Research ArticleCancer

Targeting VCP enhances anticancer activity of oncolytic virus M1 in hepatocellular carcinoma

See allHide authors and affiliations

Science Translational Medicine  23 Aug 2017:
Vol. 9, Issue 404, eaam7996
DOI: 10.1126/scitranslmed.aam7996

A virus and its reinforcement

Oncolytic viruses can be effective against a variety of cancers, including hepatocellular carcinoma, where a viral treatment is showing evidence of efficacy in people. Zhang et al. performed a high-throughput drug screen to search for compounds to pair with an oncolytic virus called M1 to further increase its effectiveness against hepatocellular carcinoma. Through this screen, they identified inhibitors of valosin-containing protein, then used them together with M1, and demonstrated the efficacy of this regimen in mouse models of cancer. In addition, the combination was well tolerated in primates, suggesting that the drug and virus combination may translate to human patients.


Oncolytic virotherapy is rapidly progressing through clinical evaluation. However, the therapeutic efficacy of oncolytic viruses in humans has been less than expected from preclinical studies. We describe an anticancer drug screen for compounds that enhance M1 oncolytic virus activity in hepatocellular carcinoma (HCC). An inhibitor of the valosin-containing protein (VCP) was identified as the top sensitizer, selectively increasing potency of the oncolytic virus up to 3600-fold. Further investigation revealed that VCP inhibitors cooperated with M1 virus–suppressed inositol-requiring enzyme 1α (IRE1α)–X-box binding protein 1 (XBP1) pathway and triggered irresolvable endoplasmic reticulum (ER) stress, subsequently promoting robust apoptosis in HCC. We show that VCP inhibitor improved the oncolytic efficacy of M1 virus in several mouse models of HCC and primary HCC tissues. Finally, this combinatorial therapeutic strategy was well tolerated in nonhuman primates. Our study identifies combined VCP inhibition and oncolytic virus as a potential treatment for HCC and demonstrates promising therapeutic potential.


Hepatocellular carcinoma (HCC) is the second leading cause of cancer-related death in men and claims more than 700,000 lives per year worldwide (1, 2). Although surgical resection, liver transplantation, and ablative therapies are curative for early stages of the disease, systemic pharmacotherapy is usually the final and main treatment for advanced-stage disease (3). Conventional chemotherapy has yielded poor response rates in patients with advanced-stage HCC. Sorafenib has been recognized as the standard systemic treatment for patients with HCC; however, the tumor response rate is unsatisfactory (4). Thus, there is an urgent need for new strategies to enhance treatment responses in HCC.

Oncolytic viruses (OVs) are self-amplifying cancer biotherapeutics that destroy malignancies without harming normal tissues (5). Oncolytic virotherapy is a potentially attractive strategy for the treatment of HCC (6). For example, the median overall survival in patients receiving high-dose [109 plaque-forming units (PFUs)] JX-594 (a vaccinia virus encoding granulocyte-macrophage colony-stimulating factor) is more than twice that of patients receiving low-dose (108 PFUs) JX-594 (14.1 months versus 6.7 months, respectively; P = 0.02) (7).

M1 virus is a strain of Getah-like alphavirus (GLV) with a small genome (11.7 kb; single positively stranded RNA), which was isolated from culicine mosquitoes collected on Hainan Island of China (8). GLV is predominantly transmitted among horses and provokes a mild, self-limited mosquito-borne illness in horses (9). M1 virus selectively kills HCC deficient in zinc finger antiviral protein (ZAP) (10). Tissue microarray analysis (TMA) showed that 69% of liver cancer tissues show low expression of ZAP compared to noncancer tissues (10), thus indicating that M1 may serve as a potential oncolytic agent for HCC treatment.

Although preclinical studies using OVs to treat cancers continue to generate great enthusiasm, the therapeutic efficacy of OVs in clinical trials has fallen short of the efficacy expectations based on preclinical models (11). Hence, improvements to therapeutic efficacy through either combination therapies or viral engineering will be critical to the success of these platforms.

Researchers have explored the concept of “conditionally enhancing” oncolytic viral efficacy by using pharmacological agents (5, 1215). We have previously shown that treatment of resistant tumor cells with cyclic adenosine monophosphate pathway activators that down-regulate interferon (IFN)–responsive genes can increase oncolysis by M1 virus (16). We then screened an anticancer small-molecule compound library to discover specific sensitizers of M1 virus–mediated oncolysis in HCC. We found that small-molecule valosin-containing protein (VCP) inhibitors (VCPIs) cooperate with M1 virus to kill HCC cells by suppressing the inositol-requiring enzyme 1α (IRE1α)–X-box binding protein 1 (XBP1) pathway and promoting endoplasmic reticulum (ER) stress–induced apoptosis in vitro and in vivo. Furthermore, our results suggest that VCP expression may be a useful biomarker to identify subsets of HCC that are sensitive or resistant to this combination treatment. This finding provides a rationale for clinical studies directed at exploring OV M1 in combination with VCPIs in the treatment of HCC.


High-throughput anticancer drug screen identifies sensitizers for OV M1

To identify molecular targets to improve the efficacy of M1 virus, we performed a combinatorial drug screen of 350 anticancer small molecules that inhibit pathways involved in growth, metabolism, and apoptosis. Cell viability was assessed after treatment with increasing doses of candidate drugs in the presence and absence of M1 virus. A low virus titer (0.001 PFU per cell) was used so that virus alone caused minimal cell death over the time of the assay. Differences in the area under the curve (DAUCs) for each compound with and without M1 virus were calculated (Fig. 1, A and B).

