Abstract

The chromatin remodeling complex SWI/SNF regulates the accessibility of target genes to transcription factors and plays a critical role in the tumorigenesis of hepatocellular carcinoma (HCC). The SWI/SNF complex is assembled from approximately 15 subunits, and most of these subunits have distinct roles and are often aberrantly expressed in HCC. A comprehensive exploration of the expression and clinical significance of these subunits would be of great value. In the present study, we obtained the gene expression profile of each SWI/SNF subunit and the corresponding clinical information from The Cancer Genome Atlas (TCGA). We found that 14 out of the 15 SWI/SNF subunits were significantly increased in HCC tissues compared with paired normal liver tissues, and 11 subunits were significantly associated with overall survival (OS). We identified a four-gene prognostic signature including actin-like 6A (ACTL6A), AT-rich interaction domain 1A (ARID1A), SWI/SNF related, matrix associated, actin dependent regulator of chromatin subfamily C member 1 (SMARCC1) and SWI/SNF related, matrix associated, actin dependent regulator of chromatin, subfamily D, member 1 (SMARCD1) that could effectively predict OS in HCC patients. Among the genes, SMARCD1 has the most prognostic value. We further conducted in vitro and in vivo experiments and revealed that SMARCD1 promotes liver cancer growth by activating the mTOR signaling pathway. In conclusion, our study has revealed that the expression of SWI/SNF complex subunits, especially SMARCD1, is highly associated with HCC development and acts as a promising prognostic predictor.

Introduction

Hepatocellular carcinoma (HCC) has long been a major public health problem, ranking as the third leading cause of cancer-related deaths with a rising incidence worldwide, especially in China [1]. Although considerable improvements in surgery and therapeutic strategies have been achieved, the survival rate of HCC remains far from satisfactory due to its late diagnosis, rapid development, and easy metastasis and the lack of precise therapeutic targets. HCC development is a consequence of a multistep process that involves complicated interplay between genetic, epigenetic, and transcriptomic alterations [2]; however, the detailed mechanics involved in HCC development are still largely unknown, and this lack of information has hindered improvements in the diagnosis, treatment, and prognosis prediction of HCC.

The SWI/SNF complex is an evolutionarily conserved multisubunit complex that utilizes the energy of ATP hydrolysis to remodel chromatin from a ‘condensed’ state to an ‘open’ state, therefore playing an important role in controlling gene expression [3,4]. The ∼2 MDa complex is assembled from 12 to 15 subunits (as shown in Figure 1A). In each complex, there is one catalytic ATPase subunit, either SWI/SNF related, matrix associated, actin dependent regulator of chromatin (SMARCA) 4 (SMARCA4) or 2 (SMARCA2). Several core subunits, such as SWI/SNF related, matrix associated, actin dependent regulator of chromatin, subfamily B, member 1 (SMARCB1) and SWI/SNF related, matrix associated, actin dependent regulator of chromatin subfamily C member 1 (SMARCC1), are present in all types of SWI/SNF complexes. Some subunits are selectively present in some variants; for example, AT-rich interaction domain 1A (ARID1A) and ARID1B are mutually exclusive subunits for the BRG1-associated factor (BAF) complex, which is a type of SWI/SNF complex, while PBRM1 and AT-rich interaction domain 2 (ARID2) are specifically assembled in the polybromo BAF (PBAF) complex [3]. By providing binding sites or facilitating the binding of other transcription factors to target DNA, BAF and PBAF either promote or suppress gene transcription, depending on the cofactors the subunits bind [5].

The expression levels of and coexpression of SWI/SNF subunits in the The Cancer Genome Atlas HCC cohort

Figure 1
The expression levels of and coexpression of SWI/SNF subunits in the The Cancer Genome Atlas HCC cohort

(A) Brief diagram of the transcriptional regulation of the SWI/SNF complex. (B) Heatmap visualizing the expression levels of SWI/SNF subunits in each clinical sample. (C) The significantly differentially expressed SWI/SNF subunits between tumor tissues and normal control tissues. (D) Pearson correlation analysis was used to determine the correlation between any two members of the SWI/SNF complex.

Figure 1
The expression levels of and coexpression of SWI/SNF subunits in the The Cancer Genome Atlas HCC cohort

(A) Brief diagram of the transcriptional regulation of the SWI/SNF complex. (B) Heatmap visualizing the expression levels of SWI/SNF subunits in each clinical sample. (C) The significantly differentially expressed SWI/SNF subunits between tumor tissues and normal control tissues. (D) Pearson correlation analysis was used to determine the correlation between any two members of the SWI/SNF complex.

Any changes in the subunits may alter the function of the SWI/SNF complex in transcription regulation; however, the individual role of each subunit is distinct but unclear. Inactivation of SMARCB1 or SMARCA4 results in embryonic lethality at embryonic day 3.5 (E3.5) [6,7]. ARID1A ablation leads to the absence of mesoderm and arrests at E6.5 [8]. Silencing of SWI/SNF related, matrix associated, actin dependent regulator of chromatin, subfamily D, member 3 (SMARCD3) results in a defect in heart development [9]. In contrast, mice can survive SMARCA2 deprivation without any obvious developmental defects [10].

Accumulated evidence has shown that the SWI/SNF complex has a widespread role in tumor suppression, as inactivating mutations in several SWI/SNF subunits have recently been identified at a high frequency in a variety of cancers, such as ARID1A mutation in ovarian cancer, SMARCA4 mutation in lung cancer, PBRM1 mutation in renal cancer, and SMARCB1 mutation in rhabdoid tumors [3]. In animal experiments, conditional Smarcb1 deletion in mice results in the formation of rhabdoid-like tumors and lymphomas [11], while SMARCA4 haploinsufficiency leads to mammary tumors [12]. Suppression or deficiency of ARID1A, the most studied subunit, leads to liver tumorigenesis [13,14]. It is obvious that the aberrant expression of SWI/SNF subunits is highly associated with the development and prognosis of various cancers, including HCC. Interestingly, the subunit mutations and aberrant expression are only modestly overlapped across different cancers, which indicates that the individual subunit has specific biological significance in various cancers, either as a tumor suppressor, promoter, or prognostic predictor.

Previous studies have revealed that a few SWI/SNF subunits are dysregulated in HCC. SMARCA2 and SMARCA4, the two catalytic ATPase subunits and critical gene expression regulators, are differentially expressed in HCC tissues [15]. Actin-like 6A (ACTL6A) and the core ATPase subunit SMARCA4 (Brg1) are often highly expressed in HCC tissues, while the down-regulation of SMARCB1 (Snf5) and SMARCA2 (Brm) is significantly associated with overall survival (OS) [15,16]. High SMARCA4 expression can promote liver cancer cell proliferation and invasion and can predict HCC recurrence, while the decrease in or loss of SMARCA2 expression is highly associated with poor OS [15,17].

The expression and clinical significance of each SWI/SNF subunit in HCC is still unclear. In the present study, we aimed to identify the significant differentially expressed SWI/SNF subunits between tumor and normal tissues based on the expression profile in the HCC dataset from The Cancer Genome Atlas (TCGA), the largest scale cancer genomics program including comprehensive and multidimensional data covering 33 types of cancer. In addition, in combination with the expression profile and clinical information, the prognostic value of each subunit was unbiasedly analyzed. We identified that a four-gene (SMARCC1, SMARCD1, ARID1A, and ACTL6A) risk signature built on the TCGA HCC cohort can effectively predict OS in HCC patients. Among these four genes, SMARCD1 was the most valuable for predicting prognosis, and we further explored its role and underlying molecular mechanism in promoting HCC.

