Btg2 Enhances Retinoic Acid-Induced Differentiation by Modulating Histone H4 Methylation and Acetylation

Abstract

Retinoic acid controls hematopoietic differentiation through the transcription factor activity of its receptors. They act on specific target genes by recruiting protein complexes that deacetylate or acetylate histones and modify chromatin status. The regulation of this process is affected by histone methyltransferases, which can inhibit or activate transcription depending on their amino acid target. We show here that retinoic acid treatment of hematopoietic cells induces the expression of BTG2. Overexpression of this protein increases RARα transcriptional activity and the differentiation response to retinoic acid of myeloid leukemia cells and CD34⁺ hematopoietic progenitors. In the absence of retinoic acid, BTG2 is present in the RARα transcriptional complex, together with the arginine methyltransferase PRMT1 and Sin3A. Overexpressed BTG2 increases PRMT1 participation in the RARα protein complex on the RARβ promoter, a target gene model, and enhances gene-specific histone H4 arginine methylation. Upon RA treatment Sin3A, BTG2, and PRMT1 detach from RARα and thereafter BGT2 and PRMT1 are driven to the cytoplasm. These events prime histone H4 demethylation and acetylation. Overall, our data show that BTG2 contributes to retinoic acid activity by favoring differentiation through a gene-specific modification of histone H4 arginine methylation and acetylation levels.

Retinoic acid (RA) is a physiologic regulator of hematopoietic differentiation (11). The significance of RA is highlighted by its effectiveness as a therapeutic agent in acute promyelocytic leukemia (APL), where RA-induced differentiation of the leukemia cells dramatically improves patient prognosis (33). The effects of RA are mediated by the transcriptional activity of its nuclear receptors. RARα is the major receptor in hematopoietic cells, where it acts as an RA-dependent transcriptional activator (11). RARα binds specific DNA responsive elements on the promoter of target genes as a heterodimer with RXR, a cofactor of several other nuclear receptors (9, 11). In the absence of ligand, the RARα/RXR heterodimer binds corepressor molecules, such as N-CoR and SMRT, through the RARα component. These corepressors, in turn, recruit histone deacetylases that deacetylate histones, thus promoting a structural modification of chromatin that represses target gene transcription. RA binding promotes detachment of the repressor complex and recruitment of coactivator molecules with histone acetyltransferase activity. Histone acetylation contributes to modify chromatin structure and allows gene transcription (9).

Several other molecules potentially participate in this regulation. Histone methylation has been recognized as a powerful regulatory mechanism (6, 26). Lysine 9 methylation of histone H3 is an alternative to acetylation and represses transcription (3, 25, 47). In contrast, lysine 4 methylation of histone H3 correlates with active gene transcription (7, 41). Arginine methylation of histones contributes to this regulation (10, 26). In transcriptional assays, methylation of the Arg 17 of histone H3 by the methyltransferase CARM1 contributes to estrogen, androgen, and thyroid receptor-mediated activation (4, 32, 42, 51). Methylation of Arg3 of histone H4 by the arginine methyltransferase PRMT1 also enhances transcription activation by coactivators of the p160 family (24, 48, 53, 55). These epigenetic modifications of histones also affect promoter DNA methylation, further contributing to generate active or inactive chromatin (55). Although histone methylation may facilitate acetylation (2, 42, 54), the exact temporal relationships among these events during transcriptional regulation are still under investigation. The picture is further complicated by the fact that histone methyltransferases are themselves bound by other proteins that have the potential to regulate their activity and ultimately the function of nuclear receptors (28, 46). BTG2/PC3/Tis21 is a PRMT1 binding protein and the product of a p53 regulated gene (1, 8, 13, 30, 40, 52). BTG2 participates in the control of G₂/M cell cycle arrest following DNA damage or the differentiation of different cell types. This action is mediated by the inhibition of cyclin D1 and cyclin E transcription in an Rb-dependent fashion (20, 29). BTG2 enhances HOXB9-mediated transcription (36) and can affect estrogen receptor function through its interaction with hCAF1 (35). Recently, the interaction of BTG2 with BMP activated SMADS and its effect on BMP-dependent transcription has been uncovered (34). Overall, BTG2 is able to behave as a transcriptional coregulator in different model systems, although the molecular mechanisms of this action are probably multiple and often unclear. BTG2 activity also plays a role in differentiation since it affects maturation of cells of neural origin and osteoclasts (1, 8, 15, 12, 23, 27). The biological and molecular activities of BTG2 may be relevant to cancer pathogenesis, since mutation or aberrant expression of this gene was found in human cancers (23, 49). On these bases, we investigated BTG2 participation to RA-induced differentiation of the hematopoietic myeloid lineage and the molecular mechanisms of its biological activity. We show here that BTG2 expression is induced by RA treatment and that BTG2 overexpression improves RA-induced differentiation. BTG2 participates in the RARα transcriptional complex and enhances histone H4 arginine methylation. However, upon RA treatment, increasingly expressed BTG2 and PRMT1 detach from RARα, allowing demethylation and acetylation of histone H4. Thus, BTG2 contributes to regulate chromatin remodeling on specific target genes during RA-induced differentiation.