Fig. 1. Combinatorial drug screen identifies VCPI EerI as the top sensitizer for M1 virus in HCC cells.

(A) An outline of drug screening protocol. Hep3B cells were seeded in 96-well plates and treated with escalating doses of each of the 350 compounds in the drug screen, either singly or in combination with M1 virus (MOI = 0.001). (B) Representative drugs for drug screening. Dose-response curves were generated for each drug in the absence or presence of M1 virus, and the DAUC (fold) was calculated according to the following formula: (AUCSingle − AUCCombined)/AUCCombined; the orange areas represent DAUC, drug “A” represents an effective drug, and drug “B” represents an ineffective drug. (C) The agents were ranked according to DAUC (fold) between two dose-response curves for Hep3B cell line. Each dot represents one candidate drug from the anticancer compound library. (D) Top 20 candidate drugs identified through this screening. EGFR, epidermal growth factor receptor; DNA-PK, DNA-dependent protein kinase; GSK-3, glycogen synthase kinase–3; TNF-α, tumor necrosis factor–α; PI3K, phosphatidylinositol 3-kinase; PDGFR, platelet-derived growth factor receptor; MEK, mitogen-activated protein kinase kinase.

Through anticancer compound library screening, we identified numerous compounds targeting different pathways as being able to cooperate with M1 virus (DAUC ≥ 0.5). We identified the VCPI Eeyarestatin I (EerI) as the strongest sensitizer for M1 virus in vitro (DAUC = 1.35; Fig. 1, C and D, and table S1). EerI targets the VCP complex and inhibits deubiquitination of VCP-associated ERAD (ER-associated degradation) substrates, which is required for protein turnover (17). An earlier report has shown that EerI has preferential cytotoxic activity against cancer cells (18). Our screening data also identified the 26S proteasome inhibitor bortezomib as the second most efficacious candidate drug (DAUC = 1.09; Fig. 1, C and D, and table S1), and it has been reported to synergize with several classes of OVs in the treatment of cancers (1922). Both VCP and the 26S proteasome are components of the ERAD pathway, thus demonstrating an importance of ERAD for the anticancer effects of OV M1. Several other drug classes including epidermal growth factor receptor inhibitors, DNA-dependent protein kinase inhibitors, and aromatase inhibitors also cooperated with alphavirus M1 (Fig. 1D and table S1).

Inhibition of VCP sensitizes HCC to M1 viral oncolysis

VCP (Cdc48p in yeast) is a ubiquitous, abundant, and essential adenosine triphosphatase. It controls protein homeostasis by acting as a molecular segregase that extracts specific ubiquitin-modified client proteins from the ER and delivers them to the proteasome for degradation (23). VCP has been increasingly linked to cancer, and evidence suggests that VCP contributes to malignant transformation by promoting the degradation of proteins in cancer-associated pathways (24, 25). Moreover, VCP knockdown triggers the accumulation of polyubiquitinated (poly-Ub) proteins and activates the unfolded protein response (UPR) (26, 27). Recently, several small molecules have been found to target VCP and induce ER stress and ER stress–associated apoptosis, thus ultimately resulting in cancer cell death (18, 2830).

Here, the highest-scoring drug in the screen was EerI, a highly specific VCPI. To confirm that VCP inhibition and M1 virus exhibit synergistic activity against HCC cells, we chose two VCPIs, NMS-873 and CB-5083, which have been reported to activate the UPR, interfere with autophagy, and induce cancer cell death (28, 29). DAUCs for VCPIs with and without M1 virus were calculated (1.32 for EerI, 0.85 for NMS-873, and 0.75 for CB-5083). All three VCPIs sensitized HCC cells to M1 virus in vitro (Fig. 2A).

Fig. 2. VCP inhibition enhances the oncolytic efficacy of M1 virus in HCC cells.

(A) Hep3B cells were treated with increasing doses of VCPIs in the absence or presence of M1 virus (MOI = 0.001) for 48 hours, cell viability was measured, and DAUC (fold) was calculated. (B) EC50 shifts were determined for Hep3B cells treated with escalating titers of M1 virus with or without VCPIs (EerI, 2 μM; NMS-873, 0.25 μM; CB-5083, 0.25 μM) for 48 hours. (C) Western blot of cells infected with siRNAs targeting VCP. siNC, negative control siRNA; GAPDH, glyceraldehyde-3-phosphate dehydrogenase. (D) Phase-contrast images of HCC cells treated with siRNA targeting VCP (48 hours), followed by M1 virus infection (0.001 PFU per cell for Hep3B and 0.1 PFU per cell for Huh 7). Scale bars, 50 μm. (E) Quantification of MTT staining from cells in (D). Data are means ± SD in (A), (B), and (E) (n = 3). **P < 0.01.

To further quantify sensitization, we tested the dose-response relationship of VCPIs and M1 virus. To this end, HCC cells were treated with VCPIs at the indicated concentrations, and after 1 hour, they were infected with increasing doses of M1 virus. Cell viability was measured with MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] assays, and survival data were plotted (Fig. 2B). The dose required to kill 50% of cells (EC50) was determined using nonlinear regression. Our data showed that pretreatment with nontoxic concentrations of VCPIs sensitized the human HCC cell line Hep3B toward M1 oncolysis (EC50 shift of 3600-fold for EerI, 1320-fold for NMS-873, and 1030-fold for CB-5083; Fig. 2B). Moreover, combination index analyses further determined that the interaction between VCPI and M1 virus was synergistic (table S2).