Materials and methods

Public data source acquisition

The RNA-seq transcriptome data and corresponding clinical information of HCC samples were downloaded from The National Cancer Institute Genomic Data Commons (NCI-GDC) (https://gdc.cancer.gov/). A total of 374 HCC cases and 50 paired normal control samples were included for subsequent analysis.

Data processing, differential expression and correlation analysis

The EdgeR package was used to screen the differentially expressed genes between paired normal samples and tumor samples. Adjusted P<0.05 and log2FC > 1 were adopted as the cut-off values. Fifteen currently known SWI/SNF complex subunits, including SMARCA2, SMARCA4, ARID1A, ARID1B, PBRM1, ARID2, SMARCB1, SMARCE1, BRD7, SMARCC2, SMARCC1, ACTL6A, SMARCD1, SMARCD2 and SMARCD3, were included and analyzed. Pearson correlation analysis was employed to reveal the expression association between each pair of subunits.

Consensus clustering analysis

To determine whether the expression level of each SWI/SNF complex subunit was associated with prognosis, the TCGA HCC cohort was clustered into different groups by consensus expression of SWI/SNF complex subunits with the ‘ConsensusClusterPlus’ R package [18]. The OS difference between different clusters was calculated by the Kaplan–Meier method and log-rank test in R. The χ2 test was used to compare the distribution of age, sex, grade and stage between different clusters.

Prognostic signature generation and prediction

Univariate Cox analysis was performed to evaluate the correlation between SWI/SNF complex subunits and OS for the TCGA HCC cohort using survival analysis in R. The hazard ratios (HRs) of genes that were larger than 1 were considered risk-conferring genes, while those with HRs less than 1 were regarded as protective genes. Multivariate Cox regression analysis and the Akaike information criterion (AIC) method were used to determine the optimal model. A risk score for each patient was calculated as the sum of the gene scores, which was obtained by multiplying the expression of each gene and its coefficient. The TCGA HCC cohort was stratified into high- and low-risk groups based on the median value of the risk scores. The difference in OS between the high- and the low-risk groups was calculated by the Kaplan–Meier method with a two-sided log-rank test. A chi-square test was performed to compare the distribution of clinicopathological parameters between the high- and low-risk groups. Heatmaps were used to visualize the difference with the pheatmap R package. Univariate and multivariate Cox regression analyses were used to identify the independent prognostic factors for the TCGA HCC cohort. The survival difference between the high- and the low-risk groups stratified by age, sex, grade and stage was further evaluated.

Cell culture and transfection

The human HCC cell line Huh7 was purchased from the ATCC, which was cultured in DMEM emented with 10% fetal bovine serum (FBS), 1% penicillin and 1% streptomycin at 37°C in a 5% CO2 humidified incubator. Plasmids encoding shSmarcd1 (shRNA1, AAGTCCTTGGTGATTGAACTGGA; shRNA2, AATGTACGGTGTACTGTCCTACT) were transfected into cells lines by Lipofectamine 3000 according to the manufacturer’s instructions. The efficiency was measured by Western blot 72 h after transfection. MHY1485 and INK128 were used as mTOR activator or inhibitor at the concentration of 10 and 1 μM, respectively.

Cell viability, proliferation and colony formation

CCK-8 (Sigma–Aldrich Co, St. Louis, MO, U.S.A.) and EdU incorporation (RiboBio, Guangzhou, China) assays were carried out to evaluate cell viability and proliferation according to the manufacturers’ instructions. For colony formation assays, cells were trypsinized, resuspended, seeded into six-well plates at a concentration of 100 cells per well, and cultured at 37°C for 1–2 weeks. At the end of the incubation, the cells were fixed with 100% methanol and stained with 0.1% Crystal Violet. Megascopic cell colonies were counted.

Wound healing assay

Approximately 2 × 105 cells were plated in six-well plates after the different treatments. A linear scratch was generated on the cell monolayer with a sterile pipette. Photomicrographs of live cells were obtained at 20× magnification, and the distance migrated was observed after 24 h. The remaining wound area was measured using ImageJ software.

Matrigel invasion assay

The Matrigel invasion assay was performed in 24-well Transwell culture plates. Cells were resuspended and then seeded in 24-well Transwell plates containing FBS-free medium in the upper chamber and complete growth medium supplemented with 10% FBS in the lower chamber for 24 h at 37°C. Non-invading cells were removed from the upper surfaces of the invasion membranes, and the cells on the lower surface were stained with Hematoxylin. The average number of cells per field was determined by counting the cells in six random fields per well. Cells were counted in four separate fields in three independent experiments.

Orthotopic nude mouse models and treatment

Male BALB/c nude mice aged 4–6 weeks (purchased from the Animal Center at the Cancer Institute at Chinese Academy of Medical Science) were raised in specific pathogen-free animal facilities, and animal experiments were conducted in the Laboratory Animal Center, West China Hospital, Sichuan University. The animals involved in the present study received humane care and were provided free access to water and food. A total of 2 × 106 Huh7 cells transfected with shSmarcd1 or shControl were injected into the abdominal subcutaneous of each nude mouse. The tumors were measured weekly. Mice were anesthetized with 1–3% isoflurane and killed by CO2 inhalation before killing. All animal work took place in in accordance with national and international laws and policies and approved by the Animal Care and Use Committee of Sichuan University.

Real-time RT-PCR analysis

RNA was extracted from tissues using TRIzol reagent (Invitrogen, CA, U.S.A.) and further purified by the RNeasy kit (Qiagen, DUS, Germany). cDNA was generated using the iScript cDNA synthesis kit (Bio-Rad, CA, U.S.A.), and qRT-PCR was performed with the Bio-Rad CFX96 System. The sequences of the indicated primers were presented in Supplementary Table S1.

Sample preparation and Western blot

Cells and tissues were lysed with RIPA buffer according to the manufacturer’s instructions (Beyotime Biotechnology, Shanghai, China). Proteins were separated with SDS/PAGE and transferred to PVDF membranes. Membranes were blocked with PBS containing 0.05% Tween and 5% non-fat milk, and probed with antibodies listed in Supplementary Table S2. Nuclear and cytoplasmic extracts were separated to detect β-catenin according to manufacturer’s instructions of NE-PER™ Nuclear and Cytoplasmic Extraction Reagents (Thermo, MA, U.S.A.).

Patients, tissue microarray and immunohistochemistry staining

We studied 150 HCC patients from a cohort of 205 patients diagnosed with HCC who had their tumors removed at West China Hospital of Sichuan University between 2013 and 2014. OS was defined as the time between the initial surgery and death. We prepared tissue microarray (TMA) cores (1.5-mm diameter) from formalin-fixed, paraffin-embedded samples. Immunohistochemistry (IHC) staining was performed on the TMA slides, and the results were interpreted by three pathologists using a blinded method. The expression level of SMARCD1 was scored according to the signal intensity and distribution. Briefly, at least five 400×-magnified areas were examined and assigned to the following categories: 0, <5%; 1, 5–25%; 2, 25–75%; 3, >75%. The intensity of staining was scored as follows: 1, weak; 2, moderate; intense. Tissues with an immunohistochemical score of 0-3 –ere designated as low expression, and those with scores of 4–9 were designated as high expression. Classic core clinical characteristics, such as age, AFP level, WHO grade, clinical stage, tumor size, nodal status, and distal metastasis, were included to analyze the correlation of SMARCD1 with HCC. Approval for the presentstudy was granted by the Ethics Committee of the West China Hospital, Sichuan University, and each patient provided written informed consent.

