TLR3 mediates synergistic inhibition of sorafenib and poly(I:C) in liver cancer cells in vitro
Firstly, we evaluated TLR3 expression in various HCC cell lines and HCC patient tissues, we found significantly higher expression in LM3 and MHCC-97H liver cancer cells than other cell lines (Figure S1A) as well as in HCC patient tumor tissue (Figure S2). To further determine the biological significance of TLR3 in NK cells, we generated a TLR3-knockdown stable LM3 cells using short hairpin RNA (shRNA) to study the effect of TLR3 on poly(I:C) function. Western blot analysis confirmed that the level of TLR3 expression was downregulated by more than 80% in TLR3-knockdown stable LM3 cells compared to parent cells (Figure S1B).
Then, we tested the direct cytotoxicity of sorafenib and poly (I:C) on LM3-Mock and LM3-shTLR3 cells in vitro. Sorafenib inhibited proliferation of both cell types at a minimal concentration of 5 µM in a dose-dependent manner (Fig. 1A). In contrast, poly(I:C) did not inhibit proliferation of LM3-shTLR3 cells. The combination of sorafenib with poly(I:C) exhibited synergistically growth inhibition in LM3-Mock cells, but not in LM3-shTLR3 cells (Fig. 1B). The results suggest that TLR3 mediates poly(I:C)’s direct suppressive effect on tumor cells.
We further determined the effect of sorafenib on cell invasion and found that sorafenib significantly inhibited cell invasion in both LM3-Mock and LM3-shTLR3 cells at the 10 µM dose (Fig. 1C). We also observed that poly(I:C) exhibited stronger inhibition of invasion in LM3-Mock cells than in LM3-shTLR3 cells. The combination of poly(I:C) and sorafenib also exhibited synergistically inhibiton of invasion only in LM3 cells but not in LM3-shTLR3 cells (Fig. 1C). The results further demonstrate that TLR3 mediates poly(I:C)’s direct effect on tumor cells.
The combination of sorafenib with poly(I:C) exhibits synergistic inhibition of tumors growth in TLR3 knockdown in vivo
We established xenograft tumor model using LM3-Mock cells and LM3-shTLR3 cells. Four weeks after treatment, sorafenib significantly reduced the tumor volume compared to the control treatment in LM3-Mock cell xenograft model (0.63 ± 0.25 cm3 versus 1.29 ± 0.41 cm3, P < 0.01) (Fig. 2A)and the LM3-shTLR3 models (1.09 ± 0.23 cm3 versus 2.05 ± 0.16 cm3, P < 0.05) (Fig. 2B). Although no statistical difference was observed in tumor volume between the poly(I:C)-treated group and the control treatment groups in the LM3-Mock model (P = 0.06), the combination of poly(I:C) and sorafenib markedly reduced tumor volume by 62% (P < 0.001) in the LM3-Mock xenograft tumor compared to sorafenib treatment alone. However, the difference between the combination group and sorafenib group was not significant in the LM3-shTLR3 xenograft tumor (Fig. 2B). To investigate the effect of NK cells on HCC growth and metastasis in vivo, the LM3-shTLR3 models were treated with human NK cells twice a week for four weeks. The tumor size was decreased significantly in mice treated with the combination of poly(I:C) and sorafenib compared to sorafenib treatment alone. The results suggest that TLR3 mediates the directly and indirectly inhibitory effect of poly(I:C) on liver cancer cells.