To evaluate the therapeutic index, we screened a panel of HCC cell lines in addition to the Hep3B cell line (n = 6), one normal liver cell line (L02), and primary human hepatocytes (HH) for responsiveness to the OV M1 and VCPIs (fig. S1A). In six of seven tumor cell lines tested, EerI reduced the EC50 of M1 virus by 10- to 3600-fold (fig. S1B). Similar results were obtained using CB-5083, which reduced the EC50 of M1 virus by 4- to 1030-fold (fig. S1B). In contrast to the results in tumor-derived cell lines, the noncancerous L02 cell line and primary human hepatocytes were unaffected by M1 and VCPI combination therapy, thus demonstrating that this enhancement was tumor cell–specific.

Furthermore, in keeping with our hypothesis, knockdown of VCP resulted in increased sensitivity to M1 virus infection (Fig. 2, C to E). Conversely, overexpression of VCP partly reversed the synthetic lethal interaction between M1 virus infection and VCPI EerI (fig. S2). Collectively, these data suggested that VCP inhibition enhanced M1 virus–mediated oncolysis in HCC.

Expression of VCP in HCC correlates with sensitization by M1/VCPI combination therapy

As the above data showed, M1/VCPI was effective in a large proportion of HCC cell lines, with the exception of one type of HCC cell line and normal cells. To test whether increased expression of VCP correlates with enhanced cell killing by the combination treatment with M1 and VCPI, we assayed the amounts of protein and calculated the DAUC (fold) in the nine cell lines in fig. S1B. Increased sensitivity to M1/VCPI correlated with relative VCP expression. That is, high VCP expression was associated with higher sensitivity to combination treatment (Fig. 3, A to C). In addition, VCP expression was generally higher in cancer cells compared with normal cells (Fig. 3A).

Fig. 3. VCP expression correlates with the sensitivity to the combination of VCPI and M1 virus in HCC and normal cells.

(A) VCP protein expression in different cell lines. L02, normal liver cells; HH, primary human hepatocytes. Quantification was done using ImageJ software. (B) Cells were infected with escalating titers of M1 virus (MOI = 0.01, 0.1, 1, 10, and 100) in the absence or presence of EerI (2 μM), and the DAUCs (fold) between two dose-response curves for each cell line were calculated. (C) Correlation of DAUC and VCP expression. r is the Pearson correlation coefficient. (D) Representative VCP immunohistochemical staining in tumor tissue cores and adjacent noncancer tissues from TMA. Scale bar, 50 μm. (E) Statistical analysis of immunohistochemical (IHC) staining intensity. Box-and-whisker plots show the median (horizontal line), interquartile range (box), and 5th to 95th percentiles of the data. T, tumor tissues (n = 93); A, adjacent noncancer tissues (n = 83). Dots indicate outliers. ***P < 0.001. (F) Kaplan-Meier survival curves of tumor-free survival according to the VCP protein expression in HCC patients (n = 93). The median value of all 93 cases was chosen as a cutoff point for separating VCP high-expressing tumors from low-expressing cases.

To further explore this finding, we examined the protein expression of VCP in tumor and adjacent nontumor tissues of 93 HCC patients. In agreement with the results from cell lines, relative VCP expression was significantly increased in primary HCC tissues compared with adjacent nontumor tissues (P < 0.001; Fig. 3, D and E). This finding might explain the tumor cell specificity of M1/VCPI treatment in HCC. In addition, we demonstrated that higher expression of VCP correlated with shorter recurrence-free survival in HCC (median survival, 17 months versus 35 months on high and low expression; P = 0.091; Fig. 3F). Together, our results suggest that VCP expression may be a useful biomarker to identify subsets of HCC that are sensitive or resistant to this combination treatment.

Inhibition of VCP does not alter cellular susceptibility to M1, enhance viral production, or promote bystander killing

Mammalian cells respond to infection with RNA viruses through a signaling cascade initiated by the members of a family of cytosolic (RIG-I–like receptors) and endosomal (Toll-like receptors) viral RNA sensors (31). Once triggered, these receptors activate IFN response factor (IRF) 3/7 and nuclear factor κB (NF-κB), thus resulting in the up-regulation of hundreds of so-called IFN-stimulated genes, including those encoding the IFNs themselves (15). This response “warns” neighboring cells of an impending virus encounter, prompting those cells to preemptively express antiviral genes. The experiments described in this section were undertaken with the intent of testing the role of VCP in the cellular antiviral response.

First, we evaluated the effect of VCPI therapy on M1 infectivity and productivity. Time-lapse microscopy showed that the VCPI EerI did not alter M1 infectivity or spread through Hep3B cells (Fig. 4A and fig. S3), and single-step growth curves and virus protein expression revealed that VCPI treatment did not affect the kinetics of M1 virus in Hep3B cells in vitro (Fig. 4, B and C). To characterize the antiviral response in cells, we measured IRF3, IRF7, IFNα, and IFNβ production in HCC cells treated with the VCPI EerI and infected with M1 virus. As shown in Fig. 4D, pretreatment of HCC cells with EerI did not alter transcriptional induction of antiviral factors by M1 virus. In addition, virus infection was severely impaired by the addition of IFNα to the culture medium, and this protective effect was not reversed by pretreatment of cells with VCPI EerI (Fig. 4, E and F). We further analyzed viral spread and replication in vivo by performing tissue virus titration. We found no difference in the kinetics of viral spread in vehicle- or EerI-treated Hep3B tumor-bearing mice (Fig. 4, G and H). We concluded from these experiments that enhanced oncolytic cell death was not due to a suppression of the innate antiviral response.