MATERIALS AND METHODS

RDA, clones, and plasmids.

RDA was performed as described previously (22) on polyadenylated mRNA extracted from U937 or PR9 cells (18) treated with 150 μM ZnSO₄ for 8 h. At 5 × 10⁻⁶ M RA was added to the culture for 3 h. PCR clones were identified by sequencing. A coding cDNA for the human BTG2 protein (NM_006763) was cloned by reverse transcription PCR using RNA extracted from the PR9 cell line. The primers used were 5′-GACATGAGCCACGGGAAGGGAAC-3′ and 5′-GCCTAGCTGGAGACTGCCATCAC-3′. The BTG2 cDNA was tagged with a hemagglutinin (HA) epitope and then cloned into the pSG5 mammalian expression vector (Stratagene). BTG2 was also subcloned into pGEX-4T1 (Pharmacia) to generate the GST-BTG2 fusion protein and into the retroviral vector PINCO (19). The reporter vector (RARE-luc) and pCMV-βgal used in luciferase assays were as previously reported (5). The PRMT1 cDNA (NM_198318) was PCR cloned by using RNA extracted from the U937 cell line. The primers used were 5′-AGGCCGCGAACTGCATCATGGAG-3′ and 5′-ACCGACTACCGGATGCGCTGAG-3′. The cDNA was cloned in the pSG5 vector for transactivation experiments and pGEX-4T1. The glutathione S-transferase (GST) fusion proteins were produced in BL21 bacteria (Invitrogen, Carlsbad, CA) and bound to glutathione-Sepharose beads (Amersham Biosciences) according to the manufacturer's instructions.

Cell culture and retroviral vector infection.

The myeloid leukemia cell lines U937, PR9, HL60, NB4, NB4-R4, and HL60R were grown in RPMI 1640 with 10% fetal calf serum. The PR9 cell line is a U937 derivative with inducible expression of the PML/RARα fusion protein (18). HL60R (38) is a derivative of HL60 carrying a mutated RARα; NB4R4 (43) is a derivative of NB4 carrying a mutated PML/RARα protein. These mutants are RA resistant. The 293T, Phoenix, and HeLa cell lines were maintained in Dulbecco modified Eagle medium with 10% bovine calf serum. CD34⁺ hematopoietic progenitor cells (HPCs) were cultured in STEMSPAM medium (Stem Cell Technologies, Vancouver, British Columbia, Canada) and differentiated with published combinations of growth factors (17). Briefly, the cells were stimulated to proliferate with 50 ng of stem cell factor/ml, 100 U of interleukin-3 (IL-3)/ml, and 50 ng of FLT3-L/ml; monocytic differentiation was obtained with 10% fetal bovine serum and 100 ng of macrophage colony-stimulating factor/ml; and granulocytic differentiation was obtained with 1 U of IL-3/ml, 500 U of granulocyte colony-stimulating factor/ml, and 0.1 ng of granulocyte-macrophage colony-stimulating factor/ml. RA-induced differentiation was performed with 10⁻⁷ M RA (Sigma).

Infection of U937 cells, HL60 cells, and HPCs and fluorescence-activated cell sorting purification of infected cells was performed as described previously (17) using a BTG2 expressing PINCO vector or an empty control vector.

Northern blotting and real-time PCR.

RNA was isolated from the PR9, U937, NB4, NB4R4, HL60, and HL60R cell lines by using TRIzol (Invitrogen, Carlsbad, CA). Northern analysis was performed with by standard procedures. The membranes were hybridized with the BTG2 cDNA or a GAPDH (glyceraldehyde-3-phosphate dehydrogenase) probe (16). Quantitative real-time PCR was performed as published (31) in an ABI Prism 7000 sequence detection system (Applied Biosystems) using the oligonucleotides 5′-GAAGGTGAAGGTCGGAGT-3′ and 5′-CATGGGTGGAATCATATTGGAA-3′ for GAPDH. BTG2 expression was detected with the primers 5′-CACAGAGCACTACAAACACCACTG-3′ and 5′-CTTGTGGTTGATGCGAATGC-3′. The RARβ primers were 5′-CTTCCTGCATGCTCCAGGA-3′ and 5′-CGCTGACCCCATAGTGGTA-3′. Gene expression, normalized for GAPDH expression serving as an endogenous control, was calculated by using the Delta-Delta CT method.

Immunoprecipitation and Western blotting.