Statistical analysis

All statistical analyses were performed using the SPSS version 19.0 software program (SPSS Inc., Chicago, U.S.A.) or GraphPad Prism 7.0 (GraphPad, San Diego, CA, U.S.A.). All statistical tests were two-sided. A P-value of less than 0.05 was considered statistically significant.

Results

Differential expression of SWI/SNF complex subunits between tumor samples and normal samples

Data from 374 HCC cases and 50 paired normal control samples were downloaded from NCI-GDC. A heatmap was generated to visualize the expression pattern of each SWI/SNF subunit between tumor and normal tissues. Red or blue color in the plots represented relatively high or low expression, respectively (Figure 1B). According to the results, all the subunits were differentially expressed between tumor and normal tissues (P≤0.001). Except for SMARCA2, which was decreased, all the other members (14 in total) were significantly increased in tumor tissues (Figure 1B,C). The Pearson correlation analysis indicated that most of these subunits were coexpressed, especially SMARCC2 and ARID2, ARID1B and ARID2, and SMARCD1 and SMARCA4, which were the most correlated (r > 0.75, Figure 1A,D).

Consensus clustering of SWI/SNF subunits identified two clusters of HCC with different clinical outcomes

To determine whether the expression level of each SWI/SNF subunit was associated with prognosis, the TCGA HCC cohort was clustered into different groups (k = 2–9) by consensus expression of SWI/SNF subunits with ‘ConsensusClusterPlus’, an algorithm for determining cluster count and membership by stability evidence in unsupervised analysis [18]. Based on the expression of the SWI/SNF complex subunits, k = 2 was demonstrated to be the most appropriate selection, so the HCC patient cohort was divided into two clusters (Figure 2A,B, Supplementary Figure S1). Principal component analysis (PCA) revealed that this method effectively distinguished between HCC patients into the two groups (Figure 2C). Significantly shorter OS was observed in cluster 2 patients than in cluster 1 patients (P=0.039) (Figure 2D). Then, the association between the clustering and clinicopathological features, together with the expression of the subunits in two different clusters, was evaluated (Figure 2E). A significant difference was found in tumor grade (P=0.024) between the two clusters, and a larger portion of patients with higher tumor grade (G3 + G4) were found in cluster 2 than that in cluster 1 (53.42 vs 33.58%), while no significant difference was observed for other parameters, such as metastasis and stage.

Differential OS and grade of TCGA HCC patients in the two different clusters according to the expression of SWI/SNF subunits

Figure 2
Differential OS and grade of TCGA HCC patients in the two different clusters according to the expression of SWI/SNF subunits

(A) Consensus clustering cumulative distribution function (CDF) for k = 2–9. (B) The TCGA HCC cohort was divided into two distinct clusters when k = 2. (C) TCGA HCC patients were divided into two distinct clusters according to PCA. (D) The OS of cluster 1 was significantly shorter than that of cluster 2. (E) A significant difference was found in the grade of cluster 1 and that of cluster 2. N, lymph node metastasis classification (N0: no regional lymph node metastases; N1: metastasis in 1–2 regional lymph nodes; NX: regional lymph node condition cannot be assessed.); M, distant metastasis classification (M0, no distant metastasis; M1, distant metastasis, MX: distant metastasis condition cannot be assessed); T, extent of the primary tumor range from T1 to T4.

Figure 2
Differential OS and grade of TCGA HCC patients in the two different clusters according to the expression of SWI/SNF subunits

(A) Consensus clustering cumulative distribution function (CDF) for k = 2–9. (B) The TCGA HCC cohort was divided into two distinct clusters when k = 2. (C) TCGA HCC patients were divided into two distinct clusters according to PCA. (D) The OS of cluster 1 was significantly shorter than that of cluster 2. (E) A significant difference was found in the grade of cluster 1 and that of cluster 2. N, lymph node metastasis classification (N0: no regional lymph node metastases; N1: metastasis in 1–2 regional lymph nodes; NX: regional lymph node condition cannot be assessed.); M, distant metastasis classification (M0, no distant metastasis; M1, distant metastasis, MX: distant metastasis condition cannot be assessed); T, extent of the primary tumor range from T1 to T4.

Identification of prognostic signature

Univariate Cox regression was used to identify the subunits that were significantly associated with OS in the TCGA HCC cohort. Except for SMARCA2, SMARCD2, SMARCD3 and SMARCE1, all the other 11 subunits were significantly associated with OS (Figure 3A). Then, these 11 genes and their expression in each HCC patient together with survival time were chosen to construct the prognostic signature, and the coefficients were obtained from the Lasso algorithm (Figure 3B,C). As a result, only ACTL6A, ARID1A, SMARCC1 and SMARCD1 were identified as prognostic markers with coefficient values of 0.0245, 0.009, 0.0314 and 0.0291, respectively. The risk score for each patient was calculated in R and presented in Supplementary Table S3. A total of 342 HCC patients were evenly divided into the high-risk group and low-risk group. The expression of these four genes in the high-risk group was higher than that in the low-risk group, and a significant difference was found in grade and sex (P<0.05) between the two groups (Figure 3F). More importantly, OS was longer in the low-risk group than in the high-risk group (Figure 3D). The prognostic signature model showed good prediction efficiency, with an area under the ROC curve (AUC) value equal to 0.732 (Figure 3E).

Construction of the prognostic signature

Figure 3
Construction of the prognostic signature

(A) Univariate analysis of the SWI/SNF subunits was conducted to identify the genes that significantly correlated with OS. (B) Deviance values of the Lasso models as a function of the regularization parameter λ. The optimal penalty, λ, determined by ten-fold cross-validation, is the value that minimizes the deviance curve (in red dotted rectangle). (C) Trace plot showing nonzero model coefficients as a function of the regularization parameter λ. As λ increases to the right, Lasso sets various coefficients to zero, removing them from the model. When λ corresponded to the minimum deviance, four variables were selected. (D) The OS was remarkably shorter in the high-risk group than in the low-risk group. (E) ROC curves were used to evaluate the prediction efficiency of the prognostic signature. ROC, receiver operating characteristic curve. (F) Significant differences were found in sex and grade between the high- and low-risk groups. N, lymph node metastasis classification (N0: no regional lymph node metastases; N1: metastasis in 1-2 regional lymph nodes; NX: regional lymph node condition cannot be assessed.); M, distant metastasis classification (M0, no distant metastasis; M1, distant metastasis, MX: distant metastasis condition cannot be assessed); T, extent of the primary tumor range from T1 to T4.

