Epigenetic regulation of asymmetric cell division by the LIBR-BRD4 axis

Abstract

Asymmetric cell division (ACD) is a mechanism used by stem cells to maintain the number of progeny. However, the epigenetic mechanisms regulating ACD remain elusive. Here we show that BRD4, a BET domain protein that binds to acetylated histone, is segregated in daughter cells together with H3K56Ac and regulates ACD. ITGB1 is regulated by BRD4 to regulate ACD. A long noncoding RNA (lncRNA), LIBR (LncRNA Inhibiting BRD4), decreases the percentage of stem cells going through ACD through interacting with the BRD4 mRNAs. LIBR inhibits the translation of BRD4 through recruiting a translation repressor, RCK, and inhibiting the binding of BRD4 mRNAs to polysomes. These results identify the epigenetic regulatory modules (BRD4, lncRNA LIBR) that regulate ACD. The regulation of ACD by BRD4 suggests the therapeutic limitation of using BRD4 inhibitors to treat cancer due to the ability of these inhibitors to promote symmetric cell division that may lead to tumor progression and treatment resistance.

Graphical Abstract

Graphical Abstract.

Introduction

Asymmetric cell division (ACD) is a mechanism used by stem cells to maintain the number of progeny and also protect against cancer (1–6). In contrast, symmetric cell division (SCD) is used to expand stem cell numbers and is permissive for aneuploidy, leading to cancer formation (4–7). It has been shown that newly synthesized histones are segregated to daughter cells under ACD (8). However, the histone mark and the chromatin reader/adaptor that mediates this daughter cell-specific segregation of newly synthesized histones are largely unknown. Long noncoding RNAs (lncRNAs) are long (>200 nt) RNAs that lack protein-coding sequences (9,10). LncRNAs regulate crucial biological processes, including cell differentiation, lineage determination, organ development, tumorigenesis etc. (9–13). LncRNAs that are mechanistically involved in asymmetric cell division remain elusive.

The transcriptional co-regulator or chromatin reader that may mediate/accompany daughter cell-specific segregation of newly synthesized histones still remains unidentified. BRD4 has been shown to be a gene bookmark for post-mitotic transcriptional re-activation (14). Differentiating daughter cells may require faster transcriptional re-activation. Therefore, we tested the role of BRD4, a chromatin reader that binds to acetylated histones (15), in post-ACD gene re-activation and its segregation into daughter cells. Here we show that BRD4 regulates the process of ACD through its target, ITGB1. A lncRNA, LIBR (LncRNA Inhibiting BRD4) represses BRD4 levels through inhibiting the translation of BRD4. The results have implications on therapeutic resistance caused by BRD4 inhibitors (e.g. JQ1, etc).

Materials and methods

Cell culture

The cell lines used were described including the human beast adenocarcinoma cell lines, MDA-MB-231 and MDA-MB-468, human induced pluripotent stem (iPS) cell line, KYOU-DXR0109B (obtained from ATCC) and human embryonic stem (hES) cell, TW1 (obtained from BCRC). Human embryonic kidney 293T cell line was used in transient transfection experiments. MDA-MB-231, MDA-MB-468 and 293T cell lines were cultured in Dubelcco's modified Eagle's medium (Gibco, Thermo Fisher Scientific Inc.) containing 10% fetal bovine serum (FBS) in a 5% CO₂/95% air incubator. iPS and hES cell lines were maintained in mTeSR1 medium (Stemcell Technologies Inc.) on Geltrex (Thermo Fisher Scientific Inc.)-coated plate in a 5% CO₂/95% air incubator.

Plasmids and antibodies

The Flag epitope-tagged human BRD4 cDNA in pcDNA3 vector (pcDNA3-Flag-BRD4) was obtained from Prof. Cheng-Ming Chiang (University of Texas Southwestern Medical Center). The shRNA plasmids of BRD4 were obtained from Prof. Shu-Chun Teng (National Taiwan University). The shRNA plasmids of ITGB1 were obtained from National RNAi Core Facility (NRC), Academia Sinica. The shRNA plasmids of lncRNA 889–003 and LIBR were designed and synthesized by TOOLSilent shRNA plasmid set in pGLV2/U6/Puro. The sequence information of these shRNA plasmids was shown (Supplementary Table S1). The BRD4 primary antibody was obtained from Prof. Cheng-Ming Chiang (University of Texas Southwestern Medical Center). The information of the other antibodies was shown (Supplementary Table S2).

Paired-cell assay

Cells were plated at a very low density (1.5 × 10⁴ cells in 12-well plate) on coated coverslip, and synchronized through a thymidine–nocodazole–blebbistatin (Sigma) procedure (Figure 1A) to control cell division for entering second mitosis and paired-cell formation. The paired cells were fixed for immunofluorescence staining.

Figure 1.