Fig. 4. VCP inhibition does not alter the infectivity and productivity of M1 virus and does not cause bystander killing.

(A) Hep3B cells were pretreated with vehicle or 2 μM EerI, followed by M1–green fluorescent protein (GFP) infection (MOI = 1), after which phase-contrast and fluorescence microscopy images were captured. Scale bar, 50 μm. (B) Hep3B cells were pretreated with vehicle or 2 μM EerI, followed by M1-GFP infection (MOI = 5). Single-step growth analyses were conducted. (C) Viral structural protein E1 and the nonstructural protein NS3 were analyzed by Western blot after treatment. (D) Huh 7 cells were treated with vehicle or 2 μM EerI and M1 virus (MOI = 10) for 12 hours and then processed for quantitative reverse transcription polymerase chain reaction to measure IFN-stimulated gene transcripts indicated on the x axis. Data are means ± SD (n = 3). (E) Huh 7 cells were pretreated with EerI (2 μM) for 1 hour and then infected with M1-GFP at 10 MOI in the presence or absence of recombinant IFNα (1000 U/ml). GFP expression was monitored 24 or 48 hours after M1 inoculation. Scale bar, 50 μm. (F) Huh 7 cells were pretreated with EerI (2 μM) for 1 hour and then infected with M1-GFP at 10 MOI in the presence or absence of recombinant IFNα (1000 U/ml). Cell viability was measured after 48 hours. Data are means ± SD (n = 3). **P < 0.01. (G) Timeline of the experimental setup for (H). (H) M1-GFP (7 × 105 PFUs per mouse) was delivered intravenously to nude mice bearing Hep3B tumors that had been pretreated (30 min) with vehicle or EerI (2 mg/kg, intraperitoneally), according to the schedule. Subcutaneous tumor titers are shown for animals sacrificed 24 hours after virus administration. N.S., no statistical differences (n = 4 per group). (I) Black and gray columns show that Hep3B cells were treated with vehicle or EerI (5 μM). Orange and red columns indicate that Hep3B cells were pretreated (1 hour) with vehicle or EerI (5 μM) before infection with M1 virus (MOI = 0.001). Purple and blue columns (bystander experiments) indicate that Hep3B cells were pretreated with vehicle or EerI (5 μM) before incubation with conditioned medium (CM) derived from cells infected with M1 virus (MOI = 0.001). Cell viability was assessed after 48 hours. Data are means ± SD (n = 3). **P < 0.01.

The bystander effect has emerged as a possible mechanism of cytotoxicity in combinations of OVs and molecular targeted therapies (5, 15). We asked whether VCPI could sensitize HCC cells to death induced by other cytokines secreted after M1 virus infection (such as tumor necrosis factor–α). We detected no cytotoxicity when conditioned medium from virus-infected Hep3B cells was transferred to Hep3B cells that had been pretreated with or without EerI (Fig. 4I).

Enhanced efficacy of a combination of VCPI and OV M1 is dependent on the IRE1α-XBP1 pathway and ER stress–induced apoptosis

Unfolded proteins accumulate in malignant cells and induce a stress response that involves the ER (32, 33). Once triggered, ER stress activates a signal transduction pathway known as the UPR. The UPR is initially engaged as a protective mechanism to reduce protein accumulation. When ER stress becomes insurmountable, cell death ensues. Induction of ER stress is a hallmark of viral infection, owing to the synthesis of large amounts of viral proteins that are required for the production of progeny virus (34). Therefore, we presumed that inhibiting a key component of the ERAD pathway might gradually increase the unfolded protein burden in the ER of cancer cells, which, in turn, would predispose the cells to ER stress–mediated apoptosis in response to a subsequent M1 virus infection. Consistent with the notion that M1 virus and VCPIs synergized to induce irresolvable ER stress, we observed severe ER swelling within 24 hours after M1/VCPI treatment (Fig. 5, A to C, and fig. S4, A and B).

Fig. 5. VCP inhibition enhances ER stress induced by M1 virus and promotes ER stress–associated apoptosis.