For immunoprecipitation, cells were lysed by sonication in E1A buffer (50 mM HEPES [pH 7], 150 mM NaCl, 0.1% NP-40, 5 mM EDTA, 1 mM dithiothreitol [DTT], and protease inhibitors). Lysates were clarified by high-speed centrifugation and quantified by using a Bradford reagent. Then, 4 mg of protein lysate was precleared with protein A-Sepharose and immunoprecipitated with the antibodies anti-Sin3A (sc-767; Santa Cruz Biologicals, Santa Cruz, CA), anti-PRMT1 (ChIP-Grade; Abcam, Cambridge, United Kingdom), anti-RAR (M-454; Santa Cruz), and anti-BTG2 (20). Immunoprecipitates were analyzed by Western blotting by standard methods with the antibodies anti-BTG2 (20), anti-PRMT1 (Abcam), anti-RARα (c-20; Santa Cruz), anti-α-tubulin (Sigma clone B-5-1-2), anti-RARβ (c-19; Santa Cruz), and anti-mSin3A as described above. Filters were treated with anti-mouse or anti-rabbit horseradish peroxidase-conjugated antibody and detected by using the ECL System (Amersham, Madison, WI).

Nuclear and cytoplasmic extracts.

Nuclear and cytoplasmic extracts were obtained from 5 × 10⁶ HL60 cells stimulated with 10⁻⁶ M RA or unstimulated. After treatment, the cells were washed three times with ice-cold phosphate-buffered saline, harvested, and resuspended in 0.5 ml of buffer A (10 mM HEPES [pH 7.9], 10 mM KCl, 1.5 mM MgCl₂, 0.5 mM DTT, 0.2 mM phenylmethylsulfonyl fluoride [PMSF]) with protease inhibitor (5 μg of aprotinin/ml, 5 μg of pepstatin/ml A, and 5 μg of leupeptin/ml). After 10 min on ice, Nonidet P-40 was added to a final concentration of 0.5%. Nuclei were separated by low-speed centrifugation, and the supernatant was used as a cytosolic fraction. The nuclear pellet was resuspended in 50 μl of buffer B (20 mM HEPES [pH 7.9], 1.5 mM MgCl₂, 0.42 M NaCl, 0.2 mM EDTA, 0.5 mM DTT, 1 mM PMSF, 10% glycerol) and protease inhibitor. After 30 min at 4°C, lysates were centrifuged at 13,000 × g for 5 min, and the supernatant was used as nuclear fraction. The protein concentration of extracts was measured by using a protein dye reagent (Bio-Rad, Richmond, CA).

Immunofluorescence.

HL60 cells were collected before and after a 6-h treatment with 10⁻⁶ M RA, cytocentrifuged, and fixed with methanol at room temperature for 5 min, followed by acetone at −20°C for 2 min. The slides were incubated for 1 h with the primary antibody against BTG2 (20) or PRMT1 (Abcam), washed, and incubated for 45 min with a secondary antibody conjugated to TRITC [tetramethyl rhodamine 5 (and 6)-isothiocyanate]. A DNA dye (DAPI [4′,6′-diamidino-2-phenylindole]) was used for nuclear staining. The cells were washed, mounted in Mowiol (Calbiochem) and photographed with an Olympus BX60 microscope equipped with a 3CCD color camera (C5810; Hamamatsu).

Cell phenotype.

In differentiation experiments cells were treated with 10⁻⁷M RA for the indicated times. Immunophenotyping was performed as published (18) using phycoerythrin-conjugated Serotech antibodies (Serotech, Oxford, United Kingdom). Nitroblue tetrazolium (NBT) assay was performed as described previously (18).

In vitro methylation assay.

Core histones were obtained by acid extraction according to the manufacturer's instructions (Upstate Biotechnology). Twenty micrograms of basic proteins was incubated with 1 μg of GST-PRMT1 alone or with an equimolar amount of GST-BTG2 in a total volume of 15 μl containing 50 mM Tris-HCl (pH 8), 0.5 mM DTT, 1 mM PMSF, and 2 μl of SAM(³H) (Amersham Biosciences) at 30°C for 1.5 h (30). The samples were fractionated by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis. Radiolabeled components were detected by autoradiography.

Transactivation experiments.

HL60 cells were transfected by electroporation (960 μF, 250 V) in 300 μl of RPMI containing the amount of each expression plasmid (pSG5, pSG5-BTG2, or pSG5 PRMT1) indicated in the figure or figure legend: 3 μg of pCMV-βgal for normalization of transfection efficiency and 7 μg of the reporter vector βRARE-Luc, carrying the RA responsive element (RARE) of the RARβ gene promoter (16) linked to a luciferase gene. HeLa cells were transfected by calcium phosphate coprecipitation with 1.5 μg of βRARE-Luc, 300 ng of pCMV-βgal, and 50 to 1,250 ng of pSG5-BTG2 or pSG5 PRMT1. At 16 h posttransfection the cells were stimulated with 10⁻⁷ M RA for an additional 8 h and then lysed for the luciferase assay. The luciferase activities were normalized for transfection efficiency by dividing the relative light units by the β-galactosidase activity expressed from cotransfected pCMV-βgal.

