BET inhibitor suppresses migration of human hepatocellular carcinoma by inhibiting SMARCA4

Hepatocellular carcinoma (HCC) is one of the most prevalent and poorly responsive cancers worldwide. Bromodomain and extraterminal (BET) inhibitors, such as JQ1 and OTX-015, inhibit BET protein binding to acetylated residues in histones. However, the physiological mechanisms and regulatory processes of BET inhibition in HCC remain unclear. To explore BET inhibitors’ potential role in the molecular mechanisms underlying their anticancer effects in HCC, we analyzed BET inhibitor-treated HCC cells’ gene expression profiles with RNA-seq and bioinformatics analysis. BET inhibitor treatment significantly downregulated genes related to bromodomain-containing proteins 4 (BRD4), such as ACSL5, SLC38A5, and ICAM2. Importantly, some cell migration-related genes, including AOC3, CCR6, SSTR5, and SCL7A11, were significantly downregulated. Additionally, bioinformatics analysis using Ingenuity Knowledge Base Ingenuity Pathway Analysis (IPA) revealed that SMARCA4 regulated migration response molecules. Furthermore, knockdown of SMARCA4 gene expression by siRNA treatment significantly reduced cell migration and the expression of migration-related genes. In summary, our results indicated that BET inhibitor treatment in HCC cell lines reduces cell migration through the downregulation of SMARCA4.


Differentially expressed genes (DEGs) of BET inhibitor-treated HCC cells. Based on the results
shown in Fig. 1, we treated HepG2 cells with BET inhibitors (JQ1 or OTX-015; 5 µM) for 24 h in cDNA library preparation for RNA-seq experiment. RNA-seq transcriptional analysis was performed using three independent samples (biological replicates) of BET inhibitor treatment. We sequenced nine libraries obtained from 24 h control (3 samples), JQ1 (5 µM) (3 samples), and OTX-015 (5 µM) (3 samples) treatments. We combined the data from all experiments for each group, and the genes identified whose expression levels significantly differ. We used a 1% FDR, P < 0.05, fold change log 2 -fold change ≥ 2, log 2 -fold change ≤ − 2 for up-or downregulation, respectively, for defining DEGs. RNA-seq analysis revealed DEGs in BET inhibitor-treated HepG2 cells at 24 h: 627 genes in JQ1-treated HepG2 cells and 605 genes in OTX-015-treated HepG2 cells were differentially regulated. Among them, 474 and 447 genes were significantly downregulated, whereas 153 and 158 genes were statistically upregulated in JQ1-or OTX-015-treated HepG2 cells, respectively, relative to those of the control HepG2 cells after 24 h (Fig. 2a). Furthermore, to investigate the common and unique up/downregulated genes between JQ1-and OTX-015-treated HepG2 cells, we used RNA-seq data to compare the transcriptome of JQ1treated HepG2 cells with that of OTX-015-treated HepG2 cells. JQ1-treated HepG2 cells contained 152 downregulated genes and 57 upregulated genes that were not common to OTX-015-treated HepG2 cells. In contrast, OTX-015-treated HepG2 cells had 125 downregulated genes and 62 upregulated genes that were not common to JQ1-treated HepG2 cells (Fig. 2b). However, JQ1-and OTX-015-treated HepG2 cells also had similarities in their transcriptomes. Of the downregulated genes, JQ1-and OTX-015-treated HepG2 cells shared 322 genes. Of the upregulated genes, JQ1-and OTX-015-treated HepG2 cells shared 96 genes (Fig. 2c).
