Dimethyl fumarate (DMF) represses the tumorigenicity of hepatocellular carcinoma cells in vivo
To investigate if DMF displays anticancer effects, eight-week-old female athymic BALB/c nude mice were engrafted with Huh7 cells (Fig. 1a). When the tumours became palpable, the mice received an intraperitoneal injection of DMF (30 mg/kg) or vehicle daily. At the end of 14 days, tumours were excised and measured. As shown in Fig. 1b-d, the tumour volumes after treatment were 605.84 ± 65.06 mm3 and 232.1± 57.14 mm3 (mean ± standard deviation) in the control and treatment groups, respectively (p ≤ 0.01). Tumour weight in the treatment group significantly reduced while both groups showed similar body weight profiles (Fig. S1). We then performed immunohistochemistry staining of tumour sections. As shown in Fig. 1e, a significant decrease in proliferation marker ki67 while active caspase 3 and E-cadherin increased after DMF treatment relative to vehicle treatment. These results suggested that DMF significantly inhibits tumour growth and metastasis in vivo.
To understand the underline antitumor effects of DMF administration, analysis of transcriptional pathways was carried out on Huh7 cells treated with DMF for 48 hours. Compared with gene expression in the control cells, 2031 upregulated genes and 1891 downregulated genes were found in the cells treated with DMF (|log2FC| ≥ 1 and P-value ≤ 0.001, Fig. 1f). Interestingly, we observed the modulation of many genes involved in DNA replication and repair, cell proliferation, metastasis, metabolism, and protein folding in DMF-treated cells (Fig. 1g, h). Next, we selected several genes in each group and evaluated their mRNA levels by RT-qPCR. Our analysis verified the altered expression of the genes in both Huh7 and HCC-LM3 cells, e.g., cell cycle regulators CDC6 and MCM7 are significantly downregulated in the DMF-treated cells (Fig. 1i). The transcriptomic results suggested that DMF exerts regulatory effects on vital pathways involved in oncogenesis and tumour progression. Taken together, these data demonstrated that DMF treatment significantly inhibited tumour growth in vivo and that variety of pathways were involved in the inhibition of tumorigenicity.
DMF represses cell proliferation and induces apoptosis
Based on signaling pathways from the above bioinformatics study, we then performed in vitro experiments to confirm the antiproliferation effects of DMF on hepatocellular carcinoma using cell counting kit-8 (CCK-8) assay. Incubation of cells with DMF resulted in reduced cell viability in a time and dose-dependent manner (Fig. 2a, Fig. S2). Next, EdU incorporation assay was performed to assess cell division after DMF treatment. We observed that DMF treatment reduced the number of dividing cells (Fig. 2b), which was accompanied by increased DNA damage and altered cell morphological features, as indicated by a dose-dependent and a time-dependent increase of p-H2A.X (Fig. 2c, d).
To clarify whether DMF-induced cell growth inhibition is associated with apoptosis, annexin V/7AAD staining was performed. Obviously, DMF treatment caused a time-dependent increase of apoptotic (annexin V+) cells, compared to control cells without DMF treatment (Fig. 2e, f). Next, we investigated the expression of apoptotic-related proteins, including active caspase-3 and c-PARP, by western blot assay (Fig. 2g, h). Similarly, DMF treatment induced the cleavage of apoptotic markers, including both caspase-3 and PARP. Furthermore, the incubation of HCC cells with the pan-caspase inhibitor Z-VAD-FMK significantly reduced the cell death induced by DMF (Fig. 1i). In summary, DMF inhibited hepatoma cell growth in vitro by repressing cell division and activating caspase-dependent apoptosis.
DMF inhibit HCC metastasis in vitro and in vivo
Given that metastasis contributes to poor survival rate and high recurrence in HCC, we explored the effects of DMF on metastasis. Increased cell motility, a gain of migratory and invasive abilities play an important role in cancer metastasis. We examined the potential inhibitory effects of DMF on cancer metastasis using transwell migration assays. Interestingly, we observed that DMF reduced the HCC cells’ migratory capacity in a dose-dependent manner (Fig. 3a, b), further supporting our earlier results. To explore its in vivo inhibitory effects on cancer metastasis, we established an HCC-zebrafish metastatic xenograft model. Dil-labelled huh7 cells were injected into the duct of Cuvier of FLK-EGFP zebrafish at 48 h post-fertilization (hpf) and then treated with either vehicle or DMF. Notably, we observed fewer metastases in the bodies of the larvae and reduced tumour size in the treatment group compared to the control, further confirming the antimetastatic potential of DMF (Fig. 3c, d).