ChIP.

A chromatin immunoprecipitation (ChIP) assay was performed according to standard protocols (Upstate Biotechnology, Lake Placid, NY) with minor modifications. Briefly, 293T or HL60 cells or their BTG2-overexpressing derivatives were cross-linked with 1% formaldehyde at room temperature for 10 min, washed in ice-cold phosphate-buffered saline, and lysed with SDS lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris-HCl [pH 8]). Cellular lysates were sonicated, diluted with ChIP dilution buffer, and immunoprecipitated with specific antibodies: anti-acetyl-H4 (Upstate Biotechnology), anti-PRMT1 (Chip-Grade; Abcam), anti-dimethyl-H4 Arg3 (Upstate Biotechnology), anti-Sin3A (Santa Cruz), anti-BTG2 (20), and anti-CD40 (Pharmingen/BD Biosciences, San Diego, CA) as a negative control. Immunoprecipitates were collected with protein A-agarose beads (Upstate Biotechnology) and washed sequentially with a low-salt and high-salt wash buffer (Upstate Biotechnology). DNA-protein complexes were eluted with 1% SDS and 0.1 M NaHCO₃, heated to 65°C overnight, and deproteinized by treatment with proteinase K at 45°C for 1 h and phenol-chloroform extraction. The DNA was recovered by ethanol precipitation and assayed by PCR with the following primers: RARβ-promoter (X56849), 5′-GTTGGGTCATTTGAAGGTTAGCAG-3′ and 5′-ACAAACCCTGCTCGGATCGCT-3′; and actin gene, 5′-CTTCTACAATGAGCTGCGTGTGG-3′ and 5′-ATGGCTGGGGTGTTGAAGGTCTCA-3′. The amplified RARβ promoter segment contains the canonical RARα responsive element (9). PCR products were Southern blotted according to standard procedures and probed with cloned and sequenced DNA fragments amplified from genomic DNA with the same oligonucleotides.

RESULTS

BTG2 expression is induced by RA.

To identify genes that are specifically linked to RA-induced differentiation, we performed a representational difference analysis (RDA) with RNA of U937 cells, which undergo a very limited differentiation when treated with RA and U937 cells expressing the APL fusion protein PML/RARα (PR9), which confers RA responsiveness to the cells (18). BTG2 is among the mRNA species that are induced within 3 h of RA treatment. To confirm the RA-dependent induction of BTG2 mRNA, we studied its expression by Northern blotting in a panel of myeloid cell lines (Fig. 1A). BTG2 mRNA is rapidly induced by RA in PR9 cells and in the APL cell line NB4 but not in its mutated RA-resistant derivative, NB4R4. U937 cells showed a modest induction, in agreement with their limited response to RA. These data were confirmed by real-time PCR (Fig. 1B), showing an early and increasing induction of BTG2 mRNA by RA in HL60, PR9, and NB4 cells but not in the RA-resistant derivatives HL60R and NB4R4. Protein expression in U937, PR9, HL60, and HL60R cells match mRNA data (Fig. 1C). Notably, BTG2 induction preceded proliferation arrest (not shown) by at least 48 h and was observed in HL60 cells despite the absence, in these cells, of a functional p53. Thus, BTG2 induction is linked to a functional RA signaling.

FIG. 1.

BTG2 expression is induced by RA treatment. (A) Northern blot analysis of BTG2 expression in the indicated cell lines, untreated or treated with RA for the indicated times. Blots were hybridized with radiolabeled BTG2 and control GAPDH probes. (B) Real-time PCR analysis of BTG2 expression in the indicated cell lines, untreated or treated with RA for the indicated times. (C) Western blot analysis of BTG2 expression in lysates from the indicated cells. C+ is a lysate from 293T cells transfected with a BTG2 expression vector. Blots were stripped and rehybridized with an anti-α-tubulin antibody as a loading control.

BTG2 expression enforces RARα transcriptional and biological activity.