To further characterize the BET inhibitor-treated HepG2 cells, we performed an upstream regulator analysis of DEGs using IPA software. The upstream regulator analysis identified 18 regulators, of which the top regulators were IFNG, MAPK1, IFNL1, and BRD4 (Fig. 2d). IPA analysis indicated that BRD4 formed a direct or indirect network with several downregulated genes commonly involved in JQ1-and OTX-015-treated HepG2 cells (Fig. 2e). The expression changes of these genes, including FOS, ACSL5, SLC38A5, and ICAM2, were validated by qPCR using GAPDH as the reference gene (Fig. 2f). To confirm BET inhibitors' distinct effects in HepG2 cells, we incubated HepG2 cells treated with JQ1 or OTX-015, which showed that ACSL5, SLC38A5, and ICAM2 were downregulated and FOS was upregulated. More importantly, BET inhibitors also suppressed genes' expression, including ACSL5, SLC38A5, and ICAM2, in Huh7 cells. www.nature.com/scientificreports/ Network analysis of the altered genes in BET inhibitor-treated HCC cells. Next, we identified the network of genes and related pathways representing the interacting genes in JQ1-or OTX-015-treated HepG2 cells using IPA software. Network-1 and Network-2 in JQ1-or OTX-015-treated HepG2 cells are illustrated in Figure 3a and b. Twenty-four hours of JQ1 treatment revealed genes in networks-1 known to be involved in cell death and survival, inflammatory response, organismal injury, and abnormalities (Fig. 3a). The genes in Network-2 in JQ1-treated HepG2 cells are involved in cellular movement, hematological system development and function, and inflammatory response. In OTX-015-treated HepG2 cells, the genes in Networks-1 are known to be involved in cell-to-cell signaling and interaction, embryonic development, and organ development. The genes in Networks-2 in OTX-015-treated HepG2 cells are known to be involved in cellular movement, hematopoiesis, and immune cell trafficking (Fig. 3b).

Biofunctional analysis of the altered genes in BET inhibitor-treated HCC cells.
To further characterize BET inhibitor-treated HepG2 cells, we determined the biofunctions of DEGs obtained from JQ1-or OTX-015-treated HepG2 cells (Fig. 3c). The gene functions activated by JQ1 or OTX-015 treatment were generally associated with apoptosis, reactive oxygen species, and cell adhesion. Interestingly, the gene functions that were inactivated by JQ1 or OTX-015 treatment were genes commonly associated with cell migration. Furthermore, to investigate the common and unique migration-regulated genes between JQ1-and OTX-015-treated HepG2 cells, we used RNA-seq data to compare the transcriptome of JQ1-treated HepG2 cells with that of OTX-015-treated HepG2 cells (Fig. 3d). JQ1-treated HepG2 cells demonstrated 26 changed genes that were not common to the OTX-015-treated HepG2 cells. In contrast, OTX-015-treated HepG2 cells showed 23 altered genes that were not common to JQ1-treated HepG2 cells. Of the changed genes, JQ1-and OTX-015treated HepG2 cells shared 44 genes. www.nature.com/scientificreports/

Downregulation of cell migration-related genes in BET inhibitor-treated HCC cells.
After a functional analysis of the altered genes in BET inhibitor-treated HepG2 cells, we focused on those associated with cell migration. Of these cell migration-related genes, the 44 commonly altered genes are listed in Figure 4a. The normalized RNA-seq read densities of cell migration-related genes (AOC3, CCR6, SSTR5, and SCL7A11) were decreased in BET inhibitor-treated HepG2 cells (Fig. 4b).
To verify the RNA-seq results, we confirmed the expression of cell migration-related genes by qPCR. ASIC1, CD9, SSTR5, and VAV3 mRNA were downregulated in BET inhibitor-treated HepG2 and Huh7 cells (Fig. 4c).
Besides, a wound healing assay was conducted to investigate the migration effects of BET inhibitor-treated cells. The number of Huh7 cells that migrated into a wound field following BET inhibitor treatment was significantly smaller than that of control cells (Fig. 4d). These data strongly suggest that BET inhibitor treatment was associated with the cell migration response.
Role of SMARCA4 in cell proliferation and migration. The initial bioinformatics analysis revealed that many genes involved in cell migration were regulated by SMARCA4 (Fig. 5a). Kaplan-Meier survival curve analysis suggested that the overall survival of liver cancer patients with high SMARCA4 expression was shorter . The color scale shown in the heat map represents the log2 fold change values. Red color indicates upregulated genes while blue color indicates downregulated genes. The heat map was created in R using the ggplot2 package version 3.3.3 (URL: https:// ggplo t2. tidyv erse. org) 69 . The p value with an asterisk attached in the cell represents *p < 0.05, **p < 0.01, and ***p < 0.001. (d) Upstream regulator analysis of alternated gene datasets in Con vs. JQ1-treated HCC and Con vs. OTX-015-treated HCC cells using Ingenuity pathway analysis (IPA; https:// www. quiag enbio infor matics. com/ produ cts/ ingen uity-pathw ay-analy sis). (e) The activity of highly connected negative regulators of BRD4, a member of the BET family of proteins, led to this network's inactivation, as assessed using the IPA molecule activity predictor in BET inhibitor-treated HCC cells. The red line indicates common genes in JQ1-and OTX-015-treated HepG2 cells. www.nature.com/scientificreports/ than that of patients with low SMARCA4 expression (Fig. 5b). We analyzed the expression of the SMARCA4 gene using qPCR. The SMARCA4 gene expression was significantly downregulated in BET inhibitor-treated HCC cell lines (Fig. 5c).