Epithelial-to-mesenchymal transition (EMT) has been implicated in HCC metastasis and is characterized by dysregulation of mesenchymal and epithelial markers. We next tested whether DMF could block the EMT program in HCC cells. Consistently with in vivo results, DMF treatment induced the downregulation of mesenchymal markers (N-cadherin and β-catenin) and the upregulation of epithelial marker (E-cadherin) (Fig. 3e). The results were further verified using immunofluorescence analysis in HCC cells after DMF treatment (Fig. 3f). Together, our data showed that DMF inhibited the metastatic potential of HCC cells both in vitro and in vivo.
DMF induces mitochondrial dysfunction and ATP depletion
Fumarates are readily transported into the mitochondria . Next, we mapped the morphological effects of DMF on various cellular organelles. Huh7 cells were treated and analyzed with transmission electron microscopy (TEM). The untreated Huh-7 cells showed intact organelles ultrastructures such as nucleus membrane, mitochondria consisting of well-defined cristae, and electron-dense mitochondrial matrix (Fig. 4a). In contrast, the DMF-treated cells exhibited the disrupted and engulfed mitochondria, swollen morphology with a visible accumulation of vacuoles in the cytosol, and severely swollen endoplasmic reticulum (ER).
The presence of disrupted mitochondria and extensive mitophagy suggested DMF-induced mitochondrial impairment. Next, we investigated the effect of DMF exposure on mitochondrial membrane potential. Consistently, after treatment with DMF for 48 hours, a significant loss of mitochondrial membrane potential (Fig. 4b) displayed in DMF-treated cells, correlated with the release of cytochrome c into the cytosol (Fig. 4c). As previously reported, compromised mitochondria are often associated with amplified superoxide production and ATP depletion. Therefore, we further examined the level of mitochondrial superoxide (MitoSOX) production upon DMF treatment. Our results from flow cytometric analyses showed that DMF triggered an increase in MitoSOX compared to control (Fig.4d). Similarly, we observed a marked reduction of ATP level (Fig.4e) and compensatory activation of AMPK/ACC in DMF-treated cells as compared to the control (Fig.4f, g), further suggesting that mitochondrial dysfunction and membrane integrity damage were induced by DMF exposure. Together, our data clearly demonstrated that DMF triggered extensive mitochondria stress and that mitochondria seemed to involve in DMF-induced cell death.
DMF induces apoptosis and metastasis inhibition through downregulating Bcl-xL
Cell apoptosis is mainly initiated through the extrinsic or intrinsic pathways. Mitochondria participate in the intrinsic apoptosis arm through the Bcl-2 family proteins. Our earlier results suggested that DMF induced caspase-dependent apoptosis and extensive mitochondrial stress. Next, we assessed the role of mitochondrial in DMF-induced apoptosis. The expression of several Bcl-2 family proteins was assessed using immunoblot assays. We observed an increased level of proapoptotic Bax and Bak accompanied with marked Bcl-xL downregulation (Fig.5a-c, Fig.S3a). Unlike other antiapoptotic Bcl-2 proteins, Bcl-xL is highly expressed in one-third of HCC patients and contributes to their therapeutic resistance and low survival rates[7, 8].
Based on this, we next examined whether DMF inhibits HCC progression by downregulating Bcl-xL. Firstly, we stably infected HCC cells with either empty- or Bcl-xL-expressing vectors and treated them with vehicle or DMF. TUNEL staining assay revealed that Bcl-xL-overexpression significantly reduced apoptosis rate in DMF-treated cells (Fig.5d Fig.S3b, c). Additionally, as Bcl-xL is known to regulate mitochondria integrity, we tested whether Bcl-xL overexpression could modulate the levels of cytosolic cytochrome c in DMF-treated cells. Interestingly, immunoblot assay revealed that DMF-induced cytosolic cytochrome c was significantly decreased in Bcl-xL overexpressing cells (Fig. 5e). Together, these findings suggested that Bcl-xL downregulation by DMF contributed to cytochrome c release, caspase activation, and ultimately apoptosis.
