constructs. After 48 hours, cells were harvested for total RNA isolation. Levels of Mim-1 mRNA were analysed by RT-qPCR and normalised to GAPDH mRNA levels. Transcript levels in empty vector-transfected cells (−) were set to 1. C: Mi2β synergises with PRMT4 in c-Myb-dependent gene expression. HD11 cells were transfected with indicated constructs. After 48 hours, cells were harvested for total RNA isolation. Levels of Mim-1 mRNA were analysed by RT-qPCR and normalised to GAPDH mRNA levels. Transcript levels in empty vector-transfected cells (−) were set to 1. D: The catalytic activity of PRMT4 and Mi2 is essential for their cooperative function. HD11 cells were transfected with tagged wild type (black) and catalytic mutant forms (grey) of PRMT4 and Mi2α (methyltransferase-dead PRMT4 mutant: VLD; helicase-dead Mi2α mutant: KA). Total RNA was isolated 48 hours after transfection. Levels of Mim-1 mRNA were analysed by RT-qPCR and normalised to GAPDH mRNA levels. Transcript levels of empty vector-transfected cells (−) were set to 1.

Figure 5. PRMT4 and Mi2 are concomitantly recruited with c-Myb to the Mim-1 gene.

A, B: HD11-C3 cells stably express a doxycycline-inducible c-Myb construct. HD11-C3 cells were either left untreated (−Dox) or treated with doxycycline (+Dox) for 30 hours. For isolation of protein extracts and total RNA cells were harvested. (A) Levels of c-Myb protein were detected by Western Blot analysis and β-Tubulin staining served as a loading control. (B) Transcript levels of Mim-1 were measured by RT-qPCR and normalised to GAPDH mRNA levels. The mRNA level in untreated cells (−) was set to 1. C: Schematic representation of the Mim-1 gene upstream of the transcriptional start site (indicated by arrow). Enhancer and promoter regions (blue rectangles) and the c-Myb binding site (red dots) are marked. The amplicon of the promoter region (−201 to −75), enhancer region (−2022 to −1916) and upstream control region (−6687 to −6544) generated in the ChIP-qPCR (D) are indicated. D–H: PRMT4 and Mi2 are direct transcriptional regulators of c-Myb target gene Mim-1. HD11-C3 cells were treated as in A, B and harvested for chromatin isolation. Chromatin was subjected to ChIP analysis using antibodies against c-Myb (D), PRMT4 (E), Mi2 (F), H3R17me2 (G, I), histone H3 (H, I) and control antibody (IgG). Immunoprecipitated DNA was analysed in triplicates by qPCR with primers spanning the regions of the Mim-1 gene indicated in C (orange = enhancer; blue = promoter; purple = upstream control region). Legend in D also applies to E–I. Mean values were expressed in % input of chromatin.

Figure 6. PRMT4 and Mi2 are direct transcriptional regulators of c-Myb target genes in human haematopoietic cells.

A–D: PRMT4 and Mi2 regulate overlapping c-Myb target genes in K562 cells. Knockdown of c-Myb, PRMT4, Mi2α and Mi2β was achieved in K562 by electroporation of the corresponding siRNAs. Cells were harvested 3 days later and protein extracts and total RNA were prepared. (A) For estimation of knockdown efficiency, protein levels of c-Myb, PRMT4 and as loading control β-Tubulin were measured by Western Blot analysis. (C) For detection of knockdown efficiency of Mi2α and Mi2β at mRNA levels, RT-qPCR was conducted using gene-specific primers for Mi2α and Mi2β and normalised to GAPDH. Transcript levels in si ctrl-treated cells were set to 1. (B, D) Transcript levels of the c-Myb target genes Cdc7, c-Myc, Gata3 and CyclinB1 (CycB1) were measured in the various knockdown conditions by RT-qPCR and normalised to GAPDH. Transcript levels in si ctrl-treated cells were set to 1. E–G: Concomitant recruitment of c-Myb, PRMT4 and Mi2 to c-Myb target genes in human haematopoietic cells. K562 cells were harvested and subjected to ChIP analysis using control antibodies (IgG) and antibodies against c-Myb, PRMT4 and Mi2. Immunoprecipitated DNA was analysed by qPCR with primers for (E) Cdc7 promoter, corresponding control region and β-Tubulin promoter, for (F) c-Myc promoter and corresponding control region and for (G) Cyclin B1 (CycB1) promoter and corresponding control region. Recruitment is displayed in % input of chromatin.

Figure 7. PRMT4 and Mi2 are important regulators of proliferation and differentiation function in human erythropoiesis.

A: PRMT4 and Mi2 have similar effects on cell cycle progression to c-Myb. K562 cells were transfected with siRNAs targeting c-Myb, PRMT4, Mi2α and Mi2β or with control siRNA (si ctrl). The DNA content of propidium iodide (PI)-stained cells was measured by flow cytometry (FACS). From each knockdown the percentage of cells in sub-G1, G1 and G2/M phase was determined. Shown are the changes in percentage (%) relative to the si ctrl condition. A representative data set is depicted. B, C: PRMT4 and Mi2 have similar pro-proliferative functions to those of c-Myb. K562 cells were transfected with siRNAs, as in A. After 3 days of transfection, cells were cultured in methylcellulose for 2 days and stained with INT (Iodonitrotetrazolium chloride) for 5 days. (B) Representative pictures of the colonies were taken for each knockdown condition at the same magnification. (C) For quantification, colonies larger than 0.1 mm were counted. The mean of colony numbers was calculated from three independent experiments and error bars are indicated accordingly. D: Depletion of PRMT4 and Mi2 recapitulates the pro-differentiating effects of c-Myb knockdown. K562 cells were transfected with the indicated plasmids for 3 days. Cells were then re-seeded and either left untreated or treated with 30 µM hemin for 3 days. Subsequently, benzidine staining of differentiated cells was performed. The mean percentage (%) of benzidine-positive K562 cells was determined from triplicate countings and error bars are indicated accordingly.