To investigate the functional role of BTG2, we studied the effects of its overexpression on RA-induced differentiation and on the transcriptional activity of RARα, the major RA receptors expressed in myeloid hematopoietic cells. We cotransfected in the HL60 myeloid cell line BTG2 and an RA-responsive element of the RARβ gene promoter linked to a luciferase reporter gene. The expression of BTG2 markedly increased the RA-dependent transcriptional activity in a dose-dependent manner (Fig. 2A). Interestingly, cotransfection of a PRMT1 cDNA increased the transcriptional stimulation of the RAR/RXR complex mediated by BTG2 (Fig. 2B). These results are not tissue specific, since the transfection of HeLa cells gave similar results (data not shown). Overall, these data support the conclusion that BTG2 interaction with RARα may increase its transcriptional activity. To verify whether this effect had a biological relevance in the hematopoietic system, we overexpressed BTG2 in the myeloid cell lines HL60 and U937 and in CD34⁺ human HPCs (Fig. 3A) and studied the differentiation potential of these cells. In both cell lines HL60 and U937, overexpression of BTG2 increased the differentiation response to RA, as shown by the expression of the differentiation marker CD11b, by the NBT reduction assay (Fig. 3B). Enforcing these data, expression of BTG2 in HPCs increased the level of differentiation obtained by adding RA to growth factors combinations leading to monocytic (M-growth factors) or granulocytic (G-growth factors) differentiation, as shown by the expression of the surface markers CD11b and CD14 (Fig. 3C). In our cell systems, the effect of BTG2 on cell growth was unnoticeable (not shown).

FIG. 2.

BTG2 expression increases RARα transcriptional activity. (A) Transcriptional activity of a luciferase-linked β-RARE responsive element cotransfected in HL60 cells with the indicated amounts of vector DNA (V) or a BTG2 expression vector. (B) Transcriptional activity of a luciferase-linked β-RARE responsive element cotransfected in HL60 cells with vector DNA (V) or a BTG2 expression vector with or without a PRMT1 expression vector. RA indicates an 8-h treatment with 10⁻⁷ M RA. The results are expressed as the transcriptional activity relative to that measured in cells transfected with the responsive element and empty vector alone. The means and standard deviations from three triplicate experiments are given.

FIG. 3.

BTG2 expression increases RA-induced differentiation of hematopoietic cells. (A) Western blot showing BTG2 overexpression in the indicated retrovirally infected cells. C+, lysates from 293T cell transiently transfected with a BTG2 expression vector; C, control uninfected cells; B, cells infected with a BTG2 expression vector. Blots were stripped and rehybridized with an anti-α-tubulin antibody as a loading control. (B) RA-induced differentiation of HL60 and U937 cells with or without BTG2 overexpression measured as the percentage of NBT- or CD11b-positive cells. Cells were treated with 10⁻⁷ M RA for the indicated times. C, cells infected with a control “empty” retroviral vector; BTG2, cells infected with a BTG2 retroviral vector. (C) Differentiation of CD34⁺ human hematopoietic progenitor cells with 10⁻⁷ M RA and growth factors cocktails inducing monocytic or granulocytic differentiation (see Materials and Methods). C, BTG2, as described above. The means and standard deviations from three independent experiments are given.

BTG2 and PRMT1 take part in the RARα transcriptional complex.

To investigate the molecular mechanisms of BTG2 activity on RA-induced differentiation, we searched for an interaction of the BTG2 protein with the arginine methyltransferase PRMT1 and with the transcriptional protein complex of RARα. Using anti-BTG2, anti-Sin3A, anti-RARα, and anti-PRMT1 antibodies in coimmunoprecipitation experiments followed by Western blotting (Fig. 4), we found that BTG2, Sin-3A, PRMT1, and RARα coimmunoprecipitate in the absence of RA. RA treatment abolished BTG2 coimmunoprecipitation with Sin3A but not with PRMT1. These data strongly suggested that BTG2 and PRMT1 take part in the basal RARα transcriptional complex in vivo but are released by RA binding to RARα. To verify this hypothesis, we performed ChIP analysis on the RA-responsive myeloid cell line HL60 using an anti-BTG2, anti-Sin3A, anti-PRMT1, or anti-RARα antibody and searched by PCR for a segment of the RARβ gene promoter containing a RARα responsive element (RARE). The results (Fig. 5A) showed that BTG2 and PRMT1, as well as Sin3A and RARα, reside on the RARβ promoter in live cells. Notably, the RARβ promoter was more represented in the chromatin immunoprecipitated with an anti-PRMT1 from cells overexpressing BTG2 than from control cells (Fig. 5A). We found high levels of PRMT1, BTG2 and Sin3A on the RARβ promoter for up to 30 min of RA treatment. After this time, BTG2 and PRMT1 are rapidly released from the receptor together with Sin 3A. In these first hours of RA treatment we did not find significant changes in the expression of the RARα protein (Fig. 5B), which maintains its location on the RARβ promoter (Fig. 5A).

FIG. 4.