Figure 3
Construction of the prognostic signature

(A) Univariate analysis of the SWI/SNF subunits was conducted to identify the genes that significantly correlated with OS. (B) Deviance values of the Lasso models as a function of the regularization parameter λ. The optimal penalty, λ, determined by ten-fold cross-validation, is the value that minimizes the deviance curve (in red dotted rectangle). (C) Trace plot showing nonzero model coefficients as a function of the regularization parameter λ. As λ increases to the right, Lasso sets various coefficients to zero, removing them from the model. When λ corresponded to the minimum deviance, four variables were selected. (D) The OS was remarkably shorter in the high-risk group than in the low-risk group. (E) ROC curves were used to evaluate the prediction efficiency of the prognostic signature. ROC, receiver operating characteristic curve. (F) Significant differences were found in sex and grade between the high- and low-risk groups. N, lymph node metastasis classification (N0: no regional lymph node metastases; N1: metastasis in 1-2 regional lymph nodes; NX: regional lymph node condition cannot be assessed.); M, distant metastasis classification (M0, no distant metastasis; M1, distant metastasis, MX: distant metastasis condition cannot be assessed); T, extent of the primary tumor range from T1 to T4.

Univariate and multivariate Cox regression analyses were performed to determine whether the prognostic signature-based risk score was an independent prognostic indicator. After deleting cases with missing values for age, sex, grade or stage, a total of 235 cases were used for subsequent analysis. The univariate analysis showed that stage (P<0.001), T stage (P<0.001), M stage (P=0.023) and risk score (P<0.001) were significantly correlated with OS (Supplementary Figure S2A). When these parameters were included in the multivariate Cox regression model, only the risk score (P<0.001, HR = 1.704, 95% CI = 1.349–2.153) was identified as an independent prognostic factor (Supplementary Figure S2B). Thus, the expression of the signature gene set including ACTL6A, ARID1A, SMARCC1 and SMARCD1 could be used as prognostic markers for HCC patients.

SMARCD1 is a prognostic subunit for HCC patients

To further evaluate the role of these four prognostic signature genes (ACTL6A, ARID1A, SMARCC1 and SMARCD1) in HCC patients, we reanalyzed their expression one by one in TCGA HCC data (Figure 4A,B). We further confirmed that in 20 paired fresh HCC tissues, SMARCD1 was identified as the highest elevated gene in tumor tissues compared with corresponding nontumor tissues (fold change > 4) (Figure 4C). Meanwhile, the increase in SMARCD1 protein in HCC tissues was further confirmed by an immunoblotting assay (Figure 4D). In addition, TMA-based IHC staining was conducted, the expression of SMARCD1 was scored according to the intensity and positive rate, and the representative images for each score are listed in Supplementary Figure S3. HCC patients were divided into two classes by high (n=84) or low (n=66) SMARCD1 expression, and representative images of staining are shown in Figure 4E. The association of SMARCD1 expression and the clinical characteristics is presented in Table 1, which indicates that SMARCD1 expression is closely related to tumor size (P=0.007) and TNM stage (P=0.005). Consistent with the relationship between SMARCD1 mRNA level and OS status, univariate and multivariate analysis revealed that the expression of SMARCD1 was highly associated with OS (Supplementary Table S4) and significantly shorter OS was observed in patients with higher SMARCD1 expression (P=0.008, Figure 4F).

SMARCD1 is a valuable prognostic subunit for HCC patients

Figure 4
SMARCD1 is a valuable prognostic subunit for HCC patients

(A) According to TCGA HCC dataset, the individual expression of SMARCD1, SMARCC1, ACTL6A and ARID1A in normal and tumor tissues. (B) The expression of these four genes in tumor and paired normal tissues (n=50). (C) mRNA levels were detected by RT-PCR in 20 paired fresh tissues. (D) The protein level of SMARCD1 was detected by immunoblotting assay and quantified by ImageJ software. (E) Representative image of high or low expression of SMARCD1 in HCC tumor tissues; scale bar = 50 μm. (F) The correlation of SMARCD1 expression with the OS time of HCC patients. *P<0.01, ***P<0.001.

Figure 4
SMARCD1 is a valuable prognostic subunit for HCC patients

(A) According to TCGA HCC dataset, the individual expression of SMARCD1, SMARCC1, ACTL6A and ARID1A in normal and tumor tissues. (B) The expression of these four genes in tumor and paired normal tissues (n=50). (C) mRNA levels were detected by RT-PCR in 20 paired fresh tissues. (D) The protein level of SMARCD1 was detected by immunoblotting assay and quantified by ImageJ software. (E) Representative image of high or low expression of SMARCD1 in HCC tumor tissues; scale bar = 50 μm. (F) The correlation of SMARCD1 expression with the OS time of HCC patients. *P<0.01, ***P<0.001.

Table 1
Correlations between SMARCD1 expression and clinicopathological features of HCC patients
VariablesCasesExpressionP
Low (n=66)High (n=84)
Age (years)     
  <50 71 27 44 0.162 
  ≥50 79 39 40  
Gender     
  Male 81 35 46 0.870 
  Female 69 31 38  
Tumor size (cm)     
  ≤5 70 39 31 0.007 
  >5 80 27 53  
AFP (ng/ml)     
  ≤20 64 29 35 0.780 
  >20 86 37 49  
Liver cirrhosis     
  Presence 77 30 47 0.202 
  Absence 73 36 37  
HBsAg     
  Positive 109 49 60 0.701 
  Negative 41 17 24  
TNM stage     
  I/II 74 41 33 0.005 
  III/IV 76 25 51  
Vascular invasion     
  Presence 66 25 41 0.181 
  Absence 84 41 43  
Multiplicity     
  Single 92 39 53 0.617 
  Multiple (≥2) 58 27 31  
VariablesCasesExpressionP
Low (n=66)High (n=84)
Age (years)     
  <50 71 27 44 0.162 
  ≥50 79 39 40  
Gender     
  Male 81 35 46 0.870 
  Female 69 31 38  
Tumor size (cm)     
  ≤5 70 39 31 0.007 
  >5 80 27 53  
AFP (ng/ml)     
  ≤20 64 29 35 0.780 
  >20 86 37 49  
Liver cirrhosis     
  Presence 77 30 47 0.202 
  Absence 73 36 37  
HBsAg     
  Positive 109 49 60 0.701 
  Negative 41 17 24  
TNM stage     
  I/II 74 41 33 0.005 
  III/IV 76 25 51  
Vascular invasion     
  Presence 66 25 41 0.181 
  Absence 84 41 43  
Multiplicity     
  Single 92 39 53 0.617 
  Multiple (≥2) 58 27 31  

SMARCD1 is crucial for cell proliferation

To investigate the biological function of SMARCD1 in HCC development, we selectively silenced SMARCD1 with two short-hairpin RNAs (shRNAs), which both effectively reduced SMARCD1 expression in the Huh7 HCC tumor cell line (Figure 5A). The CCK-8 assay results based on the first shSMARCD1 demonstrated that knockdown of SMARCD1 remarkably inhibited Huh7 cell viability (Figure 5B). In addition, the proliferation of shSmarcd1-treated cells was dramatically weakened according to EdU incorporation and colony formation assays, with 29.9% and 31.2% reductions, respectively (Figure 5C,D). Similar results were also observed when the other shSMARCD1 were used (Supplementary Figure S4). However, wound healing and Transwell analysis did not show any difference between shVector- and shSmarcd1-treated cells (Figure 5E–G). These results suggested that SMARCD1 is crucial for cell proliferation, while it is unlikely to play any notable role in tumor metastasis.

