SMYD3 expression is upregulated in HCC tissues
To explore the key regulatory factors affecting the tumorigenesis and metastasis of HCC, we investigated the transcriptomes in 5 pairs of HCC and adjacent normal tissues using a high-throughput RNA deep sequencing approach (RNA-seq), and 3333 differentially expressed genes were identified (Additional file 2). Then, the RNA-seq results were integrated with the differentially expressed genes of HCC and adjacent tissues in the TCGA database (2207 genes) and The Human Transcription Factors database (1639 genes) [33], and 24 differentially expressed regulatory factors were found (Fig. 1a). These 24 genes are shown in Fig. 1b, and among them, we were interested in SMYD3. SMYD3, which belongs to the SMYD family, could catalyze histone and nonhistone methylation and is a potential transcription factor. Analysis of the TCGA database using cBioportal (http://www.cbioportal.org/) [34] indicated that SMYD3 mutation frequencies were in nearly 8% of HCC cases, and these alterations were caused by gene amplification, mutation, and deletion, of which, gene amplification accounted for the majority (Fig. 1c). The results from TCGA database also indicated increased expression levels of SMYD3 in tumor tissues compared to noncancerous tissues, and its expression tended to be positively correlated with disease stage progression (Fig. 1d). In addition, we investigated the role of SMYD3 in tumor development and progression in vivo by subcutaneously implanting LM3 cells that had been engineered to stably express SMYD3 shRNA or control scrambled shRNA into athymic BALB/c mice. Growth of the implanted tumors was monitored by measuring tumor size every 3 days over a period of 4 weeks. The results revealed that tumor growth was substantially decreased in response to SMYD3 knockdown. The average tumor weight/volume of mice in the SMYD3 interference group was 46.4%/44.9% of that in the control group (Fig. 1e). To confirm this finding, we collected 12 pairs of HCC and adjacent noncancerous tissues and found that the mRNA/protein level of SMYD3 was upregulated in the tumor tissues (Fig. 1f. g). To assess the function of SMYD3 in tumor metastasis, LM3 cells stably expressing firefly luciferase were infected with lentiviruses carrying shSMYD3 or shSCR. Then, the cells were intravenously injected into immunocompromised severe combined immunodeficient (SCID) male mice (n = 5). Metastatic tumors were measured using quantitative bioluminescence imaging after 4 weeks using an IVIS imaging system (Xenogen). We found that SMYD3 deficiency significantly reduced the number of tumor nodules and HCC cell lung metastasis in vivo (Fig. 1h). These results indicated that SMYD3 is highly expressed in tumor tissues and may play an important role in tumor proliferation and metastasis.
SMYD3 accelerates HCC proliferation and invasion
To investigate the function of SMYD3 in HCC, we first used an enhanced BrdU (EdU) incorporation assay and found that SMYD3 depletion was associated with a decreased mitotic rate compared to controls in both LM3 and SK-HEP-1 cells (Fig. 2a, b). Using colony formation assays, we found that in LM3 cells, SMYD3 knockdown was associated with a marked decrease in colony number, whereas SMYD3 overexpression was associated with a significant increase in colony number in LM3 cells (Fig. 2c). Consistent results were obtained in SK-HEP-1 cells (Fig. 2d). Growth curve assays revealed that SMYD3 knockdown significantly reduced cell growth and that SMYD3 overexpression markedly accelerated cell growth in both LM3 cells (Fig. 2e) and SK-HEP-1 cells (Fig. 2f). In addition, the impact of loss of function of SMYD3 on the migration and invasion of these cells was assessed using wound-healing assays and transwell invasion assays. In the wound-healing assays, the remaining open distance of cells with altered SMYD3 levels was different from that of the control group 24 or 36 h after migration. SMYD3 knockdown was associated with a decreased migration rate in LM3 (Fig. 2g) and SK-HEP-1 (Fig. 2h). The results from transwell invasion assays in two HCC cell lines (LM3 and SK-HEP-1) revealed significantly decreased invasiveness in response to SMYD3 knockdown (Fig. 2i). Furthermore, using a TUNEL assay, we found that SMYD3 knockdown elevated the apoptosis rate of LM3 cells (Fig. 2j). Quality testing of the interference efficiency of SMYD3 and SMYD3 overexpression in LM3 and SK-HEP-1 cells is shown in Fig. 2k. These findings support the notion that SMYD3 accelerates HCC cell proliferation and invasion and inhibits apoptosis.