Regulation of ACD by BRD4. (A) The workflow represented the schedule of cell synchronization and analysis of paired cells. (B) The representative images of the paired-cell assay of MDA-MB-231 cells. DNA, blue; CD44, green; Numb, red; ACD, asymmetric cell division; SCD, symmetric cell division. Scale bars, 20 μm. (C) The representative images of the paired-cell assay of hESC and MDA-MB-231 cells. DNA, blue; CD44, green; BRD4, red; ACD, asymmetric cell division; SCD, symmetric cell division. Scale bars, 20 μm. (D) Western blot analysis of BRD4 in hESC treated with 10% FBS-containing medium (FBS) or serum-free and feeder-free expansion medium for human pluripotent stem cells (mTeSR1 medium; cont.). (E) The quantification of the paired-cell assay in hESC maintained in mTeSR1 medium (cont.) or going through differentiation in 10% FBS-containing medium (FBS). n (total counted cells) = 100 and three independent experiments. (F) Western blot of BRD4 in MDA-MB-231 cells treated with 100 ng/ml Wnt3a, 100 ng/ml Wnt5a, or vehicle control (PBS). (G) The quantification of the paired-cell assay in MDA-MB-231 treated with 100 ng/ml Wnt3a, 100 ng/ml Wnt5a, or vehicle control (PBS). n (total counted cells) = 100 and three independent experiments. Data represent mean ± s.d. ^∗P < 0.05 (Student's t-test). (H) Western blot analysis of BRD4 in hESC-shSCR or hESC-shBRD4 cells maintained in mTeSR1 medium (cont.) or differentiating in 10% FBS-containing medium (FBS), respectively. (I) The quantification of the paired-cell assay in hESC-shSCR or hESC-shBRD4 cells maintained in mTeSR1 medium (cont.) or going through differentiation in 10% FBS-containing medium (FBS). n (total counted cells) = 100 and three independent experiments.

Immunofluorescence staining

For immunofluorescence staining, cells on glass coverslips were fixed with 4% paraformaldehyde and permeabilized with 0.5% Triton X-100. After washing with PBS, fixed cells or slides were blocked with blocking buffer (PBS, 0.1% Tween-20, with 3% normal goat serum) for 1 h and incubated with primary antibody diluted in blocking buffer overnight at 4°C. After washing 3 times with PBS for 10 min, the fixed cells were treated with the appropriate secondary antibody (Alexa Fluor 488-conjugated anti-mouse IgG or Alexa Fluor 594 anti-rabbit IgG) (Abcam) that was diluted in blocking buffer for 1 h at room temperature. Finally, the fixed cells were washed 3 times for 10 min with PBS, and their nuclei were counterstained, mounted and observed by using fluorescence microscope or confocal microscope.

Western blot analysis

For western blot analysis, whole cell lysates were prepared using RIPA lysis buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 0.5% sodium deoxycholate, 1% NP-40, 0.1% SDS and supplemented with protease inhibitor cocktail), The protein concentration was determined by BCA Protein Assay Kit (Thermo Fisher Scientific Inc.) and using bovine serum albumin (BSA) as a standard. After fractionation of proteins by SDS-PAGE, proteins were transferred to a nitrocellulose membrane. The membrane was blocked with 5% skim milk, then probed with primary and horseradish peroxidase (HRP)-conjugated secondary antibodies according to the manufacturer's protocol. Finally, proteins were visualized by using enhanced chemiluminescence (ECL).

Real-time quantification PCR

Total RNA was isolated using Trizol reagent (Invitrogen, Thermo Fisher Scientific Inc.) and treated with RNase-free DNase I according to manufacturer's recommendations. Single-stranded cDNA was synthesized by the SuperScriptTM IV First-Strand Synthesis System (Invitrogen, Thermo Fisher Scientific Inc.). Real-time PCR was performed on a Bio-Rad iCycler iQ PCR detection system according to the manufacturer's instructions. The sequences of primer pairs were shown (Supplementary Table S3). The 2^–ΔΔCt method of relative quantification was used to estimate the copy number of gene expression.

Lentiviral infection

Lentivirus containing short hairpin RNAs (shRNAs) expressed in a lentiviral vector (pLKO.1-puro or pGLV2-puro) were generated in 293T cells as previously described. Plasmids pLKO-shSCR, pLKO-shBRD4 and pLKO-shITGB1 were provided by National RNAi Core Facility of Academia Sinica, Taipei, Taiwan. Plasmid pGLV2-shSCR, pGLV2-shLIBR and pGLV2-sh889-003 were purchased from TOOLSilent shRNA plasmid set. For lentivirus production, 293T cells were transfected with lentiviral vectors expressing different shRNAs along with envelope plasmid pMD.G and packaging plasmid pCMVΔ8.91. Virus was collected 48 h after transfection. To prepare the knockdown cells, cells were infected with lentivirus for 24 h, and were selected with appropriate antibiotics.

Public ChIP-seq analysis

ChIP-seq data were downloaded from ENCODE project containing the accession number of HepG2 BRD4 (ENCFF921PQB) or hESC (H1) H3K56Ac (ENCFF600YNW). Peaks were annotated by Goldmine, an R package, using the gene table of Gencode V36 from UCSC genome browser (16).

Co-immunoprecipitation (co-IP) assay

The cell lysates were mixed with an equal amount of immunoprecipitation buffer (150 mM NaCl, 1% NP-40, 1% deoxycholate, 0.1% SDS and 50 mM Tris–HCl at pH 7.5). The primary antibody or IgG was added to the lysates, and the reactions were incubated at 4°C in a rotary device for 2 h, then added blocked Dynabeads protein A and incubated at 4°C in a rotary device overnight. The immunoprecipitates were washed three times with the same lysis buffer, mixed with 1 × Laemmli dye, boiled for 5 min, and applied for Western blotting.