To further verify bioinformatics analysis results, we analyzed the Kaplan-Meier survival curves of genes (AREG, SPP1, MAPK13, and EREG) related to cell migration regulated by SMARCA4. Kaplan-Meier survival curve analysis suggested that liver cancer patients' overall survival with high expression of these genes was shorter than that of patients with low expression (Fig. 5d). The expression of AREG, SPP1, MAPK13, and EREG genes obtained using qPCR was significantly downregulated in BET inhibitor-treated HCC cell lines (Fig. 5e). The HCC cell lines treated with both JQ1 and OTX-015 showed downregulated SMARCA4 genes and a few target genes regulated by SMARCA4. Thus, we speculated that SMARCA4 might play an important role in HCC cell migration.
Next, we investigated whether SMARCA4 depletion affects cell proliferation and cell migration. Huh7 cells were treated with different concentrations of SMARCA4 siRNA, and siSMARCA4 led to a significant lack of SMARCA4 expression relative to that in the negative control group (scrambled siRNA-siNC; Fig. 6a). These reductions were observed at the protein level (Fig. 6b). The full-length blots are presented in Supplementary Figure S2. Using qPCR analysis, we revealed that the EREG gene expression was significantly decreased in SMARCA4 depleted cells. However, some migration-related genes, including AREG, SPP1, and MAPK13, showed increased expression in SMARCA4-depleted cells (Fig. 6c). We found that SMARCA4 directly regulates EREG gene expression. Also the amount of EREG released is significantly decreased after SMARCKA4 knockdown (Fig. 6d).
Furthermore, we performed a cell proliferation assay of Huh7 cells treated with scrambled siRNA or SMARCA4 siRNA for 48 h. We found that the depletion of SMARCA4 decreased the proportion of EdU-positive cells. The decrease in SMARCA4 expression reduced the proliferation of Huh7 cells (Fig. 6e). The shortage of SMARCA4 significantly reduced the number of Huh7 cells that migrated into a wound field relative to scrambled siRNA-treated Huh7 cells ( Fig. 6f and S3). These data strongly suggest that SMARCA4 is involved in the regulating cell proliferation and migration response in HCC cells.

Discussion
HCCs are cancers that are not easy to treat due to their heterogeneity and drug resistance. Drugs, including sorafenib, a representative prescription currently used to treat HCC, have been reported to be ineffective or www.nature.com/scientificreports/ poorly performing in some patients 44,45 . Therefore, an epigenetic modulator was recently studied to treat cancer with kinase inhibitors 18,26,46 .
In a previous study, the BRD4 inhibitor JQ1 was reported to have an inhibitory effect on HCC cell proliferation and metastasis 18,47 . OTX-015 was confirmed to be effective for acute leukemia and NUT and was subsequently used clinically 31,33 . In this study, treatment with two representative BRD4 inhibitors, JQ1 and OTX-015, was confirmed to inhibit HCC cell lines' proliferation. The effect of BET inhibitor OTX-015 had not been known for HCC cells. Here, we focused on common DEGs that were altered by JQ1 or OTX-015 treatment in HepG2 cells. Approximately 60-70% of the DEGs that were altered when treated with each BET inhibitor in HepG2 cells demonstrated the same pattern. It was found that each inhibitor had a reasonably similar effect. Additionally, IPA-based analysis confirmed that the typical DEGs mode after treatment with each inhibitor was related to a decrease in BRD4. www.nature.com/scientificreports/ The cellular movement was critical in the IPA network analysis results for genes with decreased expression among the DEGs obtained from HepG2 cells treated with JQ1 or OTX-015. The inhibition of cell migration by JQ1 in salivary adenoid cystic carcinoma (SACC), prostate cancer, etc. through the downregulation of BRD4 had been confirmed in previous studies 48,49 . However, few studies have been conducted on cell migration by JQ1 or OTX-015 in HCC cells. Through IPA analysis, we demonstrated that genes reduced by JQ1 or OTX-015 reduce cell movement, mainly biofunctions corresponding to the migration and movement of tumor cell lines. Additionally, among DEGs, genes related to cell migration accounted for approximately 11%, and about 50% of the genes showed the same results for both drugs. These results suggest that the anticancer effect of the BET inhibitor treatment of HCC cells is due to cell migration inhibition. In addition, since DEGs corresponding to more than half showed the same pattern in JQ1-treated Huh7 cells (Fig. S1), the result suggests that it is not specific to HepG2 cells but can be interpreted as a result of HCC cells.