Proteomic analysis of the SMYD3 interactomes
In an effort to better understanding the mechanistic role of SMYD3, affinity purification and mass spectrometry assays were used to identify proteins associated with SMYD3 in vivo. In these experiments, FLAG-tagged SMYD3 or an empty vector was stably expressed in LM3 cells. Cellular extracts were prepared and subjected to affinity purification using an anti-FLAG affinity gel. Immunocomplexed proteins were separated using SDS–PAGE and silver stained (Fig. 3a, b). Immunoprecipitated proteins in specific bands in comparison to the vector were gel extracted, trypsin digested, and identified using liquid chromatography tandem mass spectrometry. Surprisingly, > 700 unique proteins with a protein score equal to or higher than two were identified (Additional file 3). To identify putative functional processes associated with SMYD3-interacting proteins, we next performed KEGG pathway [35] enrichment analysis and Gene Ontology (GO) analysis [36] for these interacting proteins. KEGG pathways analysis showed that SMYD3-interacting proteins are associated with regulation of many vital biological processes and activities (Fig. 3c). The top-ranked categories of biological process (BP) analysis were nucleocytoplasmic transport, DNA replication, telomere maintenance, positive regulation of chromosome organization, histone modification, and so on, suggesting that SMYD3 is related to gene transcriptional regulation and expression (Fig. 3d). In addition, cellular component (CC) analysis showed that SMYD3-interacting proteins were related to chromosomal region, cytosolic part, cell-substrate adherens junction, NuRD complex, methyltransferase complex, PcG protein complex, Cul4-RING E3 ubiquitin ligase complex, and so on, implying that SMYD3 is likely to participate in cell adhesion and invasion (Fig. 3e). Molecular function (MF) analysis identified many predominant themes, including cell adhesion molecule binding, ubiquitin protein ligase binding, histone binding, and RNA methyltransferase activity. This indicates that SMYD3 may be involved in epigenetic regulation, such as DNA/RNA or histone modification and chromatin remodeling (Fig. 3f). To further research the functional relationship between SMYD3-interacting proteins and specific functional complexes, a PPI network of the identified proteins was constructed using the STRING online database (http://string-db.org) [37]. We are interested in the role of SMDY3 in epigenetic regulation, so we selected 76 proteins involved in epigenetic regulation among more than 700 SMYD3 interacting proteins and the PPI analysis revealed multiple SMYD3-associated complexes, which suggest previously unknown functions of SMYD3 (Additional file 4: Fig. S1). Notably, the PPI analysis identified nearly every component of the NuRD complex, which is important in histone deacetylation and chromatin remodeling.