SMARCD1 is crucial for cell proliferation

Figure 5
SMARCD1 is crucial for cell proliferation

(A) The efficiency of shSmarcd1 was detected 72 h after transfection. (B) Cell viability was measured by CCK-8 assay. (C) Representative images of the EdU incorporation assay and corresponding statistical results. (D) The colony formation ability of Smarcd1-silenced cells was dramatically reduced. (E) The wound healing assay was carried out to evaluate cell movement alterations in shSmarcd1-treated cells. (F,G) Matrigel invasion and migration assays were carried out to evaluate the migration ability of shVector- and shSmarcd1-treated cells. (H,I) Tumor size was dramatically reduced in the shSmarcd1-treated group. (J) The expression of the proliferation marker Ki-67 was dramatically decreased in shSmarcd1-treated tumors; scale bar = 50 μm. *P<0.05, **P<0.01, n.s not significant.

Figure 5
SMARCD1 is crucial for cell proliferation

(A) The efficiency of shSmarcd1 was detected 72 h after transfection. (B) Cell viability was measured by CCK-8 assay. (C) Representative images of the EdU incorporation assay and corresponding statistical results. (D) The colony formation ability of Smarcd1-silenced cells was dramatically reduced. (E) The wound healing assay was carried out to evaluate cell movement alterations in shSmarcd1-treated cells. (F,G) Matrigel invasion and migration assays were carried out to evaluate the migration ability of shVector- and shSmarcd1-treated cells. (H,I) Tumor size was dramatically reduced in the shSmarcd1-treated group. (J) The expression of the proliferation marker Ki-67 was dramatically decreased in shSmarcd1-treated tumors; scale bar = 50 μm. *P<0.05, **P<0.01, n.s not significant.

To further investigate the role of SMARCD1 in cell proliferation in vivo, a subcutaneous tumor transplantation model was established in nude mice. The tumor volume was measured weekly. Although all the mice injected with Huh7 cells harbored tumors, the volume of tumors in the shSmarcd1-treated group were significantly smaller than those in the control group, indicating that knockdown of SMARCD1 markedly delayed tumor growth (Figure 5H,I). Correspondingly, a lower proliferation rate in the shSmarcd1-treated group was observed by Ki-67 staining (Figure 5J), which further confirmed the crucial role of SMARCD1 in cell proliferation.

SMARCD1 promotes cell proliferation by activating the mTOR signaling pathway

Known as the chromatin remodeling complex, the SWI/SNF complex widely participates in gene expression regulation by interacting and facilitating transcription factor binding to target genes. To further investigate the potential molecular mechanism by which SMARCD1 regulates cell proliferation, we conducted Gene Set Enrichment Analysis (GSEA) based on the TCGA HCC dataset using KEGG gene sets (c2.cp.kegg.v6.2.symbols). According to the results, we found that high expression of SMARCD1 was closely associated with the cell cycle, Wnt signaling pathway and mTOR signaling pathway (Figure 6A). In addition, Pearson correlation analysis by R revealed that SMARCD1 was highly positively correlated with key cell cycle regulators CDK4 and E2F4 (Figure 6B). Consistently, we found that CDK4, E2F4 and cyclin D1, all of which are key cell cycle regulators, were significantly decreased in SMARCD1-silenced cells (Figure 6C).

The role of SMARCD1 in cell proliferation is Wnt/β-catenin pathway-independent

Figure 6
The role of SMARCD1 in cell proliferation is Wnt/β-catenin pathway-independent

(A) GSEA analysis revealed that the genes correlated with SMARCD1 were enriched in the cell cycle, Wnt/β-catenin and mTOR signaling pathways. (B) Pearson analysis was carried out to evaluate the correlation between CDK4, E2F4 and SMARCD1. (C) CDK4, E2F4 and cyclin D1 were decreased in shSmarcd1-treated cells. (D) Immunofluorescence staining of β-catenin (red) and SMARCD1 (green); scale bar = 25 μm. (E) The detection of nuclear and cytoplasmic β-catenin in shVector- or shSmarcd1-treated cells. (F) The detection of nuclear β-catenin in SMARCD1-high or SMARCD1-low HCC tumor tissues.

Figure 6
The role of SMARCD1 in cell proliferation is Wnt/β-catenin pathway-independent

(A) GSEA analysis revealed that the genes correlated with SMARCD1 were enriched in the cell cycle, Wnt/β-catenin and mTOR signaling pathways. (B) Pearson analysis was carried out to evaluate the correlation between CDK4, E2F4 and SMARCD1. (C) CDK4, E2F4 and cyclin D1 were decreased in shSmarcd1-treated cells. (D) Immunofluorescence staining of β-catenin (red) and SMARCD1 (green); scale bar = 25 μm. (E) The detection of nuclear and cytoplasmic β-catenin in shVector- or shSmarcd1-treated cells. (F) The detection of nuclear β-catenin in SMARCD1-high or SMARCD1-low HCC tumor tissues.

The Wnt/β-catenin pathway is one of the most powerful cell cycle-regulating signaling pathways, which is largely dependent on the translocation of β-catenin from the cytoplasm to the nucleus [19]. To explore whether SMARCD1 regulates cell proliferation through the Wnt/β-catenin pathway, we detected the expression and location of β-catenin. Minimal alteration was found in the amount of total or active β-catenin (Figure 6D,E). Consistent with the findings in vitro, no obvious difference in active β-catenin was observed between SMARCD1high and SMARCD1low clinical HCC samples (Figure 6F), which indicated that the role of SMARCD1 in cell proliferation was not Wnt/β-catenin pathway-dependent.

However, we found that the expression of p-4E-BP1 and p-P70S6K, both of which are core functional proteins in the mTOR signaling pathway, was dramatically reduced in SMARCD1-silenced cells (Figure 7A, Supplementary Figure S4A). To further confirm the relationship between SMARCD1 expression and mTOR signaling, a specific inhibitor (INK128, MedChemExpress, China) and activator (MHY1485, MedChemExpress, China) targeting mTOR signaling were used [20,21]. Although both inhibitor and activator did not alter SMARCD1 expression, they actually changed the expression of p-4E-BP1 and p-P70S6K (Figure 7B). As demonstrated, both p-4E-BP1 and p-P70S6K were similarly expressed in shSmarcd1- and inhibitor-treated cells, while the activator rescued their expression in shSmarcd1-treated cells, which was consistent with the cell proliferation changes (Figure 7B,C). Decreased expression of p-4E-BP1 and p-P70S6K was also found in xenograft tumor tissues originating from shSmarcd1-treated cells (Figure 7D). In addition, compared with those in SMARCD1low clinical samples, higher expression of p-4E-BP1 and stronger proliferative capacity were observed in SMARCD1high clinical samples (Figure 7E).

SMARCD1 promotes cell proliferation by activating the mTOR signaling pathway

Figure 7
SMARCD1 promotes cell proliferation by activating the mTOR signaling pathway

(A) The detection of core functional proteins of the mTOR pathway in shVector- or shSmarcd1-treated cells. (B) Inhibition or activation of the mTOR pathway with or without shSmarcd1 treatment. (C) Cell proliferation ability of control, inhibitor, activator and shSmarcd1 cells treated with or without activator was assessed by EdU incorporation assay. (D) The detection of core functional proteins of the mTOR pathway and SMARCD1 in tumor tissues formed by shVector- or shSmarcd1-treated Huh7 cells. (E) IHC staining of p-4E-BP1 and Ki-67 in HCC samples on TMA slides. (F) Pearson correlation analysis of the relationship between SMARCD1 and SMARCA4.