Chromatin immunoprecipitation (ChIP) assay

Briefly, cells were cross-linked with 1% formaldehyde for 10 min and stopped by adding glycine to a final concentration of 0.125 M. Fixed cells were lysed with cell lysis buffer (50 mM Tris–HCl [pH 8.0], 0.5% SDS, 100 mM NaCl, 5 mM EDTA, and protease inhibitors) for 10 min at 4°C. Cells were pelleted by centrifugation and suspended in 3 ml of nuclei lysis buffer (50 mM Tris–HCl [pH 8.0], 0.1% SDS, 1% Triton X-100, 10 mM EDTA, 150 mM NaCl, and protease inhibitors). Cells were sonicated for 15 bursts of 30 s ON and 30 s OFF using a Bioruptor UCD-200 sonicator. For each immunoprecipitation, cell lysate was precleared by adding 50 μl of Dynabeads protein G at 4°C for 1 h. The supernatants were incubated at 4°C for 2 h with antibody, and incubated with Dynabeads protein G overnight at 4°C. Beads were successively washed with high salt washing buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris–HCl [pH 8.0], 500 mM NaCl) six times and TE buffer (10 mM Tris, 1 mM EDTA) and then eluted with elution buffer (1% SDS and 0.1 M NaHCO₃. 20 μl of 5 M NaCl). The elutes were incubated at 65°C for 5 h to reverse the cross-linking. After digestion with proteinase K, the solution was phenol/chloroform-extracted and ethanol-precipitated. The precipitated-DNA was suspended in water and used in real-time quantification PCR assay. The antibodies and primers used for qChIP assay was shown (Supplementary Tables S2 and S4).

Luciferase reporter assay

The sequences of the ITGB1 promoter reporter constructs were shown (Supplementary Table S5). The BRD4 5′UTR region was constructed into pGL3-control vectors. The sequence information of the BRD5 5′UTR region was shown (Supplementary Table S6). The reporter constructs were transfected into MDA-MB-231 cells using PolyJet™ Reagent (SignaGen Laboratories) according to the manufacturer's protocol. Firefly luciferase activity was measured by the Dual-Luciferase Assay System (Promega), and normalized by the levels of Renilla fluorescence.

RNA fluorescence in situ hybridization (RNA-FISH)

RNA FISH Probe Set for LIBR and BRD4 mRNA were designed by using Stellaris Designer v4.2 and synthesized them with Fluorescein or CaL Fluor Red 610 dye labeled from Biosearch Technologies, Inc. RNA-FISH assays were performed by using Stellaris® RNA FISH kits according to manufacturer's protocol. The sequences of the RNA-FISH probe for LIBR and BRD4 mRNA were shown (Supplementary Table S7).

RNA pull-down assay

The cDNA of full-length LIBR and truncated LIBR was cloned into pcDNA3.1 vector. Their sequences were shown (Supplementary Table S8). Biotin-labeled RNAs were in vitro transcribed using the Biotin RNA Labeling Mix (Roche) and T7 RNA polymerase (Roche) and were purified using an RNeasy Mini Kit (Qiagen). The pre-washed streptavidin bead C1 (Invitrogen) were incubated with Biotin-labeled RNAs in the RNA capture buffer (20 mM Tris–HCl, pH 7.5, 1 M NaCl, and 1 mM EDTA) for 30 min at room temperature. The RNA-captured beads were then washed once with NT2 buffer (50 mM Tris–HCl, pH 7.4, 150 mM NaCl, 1 mM MgCl₂, and 0.05% NP-40) and incubated with pre-cleared cell lysates diluted in NT2 buffer supplemented with 100 U/ml RNase Out, 50 U/ml Superase IN (Ambion), 2 mM dithiothreitol, 30 mM EDTA, and 0.02 mg/ml Heparin at 4°C for at least 6 hr with gentle rotation. The RNA complexes were washed twice with an NT2 buffer, an NT2 high-salt buffer (500 mM NaCl), another NT2 high-salt buffer (1 M NaCl), an NT2-KSCN buffer (750 mM KSCN), and PBS for 5 min at 4°C and finally eluted with 2 mM D-biotin in PBS. The eluted RNAs were isolated using Trizol reagent, and then applied for quantification RT-RCR by using BRD4-5′UTR-1 primers (Supplementary Table S3). All processes were performed in RNase-free conditions.

RNA antisense purification-RNA (RAP-RNA) assay

The LIBR LncRNA was captured by using pools of single-stranded DNA (ssDNA) biotinylated oligos (Supplementary Table S9). To generate the probe sets, we synthesized pools of oligos with 100 nt complementarity to the LIBR LncRNA. Cells were prepared in 15 cm plate, then were treated with 0.5 mg/ml AMT solution (+AMT) or PBS only (-AMT) and irradiated in a tissue culture dish for 7 min with long-wave ultraviolet light. The crosslinked RNA were isolated by using TRIzol and fragmented by sonication, then residual DNA were digested by DNase I. 2 mg of the resulting purified RNAs were incubated with 15 pmol of biotinylated ssDNA probe for purification in LiCl hybridization buffer (10 mM Tris–HCl, pH7.5, 1 mM EDTA, 500 mM LiCl, 1% Triton X-100, 0.2% SDS, 0.1% sodium deoxycholate and 4 M urea) at 55°C for 2 h, shaking at 1200 rpm., then streptavidin C1 beads were added and the mixture was incubated at 37°C for 30 min, shaking at 1200 rpm. After incubation, the samples were washed three times with low stringency wash buffer (0.1% SDS, 1% NP-40 and 4M urea in 1xSSPE buffer) and high stringency wash buffer (0.1% SDS, 1% NP-40 and 4 M urea in 0.1xSSPE buffer) at 58°C for 5 min. Finally, RNAs were eluted with RNase H (Invitrogen) in RNase elution buffer (50 mM Tris–HCl, pH7.5, 75 mM NaCl, 3 mM MgCl₂, then 0.125% N-lauroylsarcosine, 0.0025% sodium deoxycholate and 2.5 mM TCEP were added freshly) at 37°C for 30 min. The eluted RNAs were isolated with phenol–chloroform method and ethanol precipitation, and then applied for quantification RT-RCR by using BRD4-5′UTR-1, BRD4-5′UTR-2 and BRD4 primers (Supplementary Table S3).