BTG2 interacts with the RARα transcriptional complex. Coimmunoprecipitation experiments from lysates of HL60 cells. Cells were treated with 10⁻⁶ M RA for the indicated times. Lysates were immunoprecipitated with the antibodies shown on the right. Samples were analyzed by Western blotting with the indicated antibodies. The positions of molecular mass markers expressed in kilodaltons are indicated on the right of each panel. C+, lysates from HL60; C−, immunoprecipitation with an anti-CD40 antibody from lysates of untreated HL60 cells. Ig, position of the immunoglobulin band when visible in the frame.

FIG. 5.

BTG2 takes part in the RARα transcriptional complex on the RARβ promoter. (A) Anti-PRMT1, anti-BTG2, anti-Sin3A, and anti-RARα ChIP experiments on HL60 cells infected with a control empty vector (C) or with a BTG2 vector (B) treated with RA for the indicated times. Input, PCR with RARβ promoter oligonucleotides on the cell lysates used for ChIP; Actin, control amplification of a β-actin gene sequence; RARβ, amplification with RARβ promoter oligonucleotides. In all of the experiments shown in panel A, PCR amplification with RARβ promoter oligonucleotides on chromatin precipitated with an anti-CD40 antibody, as a negative control, gave no signal; a representative panel is shown. (B) Western blot showing persistent RARα expression in HL60 or HL60 BTG2 cells treated with 10⁻⁶ M RA for the indicated times.

Since our coimmunoprecipitation data indicated that BTG2 and PRMT1 maintain their association when the cells are exposed to RA, we investigated whether the BTG2 and PRMT1 proteins may consensually move within the cell. To this end we analyzed the nuclear and cytoplasmic distribution of these proteins. Western blotting on cytoplasmic and nuclear cellular lysates indicated that both the BTG2 and the PRMT1 proteins are present in the nucleus and in the cytoplasm (Fig. 6A). RA treatment significantly decreased their nuclear fraction, while increasing the cytoplasmic one. Immunofluorescence studies confirmed these data (Fig. 6B).

FIG. 6.

(A) Western blot of nuclear and cytoplasmic fractions extracted from HL60 cells treated with 10⁻⁶ M RA for the indicated times. Filters were transversely cut at a molecular mass of 31 kDa on the basis of stained molecular mass standards. The upper half was hybridized with an anti-PRMT1 antibody, and the lower half was stained with an anti-BTG2 antibody. The loading amounts of the nuclear fractions were controlled by staining the filters after the electrotransfer with Ponceau S. The staining identifies the core histone bands. The loading amounts of the cytoplasmic fractions were evaluated by stripping the filters and rehybridizing them with an anti-α-tubulin antibody. (B) Immunofluorescence experiments on cytospin slides showing HL60 cells before (−RA) and after (+RA) a 6-h treatment with 10⁻⁶ M RA. The primary antibodies are indicated on the left. The secondary antibodies were labeled with TRITC, tetramethyl rhodamine 5 (and 6)-isothiocyanate. DAPI is a nuclear stain used to identify the nuclei.

Overall, our data indicate that BTG2 and PRMT1 contribute to the RARα complex on target genes promoter but are released by RA, which induces the transfer of their complex to the cytoplasm.

BTG2 overexpression increases the gene-specific arginine methylation of histone H4.

To verify whether the increased amount of PRMT1 on the RARβ promoter is functionally active, we performed ChIP experiments on HL60 cells overexpressing BTG2 using anti-methyl arginine 3 histone H4 antibodies and amplified by PCR from the immunoprecipitated chromatin the fragment of the RARβ promoter containing a RARE. Histone H4 arginine methylation levels on the RARβ promoter are much higher in unstimulated BTG2-overexpressing cells compared to control cells (Fig. 7A). When the cells were treated with RA the methylation in BTG2 cells is maintained at consistently higher levels than in control cells. After 1 h from the beginning of RA treatment, histone H4 arginine methylation on the RARβ promoter markedly decreases in both BTG2-overexpressing and control cells. These data remarkably parallel the presence of PRMT1 and BTG2 on the RARβ promoter. However, a GST-BTG2 fusion protein did not modify the in vitro histone methylation induced by a GST-PRMT1 fusion protein (Fig. 7B), suggesting that BTG2 may act by increasing the amount of PRMT1 on specific promoters.

FIG. 7.

