Novel CARM1-Interacting Protein, DZIP3, Is a Transcriptional Coactivator of Estrogen Receptor-α

Abstract

Coactivator-associated arginine methyltransferase 1 (CARM1) is known to promote estrogen receptor (ER)α-mediated transcription in breast cancer cells. To further characterize the regulation of ERα-mediated transcription by CARM1, we screened CARM1-interacting proteins by yeast two-hybrid. Here, we have identified an E3 ubiquitin ligase, DAZ (deleted in azoospermia)-interacting protein 3 (DZIP3), as a novel CARM1-binding protein. DZIP3-dependent ubiquitination of histone H2A has been associated with repression of transcription. However, ERα reporter gene assays demonstrated that DZIP3 enhanced ERα-mediated transcription and cooperated synergistically with CARM1. Interaction with CARM1 was observed with the E3 ligase RING domain of DZIP3. The methyltransferase activity of CARM1 partially contributed to the synergy with DZIP3 for transcription activation, but the E3 ubiquitin ligase activity of DZIP3 was dispensable. DZIP3 also interacted with the C-terminal activation domain 2 of glucocorticoid receptor-interacting protein 1 (GRIP1) and enhanced the interaction between GRIP1 and CARM1. Depletion of DZIP3 by small interfering RNA in MCF7 cells reduced estradiol-induced gene expression of ERα target genes, GREB1 and pS2, and DZIP3 was recruited to the estrogen response elements of the same ERα target genes. These results indicate that DZIP3 is a novel coactivator of ERα target gene expression.

The estrogen activated nuclear receptor recruits transcriptional coregulators to activate or repress transcription of estrogen receptor (ER)α target genes when it occupies ER-binding sites on chromatin (1–3). To signal for activation of transcription, coregulators employ a variety of molecular mechanisms. The earliest and most well-described transcriptional coregulators are the p160 steroid receptor coactivator (SRC) family that signals for transcriptional activation through the formation of protein scaffolds for recruitment of additional coregulators (4–7), which often harbor various enzymatic activities to modulate chromatin structure and function. These include proteins that perform posttranslational modifications of histones or nonhistone proteins, such as acetylation, methylation, phosphorylation, or ubiquitination. Coregulators can also impact protein-protein interactions (8) and modulate recruitment and activation of RNA polymerase II with the basal transcription machinery (9). These SRCs contribute to the tissue specific response to estrogen stimulation and are known to be important regulators of ERα target gene expression programs (10, 11).

The first implication that histone arginine methylation has a role in transcriptional activation by nuclear receptors was the discovery that coactivator-associated arginine methyltransferase 1 (CARM1) was recruited to steroid activated nuclear receptors by the SRC family protein, glucocorticoid receptor-interacting protein 1 (GRIP1) (12). CARM1 methylates histone H3 at arginines 2, 17, 26, and 42, and these modifications are associated with activation of transcription (12–14). H3R17 methylation recruits Tudor domain-containing protein 3, providing a scaffold for further recruitment of topoisomerase IIIB (15, 16). The Tudor domain-containing protein 3-topoisomerase IIIB complex was shown to relax negative supercoiling and reduce formation of R-loops (16). CARM1 also functions as a coactivator for other transcription factors such as p53, nuclear factor-kappa B, β-catenin, and c-Myb (17–20).

Posttranslational modifications by protein arginine methyltransferases (PRMTs) were found not to be limited to histone tail methylation but were extended to methylation of other cofactors as well (21). CARM1-dependent methylation of p300/cAMP-response element binding protein binding protein (CBP) and SRC proteins affects the activity and stability of the coactivator complex (22). A role for CARM1-dependent methylation of RNA-binding proteins HuR and HuD has also been described for mRNA splicing (23–25). In addition, methylation of the SWI/SNF subunit BAF155 was shown to enhance gene expression (26). These nonhistone substrates of CARM1 expand the role of arginine methylation in transcriptional regulation.

Methyltransferase-independent functions by CARM1 have also been observed. CARM1 contains a unique C-terminal transcriptional activation domain (AD) (27, 28). This region interacts with transcription-intermediary factor 1α and enhances nuclear receptor-mediated transcription (29). Protein-protein interactions between CARM1 and Flightless I were also shown to enhance transcription (30). Taken together, these studies reveal the interdependency of CARM1 and other transcriptional coactivators. Further exploration of CARM1-interacting proteins is required to better define the unique transcriptional regulatory mechanisms of CARM1.

Here, we demonstrate the recruitment of the E3 ubiquitin ligase DAZ (deleted in azoospermia)-interacting protein 3 (DZIP3) by CARM1 for activation of estrogen-dependent transcription. We find that DZIP3 is required for induction of 2 well-characterized ERα target genes. This work expands the context of CARM1-dependent enhancement of gene expression in breast cancer cells, and reveals a novel transcriptional coactivator function of DZIP3.