Polysome analysis

1 × 10⁷ cells were prepared in 15 cm plate. Before harvest, cells were incubated in complete medium containing a final concentration of 100 μg/ml cycloheximide for 10 min at 37°C. Polysome lysates were prepared in polysome extraction buffer 20 mM Tris–HCl at pH 8, 5 mM MgCl₂, 140m M KCl, 0.5 mM DTT, 1% Trition X-100, 0.1 mg/ml cycloheximide and 0.5 mg/ml heparin), and 10 OD_260 nm amount of lysate was loaded on top of sucrose gradient (15–50%) and performed centrifugation at 210 000g for 3.5 h. After centrifugation, th polysome profile was detected by gradient fractionation system as instructed (Brandel, BR-188 Density Gradient Fractionation System). Each sample was collected into 8 fractions. A quarter of fraction was for protein precipitation and three quarters were for RNA purification. RNA purification was performed by using PureLink™ RNA Mini Kit.

Protein precipitation for sucrose removal

Briefly add 600 μl (4×) methanol to a 150 μl (1×) of a fraction. After thorough mixing 150 μl(1x) chloroform is added. Vortex then add 450 μl (3×) water, vortex again (appears cloudy white) and centrifuge immediately for 5 min at full speed in a microfuge. A white disc of protein should form between the organic layer (bottom) and the aqueous layer (upper). Discard the upper aqueous layer. Add 650 μl of methanol to the tube and invert 3 times. Spin for 5 min at full speed in a microfuge. Remove all liquid and allow the pellet to air dry. The precipitated protein pellet can be taken up in SDS PAGE sample buffer for Western blot assay.

RNA immunoprecipitation (RIP) assay

For immunoprecipitation (IP) of endogenous ribonucleoprotein (RNP) complexes from whole-cell extracts, the cells were lysed by using RIPA lysis buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 0.5% sodium deoxycholate, 1% NP-40, 0.1% SDS and supplemented with protease inhibitor cocktail). The supernatants were incubated with protein A Mag Sepharose Xtra beads (Cytiva) coated with antibodies that recognized RCK antibody (Santa Cruz Biotechnology) or with control IgG (Santa Cruz Biotechnology) overnight at 4°C. After the beads were washed with NT2 buffer (50 mM Tris–HCl, pH 7.4, 150 mM NaCl, 1 mM MgCl₂ and 0.05% NP-40), the complexes were incubated with 20 units of RNase-free DNase I (15 min at 37°C) and further incubated with 0.5 mg/ml Proteinase K (15 min at 55°C) to remove DNA and proteins, respectively. The RNPs isolated from the IP materials was further assessed by RT-qPCR analysis with LIBR and BRD4-5′UTR-1 primers (Supplementary Table S3).

Statistical analysis

Unless otherwise mentioned, all samples were assayed in triplicates. For in vitro analyses, each experiment was repeated at least three times. The error bars represented the standard deviation (SD). Student's t test was used to compare two groups of independent samples. The level of statistical significance was set at 0.05 for all tests.

Results

BRD4 regulates the process of ACD

At first, a standard ACD protocol was established (Figure 1A) and two markers (CD44, Numb) (17) showed their proper segregation in parental and daughter cells of MDA-MB-231 cell line, respectively (Figure 1B). We then tested the segregation of BRD4 during ACD. The result showed that BRD4 was segregated to daughter cells (in contrast to CD44 which was segregated in parental cells) under ACD in both stem cells (hESC, iPSC) and tumor cells (MDA-MB-231, MDA-MB-468) (Figure 1C, Supplementary Figure S1A, B). Treatment of stem cells with fetal bovine serum (FBS) increased the expression levels of BRD4 in both hES and iPS cells (Figure 1D and Supplementary Figure S1C). The percentage of cells undergoing ACD increased after treatment of stem cells with FBS (Figure 1E and Supplementary Figure S1D). The same observations were also shown in tumor cells (MDA-MB-231, MDA-MB-468) treated with Wnt3a (18,19) (Figure 1F, G and Supplementary Figure S1E, F). In contrast, knockdown of BRD4 decrease the percentage of stem cells or tumor cells undergoing ACD (Figure 1H, I, and Supplementary Figure S1G-L). These results indicate that BRD4 segregates to the daughter cells and its levels correlate with the percentage of cells undergoing ACD.

A BRD4 downstream target, ITGB1, regulates ACD

Since BRD4 binds to acetylated histones, we tested the type of acetylated histone that binds BRD4 and correlates with its segregation to daughter cells. Since histone 3 lysine 56 acetylation (H3K56Ac) mark is incorporated into newly synthesized histones (20), we first stained H3K56Ac and examined its segregation pattern under ACD. The results showed that H3K56Ac, but not acetylated histone 4 lysine 5 (H4K5Ac), segregated to daughter cells in stem cells (hESC, iPSC) and tumor cells (MDA-MB-231) (Figure 2A and Supplementary Figure S2A–E). Co-immunoprecipitation experiments showed that using antibodies against BRD4 or H3K56Ac pulled down H3K56Ac or BRD4, respectively (Figure 2B). These results indicate that BRD4 binds to H3K56Ac during ACD.

Figure 2.