(A) Transmission electron microscopy (TEM) images of tumors after 24 hours of treatment with vehicle, EerI (2 μM), M1 (MOI = 0.001), or EerI/M1. Scale bar, 1 μm. (B) Higher-magnification images from the red boxes in (A). Orange arrows indicate the ER in tumors. Relative sizes of the ER are indicated by red lines. Scale bar, 0.5 μm. (C) Quantification of ER distension from (B). Data are means ± SD (n = 4). **P < 0.01. (D) Immunoblots of proteins after treatment with M1 (MOI = 0.001), EerI (2 μM), or CB-5083 (0.25 μM). (E) Cells were treated with vehicle, EerI (2 μM), M1 (MOI = 0.001), or EerI/M1 for 48 hours, and proteins were analyzed by Western blot. (F) Cells were treated with M1 (MOI = 0.001), EerI (2 μM), or EerI/M1 for the indicated times. Three ER stress–associated apoptosis pathways (CHOP, JNK, and caspase-12) and virus protein E1 were detected by Western blot. (G) TEM images of tumors after 24 hours of treatment with vehicle, siVCP, M1 (MOI = 0.001), or siVCP/M1. Scale bar, 1 μm. (H) Higher-magnification images from the red boxes in (G). Orange arrows indicate the ER in tumors. Relative sizes of the ER are indicated by red lines. Scale bar, 0.5 μm. (I) Quantification of ER distension from (H). Data are means ± SD (n = 4). **P < 0.01. (J) Cells were transfected with siRNA targeting VCP (48 hours), and this was followed by M1 virus infection (MOI = 0.001). Proteins were examined by Western blot. (K) Phase-contrast images of cells treated with vehicle, EerI (2 μM), M1 (MOI = 0.001), or EerI/M1 for 48 hours. Scale bar, 50 μm. (L) Hoechst 33342 staining was used to stain nuclei of cells in (K). (M) Caspase-3/7 and caspase-9 activity assays. Cells were plated on 96-well plates and infected with M1 virus (MOI = 0.001) for 48 hours in the presence or absence of EerI (2 μM). Data are means ± SD (n = 3). **P < 0.01.

We then investigated the possible implications of VCP inhibition by EerI and CB-5083 in HCC cells. For subsequent investigations, EerI and CB-5083 were applied at concentrations of 2 and 0.25 μM, respectively (about one-fifth of the EC50 dosage in Hep3B cells), because these concentrations caused ubiquitinated protein accumulation without visible cell toxicity for at least 48 hours. Furthermore, these concentrations induced amplification of ER stress markers, such as IRE1α, XBP1(s), p-PERK (phospho–protein kinase RNA–like endoplasmic reticulum kinase), and p-eIF2α (phospho–eukaryotic initiation factor 2α), thus indicating activation of UPR pathway (Fig. 5D). In contrast to VCP inhibition alone, M1 infection decreased IRE1α and XBP1(s), thus suggesting that M1 virus was able to suppress the IRE1α-XBP1(s) pathway. In addition, phosphorylation of PERK and eIF2α was also increased after M1 virus infection, and no difference was observed in the activating transcription factor 6 (ATF6) pathway in either the M1 virus– or VCPI-treated groups (Fig. 5D).

Next, we investigated the effects of M1/VCPI combination therapy on UPR pathways. We observed that M1 virus appeared to inhibit EerI-mediated up-regulation of IRE1α, XBP1(s), and BiP (binding immunoglobulin protein) (Fig. 5E and fig. S4C). PERK-eIF2α and ATF6 pathways were also analyzed after treatment, and no difference in phosphorylation of PERK or eIF2α between the M1 virus–treated and combination groups was observed. The expression of ATF6 exhibited no difference in any treatments (Fig. 5E). Additionally, we probed the status of ER stress–associated pathways after combination treatment and M1 infection alone (35). As shown in Fig. 5F, c-Jun N-terminal kinase (JNK) and caspase-12 pathways were strongly induced after combined treatment with M1 and EerI compared with either treatment alone. C/EBP-homologous protein (CHOP) was not induced after treatment (Fig. 5F and fig. S5A). As we expected, knockdown of VCP produced severe ER swelling in the presence of M1 virus infection (Fig. 5, G to I) and increased the expression of IRE1α and BiP (Fig. 5J). Similarly, in the siVCP/M1 group, M1 virus blocked the up-regulation of IRE1α and BiP induced by small interfering RNA (siRNA) against VCP. Furthermore, knockdown of VCP enhanced the activation of the JNK and caspase-12 apoptosis pathway induced by M1 virus (Fig. 5J), and CHOP was not induced in any conditions (Fig. 5J and fig. S5B). We also observed an increase in apoptotic cells after EerI treatment and M1 infection (Fig. 5, K and L). The activities of caspase-3/7 and caspase-9 were both up-regulated after combination treatment (Fig. 5M).

In summary, these data suggested that a low titer of M1 virus [multiplicity of infection (MOI) = 0.001] was able to suppress the IRE1α-XBP1 pathway, thus causing mild ER stress, but was insufficient to induce apoptosis effectively. VCP inhibition alone induced extensive ubiquitinated protein accumulation in host cells (Fig. 5E and fig. S4C) but did not trigger cell death mediated by ER stress, owing to the up-regulation of the IRE1α-XBP1 pathway, an essential protective pathway for the recovery of ER-stressed cells. When the oncolytic M1 virus and the VCPI were used in combination, ER stress–induced apoptosis was observed.

VCPIs improve the therapeutic efficacy of OV M1 in vivo and ex vivo

To evaluate the therapeutic potential of M1/VCPI combination therapy in vivo, mice with subcutaneous Hep3B and Huh 7 tumors were treated with (i) intravenous M1 virus, (ii) intraperitoneal EerI injection, or (iii) a combination of the two. In agreement with in vitro observations, the combined treatment resulted in a significantly stronger inhibition (P < 0.001) of tumor growth in HCC xenografts compared with single treatments (Fig. 6, A to C, and fig. S6).

Fig. 6. Combinatorial VCPI and OV M1 treatment is efficacious in vivo and ex vivo.