BTG2 expression increases histone H4 arginine methylation and lysine acetylation in response to RA. (A) Anti-methyl-arginine 3 histone H4 ChIP experiments on control (C) or BTG2- overexpressing (B) HL60 cells treated with RA for the indicated times. Input, Actin, and RARβ are as described for Fig. 5. (B) In vitro methylation of core histones extracted from HL60 cells by the indicated GST or GST fusion proteins. (C) Anti-acetyl-H4 ChIP experiments on control (lanes C) or BTG2-overexpressing (lanes B) HL60 cells treated with RA for the indicated times. Input, Actin, and RAR-β are as described for Fig. 5. (D) Anti-methyl-arginine 3 histone H4 and anti-acetyl-histone H4 ChIP experiments on lysates of control (lanes C) 293T cells or 293T cells transiently transfected with a BTG2 expression vector (lanes BTG2). For anti-methyl-arginine 3 histone H4 ChIP cells were not treated with RA to prevent demethylation (see Fig. 5). For anti-acetyl-H4 ChIP experiments, cells were studied before and after a 24-h RA treatment. Input, Actin, and RARβ are as described for Fig. 5. P, positive PCR control on genomic DNA; N, negative PCR control without DNA. In the experiments shown in panels A, C, and D PCR amplification with the RARβ promoter oligonucleotides on chromatin precipitated with an anti-CD40 antibody, as a negative control, gave no signal. (E) Real-time PCR showing the induction of RARβ expression by 10⁻⁶ M RA treatment of HL60 and HL60 BTG2 cells for the indicated times. (F) Western blot showing RARβ expression in HL60 or HL60 BTG2 cells treated with 10⁻⁶ M RA for the indicated times.

BTG2 and PRMT1 release from RARβ promoter is followed by histone H4 acetylation.

We next sought to determine whether these changes could correlate with modifications of histone H4 acetylation levels on the same promoter. We performed ChIP experiments with anti-acetylated histone H4 antibodies and again searched for the RARβ promoter on the precipitated chromatin. The results (Fig. 7C) showed that, when the cells are treated with RA, histone H4 acetylation overlapped a decrease in its methylation levels (see Fig. 7A). BTG2-overexpressing cells displayed a faster histone H4 acetylation response that was evident at 1 h after cells exposure to RA and increased up to 24 h after the start of treatment. Control cells displayed a slower kinetics of histone H4 acetylation on the RARβ promoter, although eventually (at 48 h) histone acetylation reached levels that are similar to those of BTG2-overexpressing cells. These epigenetic changes are reflected in a higher expression of the RARβ RNA and protein in BTG2 cells (Fig. 7E and F). Neither BTG2 overexpression nor RA treatment significantly modified histone H4 methylation or acetylation on the β-actin gene. Taken together, these data suggest that BTG2 overexpression increases histone H4 arginine methylation that in turn facilitates acetylation and allows RARβ transcription. To confirm this hypothesis, we measured by ChIP the modification of histone H4 methylation and acetylation levels on the RARβ promoter induced in 293T cells by a transient overexpression of BTG2 (Fig. 7D). Arginine methylation of histone H4, measured in the absence of RA, was markedly increased by BTG2 overexpression. This correlated with a higher level of histone H4 acetylation upon RA treatment of BTG2-overexpressing cells. Overall, these data indicate that BTG2 overexpression increases the PRMT1-mediated arginine methylation of histone H4 on the RARβ promoter. This facilitates the histone H4 acetylation response after RA stimulation.

DISCUSSION

We showed here that BTG2 participates in the regulation of RA activity on myeloid hematopoietic cells. BTG2 increases the cellular differentiation response to RA, takes part in RARα transcriptional complex, and contributes, by affecting PRMT1 function, to a gene-specific sequential increase of histone H4 arginine methylation and lysine acetylation on RA target genes. In turn, RA treatment induces the expression of BTG2 and releases BTG2 and PRMT1 from RARα, allowing a decrease in histone H4 methylation and its acetylation. These findings indicate that RA affects accessory proteins that contribute to regulate the activity of its receptors favoring cell differentiation.

BTG2 is induced by RA. Induction was not the consequence of growth arrest, since it occurred before any significant decrease in cell proliferation rate. Although BTG2 is recognized as a p53-regulated gene, its induction by RA was evident in the p53-negative cell line HL60. We did not specifically address the mechanism of BTG2 regulation by RA. However, the structural basis for RA induction of BTG2 expression does exist within the BTG2 promoter. Although it does not contain canonical palindromic RARα responsive elements (23; our unpublished results), an analysis of potential transcription factors binding sites showed the presence of three Sp1 sites (14). The Sp1 transcription factor can be activated by the RAR/RXR heterodimer by direct protein interaction, in the absence of typical RARα responsive elements (45, 50).