ITGB1, a BRD4 target, regulates ACD. (A) The representative images of the paired-cell assay of hES cells. DNA, blue; CD44, green; H3K56ac, red; ACD, asymmetric cell division; SCD, symmetric cell division. Scale bars, 20 μm. (B) Immunoprecipitation–western blot to show the interaction between the BRD4 and H3K56Ac in MDA-MB-231 cells. (C) Venn diagram analysis showed the overlap between the genes whose promoters were bound by BRD4 (HepG2) and those whose promoters were deposited with H3K56Ac histone mark (hESC (H1)). Among the 3401 transcripts potentially regulated by BRD4 and H3K56Ac, ITGB1 was one of the overlapped genes. (D) Western blot analysis of BRD4 and ITGB1 in MDA-MB-231-shSCR and MDA-MB-231-shBRD4 cells. (E) The representative images of the paired-cell assay of hES cells. DNA, blue; CD44, green; ITGB1, red; ACD, asymmetric cell division; SCD, symmetric cell division. Scale bars, 20 μm. (F) Western blot analysis of BRD4 in hESC-shSCR or hESC-shITGB1 cells maintained in mTeSR1 medium (cont.) or going through differentiation in 10% FBS-containing medium (FBS), respectively. (G) The quantification of the paired-cell assay in hESC-shSCR or hESC-shITGB1 cells maintained in mTeSR1 medium (cont.) or going through differentiation in 10% FBS-containing medium (FBS). n (total counted cells) = 100 and three independent experiments. (H) The schematic representation of the ITGB1 promoter region and the amplified region in the ChIP assay is also indicated. (I) BRD4 qChIP analysis was performed in MDA-MB-231 cells expressing BRD4 and vector control. Data represent mean ± s.d. ^∗P < 0.05 (Student's t-test). (J) Immunoprecipitation–western blot analysis showed the interaction between the Numb and ITGB1 in MDA-MB-231 cells. (K) The representative images of the paired-cell assay of MDA-MB-231 cells. DNA, blue; Numb, green; ITGB1, red; ACD, asymmetric cell division; SCD, symmetric cell division. Scale bars, 20 μm.

In order to search for the downstream targets that may mediate BRD4-induced ACD, we overlapped the ChIP-seq results from antibodies against BRD4 and H3K56Ac (ENCODE project: ENCFF921PQB and ENCFF600YNW). Among the overlapping genes (Figure 2C), integrin subunit β1 (ITGB1) is linked to ACD since ITGB1 has been shown to be located asymmetrically during cell division (21,22). Ligand-independent ITGB1 regulates spindle orientation through aligning with spindles and has a function of apically orienting atypical PKC and dictates the directionality of the ensuing cell division (21,22). To confirm the role of ITGB1, Western blot analysis and qPCR assay showed that knockdown of BRD4 decreased the protein and RNA levels of ITGB1 in MDA-MB-231 cells (Figure 2D and Supplementary Figure S2F). BRD4 overexpression increased the protein levels of ITGB1 in MDA-MB-231 cells (Supplementary Figure S2G). The distribution of ITGB1 was localized to the daughter cells, similar to the location of BRD4 during ACD in hESC and MDA-MB-231 cells (Figure 2E and Supplementary Figure S2H). In addition, knockdown of ITGB1 in hES cells treated with FBS or MDA-MB-231 cells treated with Wnt3a decreased the percentage of cells undergoing ACD (Figure 2F, G, and Supplementary Figure S2I, J), indicating that ITGB1 is a crucial BRD4 target that mediates ACD. We further tested the binding of BRD4 to the promoter region of ITGB1. The result showed that the region of ITGB1 promoter that could bind BRD4 was mapped to the region of –336 to –407 bp (the strongest) and -948 to -1050 bp (less strong but still significant) upstream of the transcription start site of ITGB1 through BRD4 overexpression or knockdown experiments using MDA-MB-231 cells (Figure 2H, I, and Supplementary Figure S2K–N). In vitro reporter gene assays also showed the activation of a reporter gene construct driven by the ITGB1 promoter containing these two regions that could be activated by Wnt3a treatment or overexpression of BRD4 (Supplementary Figure S2O–R). The activity of this construct was decreased under knockdown of BRD4 (Supplementary Figure S2S). The levels of H3K56Ac were still higher in these two regions but were not subject to BRD4 regulation in MDA-MB-231 cells (Supplementary Figure S2T, U). Since Numb is segregated to daughter cells during ACD and ITGB1 aligns with atypical PKC and dictates the directionality of the ensuing cell division (17,22), we tested whether ITGB1 interacts with Numb. The results showed that ITGB1 interacted with Numb by co-immunoprecipitation experiments (Figure 2J). Immunofluorescence staining experiments also showed the co-localization of ITGB1 and Numb in MDA-MB-231 cells (Figure 2K), further supporting the role of ITGB1 in the regulation of ACD.