Materials and Methods

Yeast two-hybrid screening

Full-length CARM1 cDNA was cloned into EcoRI sites downstream of the Gal4DBD in pGBT9 (Clontech) to generate a bait plasmid (pGBT9.CARM1). The yeast strain HF7c (containing his3 and lacZ genes controlled by Gal4-responsive elements) was sequentially transformed with the pGBT9.CARM1 plasmid and a 17-day mouse embryo cDNA library in pGAD10 (Clontech). Transformants (10⁶) were first plated on synthetic complete media plates lacking leucine and tryptophan, incubated until colonies appeared, and harvested. The amplified transformants (∼10⁷) were plated onto synthetic complete media lacking histidine, leucine, and tryptophan and containing 50mM 3-amino-1,2,4-triazole to suppress low-level expression of HIS3. This selection resulted in 5 positive clones of DZIP3 containing various C-terminal fragments.

Plasmids

Mammalian expression vectors encoding ERα, androgen receptor (AR), MMTV(ERE)-LUC, MMTV-LUC, GRIP1, p300, CARM1, GRIP1.N (amino acids [aa] 5–765), GRIP1.M (aa 563–1121), and GRIP1.C (aa 1122–1462) were described previously (31). Mammalian expression vectors for GK1, pM, pVP16, pM.GRIP1, pM.GRIP1.N (amino acids 5–765), pM.GRIP1.C (amino acids 1122–1462), and pVP16.CARM1 were also described previously (29). DZIP3 expression vectors were cloned with hemagglutinin A (HA) or FLAG epitope tags, encoded by pSG5.HA or pSG5.FLAG vectors containing SV40 and T7 promoters for expression in human cells or in vitro transcription/translation.

Cell culture and transfections

DMEM with 10% fetal bovine serum (FBS) was used to maintain CV-1, Cos-7, HEK293T, CARM1 knockout mouse embryonic fibroblasts (KO MEFs), and MCF-7 cells at 37°C and 5% CO₂.

CV-1 cells or CARM1 KO MEFs were maintained in DMEM supplemented with 10% FBS at 37°C and 5% CO₂. Cells were seeded at 1 × 10⁵ cells/well in 12-well plates the day before transfection. Transfection of expression plasmids was performed using BioT transfection agent (Bioland) according to the manufacturer's protocol. After transfection, the cells were grown in phenol red-free DMEM supplemented with 5% charcoal-stripped FBS for 48 hours in the presence or absence of 100nM 17β-estradiol (E2) for ER or 20nM dihydrotestosterone for AR. For mammalian 2-hybrid assays, CV-1 cells were mainintained in DMEM supplemented with 10% FBS. Cell lysis and luciferase assays on cell extracts were performed with Promega luciferase assay kit. The empty vector was used as controls for transfections of coactivator plasmids, and equal total DNA amounts were used in all samples in a given experiment. To control for variations in transfection efficiency, multiple independent transfection experiments were performed, and multiple plasmid preps for key plasmids were tested in these assays. The results are presented as the mean ± range of variation of 2 transfected wells from a single experiment, representative of at least 3 independent experiments. Paired 2-tailed t test was performed where indicated; *, P < .05; **, P < .01; ***, P < .001.

Protein-protein interactions

Glutathione S-transferase (GST) fusion proteins were produced in Escherichia coli strain BL21 by standard methods using glutathione agarose (Sigma) affinity chromatography. CARM1 and GRIP1.C were cloned into the vector pGEX4T1 (Amersham Biosciences) for protein expression and purification. HA-DZIP3 was synthesized in vitro by transcription and translation using the TNT-T7 coupled reticulocyte lysate system (Promega). GST pull-down assays were performed as described previously (see reference 31 below). For coimmunoprecipitation (co-IP), Cos-7 cells were seeded 1 × 10⁶ cells/10-cm dish the day before transfection. Transfection of 2.5 μg of expression vectors was performed with Bio-T (Bioland) according to the manufacturer's protocol. Whole-cell extracts were made in 1 mL of radioimmunoprecipitation assay buffer (50mM Tris-HCl [pH 8.0], 150mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, and 0.1% sodium dodecyl sulfate). co-IP and immunoblotting were performed as described previously (32). The lysate was precleared by incubation with agarose protein A/G beads for 1 hour and then immunoprecipitated overnight with 1–2 μg of the indicated antibodies and agarose protein A/G beads. Immunoblotting was performed with indicated antibodies. For endogenous co-IP, confluent HEK293T cells maintained in DMEM supplemented with 10% FBS were used. Whole-cell extracts were prepared from 1 mL of radioimmunoprecipitation assay buffer and immunoprecipitated with 2 μg of anti-DZIP3 antibody (Y18; Santa Cruz Biotechnology, Inc) and agarose protein A/G beads overnight at 4°C. Immunoblotting was performed with anti-CARM1 antibody (Millipore).