Next, we further validated the role of Bcl-xL downregulation in DMF’s anticancer effects in vivo using the zebrafish model. To do this, we established zebrafish xenografts using HCC cells infected with Bcl-xL or control lentiviruses. Bcl-xL-overexpression significantly attenuated the inhibition of tumour growth induced by DMF treatment (Fig. 5f, g). Surprisingly, we observed that Bcl-xL overexpression in DMF-treated cells abrogated the metastasis inhibition induced by DMF (Fig. 5h), further confirming that Bcl-xL downregulation plays a vital role in DMF’s mechanism of action. To decipher the molecular mechanism underlying Bcl-xL-dependent suppression of HCCmetastasis in vivo, the expression of epithelial marker E-cadherin, mesenchymal markers such as N-cadherin and β-catenin was analyzed by immunoblot assay. Compared with the control group, Bcl-xL overexpression significantly increased the expression of N-cadherin, β-catenin and decreased the E-cadherin expression in HCC cells(Fig.S3d,e). Similarly, immunofluorescence analysis revealed an increased N-cadherin expression in tandem with decreased E-cadherin expression in DMF-treated Bcl-xL overexpression cells as compared to the control cells (Fig.S3d). Together, these findings showed that Bcl-xL downregulation plays a vital role in DMF-induced tumour growth and metastasis inhibition both in vitro and in vivo.
DMF downregulates Bcl-xL through NRF2 regulation
The Bcl-xL decrease can be caused by either less production or increased protein degradation. Next, we investigated how Bcl-xL was downregulated in DMF-treated cells. Protein degradation is dependent on the proteasome or lysosomal degradation pathway. To clarify the role of proteolysis, we examined the effect of MG132 (ubiquitin-proteasome inhibitor) or the lysosome inhibitor NH4Cl on DMF-induced Bcl-xL reduction. Treatment with either MG132 or NH4Cl could not significantly reverse DMF-induced Bcl-xL downregulation or cell death (Fig. 6a,b). Then we examined the Bcl-xL mRNA level in DMF-treated HCC cells and observed significant downregulation (Fig. 6c, Fig.S4a). Nrf2 is a redox-sensitive transcriptional factor of Bcl-xL and has been reported as a target of DMF. We carried out immunoblotting analyses and observed Nrf2 reduction in the DMF-treated cells (Fig. 6d,e). Next, we examined whether Bcl-xL downregulation induced by DMF was correlated with Nrf2 depletion by overexpressing Nrf2 in HCC cells (Fig. 6f). To do this, cells transfected with Nrf2 or control plasmid were treated with DMF and subjected to immunoblot analysis. Interestingly, the inhibitory effect of DMF on the protein expression of Bcl-xL was reversed by Nrf2 overexpression (Fig. 6g, h). Similarly, we observed the reduced apoptosis and metastasis inhibition in Nrf2-overexpressed HCC cells. Taken together, our results suggest that Nrf2 acts as an important upstream regulator in DMF-induced Bcl-xL downregulation.
Endoplasmic reticulum (ER) stress protects from DMF-induced apoptosis
We also observed endoplasmic reticulum (ER) with dilated/swollen lumen and accumulation of vacuoles (Fig S5a). The structural changes were accompanied by dysregulation of ER stress genes in RNA-Seq results, indicating a possible ER stress and activation of unfolded protein response (UPR). To confirm whether ER stress was induced, a western blot assay was performed to measure the levels of UPR proteins after DMF treatment. The expression of several UPR related genes, such as GRP78/BiP, CHOP, PERK and IRE-1, was significantly increased in a concentration- and time-dependent manner as compared with control (Fig. 7a). Activation of UPR is involved in both cell survival and cell death in cancer cells. Several studies suggest that UPR activation correlates with poor treatment outcomes in clinical settings and sorafenib-treated HCC cells[12-15]. We, therefore, examined how ER stress-adaptive cells respond to DMF treatment. Huh7 and HepG2 cells were maintained in low doses of ER stress inducer and then treated with DMF. Firstly, we observed that ER-adapted cells had a higher level of Nrf2 (Fig. 7b). This observation agreed with a recent study in hepatoma cells showing that persistent ER endowed cells with higher Bcl-xL/bad ratio. We observed the reduced cell death, and caspase activation in ER stress adapted cells (Fig. 7c,d), suggesting that ER stress activation is a prosurvival mechanism.