LncRNA LIBR regulates ACD by downregulating BRD4 levels

Since lncRNAs are capable of inhibiting the translation of certain mRNAs to decrease their protein levels (11–13), we wanted to search for lncRNAs that could regulate the protein levels of BRD4 through inhibiting protein translation. We first screened for the lncRNAs that have high sequence complementarity to the mRNA of BRD4. A group of 12 lncRNAs fulfilled this criteria (Supplementary Figure S3A). We further tested whether one of these lncRNAs could be regulated by Wnt3a as Wnt3a regulates ACD. The result showed that only two lncRNAs fulfilled the criteria of repression under Wnt3a treatment using MDA-MB-231 cells (Supplementary Figure S3B). We further knocked down these two lncRNAs individually and tested whether either one correlated with BRD4 expression. Only knockdown of lncRNA CTD-2036P10.5 (which is named LIBR: LncRNAInhibitingBRD4) increased the protein levels of BRD4 in MDA-MB-231 cells (Figure 3A, B, and Supplementary Figure S3C, D), indicating that LIBR most likely regulates the protein levels of BRD4. The genomic organization and expression levels of LIBR in different tissues were shown (Supplementary Figure S3E, F; see description of LIBR and its tissue expression pattern). Further experiments using Western blot analysis and immunofluorescence staining assays showed that knockdown of LIBR increased the BRD4 protein levels and increased the percentage of ACD in three cell lines (MDA-MB-231, iPSC, and hESC) (Figure 3C-E and Supplementary Figure S3G-O). Specifically, immunofluorescence staining of BRD4 in LIBR knockdown MDA-MB-231 cells (SCD versus ACD) showed the increased staining intensity of BRD4 in both SCD and ACD stages (Supplementary Figure S3I). The localization of LIBR was shown to be in the cytoplasm (Figure 3F and Supplementary Figure S3P). The specificity of LIBR and BRD4 mRNA signals by RNA-fluorescence in situ hybridization assay (RNA-FISH) was shown by knockdown of LIBR or BRD4, respectively (Supplementary Figure S3Q, R). The co-localization of LIBR and BRD4 mRNA was shown in cells not undergoing ACD (Supplementary Figure S3S). We further measured the copy number of LIBR and the results showed that ∼75 copies existed in MDA-MB-231 cells and the copy numbers were decreased to ∼34 copies under Wnt3a treatment (Figure 3G, H). The copy numbers were ∼60 copies in iPSC and decreased to ∼26 copies after FBS treatment (Supplementary Figure S3T, U). In addition, the localization of LIBR was opposite the localization of BRD4 by immunofluorescence staining of BRD4 and RNA-FISH of LIBR under ACD, consistent with the repression of BRD4 protein levels by LIBR in MDA-MB-231 and hES cells (Figure 3I and Supplementary Figure S3V). The localization of LIBR and CD44 was segregated into the parental cells under ACD in the two cell lines (MDA-MB-231, hESC) tested (Figure 3J and Supplementary Figure S3W). All the above results were consistent with the role of LIBR in regulating BRD4 levels.

Figure 3.

LncRNA LIBR regulates ACD through decreasing BRD4 levels. (A) RT-qPCR of LIBR in MDA-MB-231 cells treated with 100 ng/ml Wnt3a, 100 ng/ml Wnt5a, or vehicle control (PBS). (B) Western blot analysis of BRD4 and β-catenin in MDA-MB-231 treated with 100 ng/ml Wnt3a, 100 ng/ml Wnt5a, or vehicle control (PBS). (C) Western blot analysis of BRD4 in MDA-MB-231-shSCR and MDA-MB-231-shLIBR cells. (D) RT-qPCR of LIBR in MDA-MB-231-shSCR and MDA-MB-231-shLIBR cells. (E) The quantification of the paired-cell assay in MDA-MB-231-shSCR and MDA-MB-231-shLIBR cells. n (total counted cells) = 100 and three independent experiments. (F) RNA-FISH assay for staining LIBR in MDA-MB-231 cells treated with 100 ng/ml Wnt3a, 100 ng/ml Wnt5a or vehicle control (PBS). Scale bars, 10 μm. (G) Titration standard curve was used for the measurement of the copy number of LIBR per 500 000 cells. The red arrows represent the RT-qPCR value from samples of MDA-MB-231 cells treated with 100 ng/ml Wnt3a, 100 ng/ml Wnt5a, or vehicle control (PBS), respectively. (H) Measurement of the copy number of LIBR in MDA-MB-231 cells treated with 100 ng/ml Wnt3a, 100 ng/ml Wnt5a or vehicle control (PBS). Data are shown as mean ± SD, n = 3 in all cases. (I) The representative images of the paired-cell assay of MDA-MB-231 cells. DNA, blue; BRD4, red; LIBR, green; ACD, asymmetric cell division; SCD, symmetric cell division. Scale bars, 20 μm. (J) The representative images of the paired-cell assay of MDA-MB-231 cells. DNA, blue; CD44, red; LIBR, green; ACD, asymmetric cell division; SCD, symmetric cell division. Scale bars, 20 μm.

LIBR represses BRD4 protein levels by inhibiting its translation

We used a SV40-luciferease construct to further map the region in the mRNA of BRD4 that could bind LIBR. Since the complementarity between BRD4 and LIBR was located in the 5′UTR of BRD4, the 5′UTR of BRD4 was inserted between the SV40 promoter and the luciferase coding sequence (SV40-BRD4-Luc construct) (Figure 4A). First, knockdown of LIBR increased the SV40-BRD4-Luc activity using a transient transfection assay (Figure 4A, left lower panel). Further transfection assays transfecting the control SV40-Luc vs. SV40-BRD4-luc together with LIBR showed that expression of LIBR decreased the luciferase activity of SV40-BRD4-Luc, but not the control SV40-Luc, supporting that LIBR bound to the 5′UTR of BRD4 to inhibit the luciferase activity (Figure 4A, right lower panel). We further mapped the region in LIBR that could interact with the 5′UTR of BRD4 to decrease its protein levels. RNA pull down assays showed that the region of nucleotides 241–821 of LIBR interacted stronger with the 5′UTR of BRD4 (Supplementary Figure 4A, B). Indeed, transfecting the LIBR^241-821 versus LIBR^1-240 construct showed that the LIBR^241-821 construct significantly decreased the protein levels of BRD4 using MBD-MB-231 cells, similar to transfecting the full length LIBR (Supplementary Figure 4A, lower panel). Further mapping experiments showed that LIBR nt. 241–440 decreased the BRD4 protein levels in MDA-MB-231 cells and interacted stronger with the 5′UTR of BRD4 (Figure 4B, C) by using RNA pull down assay. Finally, we mapped the region in the 5′UTR of BRD4 that interacted with LIBR by using a RAP-RNA (RNA antisense purification-RNA) assay. The RAP-RNA assay enabled the comprehensive mapping of RNA-RNA interactions in living cells. UV-induced AMT crosslinking is the method available to directly investigate RNA–RNA interactions (23). RAP-RNA assays showed that the regions of both nt. 85–171 and nt. 126–184 of the 5′UTR of BRD4 interacted strongly with LIBR^241–821 (Figure 4D). The potential base-pairing formed by LIBR and 5′UTR of BRD4 was predicted by IntaRNA 2.0 (24). It was predicted that there are two putative complementary regions between LIBR and 5′UTR of BRD4 which could form multiple matching base pairs (Figure 4E and Supplementary Figure S4C). These results showed that the nt. 241–440 of LIBR interacted with the 5′UTR (nt. 150–186) of BRD4 to decrease its protein levels.