SMYD3 physically associates with the NuRD complex
To further confirm the in vivo interaction between SMYD3 and its interacting proteins, total proteins from LM3 and SK-HEP-1 cells were extracted, and co-IP was performed using antibodies detecting endogenous proteins. IP with antibodies against SMYD3 followed by IB with antibodies against DDB1, CUL4B, LDHA, METTL3, METTL1, and SIRT1 demonstrated that these proteins were efficiently coimmunoprecipitated with SMYD3 in LM3 and SK-HEP-1 cells (Fig. 4a, b). As previously mentioned, almost every component of the NuRD complex was identified in the mass spectrometry results. Therefore, we conducted IP using antibodies against SMYD3 followed by IB with antibodies against components of the NuRD complex and found that they were efficiently coimmunoprecipitated with SMYD3 in LM3 cells. In turn, IP with antibodies against representative components of the NuRD complex and IB with antibodies against SMYD3 reinforced the finding that SMYD3 was efficiently coimmunoprecipitated with the NuRD complex components (Fig. 4c). The same results were obtained in SK-HEP-1 cells (Fig. 4d). In addition, we explored the molecular basis for the interaction between SMYD3 and the NuRD complex by using GST pull-down assays. We used GST-fused SMYD3 (Fig. 4e) and in vitro transcribed/translated components of the NuRD complex. These experiments revealed that SMYD3 directly interacts with MTA1/2 and RBBP4 (Fig. 4f). To further investigate the physical association and to examine the functional connection between SMYD3 and the NuRD complex, enzymatic activity assay was performed to investigate whether SMYD3 associated with an HDAC enzymatic activity. The immunoprecipitates (IPs) were first incubated with bulk histones isolated from LM3 cells, and the levels of acetylated histones in the reactions were then analyzed by western blotting. Notably, the SMYD3-containing complex indeed possessed an enzymatic activity that led to a significant decrease in the acetylation level of H4 (Fig. 4g). The results indicate that the SMYD3-containing complex possesses histone deacetylation activities by interacting with the NuRD complex.
Transcription target analysis for the SMYD3/NuRD complex
To further investigate and understand the biological significance of the interaction between SMYD3 and the NuRD (MTA1/MTA2) complex, we investigated the transcriptomes of SMYD3-, MTA1-, or MTA2-deficient LM3 cells using a high-throughput RNA deep sequencing approach (RNA-seq). Total RNA was extracted from LM3 cells transfected with control shRNA (shSCR) or shRNA-targeting SMYD3, MTA1, or MTA2. RNA-seq analysis identified 862, 693, and 941 genes whose expression was altered in response to SMYD3, MTA1, and MTA2 depletion, respectively (Fig. 5a, Additional file 5). Cross-analysis of the transcriptomes from SMYD3-, MTA1-, and MTA2-deficient cells identified 262 genes whose expression was altered in SMYD3-, MTA1-, and MTA2-depleted cells. These target genes were coregulated by SMYD3, MTA1, and MTA2 (Fig. 5b). To identify the putative functional processes associated with the targets that were coregulated by SMYD3, MTA1, and MTA2, we performed KEGG pathway enrichment analysis and classified the genes into various cellular signaling pathways, including ECM-receptor interaction, protein digestion and absorption, FoxO signaling pathway, focal adhesion, and p53 signaling pathway, which are all critically involved in cell growth, migration, and invasion (Fig. 5c). We also performed Gene Ontology (GO) analysis. The top-ranked categories of the biological process (BP) analysis included negative regulation of cell growth, negative regulation of canonical Wnt signaling pathway, regulation of insulin-like growth factor receptor signaling pathway, cell adhesion, and regulation of transcription. (Fig. 5d), suggesting that the association of SMYD3, MTA1, and MTA2 may be related to proliferation, epithelial-mesenchymal transition, and gene transcription regulation. Gene set enrichment analyses (GSEAs) [38] revealed enrichment in the Wnt β-catenin signaling pathway and Notch signaling pathway in response to SMYD3, MTA1, or MTA2 knockdown (Fig. 5e).