In HCC, the treatment method depends on the size and metastasis of the tumor. Available treatments for HCC metastasis are limited 1 . In this study, it was confirmed that cell migration, which is very important for metastasis, can be suppressed through inhibition of BRD4. Cell mobility decreased after treatment with JQ1 or OTX-015. This result is consistent with previous studies on BRD4 inhibition in HCC cells 50 .
This study studied the SMARCA4 gene to confirm that cell migration is inhibited when HCC cells are treated with JQ1 or OTX-015. SMARCA4, known as Brahma-related gene-1 (BRG1), is a component of the SWI/SWF complex (a large ATP-dependent chromatin remodeling complex) along with Brahma (BRM) and is mutated in several cancers 51 . Additionally, it is known that the SWI/SNF complex cause chromatin remodeling and affects cell differentiation and cell proliferation 51,52 . In fact, in cancers such as prostate cancer, colon cancer, and lung cancer, SMARCA4 is an epigenetic regulator and has been reported to promote metastasis through cancer migration and invasion [53][54][55] . It has been reported that the SMARCA4 gene is highly expressed in HCC and increases cell proliferation 52 . Our results also showed that SMARCA4 regulates a significant number of genes involved in cell migration. When the SMARCA4 gene was knocked down with siRNA, cell proliferation and mobility were reduced. www.nature.com/scientificreports/ Furthermore, as a result of knocking down the SMARCA4 gene to confirm its relationship to HCC cell migration, the expression of the target gene EREG gene was reduced. EREG is a gene that encodes epiregulin, an epidermal growth factor (EGF) family member that binds to the EGF receptor (ErbB) member of the receptor tyrosine kinase family. Epiregulin's role in tumors is to activate the Erk and PI3K kinase/Akt signaling pathways by binding to the EGF receptor. It generates proliferation, invasion, metastasis, angiogenesis, and resistance to apoptosis 56 . EREG is expressed at shallow levels in healthy cells, but it is also increased in various cancers 57 . In fact, in colorectal cancer (CRC), it has been reported that activation of the EGFR pathway through EREG demethylation may be a mechanism of several types of malignancies 58 . Especially, in salivary adenoid cystic carcinoma (SACC), ERE-induced EGFR activation has been reported as a significant cause of metastasis 59 . In HCC, an increase in epiregulin expression has been shown to act as a compensation mechanism for N-RAS's inhibitory effect. Therefore, when N-RAS and epiregulin are simultaneously inhibited, HCC cells' growth can be effectively suppressed 60 .
Collectively, the RNA-seq data analysis results and the SMARCA4 knockdown experiments can expect two possibilities in the regulation of migration by SMARCA4 in HCC cells. The first is when a BET inhibitor suppresses the expression of SMARCA4, an upstream regulator of migration-related genes, and the second is when the recruitment of the chromatin remodeler SMARCA4 for the expression of the migration-related genes such as EREG is inhibited by inhibition of BET protein. Our models of SMARCA4 working mechanism are described in Figure 7.
The BRD4 active site is composed of two domains, BD1 and BD2. BET inhibitors such as JQ1 or OTX-015 mainly target the BD1 domain. These BET inhibitors are severely toxic and have limited clinical use 61 . Recently developed selective BD1 or BD2 inhibitors have been shown to be effective in cancer treatment or infection 62,63 . We are currently studying the pathway of genes regulated by BD2 inhibitors (such as ABBV-744) in HCC cells.