DZIP3 small interfering RNA (siRNA) depletion and quantitative real-time-PCR

For depletion of DZIP3, MCF7 cells were cultured in hormone-free medium (phenol red-free DMEM supplemented with 5% charcoal-stripped serum) for 24 hours. Cells were then transfected with siRNA targeting DZIP3 (from Dharmacon or Gene Pharma) or nontargeting siRNA control using lipofectamine RNAimax (Invitrogen) according to the manufacturer's specifications (forward transfection at 40nM, or reverse and forward transfections at 20nM and 40nM, respectively). Cells were then cultured in hormone-free medium for a total of 72 hours before being treated with E2 or ethanol for 16 hours. Total cellular RNA was prepared using TRIzol reagent (Invitrogen). Reverse transcription of 0.9 μg of total RNA from each sample was performed with iScript (Bio-Rad) according to the manufacturer's specifications. Real-time PCR amplification of this cDNA was performed on a StepOnePlus (Life Technologies) using KAPA SYBR FAST quantitative PCR (qPCR) master mix (Kapa Biosystems) and specified primers. All mRNA levels were normalized to that of 18s mRNA. The primer sequences for qPCR are as follows: human DZIP3, 5′-GGCATCTGCTTTATCCTCCTA-3′ (forward) and 5′-GGGTGCAGGGAACACATTTA-3′ (reverse); human 18s, 5′-GAGGATGAGGTGGAACGTGT-3′ (forward) and 5′-TCTTCAGTCGCTCCAGGTCT-3′ (reverse). Primer sequences for GREB1 and TFF1 were described previously (33).

siRNA sequences are as follows (from Gene Pharma or Dharmacon): siDZIP3 #1, 5′-GCCUGGAUGAAUUGCAUAUdTdT-3′ (sense) and 5′-AUAUGCAAUUCAUCCAGGCdTdT-3′ (antisense). Other siDZIP3 sequences that were used: siDZIP3 #2, 5′-GACCGAAGAUCAGUUUAAAdTdT-3′ (sense) and 5′-UUUAAACUGAUCUUCGCUCdTdT-3′ (antisense); siDZIP3 #3, 5′-GGGCUCAGCUGGCAAAGUAdTdT-3′ (sense) and 5′-UACUUUGCCAGCUGAGCCCdTdT-3′ (antisense); and siControl (negative control siRNA having no perfect matches to known human or mouse genes), 5′-UUCUCCGAACGUGUCACGUdTdT-3′ (sense) and 5′-ACGUGACACGUUCGGAGAAdTdT-3′ (antisense).

Chromatin IP (ChIP) assays

ChIP assays were performed as described previously (31). Briefly, MCF-7 cells were cultured in phenol red-free DMEM supplemented with 5% charcoal-stripped FBS for 3 days before addition of E2. After exposure to 100nM E2 for the indicated time, cells were fixed in formaldehyde for 10 minutes before addition of glycine solution to stop cross-linking. After washing with PBS, cells were pelleted by centrifugation and resuspended in hypotonic buffer before being passed through a 25-gauge needle 5 times. Chromatin was pelleted and resuspended for sonication by a Branson Sonifier level 3 for 10-second sonication increments until chromatin fragment sizes were less than or equal to 0.5 kb as determined by agarose gel electrophoresis. Sonicated chromatin was diluted 1:10 and precleared by incubation with agarose protein A/G beads for 1 hour. IP was performed with the indicated antibodies by incubation overnight at 4°C. Agarose protein A/G beads were added to the incubation for 2 hours, and the chromatin-antibody bound fraction was precipitated overnight at −80°C. Washing and elution of the beads was followed by overnight incubation at 65°C to reverse cross-linking. DNA purification was performed by extraction with phenol and chloroform followed by ethanol precipitation, and the DNA was analyzed for qPCR. For Re-ChIP assays, the initial immunoprecipitated complexes from the ChIP procedure were eluted by incubation with 10mM dithiothreitol at 37°C for 30 minutes and diluted 1:50 in IP dilution buffer. The eluates were reimmunoprecipitated with the indicated second antibodies. Primer sequences for the amplification of the GREB1 ERE1 were previously published as forward 5′GTGGCAACTGGGTCATTCTGA and reverse 5′CGACCCACAGAAATGAAAAGG (34). DZIP3 antibodies used were as follows: 6 μg (Y18; Santa Cruz Biotechnology, Inc) or 4 μg (ab56657; Abcam). Other antibodies include ERα (HC20; Santa Cruz Biotechnology, Inc), CARM1 (Millipore), and mouse IgG (Santa Cruz Biotechnology, Inc). The signal from the immunoprecipitated DNA was normalized to the signal from 1% of the input fraction of DNA.