Figure 4.

LIBR decreased the percentage of BRD4 binding to polysomes. (A) Upper: schematic representation of reporter constructs used in transient transfection assays. Lower: relative activity of different reporters in MDA-MB-231-shLIBR cells versus MDA-MB-231-shSCR cells and MDA-MB-231-LIBR cells versus MDA-MB-231-cont cells.(B) Upper: schematic representation of full-length (FL) and five truncated mutants of LIBR in transient transfection assays. Lower: Western blot analysis of BRD4 in MDA-MB-231 overexpressed with full-length (FL) and five truncated mutants of LIBR. (C) Upper: RNA pull-down assay showed the interaction between BRD4 5′UTR regions and biotinylated full-length (FL) or five truncated mutants of LIBR. Lower: the corresponding RNA fragments were analyzed by electrophoresis on a 1.2% agarose gel. (D) RAP-RNA assays for mapping the interaction region of endogenous BRD4 mRNA by using LIBR probes in the cells expressing full-length LIBR, LIBR^1-240, LIBR^241-821 or control. (E) A potential base-pairing formed by BRD4 5′UTR and LIBR was predicted by IntaRNA 2.0. (F) RT-qPCR of LIBR in MDA-MB-231 cells overexpressing LIBR or control. (G) Upper: polysome-fractionated samples of MDA-MB-231 cells analyzed by OD254. Lower: Western blot of CBP80, eIF4E and RPL23A in series of polysome-fractionated samples. (H) Polysome profiles of MDA-MB-231 cells overexpressing LIBR or control. The polysomal distribution of the BRD4 (upper) and GAPDH (lower) mRNAs was determined by isolating the RNA from each fraction collected from a 15–50% sucrose gradient. (I) RIP analysis of the interaction of LIBR (upper) or BRD4 mRNAs (lower) with RCK in MDA-MB-231 cells overexpressing LIBR versus control. Data represent mean ± s.d. ^∗P < 0.05 (Student's t-test).

We further used the sucrose density gradient assay to analyze the binding of BRD4 to the polysome fractions in order to measure the efficiency of translation under LIBR overexpression (25–27) (Figure 4F–H). Polysome density was measured by OD254 and Western blot analysis was used to indicate the positions of CBP80, eIF4E, and RPL23A using MDA-MB-231 cells (Figure 4G). The percentage of BRD4 mRNAs binding to the polysomes were decreased under LIBR overexpression in MDA-MB-231 cells (Figure 4H, upper panel), whereas the percentage of GAPDH mRNAs binding to the polysomes did not show significant difference under LIBR overexpression (Figure 4H, lower panel). In addition, knockdown of LIBR increased the percentage of BRD4 mRNAs binding to the polysomes in MDA-MB-231 cells (Supplementary Figure S4D–F). Since RCK, a translation repressor, was shown to be recruited by a lncRNA to cause translation repression (26), we performed a RNA immunoprecipitation (RIP) assay and showed that anti-RCK antibodies pulled down ribonucleoprotein complexes (RNPs) containing more LIBR or BRD4 mRNAs, respectively in MDA-MB-231 cells overexpressing LIBR (Figure 4I). In contrast, the amount of LIBR or BRD4 mRNAs pulled down by anti-RCK antibodies were significantly less in MDA-MB-231 cells with LIBR knockdown (Supplementary Figure S4G). All the above results indicate that LIBR recruits RCK to suppress the translation of BRD4 through inhibiting the binding of BRD4 mRNAs to polysomes.

BRD4 inhibition by JQ1 blocks ACD

Although BRD4 inhibitors (I-BET or JQ1) have been proposed to treat acute myeloid leukemia (28,29), BET inhibitor resistance could emerge from leukemia stem cells (30,31). To test the relationship between BET inhibitor resistance and the mechanism of ACD regulation, we first examined whether knockdown of BRD4 could decrease the percentage of hESC going through ACD under the medium that favored embryoid body formation. The treatment protocol was shown (Figure 5A). The results showed that indeed knockdown of BRD4 decreased the percentage of hESC going through ACD (Figure 5B–D). We further treated hESC and examined the percentage of hESC going through ACD after JQ1 treatment. Western blot analysis showed the decreased BRD4 levels in hESC after JQ1 treatment (Figure 5E, a representative panel with hESC going through SCD or ACD was shown in Figure 5F). Furthermore, treatment of hESC with JQ1 decreased the percentage of hESC going through ACD (Figure 5G). Similar experiments were performed in JQ1-treated MDA-MB-231 cells that showed decreased BRD4 levels (Supplementary Figure S5A). Similar treatment of MDA-MB-231 cells stimulated with Wnt3a to promote ACD showed the decreased BRD4 levels under JQ1 treatment (Supplementary Figure S5B, C; a representative panel with MDA-MB-231 cells going through SCD or ACD was shown in Supplementary Figure S5D). Under this treatment condition, the percentage of MDA-MB-231 cells going through ACD was decreased under JQ1 treatment (Supplementary Figure S5E), indicating that BRD4 inhibition by JQ1 could promote symmetric cell division that may lead to tumor progression.