Figure 7
SMARCD1 promotes cell proliferation by activating the mTOR signaling pathway

(A) The detection of core functional proteins of the mTOR pathway in shVector- or shSmarcd1-treated cells. (B) Inhibition or activation of the mTOR pathway with or without shSmarcd1 treatment. (C) Cell proliferation ability of control, inhibitor, activator and shSmarcd1 cells treated with or without activator was assessed by EdU incorporation assay. (D) The detection of core functional proteins of the mTOR pathway and SMARCD1 in tumor tissues formed by shVector- or shSmarcd1-treated Huh7 cells. (E) IHC staining of p-4E-BP1 and Ki-67 in HCC samples on TMA slides. (F) Pearson correlation analysis of the relationship between SMARCD1 and SMARCA4.

It has been validated that SMARCA4, a key subunit containing the ATPase domain in the SWI/SNF complex, can activate PI3K/AKT signaling, which is upstream of the mTOR pathway [22,23]. According to the TCGA HCC dataset, SMARCD1 was highly co-related to SMARCA4 and physically bound to SMARCA4 (Figures 1D and 7F), which indicated that the role of SMARCD1 in cell proliferation might rely on SMARCA4 through activating the PI3K/mTOR signaling pathway.

Discussion

A large number of genes are aberrantly expressed in HCC, indicating severe dysregulation of transcription regulators [2]. In this study, we analyzed the expression of each subunit of the SWI/SNF complex, which plays a comprehensive role in transcription control [24]. We reported that 14 out of the 15 SWI/SNF subunits were significantly increased in HCC tissues compared with normal tissues, and 11 members were significantly associated with the OS of patients. We identified a four-gene signature that can effectively predict OS in HCC patients. SMARCD1 was the most increased and most prognostically valuable subunit. We further revealed that SMARCD1 promotes liver cancer growth through activating the mTOR signaling pathway.

Although the SWI/SNF subunits are mutated in up to 40% of HCC patients, the mutation frequency individual member is very low, and the highest subunit with the highest rate, ARID1A, was mutated in only 10% of patients (Supplementary Figure S5). Growing evidence has established that aberrant expression but not mutation of SWI/SNF subunits is highly associated with the tumorigenesis of cancers, including HCC [17,25,26]. However, most existing studies focused on an individual gene, so there is still a lack of systematic and comprehensive analysis of the coordinated variation between SWI/SNF subunits, considering that the subunits exert separate but coordinative roles in controlling gene transcription. Based on the data downloaded from the TCGA HCC dataset, we found that 14/15 subunits of the SWI/SNF complex are increased in HCC tissues, which suggested a functional elevation of this complex. In addition, most of these subunits were highly correlated between any two factors based on their expression levels, which added further evidence of the concerted action between a single subunit and the entire complex. The SWI/SNF complex primarily remodels chromatin from a ‘condensed’ state to a ‘open’ state, promoting the binding of transcription factors and genes, which is consistent with the overall enhancement of transcriptional activity in malignancies [4].

Eleven subunits were found to be significantly associated with the OS of HCC patients; furthermore, we identified a four-gene prognostic signature including ACTL6A, ARID1A, SMARCC1 and SMARCD1, which could effectively predict OS in HCC patients. Among these genes, SMARCD1 was the most valuable for predicting prognosis, and overexpression of SMARCD1 was closely related to tumor size, TNM stage and shorter OS time. Interestingly, according to the TCGA HCC dataset and our fresh samples, SMARCD1 was also found to be the most significantly increased gene. Thus, we investigated the role of SMARCD1 in promoting HCC.

SMARCD1, also called BAF60a, belongs to the BAF60 family and is quite different from the other two family members BAF60b and BAF60c [27]. SMARCD1 plays a critical role in liver homeostasis maintenance. Genome-wide coactivation analysis identified SMARCD1 as an important regulator of lipid metabolism through its binding with peroxisome proliferator-activated receptor γ coactivator 1-α (PGC-1α), and silencing SMARCD1 results in liver steatosis [28]. In addition, SMARCD1 is rhythmically expressed in the liver, is regulated by miR-122, and acts as an integrator of the hepatic circadian clock and energy metabolism [29,30]. Given its important role in the liver, its abnormal expression is related to the impairment of liver homeostasis and tumorigenesis. Although SMARCD1 has been reported to be associated with stem cell differentiation and neuron development, its biological function in carcinogenesis remains largely unknown [31,32]. A few SWI/SNF subunits regulate cell proliferation. SMARCA2 is a master cell proliferation suppressor and is often epigenetically silenced in cancers [33]. Restored expression of SMARCA2 in SMARCA2-deficient cells results in cell cycle arrest. Most interestingly, SMARCA2 null mice are larger in size than wild-type littermates [34]. The SWI/SNF complex is identified as a mechano-regulated inhibitor of YAP and TAZ, both of which are critical for cell proliferation [35]. In addition, different subunits probably exhibit cell type-specific expression. For example, SMARCA4 is always highly expressed in cell types that constantly undergo proliferation or self-renewal, such as hepatoblasts, cholangiocytes or hepatic progenitor cells in the liver, whereas BRM is preferentially expressed in the brain, mature hepatocytes, fibromuscular stroma and endothelial cells, cell types not constantly engaged in proliferation or self-renewal [36]. Previous work has shown the role of SMARCA2, SMARCA4 and ARID1A in HCC development [14,15,17]. In the present work, we reported that high SMARCD1 expression promotes cell proliferation. In contrast with its effects in pancreatic cancer, SMARCD1 did not notably influence on cell invasion in HCC.

The Wnt/β-catenin and mTOR pathways are critical for cell survival, metabolism, proliferation, etc., and both are related to the SWI/SNF complex. According to our analysis, the genes of these pathways are enriched in samples highly expressing SMARCD1. Canonical Wnt/β-catenin signaling activation is achieved by the translocation of β-catenin from the cytoplasm to the nucleus; however, we failed to detect an alteration of nuclear β-catenin expression following SMARCD1 knockdown or between SMARCD1high and SMARCD1low HCC samples, and this finding is also different from that in pancreatic cancer [37]. Instead, we confirmed that the role of SMARCD1 in cell proliferation was largely dependent on mTOR signaling activation.

Conclusion

Collectively, our study profiled the differential expression of SWI/SNF subunits between HCC and normal controls. Accordingly, we constructed a robust prognostic signature that might serve as a promising prognostic predictor for HCC and provide guidance for selecting therapeutic strategies. Finally, we reported that the elevation of SMARCD1 in tumor tissues promotes cell proliferation by activating the mTOR signaling pathway, providing new insights into the molecular mechanism underlying HCC development.

Clinical perspectives

  • The alteration of SWI/SNF complex subunits plays critical role in tumorigenesis, including HCC, while the individual role and clinical significance of each subunit remains largely unknown.

  • Fourteen out of the fifteen SWI/SNF subunits are significantly increased in HCC tissues, eleven of them are significantly associated with OS and a four-gene signature (ACTL6A, ARID1A, SMARCC1 and SMARCD1) effectively predicts OS in HCC patients.

  • SMARCD1, the most elevated gene, promotes liver cancer cell proliferation by activating mTOR signaling pathway.

Data Availability

The data in the present study are available from the corresponding authors upon reasonable request.