Results

DZIP3 interacts with CARM1

To identify novel CARM1-interacting proteins that potentially affect its function, a yeast two-hybrid screen was performed using CARM1 as a bait protein. A mouse 17-day embryo cDNA library (Clontech) was screened, and an E3 ubiquitin ligase, DZIP3, containing amino acid residues from 612 to 1208, was isolated as an interacting partner (Figure 1A). To validate the interaction between DZIP3 and CARM1 detected by the yeast two-hybrid assay, GST pull-down and co-IP assays were performed. An in vitro-translated DZIP3 fragment (aa 612–1208) interacted directly with GST-CARM1 but not with GST alone (Figure 1B). A co-IP assay was performed to confirm the interaction in vivo (Figure 1C). Flag-tagged full-length DZIP3 and HA-tagged full-length CARM1 were overexpressed in Cos-7 cells and immunoprecipitated with anti-Flag antibody. The IP was then immunoblotted with anti-HA antibody. CARM1 specifically interacted with DZIP3 but not with an IgG antibody control. We also confirmed the interaction between DZIP3 and CARM1 at endogenous level by co-IP (Figure 1D).

Figure 1.

CARM1 interacts with E3 ubiquitin ligase DZIP3. A, Schematic representation of DZIP3 domains. B, HA-tagged DZIP3 612–1208 was synthesized in vitro and incubated with GST or GST-CARM1 bound to glutathione-agarose beads. The bead bound fraction was analyzed by SDS-PAGE and immunoblotted. C and F, Cos-7 cells were transfected with expression vectors (2.5 μg of each) encoding the indicated FLAG or HA fusion proteins. After 48 hours, the cells were harvested and the extracts subjected to IP with anti-FLAG or anti-HA antibody followed by immunoblot with anti-HA or anti-Flag antibody. D, Endogenous co-IP. Cell extracts from HEK293T cells were immunoprecipitated with anti-DZIP3 antibody followed by immunoblotting with anti-CARM1 antibody. The location of the immunoprecipitated CARM1 is denoted by the *. E, Schematic representation of DZIP3 fragments. G, Immunoblot showing expression levels of HA-DZIP3 deletion mutants from F.

DZIP3 contains a central lysine-rich region (amino acids 656–674), a coiled-coil domain (aa 794–852), and a C-terminal really interesting new gene (RING) domain (aa 1144–1183), which harbors the ubiquitin ligase activity. DZIP3 deletion mutants were constructed to define the interaction domains with CARM1 (Figure 1, E–G). The indicated fragments of HA-tagged DZIP3 were cotransfected in Cos-7 cells with Flag-tagged full-length CARM1. IP of DZIP3 fragments with anti-HA antibody, and subsequent immunoblotting with anti-Flag antibody revealed the association of the C-terminal 913–1208 RING domain of DZIP3 with CARM1 (Figure 1F). The protein expression levels of DZIP3 deletion mutants are shown in Figure 1G.

DZIP3 is a coactivator of ERα

CARM1 is recruited to ligand activated nuclear receptors, including ERα, by the p160 SRC proteins (12, 35). In contrast, DZIP3 has been shown to repress transcription of a subset of chemokine genes in macrophages through monoubiquitination of histone H2A (36). To our knowledge, no role has been described for DZIP3 in ERα signaling. To test the effect of DZIP3 on ERα-mediated transcription, we used previously established conditions, where ERα, GRIP1, CARM1, and/or p300 were cotransfected, to observe synergistic effects of multiple coactivators (12, 28, 31–33, 35). CV-1 cells with an estrogen-responsive element (ERE) containing luciferase reporter were transiently transfected with various combinations of ERα and its coactivators. After induction with estradiol, reporter gene activation was enhanced by GRIP1, and the addition of CARM1 further enhanced the reporter activity (Figure 2A, lanes 3–4). DZIP3 alone showed minimal coactivator activity for ERα, but it cooperated strongly with GRIP1. Furthermore, the combination of DZIP3, GRIP1, and CARM1 was highly synergistic, activating transcription 18-fold greater than that achieved with GRIP1 and CARM1 (P < .001) (Figure 2A, compare lane 4 vs 13). This synergy was estradiol (E2) dependent, because in the absence of E2, minimal luciferase activity was observed. This synergistic activation of CARM1 and DZIP3 was replicated in an androgen-dependent system, thus showing that DZIP3 functions as a coactivator for multiple nuclear receptor-dependent signaling pathways (Figure 2D).

Figure 2.