Conclusions
In this study, we performed a transcriptome analysis of BET inhibitor-treated HCC cells. We found that BET inhibitors JQ1 and OTX-015 have a significant cancer inhibitory effect on cell proliferation and cell migration in HCC. We confirmed that BET inhibitors attenuate BRD4-related and cell migration-related genes. Primarily, cell migration-related genes regulated by SMARCA4 were reduced in BET inhibitor-treated HCC cells. These findings suggest that BET inhibitors modulate the cell migration effects of HCC and selectively inhibit the expression of cell migration-related genes through SMARCA4. This result indicates that JQ1 or OTX-015 can be used as a drug to improve HCC treatment.  Transcriptome analysis using RNA-seq. RNA sequencing (RNA-seq) was performed as previously described 64 . Total RNA was extracted from HCC cells using RNAiso Plus (Takara, Shiga, Japan) and a Qiagen RNeasy Mini kit (Qiagen, Hilden, Germany). RiboMinus Eukaryote kit (Invitrogen, Carlsbad, CA, USA) was used for Ribosomal RNA (rRNA) depletion. An RNA library was created by a NEBNext Ultra directional RNA library preparation kit from Illumina (New England BioLabs, Ipswich, MA, USA). RNA library sequencing was performed on the Illumina HiSeq2500 platform (Macrogen, Seoul, Korea). Transcriptome sequencing was performed on independent RNA samples form DMSO-treated (3 samples), JQ1-treated (3 samples), or OTX-015treated (3 samples) HepG2 cells in biological triplicate. FASTQ files from RNA-seq were clipped and trimmed of adapters, and low-quality reads were removed using Trimmomatic 65 . These FASTQ files were aligned using STAR (version 2.7.1) aligner software with a UCSC hg38 reference 66 . Differentially expressed genes (DEGs) was analyzed using DESeq2 with the default parameters 67 . DEGs identified by RNA-seq that had an absolute log 2 -fold change larger than 2 or smaller than -2 (log 2 -fold change ≥ 2 and log 2 -fold change ≤ − 2, p adjusted < 0.05) were selected as DEGs in JQ1-or OTX-015-treated HepG2 cells. Heat maps were visualized in R software (3.6.2) 68 using the ggplot2 package (3.3.3) 69 . The acquired data were deposited in the Gene Expression Omnibus database. The dataset accession number GSE158552.

Materials and methods
Graphical representation of the networks and pathways. The RNA-seq dataset were analyzed through the use Ingenuity Pathway Analysis (IPA; QIAGEN lnc., https:// www. quiag enbio infor matics. com/ produ cts/ ingen uity-pathw ay-analy sis; Ingenuity Systems, Mountain View, CA) to analyze the networks and pathways 70 . RNA-seq data were cut off at the fold-change (log 2 -fold change ≥ 2 and log 2 -fold change ≤ -2, p value < 0.05) in JQ1-or OTX-015-treated HepgG2 cells. The IPA software presented a functional analysis that showed genes involved in upstream regulators, biological functions/disease, and network analysis.