Competing Interests

The authors declare that there are no competing interests associated with the manuscript.

Funding

This work was supported by the Natural Science Foundation of China [grant numbers 81772538, 81800449]; the ‘Post-Doctor Research Project, West China Hospital, Sichuan University [grant number 2018HXBH019]’; the ‘Postdoctoral Interdisciplinary Innovation Incubation Fund, Sichuan University’; and the Key Research and Development program of Sichuan Provincial Department of Science and Technology [grant number 20ZDYF2916].

Author Contribution

Conceptualization: Yongjie Zhou and Yujun Shi. TCGA data analysis: Yongjie Zhou and Qing Xu. Clinical sample collection and analysis: Lv Tao and Changli Lu. Funding acquisition: Yongjie Zhou, Changli Lu and Lv Tao. Investigation: Yongjie Zhou, Qi Xu, Yuwei Chen, and Yuke Shu. Methodology: Yongjie Zhou and Zhenru Wu. Project administration: Changli Lu and Yujun Shi. Supervision: Yujun Shi and Hong Bu. Visualization: Yongjie Zhou. Manuscript preparing: Yongjie Zhou and Yujun Shi.

Ethics Approval

The research protocol was reviewed and approved by the Ethics Review Committees of West China Hospital, Sichuan University, and written informed consent was obtained from each patient included in the study. The animal handling and care procedures were conducted in accordance with national and international laws and policies and approved by the Animal Care and Use Committee of Sichuan University.

Consent for Publication

We received consent for publication from the individual patients who participated in the present study.

Acknowledgements

We thank Mingyang Shao, Xiaoyue Cao and Menglin Chen for the review of pathological sections.

Abbreviations

     
  • ACTL6A

    actin-like 6A

  •  
  • ARID1A

    AT-rich interaction domain 1A

  •  
  • ARID2

    AT-rich interaction domain 2

  •  
  • BAF

    BRG1-associated factor

  •  
  • HCC

    hepatocellular carcinoma

  •  
  • HR

    hazard ratio

  •  
  • IHC

    immunohistochemistry

  •  
  • OS

    overall survival

  •  
  • PBAF

    polybromo BAF

  •  
  • PGC-1α

    peroxisome proliferator-activated receptor γ coactivator 1-α

  •  
  • shRNA

    short-hairpin RNA

  •  
  • SMARCA2/4

    SWI/SNF related, matrix associated, actin dependent regulator of chromatin 2/4

  •  
  • SMARCB1

    SWI/SNF related, matrix associated, actin dependent regulator of chromatin, subfamily B, member 1

  •  
  • SMARCC1

    SWI/SNF related, matrix associated, actin dependent regulator of chromatin subfamily C member 1

  •  
  • SMARCD1

    SWI/SNF related, matrix associated, actin dependent regulator of chromatin, subfamily D, member 1