DZIP3 is a coactivator for ERα. A–D, CV-1 cells were transfected with MMTV(ERE)-LUC reporter plasmid (125 ng) and expression vectors encoding ERα (1 ng), GRIP1(50 ng), CARM1 (200 ng), p300 (200 ng), and DZIP3 (100, 200, and 400 ng) as indicated. Cells were incubated for 48 hours in the presence or absence of estradiol (E2), and luciferase assays were performed. For D, MMTV-LUC reporter plasmid (125 ng) and AR (1 ng) expression vectors were used, and dihydrotestosteron was added for 48 hours. Results shown are the mean ± range of variation of 2 transfected wells from a single experiment, representative of 5 independent experiments. ***, P < .001 (paired 2-tailed t test).

Previously, it was shown that CARM1 cooperates with the histone acetyltransferase, p300, to enhance ERα transcriptional activity (35, 38). DZIP3 also cooperated with p300 to enhance ER-driven reporter activity (Figure 2B, lanes 6–7). Furthermore, a strong synergy was observed among CARM1, p300, and DZIP3 in the presence of GRIP1 (Figure 2C, lanes 6–8).

Because DZIP3 was able to cooperate with GRIP1 to enhance ER-driven reporter activity (Figure 2A, lanes 8–10), we explored the possibility that DZIP3 can act as a coactivator in the absence of CARM1 and whether DZIP3 can interact with GRIP1. In CARM1 KO embryonic fibroblasts (CARM1 KO MEFs), DZIP3 was able to activate the ER reporter with GRIP1 (Figure 3A, compare lane 3 vs 5). Furthermore, GRIP1, DZIP3, and p300 enhanced ERα transcriptional activity in an additive manner (Figure 3A, lane 8), whereas GRIP1, DZIP3, and CARM1 cooperated more strongly (Figure 3A, lane 7). In a co-IP assay, Flag-tagged DZIP3 interacted with HA-tagged GRIP1 but not with an IgG antibody control (Figure 3B). We further defined the interaction domains of GRIP1 by co-IP assay. The indicated GRIP1 fragments were cotransfected with Flag-tagged DZIP3 in HEK293T cells. IP of DZIP3 with anti-Flag antibody, and subsequent immunoblotting with anti-HA antibody indicated an interaction between DZIP3 and the C-terminal AD2 of GRIP1 (amino acids 1122–1462 of GRIP1) but not with the HA-GRIP1.N or HA GRIP1.M (Figure 3C). The protein expression levels of Flag-DZIP3 were similar in each of the co-IP conditions (Figure 3E). In addition, GST pull-down assays showed that an in vitro-translated DZIP3 fragment (aa 612–1208) interacted directly with GST-GRIP1.C but not with GST alone (Figure 3D).

Figure 3.

DZIP3 interacts with GRIP1. A, CARM1 KO MEFs were transfected with MMTV(ERE)-LUC reporter plasmid (125 ng) and expression vectors encoding ERα (1 ng), GRIP1(25 ng), CARM1 (200 ng), p300 (200 ng), and DZIP3(200 ng) as indicated. Cells were incubated for 48 hours in the presence or absence of estradiol (E2), and luciferase assays were performed. Results shown are the mean ± range of variation of 2 transfected wells from a single experiment, representative of 3 independent experiments. B and C, HEK293T cells were transfected with expression vectors (2.5 μg of each) encoding the indicated FLAG or HA fusion proteins. After 48 hours, the cells were harvested and the extracts subjected to IP with anti-FLAG antibody followed by immunoblot with anti-HA antibody. D, HA-tagged DZIP3 612–1208 was synthesized in vitro and incubated with GST or GST-GRIP1.C (amino acids 1122–1462) bound to glutathione-agarose beads. The bead bound fraction was analyzed by SDS-PAGE and immunoblotted with anti-HA antibody. E, Immunoblot showing expression levels of Flag-DZIP3 proteins from C.

Requirement of specific functional domains of CARM1 in coactivator synergy

To investigate the mechanisms of transcriptional synergy between CARM1 and DZIP3, we tested the effects of mutations or deletions of key functional domains of CARM1. CARM1 contains a central catalytic domain, which is highly conserved among PRMT family members, and contains the S-adenosyl-methionine and arginine-binding sites (Figure 4A). CARM1 also harbors a unique C-terminal AD (28). We first tested the role of the CARM1 methyltransferase activity by using CARM1 E256Q mutant (12, 32). The E256Q mutation is located in the arginine-binding pocket (Figure 4A). This mutation abolishes CARM1's enzymatic activity while maintaining its ability to bind GRIP1 but do not affect CARM1 protein levels (12, 32). The synergistic enhancement of ERα function by GRIP1, wild-type CARM1, and DZIP3 was partially lost when CARM1 E256Q mutant was substituted for wild-type CARM1 (P < .001) (Figure 4B, compare lane 6 vs 8), indicating that the enzymatic activity of CARM1 is partially required for its coactivator synergy. This suggests that domains other than the methyltransferase activity may also be important for the cooperativity between CARM1 and DZIP3.