Gene expression analysis using quantitative PCR (qPCR). Gene expression analysis was performed as previously described 64 . Total RNA was extracted from HepG2 or Huh7 cells using RNAiso Plus (Takara, Shiga, Japan) according to the manufacturer's instructions. cDNA was synthesized by PrimeScript reverse transcriptase (Takara, Shiga, Japan) and amplified using gene-specific primers ( Table 1). The primers were designed by Primer Bank (https:// pga. mgh. harva rd. edu/ prime rbank/). qPCR was performed with TBGreen Premix Ex Taq II (Takara, Shiga, Japan). We used Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as an internal control. The data represent three independent experiments (n = 3). After performing qPCR, the results were analyzed using the critical threshold (△C T ) and the comparative critical threshold (△△C T ) methods in ABI 7500 (Applied Biosystems, Foster City, CA, USA) software with the NormFinder and geNorm PLUS algorithms.  GCT TCA ACG CAG ACT A  GGT CCG TGC AGA AGT CCT G   ACSL5  CTC AAC CCG TCT TAC CTC TTCT  GCA GCA ACT TGT TAG GTC ATTG   SLC38A5  GCT ACA GGC AAG AAC GTG AGG  ATT CCA AAC GAT GTC TTC CCC   ICAM2  CGG ATG AGA AGG TAT TCG AGGT  CAC CCA CTT CAG GCT GGT TAC   ASIC1  ATG GAA AGT GCT ACA CGT TCAA  GTT CAT CCT GAC TAT GGT TCTGC   CD9  AGC CAT CCA CTA TGC GTT GA  ATG GCA TCA GGA CAG GAC TTC   SSTR5  TGT TTG CGG GAT GTT GGC T  CTG TTG GCG TAG GAG AGG A   VAV3  AGA GAA ACG GAC CAA TGG ACT  GGT GGT GTT CCA GAA TAG TTCC   SMARCA4  AAT GCC AAG CAA GAT GTC GAT  GTT TGA GGA CAC CAT TGA CCATA   AREG  GTG GTG CTG TCG CTC TTG  www.nature.com/scientificreports/ In vitro wound-healing assay. HCC cells were seeded into each Culture-Insert (Ibidi, Martinsried, Germany) and incubated for 24 h. The cells were treated with JQ1 or OTX-015 for 24 h or transfected with scrambled siRNA or SMARCA4 siRNA for 48 h. After incubation, the inserts were removed to create a "wound field. " The cells were washed once and incubated with growth media for 24 h or 48 h. The cells were then visualized using a JuLi BR real-time cell history recorder (NanoEnTek, Seoul, Korea).

Kaplan-Meier plotter analysis for overall survival of HCC.
We used Kaplan-Meier plotter (KM plotter; http:// kmplot. com/ analy sis/), an online biomarker analysis tool, that evaluates the prognostic value of biomarkers in various cancers. In liver cancer data, 364 cases were analyzed, and the false discovery rate (FDR) cutoff was set to 1% 71 . Briefly, to obtain KM survival plots of the EREG, AREG, SPP1, MAPK13 genes in HCC, those genes were entered into the database.
Knockdown of SMARCA4 gene expression using siRNA treatment. Knockdown (KD) of gene expression was performed using small interfering RNA (siRNA). After seeding the cells, transfection was performed using Lipofectamine RNAiMax (Invitrogen, CA, USA) transfection agent according to the manufacturer's instructions with siRNA constructs and scrambled siRNAs. SMARCA4 siRNA (ID numbers 4677; 5′-GCA UUU CAA GGA AUA UCA Ctt-3′) and Silencer Negative Control siRNA (AM4611) were purchased from Thermo Fisher Scientific. SMARCA4 siRNA and scrambled siRNA were used at a 10 nM concentration for 48 h in a growth medium.
Western blotting assay. According to the manufacturer's instructions, nuclear or cytoplasmic proteins from the cells were isolated using a NE-PER Nuclear Cytoplasmic Extraction Reagent Kit (Thermo Fisher Scientific, Waltham, USA). Nuclear protein was separated by sodium dodecyl sulfate (SDS) polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidene difluoride membranes (Schleicher & Schuell Bioscience, Inc., Keene, NH, USA). Western blot analysis was performed using anti-SMARCA4 (Abcam, Cambridge, UK; ab110641) and anti-histone H3 (Abcam, Cambridge, UK; ab1791) antibodies. Histone H3 protein was used as an internal control.
Chemokine measurements with enzyme-linked immunosorbent assay (ELISA). Huh7 cells were transfected with siRNA for 48 h. After transfected, cell culture supernatants were concentrated 20-fold using Pierce Protein Concentrator (Thermo Fisher Scientific, Waltham, MA, USA). According to the manufacturer's instructions, the concentration of the EREG in concentrated supernatants was determined using human EREG ELISA kits (Abcam, Cambridge, UK; ab277077). The data represent three independent experiments (n = 3).

Statistical analysis.
Data are presented as the mean ± standard deviation (SD) of the mean. All statistical analyses were performed using IBM SPSS Statistics 26.0 program (IBM corp., Armonk, NY). We used one-way analysis of variance followed by Tukey's honestly significant difference post hoc test. p values < 0.05 were considered significant.