We also observed the effect of the combination of sorafenib and poly(I:C) on the number of metastatic tumors in the LM3-Mock xenograft model. Sorafenib alone significantly reduced the number of pulmonary metastatic tumor foci (89 ± 11 versus 112 ± 10 per lung, P < 0.01). However, when the number of foci was standardized by liver tumor size, sorafenib significantly increased standardized number of lung metastases (SNLM) compared to the control treatment (67 ± 7 versus 52 ± 4 per lung, P < 0.05) as we previously reported 7. The poly(I:C)-treatment alone also did not reduce SNLM compared to the control group (P = 0.41). However, the combined poly(I:C) and sorafenib reduced both the number of lung metastases (21 ± 6 versus 80 ± 12 per lung, P < 0.001) and SNLM (32 ± 7 versus 68 ± 7 per lung, P < 0.05) compared to sorafenib and poly(I:C) alone. Notably, the potential anti-metastatic effect of sorafenib was abolished when the drug was administrated in combination with poly(I:C) (SNLM: 32 ± 7 in the combination group versus 51 ± 4 in the control group, P < 0.01; Fig. 2B). The synergistic inhibitory effect of the drugs on abdominal metastases was similar to that observed in the lung when mice were subjected to combination therapy (5 ± 2 in the combination group versus 10 ± 3 in the sorafenib group, P < 0.05; Fig. 2C). The results futher demonstrate that the combination of sorafenib and poly(I:C) has synergistic inhibitory effect on liver cancer in vivo.
Furthermore, combination therapy signicantly prolonged the overall survival in both in both LM3-Mock (88.0 ± 1.6 versus 81.7 ± 2.3, P < 0.05) and LM3-shTLR3 (93.3 ± 1.7 versus 85.7 ± 2.0, P < 0.05) xenograft model compared to the sorafenib alone (Fig. 2D).
Poly(I:C) attenuates sorafenib-mediated suppression of NK cells in vivo
Because the TLR3 agonist poly(I:C) is a dsRNA analog, and treatment of the poly(I:C) induces strong anti-neoplastic effects accompanied by activation of NK cells19, we next evaluated if poly(I:C)-mediated NK cell activated the anti-tumor activity of combination therapy. Previous studies showed that sorafenib inhibits reactivity of NK cells against tumor cells in vitro and in vivo25. We also found that sorafenib treatment alone decreased the number of splenic NK cells in tumor-bearing C57B6 mice compared to the control treatment (3.26 ± 0.31% versus 2.24 ± 0.96%, P < 0.01). However, the combination of sorafenib and poly(I:C) treatment increased significantly the number of NK cells in tumor-bearing C57B6 mice compared to the sorafenib alone ( P < 0.05; Fig. 3A).
Furthermore, we performed degranulation assays in co-cultures of isolated human NK and K562 cells. Similar to our previous results, sorafenib significantly reduced CD107a expression in NK cells in the presence of K562 cells. Combination treatment of sorafenib and poly(I:C) upregulated expression of CD107a compared to cells treated with sorafenib alone (Fig. 3B)
Next, we determined the cytotoxicity of NK cells isolated from tumor-bearing C57BL/6 mice in different treatment groups. Firstly, we isolated NK cells from spleen by magnetic-activated cell sorting (> 90% purity; Fig. 3C) and determined the reactivity of NK cells. We found that splenic NK activity cells from the sorafenib-treated mice was significantly lower than that from control group, but the splenic NK activity cells was partly recovered from combination-treated mice compared to that from sorafenib-treated mice(P < 0.05;Fig. 3D). Our results indicate that poly(I:C) somewhat protects NK cells from impairment by sorafenib.
To verify the protective effect of poly(I:C) on NK cells, we used a unique model in which two agents were given separately or in combination before tumor implantation in nude mice. We found that sorafenib-treated nude mice had a lower number of CD49b + NK cells compared to the control group, while the reduction was not significant in the combination-treated mice (Fig. 4A). Consistent with our hypothesis, more lung metastatic foci were found in sorafenib-pretreated mice compared to controls when measured by the number of tumors (87.1 ± 22.1 versus 46.3 ± 12.3 per lung, P < 0.05) or area (15.1 ± 5.0% versus 9.6 ± 2.1%, P < 0.05).Interestingly, we observed no statistical difference between poly(I:C)- and vehicle-pretreated mice for the number of metastatic foci (42.0 ± 15.6 versus 45.5 ± 12.1 per lung, P = 0.643). However, if measured by area [7], pulmonary metastatic foci in poly(I:C)-pretreated mice were much fewer than in the control group (4.2 ± 2.9% versus 11.2 ± 2.9%, P < 0.05), which could be attributed to poly(I:C)-induced increased reactivity of NK cells. In combination-pretreated mice, much fewer pulmonary metastatic foci were found compared to the number in sorafenib-pretreated mice (48.5 ± 20.1 versus 86.3 ± 21.6 per lung, P < 0.05), although the difference was not significant when compared to the control group (P > 0.05) (Fig. 4B, 4C), suggesting poly(I:C) protects NK cells to some extent in vivo.