Figure 4.

Role of CARM1 methyltransferase activity and its domains in coactivator synergy. A, Schematic representation of CARM1 domains indicating the location of CARM1 mutants and fragments tested. B and C, CV-1 cells were transfected with MMTV(ERE)-LUC reporter plasmid (125 ng) and expression vectors for ERα (1 ng), GRIP1 (50 ng), CARM1 (250 ng), CARM1 E256Q mutant, or CARM1 deletion mutants (200 ng), CARM1 461–608 (200 and 400 ng), and DZIP3 (250 ng) as indicated. The luciferase activities shown are from a single experiment, which is representative of 3 independent experiments. ***, P < .001 (paired 2-tailed t test). D, Cos-7 cells were transfected with expression vectors encoding FLAG.DZIP3 and either HA.CARM1 full length (fl), HA.CARM1 3–460, HA.CARM1 461–608, or HA.CARM1 3–200. Cell extracts were incubated with anti-FLAG antibody or IgG for IP and subsequently analyzed by SDS-PAGE and immunoblotted with anti-HA antibody. E, Immunoblot showing expression levels of HA-CARM1 deletion mutants from C. Numbers indicate corresponding cell extracts from C. F, CV-1 cells were transfected with GK1 reporter plasmid (250 ng) and expression vectors for pM, pVP16, pM.GRIP1, pM.GRIP1.N (amino acids 5–765), pM.GRIP1.C (amino acids 1122–1462), VP16.CARM1 at 125ng, and pSG5HA.DZIP3 at 200 ng. Cells were incubated for 48 hours, and luciferase assays were performed. Results shown are from a single experiment, which is representative of 3 independent experiments. G, Immunoblot showing expression levels of DZIP3 proteins from D.

We next tested CARM1 mutants deleted for various structural regions of the CARM1 protein, all of which are expressed at similar levels to wild-type CARM1 except CARM1 461–608 fragment, which is expressed at a slightly lower level compared with the wild-type CARM1 (Figure 4D) (28). None of the 3 CARM1 deletion mutants was able to cooperate with DZIP3 to activate a reporter gene fully, indicating the importance of each of CARM1's structural domains for its coactivator function. The CARM1 3–500 fragment, which does not contain the C-terminal transcriptional AD, showed some activity with GRIP1 but was not able to cooperate with DZIP3 (Figure 4C, compare lane 8 vs 9), suggesting that the C-terminal AD of CARM1 is important for the synergy between CARM1 and DZIP3. The CARM1 461–608 and 3–200 fragments also showed minimal activity with GRIP1 and DZIP3, indicating that the C-terminal AD or the N-terminal domain of CARM1 are insufficient for the transcriptional synergy between CARM1 and DZIP3. Because the CARM1 461–608 fragment was expressed at a lower level, we transfected higher amounts of CARM1 461–608 fragment for comparable expression. This fragment which contains the AD of CARM1, however, did not result in reporter activation (Figure 4C, lanes 10–11). The relative protein expression levels of CARM1 deletion mutants are shown in Figure 4E.

We then determined the domain of CARM1 responsible for interaction with DZIP3 by co-IP in Cos7 cells. The region spanning 3–460 containing the catalytic domain and 3–200 of CARM1 interacted with DZIP3, but the AD (aa 461–608) of CARM1 did not (Figure 4D). These findings support the notion that both the methyltransferase activity and the C-terminal AD of CARM1 contribute to the coactivator synergy between GRIP1, DZIP3, and CARM1.

Because DZIP3 also interacts with GRIP1 (Figure 3, B–D), we tested the possibility that DZIP3 may enhance the interaction between GRIP1 and CARM1 by using mammalian two-hybrid assays. Gal4 DBD fused to GRIP1 interacted with VP16 fused CARM1 (Figure 4F, lanes 8–9). Coexpression of DZIP3 further enhanced the interaction between Gal4DBD-GRIP1 and VP16-CARM1 (Figure 4F, lane 10) while having no effect on the interaction between Gal4DBD-GRIP1.N (amino acids 5–765) and VP16-CARM1 (Figure 4F, lanes 11–13). Coexpression of DZIP3 also enhanced the interaction between Gal4DBD-GRIP1.C (amino acids 1122–1462) and VP16-CARM1 (Figure 4F, lanes 14–16).