(A) Timeline of the experimental setup for (B) and (C). (B) Hep3B xenografts were treated with vehicle, M1 virus (5 × 105 PFUs, intravenously), EerI (2 mg/kg, intraperitoneally), or a combination. Data are mean tumor volume ± SD (eight mice per group). ***P < 0.001 for M1-EerI combination versus single agents. (C) Huh 7 xenografts were treated with vehicle, M1 virus (5 × 105 PFUs, intravenously), EerI (2 mg/kg, intraperitoneally), or a combination. Data are mean tumor volume ± SD (eight mice per group). ***P < 0.001 for M1-EerI combination versus single agents. (D) Timeline of the experimental setup for (E). (E) Kaplan-Meier survival curve of mice bearing Hep3B tumors treated with vehicle, M1 virus (1 × 106 PFUs, intravenously), EerI (2 mg/kg, intraperitoneally), or a combination. CTL, n = 7; EerI, n = 7; M1, n = 8; M1 + EerI, n = 8. ***P < 0.001, log-rank with Holm-Sidak multiple comparisons. (F) Representative images of livers from each group in (E) at day 15. Dashed blue lines represent tumor areas. (G) Timeline of experimental setup for (H). (H) Kaplan-Meier survival curves of immunocompetent mice bearing Hepa1-6 tumors treated with vehicle, M1 virus (1 × 106 PFUs, intravenously), CB-5083 (25 mg/kg, orally), or a combination. CTL, n = 6; CB-5083, n = 5; M1, n = 6; M1 + CB-5083, n = 6. ***P < 0.001, log-rank with Holm-Sidak multiple comparisons. (I) Representative images of livers from each group in (H) at day 15. Dashed blue lines represent tumor areas. (J) Tumor tissues from (F) were evaluated through immunohistochemistry for Ki-67 (a marker of proliferation), IRE1α (a marker of the UPR pathway), p-JNK (a marker of ER stress–induced apoptosis), cleaved caspase-3, and E1 (structural protein of M1 virus). N, nontumor area; T, tumor area. Scale bar, 50 μm. (K) HCC tissues from five patients were treated with vehicle, EerI (10 μM), M1 (2 × 106 PFUs), or a combination for 96 hours, and cell viability was assessed. (L) Amounts of VCP protein in five patients. P, patient. Quantification (indicated by the numbers at the top) was performed using ImageJ software. (M) Correlation of cytotoxicity and relative VCP expression. r is the Pearson correlation coefficient.

We next developed an orthotopic HCC xenograft model to further investigate the antitumor efficacy of M1/VCPI combination therapy. Consistent with the earlier results, M1 virus and EerI alone had minimal effects on tumor growth, but combination therapy markedly reduced tumor burden in orthotopic HCC xenograft models. Mice that received combination treatment survived longer than those that were untreated or received single-agent therapy (Fig. 6, D to F).

Although the outcomes of M1/EerI combination treatments were encouraging in two immunodeficient mouse models (ectopic and orthotopic HCC tumors), the efficacy in an immunocompetent HCC model would be more convincing for systemic delivery of therapeutic viruses. Thus, we established orthotopic Hepa1-6 tumors in immunocompetent mice. Coadministration of M1 virus and CB-5083 or EerI delayed tumor progression and significantly prolonged survival (P < 0.001) compared with monotherapies alone (Fig. 6, G to I, and fig. S7).

Tumor sections were further examined in orthotopic/subcutaneous xenografts by immunohistochemical staining across treatment groups. M1 virus, in combination with VCPIs, (i) inhibited tumor cell proliferation, (ii) down-regulated IRE1α expression, (iii) activated the JNK pathway, and (iv) increased apoptosis (Fig. 6J and figs. S8 and S9).

To evaluate the clinical relevance of this therapeutic strategy, we examined whether EerI pretreatment could sensitize freshly derived patient tumor samples to M1 virus–mediated oncolysis. Tissue slices isolated from five patients with primary human HCC were treated with EerI before M1 infection or were left untreated and were assayed for viability 96 hours later. In line with our previous observations, EerI sensitized the slices to M1 virus–mediated tumor cell death (Fig. 6K). We also quantified VCP in the above primary tissues from HCC patients. In agreement with the in vitro results, relative VCP expression correlated with the sensitivity to the combination treatment in those HCC tissues (Fig. 6, L and M), thus further supporting the notion that VCP may serve as a potential biomarker predicting which HCC cancers are most likely to respond to M1/VCPI combination therapy. Together, these data demonstrate that VCPIs can be exploited to enhance alphavirus M1 oncolytic activity in the treatment of HCC.

Combination of OV M1 and VCPI is safe in nonhuman primates

In a previous study, M1 virus has been determined to be safe in a mouse model (10), and our data indicated that OV M1 cooperated with VCPI therapy specifically in tumor cells (fig. S1B), thus suggesting that this combination therapy would be safe. To test this hypothesis, we evaluated the toxicity of intravenously injected OV M1 and EerI in the nonhuman primate Macaca fascicularis.

In our efforts to further translate OV M1 to the clinic, we have completed a safety evaluation of intravenously injected OV M1 in nonhuman primates (36). Here, we evaluated the safety of EerI alone and EerI/M1 in combination in treatment groups of six animals each. The dosage schedule is shown in Fig. 7A. Six injections were performed every other day. During the course of the experiments, two treatment-related adverse events including slight anorexia (two of six in both groups) and diarrhea (one of six in both groups) were observed (table S3). No other behavioral changes or overt signs of illness occurred (table S3). Using an end point dilution method, we performed a neutralization assay to quantify neutralizing antibody in both treatment groups. As shown in table S4, neutralizing activity began at day 9 (for EM3 to EM6) or day 15 (for EM1 and EM2) after the first intravenous injection of M1 virus.