Figure 5.

Treatment of hESC and EBs with BRD4 inhibitor, JQ1, decreased the percentage of cells going through ACD. (A) The workflow represented the schedule of embryoid bodies (EBs) production, cell synchronization and analysis of paired cells. (B) Western blot of BRD4 in hES cells and embryoid bodies under knockdown of BRD4. (C) The representative images of the paired-cell assay of hES cells and embryoid bodies under knockdown of BRD4. DNA, blue; CD44, green; BRD4, red. (D) The quantification of the paired-cell assay in hES cells and embryoid bodies under knockdown of BRD4. n (total counted cells) = 40 and three independent experiments. Data represent mean ± s.d. (E) Western blot of BRD4 and in hESC cells and EBs treated with 100 nM JQ1, 200 nM JQ1 or vehicle control (DMSO). (F) The representative images of the paired-cell assay of hESC cells and EBs treated with 100 nM JQ1, 200 nM JQ1 or vehicle control (DMSO). DNA, blue; CD44, green; BRD4, red; ACD, asymmetric cell division; SCD, symmetric cell division. (G) The quantification of the paired-cell assay in hESC cells and EBs treated with 100 nM JQ1, 200 nM JQ1 or vehicle control (DMSO). n (total counted cells with two independent experiments) = 47, 48, 42, 35, 45 and 27, respectively.

Discussion

Asymmetric cell division is a cellular biological process (1–3) and whether there is any epigenetic mechanism that regulates ACD remains unknown. In this report, we demonstrated the role of LIBR-BRD4 signaling axis in the regulation of ACD, highlighting the unique epigenetic mechanism of ACD regulation. The segregation of H3K56Ac histone mark and ITGB1 in daughter cells during ACD further delineated the mechanistic process of ACD. The specific binding of BRD4 to H3K56Ac during ACD needs to be further delineated mechanistically.

It has been shown that the Mrhl lncRNA is down regulated by Wnt3a through the occupancy of β-catenin at the TCF4 binding site located in the upstream promoter region of Mrh1 lncRNA that is dependent upon the recruitment of the co-repressor Ctbp1 at the promoter region (32). Therefore, LIBR downregulation by Wnt3a might be regulated by TCF4. Indeed, an online tool, PROMO, was used to identify putative TCF4 binding sites site on the promoter region of LIBR (33). We found that there are several putative TCF4 binding sites site inside the 1kb upstream promoter region of LIBR (–271∼–277; –361∼–367; –665∼–671 upstream of the transcription start site), indicating the possible regulation of LIBR by Wnt3a through TCF4.

ITGB1 has been shown to be located asymmetrically during cell division (21,22). Ligand-independent ITGB1 regulates spindle orientation (through aligning with spindles) (21). ITGB1 has a function of apically orienting atypical PKC and dictates the directionality of the ensuing cell division (22). Therefore, the asymmetric distribution of ITGB1 would coincide with the asymmetric distribution of Numb and regulates ACD. The function of ITGB1 may be similar to PON that is a partner of Numb and also regulates ACD (34). Increased activated-Itgb1 also increases asymmetric cell division of EpSCs (35). All the results described are consistent with our observation in this report.

The recruitment of RCK by LIBR to target mRNA (i.e. BRD4 mRNAs) causes the decreased binding of BRD4 mRNAs to the polysome fractions similar to the results that lncRNA-p21 recruits Rck to enhance its inhibition of JUNB and CTNNB1 translation, causing the decrease in the size of polysomes associated with JUNB and CTNNB1 (25). Therefore, binding of LIBR to BRD4 mRNAs recruits RCK to decrease BRD4 translation through decreasing the binding of BRD4 mRNAs to the polysome fractions (Figure 4G, upper panel).

The ACD process is used by stem cells to protect against cancer (1–6). From our findings, the targeting of BRD4 by BET inhibitors (e.g. JQ1) may decrease their percentage of cells going through ACD (28,29) and should caution the therapeutic usage of BET inhibitors to treat cancer due to the ability of these inhibitors to promote symmetric cell division that may lead to tumor progression and therapeutic resistance to BET inhibitors (30,31). Although our results present a unique epigenetic mechanism that may explain the therapeutic resistance induced by BET inhibitors, the findings would need to be confirmed in certain types of tumors that receive BET inhibitors. Clinical correlations will further strengthen the significance of this epigenetic control of ACD.

Data availability

ChIP-seq datasets used for overlapping gene search were referenced to the ENCODE project under accession numbers ENCFF921PQB and ENCFF600YNW. The authors declare that the data supporting the findings of this study are available within the manuscript. No restriction on data availability applies. All other data supporting the findings of this study are available from the corresponding author upon reasonable request.

Supplementary data

Supplementary Data are available at NAR Online.

Funding

National Science and Technology Council Special research fellow grant [MOST 108-2320-B-182A-022 in part to K.J.W.]; Frontier grants (MOST 108-2321-B-182A-005, MOST 109-2326-B-182A-002, MOST 110-2326-B-182A-004, MOST 111-2326-B-182A-002, NSTC 112-2326-B-182A-004]; Chang Gung Memorial Hospital [OMRPG3I0011, OMRPG3I0012, OMRPG3I0013, NMRPG3J0671, NMRPG3J0672, NMRPG3J0673, CORPG3J0231, CORPG3J0232 to K.W.H. (the ‘Drug Development Center, China Medical University’ from The Featured Areas Research Center Program within the framework of the Higher Education Sprout Project (Ministry of Education, Taiwan)]. Funding for open access charge: National Science and Technology Council of Taiwan.

Conflict of interest statement. None declared.