Role of DZIP3 E3 ubiquitin ligase domain in coactivator synergy

To determine whether the ubiquitin ligase activity of DZIP3 is required for its synergistic cooperation with CARM1 in ERα transcriptional activation, we used DZIP3 RING domain mutants (C1165A and C1187S) previously shown to abolish its ubiquitin ligase activity (Figure 1A) (39). The coactivator synergy observed between GRIP1, CARM1 and DZIP3 was maintained when DZIP3 harboring the 2 RING domain mutations was examined (Figure 5A). This indicates that the E3 ligase activity of DZIP3 is not absolutely required for coactivator synergy between CARM1 and DZIP3. We, however, cannot exclude the contribution of endogenous wild-type DZIP3 in this assay.

Figure 5.

DZIP3 ubiquitin ligase activity is not required for its coactivator activity. A and B, CV-1 cells were transfected with MMTV(ERE)-LUC reporter plasmid (125 ng) and plasmids encoding ERα (1 ng), GRIP1 (50 ng), CARM1 (200 ng), DZIP3 (100, 200, and 400 ng), catalytic mutants DZIP3 C1165A or DZIP3 C1187S, and PRMT1, PRMT2, PRMT3, or yeast RMT (200 and 400 ng). Cells were treated with 100nM E2, and luciferase assays were performed. Results shown are from a single experiment representative of 3 independent experiments.

To further define the specificity of the synergy between CARM1 and DZIP3, we tested several different arginine methyltransferases in transient transfection assays. The coactivator synergy between DZIP3 and CARM1 was not observed when mammalian arginine methyltransferases PRMT1 or PRMT2 were substituted for CARM1 (Figure 5B). PRMT2 showed slight activation, and PRMT3 and yeast RMT1 showed half of the activity achieved by DZIP3 and CARM1, indicating that PRMT3 and yeast RMT1 may have functional relationships with DZIP3. These results show enhanced specificity between CARM1 and DZIP3 among some of the arginine methyltransferases tested.

DZIP3 is required for the endogenous expression of ERα target genes

To determine the importance of DZIP3 in ERα-mediated gene expression, we measured the expression of known ERα target genes in DZIP3 depleted MCF7 breast cancer cells in the presence and absence of estradiol (40). The siRNA against DZIP3 (siDZIP3) reduced endogenous DZIP3 mRNA levels in MCF7 cells by 70%–80% compared with MCF7s treated with nonspecific siRNAs (siControl) (Figure 6A). The reduction in mRNA corresponded to decreased DZIP3 protein levels as detected by Western blot analysis (Figure 6B). Upon addition of E2, GREB1 and pS2/TFF1 mRNA levels were strongly induced (Figure 6, C and D). Induction was reduced in siDZIP3-treated MCF7 cells compared with siControl-treated cells (*, P < .05; **, P < .01) (Figure 6, C and D). This indicates that endogenous DZIP3 is required for efficient induction of GREB1 and pS2 genes in response to hormone.

Figure 6.

DZIP3 is necessary for efficient transcriptional activation by ERα and is recruited to estrogen responsive promoter of the GREB1 gene. A, C, and D, MCF7 cells were transfected with siDZIP3 #1 or nonspecific siRNA before overnight incubation with either ethanol or E2. DZIP3, GREB1, and pS2/TFF1 mRNA levels were analyzed by qRT-PCR and normalized to 18s mRNA levels. Results shown are the mean and SD of 3 PCRs performed with cDNA samples from a single experiment and are representative of 5 independent experiments. *, P < .05; **, P < .01 (paired 2-tailed t test). B, Protein levels of DZIP3 were assessed by immunoblot analysis with β-actin levels serving as loading control. E and F, MCF7 cells were treated with E2 for the indicated time. ChIP assays were performed with the indicated antibodies, and amplicons surrounding the GREB1 ERE were analyzed by qPCR. The results are shown as a percentage of the input chromatin and are the mean and SD of triplicate PCRs. Results shown are from a single experiment representative of 5 independent experiments. **, P < .01 (paired 2-tailed t test). G, ChIP/ReChIP assay. MCF7 cells were treated with E2 for the indicated time. ChIP assays were performed with the anti-CARM1 antibody first, followed by ReChIP with anti-DZIP3 antibody or control IgG antibody. Amplicons surrounding the GREB1 ERE were analyzed by qPCR. The results are shown as a fold change of E2+ vs E2- after normalizing to percentage of the input chromatin and are the mean and SD of triplicate PCRs. Results shown are from 3 independent experiments. **, P < .01 (paired 2-tailed t test).

DZIP3 occupancy of endogenous ERα regulated gene

To determine whether DZIP3 is recruited to ERα target genes, we performed ChIP assays. We observed estradiol-dependent occupancy of endogenous ERα, CARM1 and DZIP3 to the ERE I of GREB1, whereas there was minimal ChIP signal for an IgG control (**, P < .01) (Figure 6, E and F). In addition, ChIP and Re-ChIP assays demonstrate that CARM1 and DZIP3 are simultaneously enriched on the same promoter region of GREB1 (**, P < .01) (Figure 6G). These findings support a role for DZIP3 as a transcriptional coactivator for ERα.