Direct activiation of NK cells by poly(I:C) facilitates its anti-tumor activity
We also investigated if sorafenib or poly(I:C) directly affected the activity of splenic NK cells from C57BL/6 mice. Treatment with sorafenib caused a substantial reduction of NK cell cytotoxicity against YAC-1 cells, while poly(I:C) marginally enhanced the activity of NK cells; the difference was not statistically significant compared to the control (Fig. 5A). However, the combination of sorafenib and poly(I:C) significantly enhanced the activity of NK cells compared to sorafenib alone. Bay 11-7082, a NF-κB-specific inhibitor, showed a modest effect on NK cell activity alone or combined with poly(I:C), which indicated poly(I:C) enhanced cytotoxicity through the Raf/MEK/ERK pathway but not the NF-κB pathway. Notably, both sorafenib and Bay 11-7082 drastically impaired the production of IFN-γ. In contrast, poly(I:C) significantly upregulated production of IFN-γ by NK cells when combined with either sorafenib or Bay 11-7082, suggesting both Raf/MEK/ERK and NF-κB pathways are involved in increased IFN-γ production. Because IFN-γ and MCP-1 are important cytokines for NK cells, we examined their expression levels in plasma by ELISA in the LM3-Mock and LM3-shTLR3 models. Levels of MCP-1 were reduced by 42.1% in sorafenib-treated mice in the LM3-Mock model (P < 0.05) and by 39.2% (P < 0.05) in the LM3-shTLR3 model compared to vehicle-treated mice. Levels of IFN-γ were also lower in plasma of the sorafenib-treated group than in the vehicle-treated mice in both models (P < 0.05). In the combination-treated mice, plasma levels of MCP-1 were elevated significantly compared to sorafenib-treated mice in the LM3-Mock model (P < 0.05), although the difference was not significant (P = 0.35) in the LM3-shTLR3 tumors (Fig. 5B). These results indicate that the immune-potentiating activity of NK cells by poly(I:C) contributes to its increased anti-tumor efficacy when combined with sorafenib.
ERK and NF-κB pathways are involved in poly(I:C)-induced activation of NK cells.
As we previously reported, sorafenib impairs the reactivity of NK cells primarily by inhibiting the ERK signaling pathway10, but not the PI3K/AKT or JAK/STAT3 pathway. Thus, we examined the signaling molecules involved in this impairment by Western blot analysis and investigated NF-κB and ERK activation.
Isolated human NK cells were treated with sorafenib and/or poly(I:C) followed by IL-2. Subsequently, cell lysates were analyzed for phosphorylated ERK (pERK1/2) and IκB. Addition of 2 µM sorafenib alone reversed IL-2-induced ERK1/2 phosphorylation and almost completely eliminated phosphorylation of ERK1/2 at 10 µM. However, poly(I:C) stimulated substantial ERK1/2 phosphorylation and necessitated 20 µM sorafenib to abolish ERK1/2 phosphorylation, demonstrating that poly(I:C) attenuates sorafenib-inhibition of ERK1/2 phosphorylation in NK cells. Sorafenib also inhibited ERK1/2 phosphorylation in LM3-shTLR3 cells in a dose-dependent manner (Fig. 6A). In contrast, poly(I:C) limited this effect in LM3−shTLR3 cells and induced considerable Iκβ phosphorylation in isolated human NK cells in dose-dependent manner, although the phosphorylation level was not influenced by sorafenib (Fig. 6B). Taken together, these findings suggest that both ERK and NF-κB signaling pathways are associated with activation of poly(I:C)-stimulated NK cells.