BTG2 contributes to the differentiation effect of RA by participating in the RARα transcriptional complex. The interaction of BTG2 with RARα probably mediates this effect, since a BTG2 and RARα protein coprecipitate from cell lysates. The interaction is maintained in vivo, as shown by the ChIP data that indicate that RARα and BTG2 are part of the same transcriptional complex on RARα responsive genes. Probably, these findings do not reflect a direct protein contact between BTG2 and RARα, since we were only able to demonstrate a weak direct in vitro interaction between BTG2 and RARα or RXRα (results not shown). Their interaction is likely to be mediated or stabilized by other proteins. We hypothesize that the stability of the association between BTG2 and RARα may be improved in vivo by BTG2 interaction with the arginine methyltransferase PRMT1. Indeed, we show that BTG2 and PRMT1 coimmunoprecipitate and PRMT1 is found on the RARE-containing region of the RARβ promoter, a model of the RARα responsive gene. Moreover, we observed increased arginine methylation levels of histone H4 on the RARβ promoter when BTG2 was overexpressed, possibly as a result of an increased amount of PRMT1 protein, since the in vitro activity of PRMT1 did not seem to be modified by BTG2. Thus, the involvement of BTG2 in the RARα transcriptional complex is functionally relevant. Confirming the participation of PRMT1 and BTG2 to the basal RARα protein complex, these proteins coimmunoprecipitate with Sin3A, a corepressor that is linked to RARα in the absence of RA (44). RA stimulation significantly modifies this complex with very rapid kinetics. Histone H4 arginine methylation on the RARβ promoter quickly falls to low levels after 1 h of RA treatment. This change is paralleled by a marked decrease in the amounts of PRMT1, BTG2, and Sin3A, but not RARα, on the RARβ promoter. BTG2 and PRMT1 still coimmunoprecipitate and are actually found mostly in the cytoplasm, as previously shown for PRMT1 and the BTG1 protein (21, 39) Thus, RA seems to release BTG2 and PRMT1 from its receptor, priming a decrease in histone H4 methylation levels. The velocity of the drop in arginine methylation levels may imply that active demethylation is taking place. These modifications are seen both in control cells and in BTG2 cells, suggesting that the overexpression of BTG2 is not able to anchor PRMT1 on the RARβ promoter. The increased BTG2 expression induced by RA does not seem to participate directly in these events, since it occurs when BTG2 is no longer on the RA target gene promoter. RA-induced BTG2 is actually mostly in the cytoplasm, possibly still linked to PRMT1, as shown by the immunoprecipitation data. In agreement with this, we could not find evidence of physical or functional interactions of BTG2 with RARα coactivators such as p300, TIF2, and ACTR (our unpublished results). Since BTG2 can affect PRMT1 binding to partner proteins, we hypothesize that the role of BTG2 induction by RA could be to modify PRMT1 interactions with different substrates. In addition, it could prevent the return of PRMT1 on RARα after the beginning of histone acetylation, when histone H4 is a poor substrate for arginine methylation (53). Actually, the decrease in histone H4 arginine methylation seems to herald lysine acetylation, which increases after 1 to 3 h of RA stimulation, when arginine methylation decreases. Upon RA treatment, histone H4 acetylation occurs more rapidly in cells that overexpress BTG2, where arginine methylation is higher, than in control cells. These data suggest that histone H4 arginine methylation precedes acetylation and increases its efficiency, inducing quicker acetylation kinetics, but is then blocked as acetylation proceeds. This is in agreement with the reported increased efficiency of p300-mediated acetylation of arginine methylated histone H4 (53). Histone H4 acetylation levels in control cells eventually “catch up” to those of BTG2-overexpressing cells. Overall, our data indicate that BTG2 serves to increase the effects of PRMT1 on RARα target promoters, facilitating RARα transcriptional activity. Our transactivation experiments consolidate this concept, indicating that BTG2 expression increases RARα transcriptional effects and that this effect is potentiated by the cotransfection of PRMT1.

These data support the notion that PRMT1-mediated arginine methylation of histone H4, together with CARM1-mediated arginine methylation of histone H3, is part of the “histone code” affecting the function of nuclear receptors. We show here that this code itself is regulated by RA, confirming the view (37) that histone arginine methylation can be driven by specific transcription factors and occur as a targeted event to activate transcription.

We also investigated the biological counterparts of these molecular findings. The response to RA-induced differentiation is increased in cells that overexpress BTG2, both in cell lines and in normal hematopoietic progenitors. The promotion of differentiation by BTG2 was previously observed in cells of neural origin. In the present study we provide the molecular basis for this effect within the hematopoietic system. Surprisingly, we found that overexpression of BTG2 does not alter the basal growth rate of the cell types studied here. In agreement with the differentiation data, a decreased proliferation rate was seen after differentiation induction in cell lines and normal hematopoietic progenitors. BTG2 is well recognized as a growth inhibitor and cell cycle regulator (52). However, its effects were not previously studied in the hematopoietic system. It is possible that tissue-specific effects contribute to our and previous observations. Alternatively, since BTG2 expression increases steadily up to 48 h after RA induction, it is possible that lower BTG2 protein levels are sufficient for the transcriptional effects and differentiation induction, whereas a higher expression may induce growth arrest. Nevertheless, the differentiation effect that we observed could contribute physiologically to growth limitation and provide further support to the role of tumor suppressor (14, 23, 49) hypothesized for BTG2.