Fig. 7. VCPI/M1 combination treatment is well tolerated in nonhuman primates.

(A) Timeline of the experimental setup. iv, intravenously. (B) Animals were weighed at the indicated times. EerI, n = 6; M1 + EerI, n = 6. (C) Body temperature was recorded at the indicated times. Dashed lines represent reference values: 37.0° to 39.5°C (53). (D) Hematological parameters for animals injected with EerI or EerI/M1 combination. Dashed lines represent reference values (54): white blood cells (WBC), 7.39 × 109/liter to 22.11 × 109/liter; platelets, 112.43 × 109/liter to 632.27 × 109/liter; lymphocytes: 15.09 to 81.83%; neutrophils, 7.14 to 78.94%. (E) Serum chemistry parameters for animals injected with EerI or EerI/M1 combination. Dashed lines represent reference values (54): ALT (alanine transaminase), 6.15 to 83.39 U/liter; AST (aspartate aminotransferase), 18.20 to 71.80 U/liter; creatinine, 36.84 to 87.16 μM; BUN (blood urea nitrogen), 4.24 to 8.42 mM.

Animals maintained their body weight in both EerI and combination treatment groups (Fig. 7B). In the EerI/M1 group, three animals (EM1, EM2, and EM3) injected with EerI/M1 had a short-term fever, peaking on day 2 (39.7°C for EM3) or day 5 (39.9°C for EM1 and 40.0°C for EM2), which subsided quickly. No individuals in the EerI group had an abnormal increase in body temperature during the experiment (Fig. 7C).

Hematology studies showed no overt leukocytopenia or leukocytosis. Platelets and percentage of neutrophil counts did not show any abnormalities. We observed decreased lymphocyte counts in both groups, which returned to normal after several days (Fig. 7D).

Animals injected with EerI had slight increases in liver enzyme concentrations (ALT and AST), which decreased spontaneously on day 14 (Fig. 7E). In all injected animals, kidney function, as measured by blood urea nitrogen and creatinine concentration, did not deteriorate during the experiment (Fig. 7E). There were no significant differences observed between the EerI and EerI/M1 groups at the end point of the experiment (table S5). Overall, the measurements of mortality, body weight, body temperature, hematology, and serum chemistry supported the conclusion that EerI combined with M1 virus was safe for nonhuman primates, thus providing important safety evidence for the clinical use of VCPI/OV combination therapy in the future.


The development of OVs as anticancer therapeutics has recently accelerated (3741). The most notable example is the U.S. Food and Drug Administration–approved oncolytic herpesvirus T-Vec for the treatment of melanoma. However, the therapeutic efficacy of OVs as single agents in patients is limited, thus indicating the necessity to create new strategies to maximize the potential of oncolytic virotherapy.

It is anticipated that systemic delivery will be imperative for treating disseminated cancers (42). Therefore, our purpose was to develop a virus that could be delivered intravenously and initiate an infection at disparate tumor sites. One of the in vivo limitations to effective therapy could be virus delivery to the tumor bed. Thus, we are interested in discovering specific sensitizers for OVs to enable viruses to induce tumor cell death at low MOI. Here, using a high-throughput anticancer compound library, we discovered that the combination of M1 virus with VCPIs was well tolerated and efficiently killed HCC cells in vitro, in vivo, and ex vivo by triggering ER stress–induced apoptosis.

ER stress is induced during viral infection (43), including that with OVs, such as Maraba virus, vesicular stomatitis virus, and vaccinia virus (14, 44). Viruses have different mechanisms to modulate UPR mediated by one master control protein (BiP) and three sensors (IRE1α, PERK, and ATF6) to overcome ER stress. Here, we found that an extremely low dose of M1 virus (MOI = 0.001) was able to suppress one major branch of the UPR pathway in host cells: the IRE1α-XBP1 pathway, which plays a key role in rescuing the ER from overload by unfolded proteins by increasing production of chaperones (such as BiP) and ER lipid biogenesis (45). The low concentration of M1 virus did not induce severe ER stress, owing to limited production of virus proteins.

VCP is a key regulator of the ERAD pathway, the main protein quality control for proteins processed in the ER (4648). Many cancer cells rely on the ERAD pathway as a result of their high protein synthesis burden and possibly aneuploidy (49, 50). For this reason, the inhibition of VCP function in cancer cells is a promising therapeutic strategy (18, 29, 51). The potent and selective VCPI, CB-5083, which was used in this study, is currently being evaluated in two clinical trials: in patients with relapsed and refractory multiple myeloma and in patients with advanced solid tumors (29, 52). Here, we showed that low dosages of VCPIs induced accumulation of poly-Ub proteins. In response to this stress, the IRE1α-XBP1 pathway was activated in host cells to resolve the overload of unfolded proteins, and low concentrations of VCPIs were consequently insufficient to trigger cell death. Our data suggest that HCC cells pretreated (30 to 60 min) with a VCPI are more sensitive to cell death after infection by M1 virus. We interpret our findings to indicate that the VCPI and M1 infection promote cell death by overloading the UPR. The data in support of this are twofold: First, M1 virus produced proteins for viral replication and spread, thus increasing the unfolded protein burden in VCP-pretreated cancer cells. Second, M1 virus was able to abrogate the prosurvival IRE1α-XBP1 pathway activated by VCPIs, thus further promoting irresolvable ER stress and resulting in robust apoptosis. Together, our results establish a viable strategy for the rational design of “chemical-biological combination therapy” by targeting proteostasis.