To validate the RNA-seq analysis, we chose 8 representative genes and validated their expression in LM3 cells using quantitative real-time PCR (qPCR). Figure 6a shows the 8 representative genes whose expression were altered in RNA-seq data. qPCR results indicated that mRNA levels of IGFBP4, CDH10, CHD5, DLG5, CASP7, TPM4, NOLC1, and OGFR were increased upon the knockdown of either SMYD3 or MTA1/MTA2 compared to control siRNA (Fig. 6b). Then, we examined whether SMYD3 and MTA1/MTA2 regulated target gene expression by binding to their promoters. qChIP assays in LM3 cells using antibodies against SMYD3, MTA1, MTA2, or control IgG revealed that SMYD3 and MTA1/MTA2 co-occupied the promoters of IGFBP4, CASP7, and NOLC1. The ChIP results were quantitated using qPCR (Fig. 6c). To further test our proposition that SMYD3 and MTA1/MTA2 function in the same protein complex at target promoters, we performed sequential ChIP/Re-ChIP using antibodies against SMYD3, MTA1, MTA2, HDAC1, HDAC2, and RBBP4 for IGFBP4. The results demonstrated that the IGFBP4 promoter that was immunoprecipitated with antibodies against SMYD3 could be reimmunoprecipitated with antibodies against MTA1, MTA2, HDAC1, HDAC2, and RBBP4. Similar results were obtained when the initial ChIP was performed using antibodies against MTA1 or MTA2 (Fig. 6d). These results support the conclusion that SMYD3 and MTA1 or MTA2 occupy co-target gene promoters as functionally collaborative protein complexes. qChIP analyses showed that SMYD3 knockdown led to a significant reduction in the binding of MTA1 or MTA2 to the promoters of IGFBP4, suggesting that SMYD3 could specifically affect NuRD complex to the target gene promoters. Notably, knockdown of either SMYD3 or MTA1 or MTA2 led to a significant decrease in H4K20me3 and a marked increase in H4Ac at the IGFBP4 promoter (Fig. 6e). When we interfered with SMYD3, MTA1, or MTA2 using the corresponding siRNA, protein levels of IGFBP4 were remarkably increased (Fig. 6f).
To further investigate the target genes specifically regulated by the SMYD3/NuRD (MTA1) complex or the SMYD3/NuRD (MTA2) complex, we next analyzed RNA-seq result and found that there are 83 target genes specifically regulated by SMYD3 and MTA1, and 142 target genes specifically regulated by SMYD3 and MTA2. To validate the RNA-seq analysis, we chose 6 representative genes which were negatively regulated by SMYD3/MTA1 or SMYD3/MTA2 (Fig. 7a), and validated their expression in LM3 cells using quantitative real-time PCR (qPCR). The results indicate that the mRNA levels of GRAMD4, MYO9B, and LOXL4 increased upon the knockdown of either SMYD3 or MTA1, but no significant change upon the knockdown of MTA2. Similarly, the mRNA levels of PCDH9, PRDM5, and DACH1 increased upon the knockdown of either SMYD3 or MTA2, but no significant change upon the knockdown of MTA1 (Fig. 7b). qChIP assays with antibodies against SMYD3, MTA1, MTA2, or control IgG revealed that SMYD3 and MTA1 specifically co-occupied the promoters of GRAMD4, MYO9B, and LOXL4. SMYD3 and MTA2 specifically co-occupied the promoters of PCDH9 and PRDM5, but not DACH1 (Fig. 7c). To further test whether SMYD3 affects the NuRD complex binding to chromatin, chromatin-binding proteins were extracted from the SMYD3 knockdown LM3 cells. The results showed that as compared to control, the amount of chromatin-bound MTA1, MTA2, HDAC1, and HDAC2 was slightly decreased with SMYD3 knockdown (Fig. 7d). In addition, knockdown of HDAC1 or HDAC2 led to an increased expression of global H4 acetylation level in LM3 cells (Additional file 6: Fig. S2a). To further investigate the function of SMYD3 independent of NuRD complex, we performed KEGG pathway analysis (Fig. 7e) and GO analysis (Fig. 7f) on 375 target genes exclusively regulated by SMYD3. The results indicated that SMYD3 was associated with regulation of many vital cell processes and activities, such as signaling pathways regulating pluripotency of stem cells and response to hypoxia, which was different from the signaling pathway regulated by coordinating with the NuRD complex, suggesting that SMYD3 is also involved in vital biological processes that are independent of the NuRD complex.