  •  
  • TCGA

    The Cancer Genome Atlas

  •  
  • TMA

    tissue microarray

References

References
1.
Chen
W.
,
Zheng
R.
,
Baade
P.D.
,
Zhang
S.
,
Zeng
H.
,
Bray
F.I.
et al.
(
2016
)
Cancer statistics in China, 2015
.
CA Cancer J. Clin.
66
,
115
132
[PubMed]
2.
Ally
A.
,
Balasundaram
M.
,
Carlsen
R.
,
Chuah
E.
,
Clarke
A.
,
Dhalla
N.
et al.
(
2017
)
Comprehensive and integrative genomic characterization of hepatocellular carcinoma
.
Cell
169
,
1327
1341
[PubMed]
3.
Helming
K.C.
,
Wang
X.
and
Roberts
C.
(
2014
)
Vulnerabilities of mutant SWI/SNF complexes in cancer
.
Cancer Cell
26
,
309
317
[PubMed]
4.
Kadoch
C.
and
Crabtree
G.R.
(
2015
)
Mammalian SWI/SNF chromatin remodeling complexes and cancer: Mechanistic insights gained from human genomics
.
Sci. Adv.
1
,
e1500447
[PubMed]
5.
Trotter
K.W.
and
Archer
T.K.
(
2008
)
The BRG1 transcriptional coregulator
.
Nucl. Recept. Signal.
6
,
e004
[PubMed]
6.
Guidi
C.J.
,
Sands
A.T.
,
Zambrowicz
B.P.
,
Turner
T.K.
,
Demers
D.A.
,
Webster
W.
et al.
(
2001
)
Disruption of Ini1 leads to peri-implantation lethality and tumorigenesis in mice
.
Mol. Cell. Biol.
21
,
3598
3603
[PubMed]
7.
Klochendler-Yeivin
A.
,
Fiette
L.
,
Barra
J.
,
Muchardt
C.
,
Babinet
C.
and
Yaniv
M.
(
2000
)
The murine SNF5/INI1 chromatin remodeling factor is essential for embryonic development and tumor suppression
.
EMBO Rep.
1
,
500
506
[PubMed]
8.
Gao
X.
,
Tate
P.
,
Hu
P.
,
Tjian
R.
,
Skarnes
W.C.
and
Wang
Z.
(
2008
)
ES cell pluripotency and germ-layer formation require the SWI/SNF chromatin remodeling component BAF250a
.
Proc. Natl. Acad. Sci. U.S.A.
105
,
6656
6661
[PubMed]
9.
Lickert
H.
,
Takeuchi
J.K.
,
Von Both
I.
,
Walls
J.R.
,
McAuliffe
F.
,
Adamson
S.L.
et al.
(
2004
)
Baf60c is essential for function of BAF chromatin remodelling complexes in heart development
.
Nature
432
,
107
112
[PubMed]
10.
Reyes
J.C.
,
Barra
J.
,
Muchardt
C.
,
Camus
A.
,
Babinet
C.
and
Yaniv
M.
(
1998
)
Altered control of cellular proliferation in the absence of mammalian brahma (SNF2alpha)
.
EMBO J.
17
,
6979
6991
[PubMed]
11.
Roberts
C.W.
,
Leroux
M.M.
,
Fleming
M.D.
and
Orkin
S.H.
(
2002
)
Highly penetrant, rapid tumorigenesis through conditional inversion of the tumor suppressor gene Snf5
.
Cancer Cell
2
,
415
425
[PubMed]
12.
Bultman
S.J.
,
Herschkowitz
J.I.
,
Godfrey
V.
,
Gebuhr
T.C.
,
Yaniv
M.
,
Perou
C.M.
et al.
(
2008
)
Characterization of mammary tumors from Brg1 heterozygous mice
.
Oncogene
27
,
460
468
[PubMed]
13.
Li
W.
,
Yang
L.
,
He
Q.
,
Hu
C.
,
Zhu
L.
,
Ma
X.
et al.
(
2019
)
A homeostatic Arid1a-dependent permissive chromatin state licenses hepatocyte responsiveness to liver-injury-associated YAP signaling
.
Cell Stem Cell
25
,
54.e5
68.e5
14.
Sun
X.
,
Wang
S.C.
,
Wei
Y.
,
Luo
X.
,
Jia
Y.
,
Li
L.
et al.
(
2017
)
Arid1a has context-dependent oncogenic and tumor suppressor functions in liver cancer
.
Cancer Cell
32
,
574.e6
589.e6
15.
Endo
M.
,
Yasui
K.
,
Zen
Y.
,
Gen
Y.
,
Zen
K.
,
Tsuji
K.
et al.
(
2013
)
Alterations of the SWI/SNF chromatin remodelling subunit-BRG1 and BRM in hepatocellular carcinoma
.
Liver Int.
33
,
105
117
[PubMed]
16.
Xiao
S.
,
Chang
R.M.
,
Yang
M.Y.
,
Lei
X.
,
Liu
X.
,
Gao
W.B.
et al.
(
2016
)
Actin-like 6A predicts poor prognosis of hepatocellular carcinoma and promotes metastasis and epithelial-mesenchymal transition
.
Hepatology
63
,
1256
1271
[PubMed]
17.
Chen
Z.
,
Lu
X.
,
Jia
D.
,
Jing
Y.
,
Chen
D.
,
Wang
Q.
et al.
(
2018
)
Hepatic SMARCA4 predicts HCC recurrence and promotes tumour cell proliferation by regulating SMAD6 expression
.
Cell Death Dis.
9
,
59
[PubMed]
18.
Wilkerson
M.D.
and
Hayes
D.N.
(
2010
)
ConsensusClusterPlus: a class discovery tool with confidence assessments and item tracking
.
Bioinformatics
26
,
1572
1573
[PubMed]
19.
Davidson
G.
and
Niehrs
C.
(
2010
)
Emerging links between CDK cell cycle regulators and Wnt signaling
.
Trends Cell Biol.
20
,
453
460
[PubMed]
20.
Hsieh
A.C.
,
Liu
Y.
,
Edlind
M.P.
,
Ingolia
N.T.
,
Janes
M.R.
,
Sher
A.
et al.
(
2012
)
The translational landscape of mTOR signalling steers cancer initiation and metastasis
.
Nature
485
,
55
61
[PubMed]
21.
Gao
L.
,
Lv
G.
,
Li
R.
,
Liu
W.T.
,
Zong
C.
,
Ye
F.
et al.
(
2019
)
Glycochenodeoxycholate promotes hepatocellular carcinoma invasion and migration by AMPK/mTOR dependent autophagy activation
.
Cancer Lett.
454
,
215
223
[PubMed]
22.
Watanabe
T.
,
Semba
S.
and
Yokozaki
H.
(
2011
)
Regulation of PTEN expression by the SWI/SNF chromatin-remodelling protein BRG1 in human colorectal carcinoma cells
.
Br. J. Cancer
104
,
146
154
[PubMed]
23.
Jubierre
L.
,
Soriano
A.
,
Planells-Ferrer
L.
,
París-Coderch
L.
,
Tenbaum
S.P.
,
Romero
O.A.
et al.
(
2016
)
BRG1/SMARCA4 is essential for neuroblastoma cell viability through modulation of cell death and survival pathways
.
Oncogene
35
,
5179
5190
[PubMed]
24.
Wang
Y.
,
Hoang
L.
,
Ji
J.X.
and
Huntsman
D.G.
(
2020
)
SWI/SNF complex mutations in gynecologic cancers: molecular mechanisms and models
.
Annu Rev. Pathol.
15
,
467
492
[PubMed]
25.
Guerrero-Martínez
J.A.
and
Reyes
J.C.
(
2018
)
High expression of SMARCA4 or SMARCA2 is frequently associated with an opposite prognosis in cancer
.
Sci. Rep.
8
,
2043
[PubMed]
26.
Wang
P.
,
Song
X.
,
Cao
D.
,
Cui
K.
,
Wang
J.
,
Utpatel
K.
et al.
(
2020
)
Oncogene-dependent function of BRG1 in hepatocarcinogenesis
.
Cell Death Dis.
11
,
91
[PubMed]
27.
Zhang
P.
,
Li
L.
,
Bao
Z.
and
Huang
F.
(
2016
)
Role of BAF60a/BAF60c in chromatin remodeling and hepatic lipid metabolism
.
Nutr. Metab.
13
,
30
28.
Li
S.
,
Liu
C.
,
Li
N.
,
Hao
T.
,
Han
T.
,
Hill
D.E.
et al.
(
2008
)
Genome-wide coactivation analysis of PGC-1alpha identifies BAF60a as a regulator of hepatic lipid metabolism
.
Cell Metab.
8
,
105
117
[PubMed]
29.
Gatfield
D.
,
Le Martelot
G.
,
Vejnar
C.E.
,
Gerlach
D.
,
Schaad
O.
,
Fleury-Olela
F.
et al.
(
2009
)
Integration of microRNA miR-122 in hepatic circadian gene expression
.
Genes Dev.
23
,
1313
1326
[PubMed]
30.
Tao
W.
,
Chen
S.
,
Shi
G.
,
Guo
J.
,
Xu
Y.
and
Liu
C.
(
2011
)
SWItch/sucrose nonfermentable (SWI/SNF) complex subunit BAF60a integrates hepatic circadian clock and energy metabolism
.
Hepatology
54
,
1410
1420
[PubMed]
31.
Alajem
A.
,
Biran
A.
,
Harikumar
A.
,
Sailaja
B.S.
,
Aaronson
Y.
,
Livyatan
I.
et al.
(
2015
)
Differential association of chromatin proteins identifies BAF60a/SMARCD1 as a regulator of embryonic stem cell differentiation
.
Cell Rep.
10
,
2019
2031
[PubMed]
32.
Nixon
K.
,
Rousseau
J.
,
Stone
M.H.
,
Sarikahya
M.
,
Ehresmann
S.
,
Mizuno
S.
et al.
(
2019
)
A syndromic neurodevelopmental disorder caused by mutations in SMARCD1, a core SWI/SNF subunit needed for context-dependent neuronal gene regulation in flies
.
Am. J. Hum. Genet.
104
,
596
610
[PubMed]
33.
Coisy-Quivy
M.
,
Disson
O.
,
Roure
V.
,
Muchardt
C.
,
Blanchard
J.M.
and
Dantonel
J.C.
(
2006
)
Role for Brm in cell growth control
.
Cancer Res.
66
,
5069
5076
[PubMed]
34.
Reyes
J.C.
,
Barra
J.
,
Muchardt
C.
,
Camus
A.
,
Babinet
C.
and
Yaniv
M.
(
1998
)
Altered control of cellular proliferation in the absence of mammalian brahma (SNF2alpha)
.
EMBO J.
17
,
6979
6991
[PubMed]
35.
Chang
L.
,
Azzolin
L.
,
Di Biagio
D.
,
Zanconato
F.
,
Battilana
G.
,
Lucon Xiccato
R.
et al.
(
2018
)
The SWI/SNF complex is a mechanoregulated inhibitor of YAP and TAZ
.
Nature
563
,
265
269
[PubMed]
36.
Reisman
D.N.
,
Sciarrotta
J.
,
Bouldin
T.W.
,
Weissman
B.E.
and
Funkhouser
W.K.
(
2005
)
The expression of the SWI/SNF ATPase subunits BRG1 and BRM in normal human tissues
.
Appl. Immunohistochem. Mol. Morphol.
13
,
66
74
[PubMed]
37.
Liu
F.
,
Xia
Z.
,
Zhang
M.
,
Ding
J.
,
Feng
Y.
,
Wu
J.
et al.
(
2019
)
SMARCAD1 promotes pancreatic cancer cell growth and metastasis through Wnt/β-catenin-mediated EMT
.
Int. J. Biol. Sci.
15
,
636
646
[PubMed]