Discussion

In this study, we investigated the interaction of the E3 ubiquitin ligase, DZIP3, and CARM1, and the potential role of DZIP3 in ERα activated transcription. We show that DZIP3 directly binds CARM1 and GRIP1 and that DZIP3 functions as a coactivator of ERα-dependent transcription. The estradiol-mediated expression of endogenous ERα target genes, GREB1 and pS2, was reduced by DZIP3 depletion. We also observed DZIP3 occupancy at the promoter of ERα target gene, GREB1. Taken together, our data indicate that DZIP3 is a coactivator for modulation of ERα-mediated gene expression in breast cancer cells.

DZIP3 has previously been shown to be recruited to the nuclear receptor corepressor (N-CoR) complex, where it monoubiquitinates H2A K119 (41). In these studies, the E3 ligase activity of DZIP3 was important for transcriptional repression of a subset of chemokine genes in RAW 264.7 monocyte/macrophage cells. However, the authors noted that repression of transcription by DZIP3 is both gene and enzyme specific, citing unpublished data that Ring1b had no effect on the same subset of chemokine genes (36). The authors also noted that, even though DZIP3 interacts with N-CoR and histone deacetylase 1/3, DZIP3 was not detected in other biochemically purified N-CoR complexes. They then suggested that the transcriptionally repressive activity of DZIP3 was tightly controlled and dependent upon a specific promoter context (36). Interestingly, studies in MCF7 cells of the well characterized ERα target gene, pS2/TFF1, have shown that N-CoR corepressor protein does not associate with the promoter during E2 stimulation (3).

Our work raises the question of how DZIP3 enhances transcription of ERα target genes. Several potential mechanisms for the coactivator function of DZIP3 include posttranslational modifications, protein-protein interactions, and noncoding RNA interactions.

Our data indicate that the E3 ligase activity of DZIP3 is not absolutely required for its coactivator function (Figure 5A). This suggests that the inactivation of the E3 ubiquitin ligase activity of DZIP3 may be a potential mechanism for switching DZIP3 from a corepressor to a coactivator. This may occur through posttranslational modifications or by protein-protein interactions. A recently described precedent for regulation of E3 ubiquitin ligase activity was shown by the identification of autism susceptibility candidate 2 as a cofactor in a subset of Polycomb repressor complex 1 (42). This noncanonical Polycomb repressor complex 1 was phosphorylated by casein kinase 2 to neutralize its repressive E3 ligase activity of RING1. Autism susceptibility candidate 2 also recruited p300 to activate transcription. Similarly, DZIP3 could undergo various posttranslational modifications such as phosphorylation or methylation that may impact its E3 ligase activity. Future studies will focus on identification of posttranslational modifications of DZIP3 that may regulate its ligase activity in the context of estrogen activated transcription.

In addition to posttranslational modifications, cell specific protein complexes or gene specific promoter contexts may determine DZIP3's activity. It was shown previously that Mi2α, a component of nucleosome remodeling and deacetylation complex, interacts with CARM1 and enhances transcription of c-Myb target genes in hematopoietic cells (20). The histone deacetylase 1 and methyl-CpG binding domain protein 3 (MBD3) components of the nucleosome remodeling and deacetylation complex no longer associated with CARM1 in the CARM1- Mi2α complex. Similarly, the complex of proteins with which DZIP3 is associated likely depends on cell type and could regulate its coactivator or corepressor activity. Furthermore, DZIP3 could employ additional domains to recruit other proteins for transcriptional activation. It could use its coiled-coil domain or its RING domain to recruit additional cofactors that may be important for its coactivator function.

The effect of DZIP3-mediated RNA binding on ERα-mediated transcription remains unexplored. Kreft and Nassal isolated DZIP3 as a Hepatitis B virus RNA-binding ubiquitin ligase (39). They also identified a unique lysine-rich RNA-binding region located centrally. Recently, it was discovered that the RNA-binding domain of DZIP3 was recruited by long noncoding RNA, HOX antisense intergenic RNA, for the targeting of Ataxin-1 for ubiquitination and degradation (43). It is known that CARM1 also methylates proteins involved in mRNA splicing and processing, and it remains to be determined whether the RNA-binding domains of DZIP3 play a role in the coactivator function of DZIP3 (25, 37, 44).

In summary, we provide evidence for a novel coactivator function of DZIP3 in cooperation with GRIP1 and CARM1 in ERα-mediated transcriptional activation. The activity of DZIP3 is likely regulated by the proteins that it interacts with, depending on different promoter contexts, adding another dimension of complexity to transcriptional regulation by this cofactor.