Our study does present some limitations. Here, we found that HCC cells have higher VCP expression compared with that in the surrounding normal tissues, thus indicating that VCP may be an attractive target for the treatment of HCC, but the potential causes of increased VCP expression in HCC cells should be explored in the future. How viruses interact with each component of UPR (such as IRE1α-XBP1 pathway in this study) over the course of replication still remains unclear. Further investigation on the molecular interaction between OVs and the UPR pathway may yield important information.

From our point of view, although new therapeutics for the treatment of cancers, including cellular therapy and oncolytic virotherapy, are being developed, a combination therapy strategy including small molecules seems reasonable. However, the safety concerns of combination treatments should be thoroughly considered. Nevertheless, the favorable outcomes of pharmacological safety study in nonhuman primates should encourage the clinical translation of VCPI/OV combination therapies in the near future. Finally, our results suggest that small molecular anticancer compounds that directly target VCP may be particularly effective and safe in combination with OV M1 in the treatment of HCC.


Study design

The objective of our study was to identify sensitizers for OV M1 for the treatment of HCC. Through a focused anticancer drug screen, we identified a VCPI as the top sensitizer for OV M1 for HCC treatment. We next evaluated combinatorial strategies of M1 virus and VCP inhibition in HCC cell lines, L02 normal liver cell line, and primary human hepatocytes. The effect of combination therapy on cell death (by ER stress–associated apoptosis) and cell growth was assessed by Western blot analysis of ER stress markers and MTT assays. This study was extended to three mouse models of HCC and freshly derived patient tumor samples to analyze the effects of combination therapy in vivo and ex vivo. Tolerability and biosafety of the combined M1/EerI therapy were tested in nonhuman primates. In all experiments, animals were randomized to different treatment groups without blinding. Sample size in experiments was specified in each figure legend.

Statistical analysis

All statistical analyses were performed using GraphPad Prism software. Comparisons between different groups were made using Student’s t test or analysis of variance (ANOVA) as appropriate in the in vitro study. Values of tumor volume were analyzed by repeated-measures one-way ANOVA. The Kaplan-Meier curves were analyzed by log-rank test. For safety study, repeated-measures ANOVA was generally not used because of the relatively large variation among macaques during the time course and limit of detection, which precluded reliable statistical analyses. All error bars indicate SD. Differences were considered significant if the P value was less than 0.05.


Materials and Methods

Fig. S1. VCPIs selectively improve the oncolytic efficacy of M1 virus in HCC cells.

Fig. S2. Overexpression of VCP reverses the synthetic lethal interaction between M1 virus infection and VCPI EerI.

Fig. S3. VCPI EerI does not alter the infectivity and spread of M1 virus.

Fig. S4. VCP inhibition enhances ER stress induced by M1 virus and promotes ER stress–associated apoptosis.

Fig. S5. Combined VCP inhibition and M1 virus do not induce CHOP expression.

Fig. S6. VCPI EerI improves the therapeutic efficacy of OV M1 in human HCC xenotransplants.

Fig. S7. VCPI EerI improves the therapeutic efficacy of OV M1 in an immunocompetent HCC mouse model.

Fig. S8. VCPI/M1 combination treatment inhibits tumor cell proliferation and promotes ER stress–associated apoptosis in Hep3B xenografts.

Fig. S9. VCPI CB-5083 improves the therapeutic efficacy of OV M1 in an immunocompetent HCC mouse model.

Table S1. Composite list of synthetic lethal scores from 350 anticancer compounds (provided as an Excel file).

Table S2. Interaction between VCPIs and M1 virus on cell killing.

Table S3. Clinical symptoms of macaques in the safety study of intravenous injection of EerI or EerI/M1.

Table S4. Titers of anti-OV M1 antibodies in macaque sera determined by infection reduction neutralization assay.

Table S5. Comparison of main outcome measures between two groups of macaques at the end of the safety study.

References (5557)


  1. Acknowledgments: We thank M. Shu (Marjorie B. Kovler Viral Oncology Laboratories, University of Chicago) for technical assistance. Funding: This work was supported by the National Natural Science Foundation of China (81503088, 81273531, and 81373428) and the Natural Science Foundation of Guangdong Province (2016A030310160 and 2016A030310146). Author contributions: H.Z., K.L., Y.L., F.X., J.H., X.X., J.C., and Y.T. performed the experiments. L.F., F.W., K.L., and H.Z. carried out the tumor histoculture end point staining computer image analysis. H.Z. and K.L. analyzed the TMA results. J.H., Y.L., W.Z., W.Y., J.L., B.L., P.Q., X.B., S.G., and X.S. provided ideas and critical comments. H.Z. and G.Y. designed the research, analyzed and interpreted data, and cowrote the paper. G.Y. supervised the research. Competing interests: H.Z., Y.L., J.C., S.G., J.H., X.B., X.X., K.L., J.L., Y.T., W.Z., W.Y., and G.Y. are inventors on patent application (201610688100.7) submitted by Guangzhou Virotech Pharmaceutical Co. Ltd. that covers the concept of combining VCPIs and OVs for cancer therapy.

Stay Connected to Science Translational Medicine

Navigate This Article