Next, we examined the effect of UPR inhibition in DMF-treated cells. We pretreated cells with GSK-2606414 (PERK), 4µ8c (IRE-Iα) or vehicle for 1hour and further treated them with vehicle or DMF for another 48 hours. Notably, we found that PERK inhibition exacerbated the cytotoxic effects of DMF as evidenced by increased apoptotic (annexin V+) cells. In contrast, IRE-Iα inhibition didn’t significantly influence DMF-induced cell death (Fig. 7e,f). Similar results were observed in IRE1 and PERK knockdown cells Fig. 7g,h). Our result agreed with the clinical observation that the activation of the PERK signalling arm correlated with the poor prognosis of HCC. Although in vitro ER-adapted cells might not fully portray the condition in HCC patients, our results show increased Nrf2 and reduced response to DMF in ER-adapted cells. Taken together, our data clearly demonstrate that DMF triggers activation of the UPR pathway as a prosurvival mechanism and that ER stress adapted cells might respond poorly to DMF treatment.
DMF enhances the efficacy of sorafenib both in vitro and in vivo
HCC exhibits remarkable intra-and inter-tumour molecular heterogeneity. This characteristic has been linked to poor treatment response and further suggests that targeting a single pathway may not be an optimal approach[16, 17]. Next, we examined whether DMF could enhance the therapeutic effects of sorafenib. Firstly, we analyzed the cell apoptosis induced by combined sorafenib and DMF treatment in HCC cell lines. As shown in Fig. 8a, sorafenib alone at 5 μM increased the percentage of Annexin V-positive cells by 2.4%. Co-treatment of Huh7 cells with sorafenib and DMF significantly increased apoptotic (annexin V+) cells by 18.3% compared to DMF only. Similarly, we observed increased expression of c-PARP after co-treatment (Fig. 8b). To explore the mechanism underlying the observed synergistic effect, we examined the signal transduction of AKT and STAT3. Increased levels of p-AKT correlate positively with poor survival rates in HCC patients receiving sorafenib treatment, while STAT3 is a major target of sorafenib[18-20]. Interestingly, immunoblot assay revealed that co-treatment sorafenib with DMF decreased the expression of p-AKT and p-STAT3 compared to sorafenib alone (Fig. 8c).
To assess the in vivo efficacy of the sorafenib and DMF combination, a zebrafish xenograft model was established and randomly assigned to four groups: control, sorafenib, DMF, and the combination of sorafenib plus DMF. Notably, treatment with DMF and sorafenib markedly enhanced the inhibitory effect on tumour growth compared with the sorafenib treatment alone in the xenograft model (Fig.S6a). For further validation, we established Huh7-C57BL/6 xenograft. First, mice bearing HCC cells were divided into 4 groups (control, DMF, sorafenib, sorafenib plus DMF) and treated accordingly for 14 days. After completion of treatment, tumour volume in all groups significantly reduced compared with the control group. In comparison with the sorafenib group, co-treatment of DMF and sorafenib markedly decreased tumour volumes (Fig. 8d-e). There were no significant differences in body weight between the treatment groups (Fig. 8f). Furthermore, we performed hepatotoxicity and nephrotoxicity assay to the impact of the combination therapy. The co-treatment significantly lowered the levels of AST and ALT, while the sorafenib only group showed significantly lower levels of creatinine compared to the control group (Table 1). These biochemical results showed that the combination therapy did not induce renal or hepatotoxicity in the treated mice, demonstrating the combined treatment of DMF and sorafenib as a potential therapeutic approach with low toxicity for the treatment of HCC.