Based on our observations that SMYD3 and the NuRD (MTA1/MTA2) complex are physically and functionally associated, we next investigated if MTA1 or MTA2 have the same role of SMYD3 in tumor proliferation and invasion. We performed EdU assays and found that knockdown of MTA1 or MTA2 were associated with a decreased mitotic rate compared to the control, which have the same effects as SMYD3 knockdown (Fig. 8a). In addition, using transwell invasion assays, we found that knockdown of MTA1 or MTA2 was associated with a marked reduction in cell invasion, consistent with the effects as SMYD3 knockdown (Fig. 8b). To assess whether the documented phenotypic effects were due to SMYD3 methyltransferase activity, we constructed catalytic mutant SMYD3 vector (SMYD3 ΔSET). Using growth curve assays and transwell invasion assays, we found that the decreased cell viability and invasiveness upon SMYD3 knockdown was retrieved by the overexpression of WT SMYD3, but could not be rescued by the ectopic expression of SMYD3 ΔSET (Fig. 8c, d), indicating that SMYD3 putative oncogenic function is associated with the methyltransferase activity. The western blots verified the efficacies of the siRNAs and plasmids used in these experiments (Additional file 6: Fig. S2b).
Moreover, we next performed rescue experiments. Using an EdU assay, we found that deletion of SMYD3 markedly decreased LM3 cell proliferation, while deletion of IGFBP4 had the opposite effect. In addition, suppression of cell proliferation caused by SMYD3 deletion could be rescued by reducing the expression of IGFBP4 (Additional file 7: Fig. S3a). In addition, results from transwell assays performed using LM3 cells demonstrated that deletion of SMYD3 caused a substantial reduction in cell invasion, and deletion of IGFBP4 significantly increased invasion. However, inhibition of the invasive potential of LM3 cells associated with deletion of SMYD3 was diminished by reducing the expression of IGFBP4 (Additional file 7: Fig. S3b). We confirmed the efficiency of the siRNAs using western blot analysis (Additional file 7: Fig. S3c).
Inhibitors of SMYD3 effectively inhibit the proliferation of HCC cells
Because SMYD3 plays an important role in the progression of HCC, BCI-121 and EPZ031686, which are selective SMYD3 inhibitors, were used to treat HCC cells. We performed MTT assays and found that cell viability was markedly reduced after treatment with BCI-121 or EPZ031686; in addition, cell viability was time- and concentration-dependent. The same results were obtained in LM3 cells and SK-HEP-1 cells (Fig. 8e, f). In addition, using colony formation assays, we revealed that colony numbers were significantly reduced when LM3 cells and SK-HEP-1 cells were treated with BCI-121 or EPZ031686 (Fig. 8g, h). These findings confirmed that SMYD3 inhibition using BCI-121 or EPZ031686 effectively alleviated HCC cell viability and proliferative potential.
We analyzed expression levels of SMYD3 and IGFBP4 in the TCGA database and in two published clinical datasets (GSE45436 and GSE121248) from the GEO database (Fig. 9a). The results revealed that expression of SMYD3 was upregulated, while IGFBP4 was downregulated in liver cancer tissues compared to noncancerous liver tissues. In addition, we found the expression of SMYD3 is significantly positively correlated with PCNA and MKI67 expression, which is a well-known marker of proliferation, in TCGA database (Fig. 9b). A reanalysis of the data sourced from published clinical datasets, such as GSE45436, GSE102079, GSE104580, GSE121248, GSE107170, and TCGA datasets, indicated that expression of SMYD3 is significantly negatively correlated with IGFBP4 expression, supporting our finding that IGFBP4 is transcriptionally inhibited by SMYD3 (Fig. 9c). In addition, Kaplan–Meier survival analysis (http://kmplot.com/analysis/) [39] revealed that lower expression of SMYD3 or overexpression of IGFBP4 was associated with improved overall survival in liver cancer patients. In addition, poor overall survival was associated with high expression of SMYD3 combined with low expression of IGFBP4 (Fig. 9d). In summary, these data support our overall hypothesis that SMYD3 cooperates with the NuRD (MTA1/MTA2) complex to inhibit expression of a series of tumor suppressor genes, leading to tumor progression (Fig. 9e).