Article Text

Antitumoural immunity by virus-mediated immunogenic apoptosis inhibits metastatic growth of hepatocellular carcinoma
  1. Bita Boozari,
  2. Bettina Mundt,
  3. Norman Woller,
  4. Nina Strüver,
  5. Engin Gürlevik,
  6. Peter Schache,
  7. Arnold Kloos,
  8. Sarah Knocke,
  9. Michael P Manns,
  10. Thomas C Wirth,
  11. Stefan Kubicka,
  12. Florian Kühnel
  1. Department of Gastroenterology, Hepatology and Endocrinology, Medical School Hannover, Germany
  1. Correspondence to Dr Florian Kuehnel, Department of Gastroenterology, Hepatology and Endocrinology, Medical School Hannover, Carl Neuberg Str. 1, D-30625 Hannover, Germany; kuehnel.florian{at}


Background and aims Viral infection of a dying cell dictates the immune response against intracellular antigens, suggesting that virotherapy may be an effective tool to induce immunogenic cell death during systemic cancer treatment. Since viruses and proteasome inhibitors both induce accumulation of misfolded proteins, endoplasmic reticulum (ER) stress and immune responses during treatment of hepatocellular carcinoma (HCC) with bortezomib and the tumour-specifically replicating virus hTert-Ad (human telomerase reverse transcriptase promoter-regulated adenovirus) were investigated.

Methods Unfolded protein response (UPR) pathways and ER stress-mediated apoptosis were investigated by western blots, caspase-3 assays, 4',6-diamidino-2-phenylindole (DAPI) and Annexin V staining in HCC cells following hTert-Ad/bortezomib treatment. Oncolysis was assessed in subcutaneous HCC mouse models. Antiviral/antitumoural immune responses were characterised in immunocompetent HCC mouse models by ELISA, ELISpot assays and pentamer staining. Systemic efficacy of antitumoural immunity was investigated by determination of lung metastases burden.

Results Bortezomib and hTert-Ad trigger complementary UPR pathways but negatively interfere with important recovery checkpoints, resulting in enhanced apoptosis of HCC cells in vitro and improved oncolysis in vivo. In immunocompetent mice, bortezomib inhibited antiviral immune responses, whereas ER stress-induced apoptosis of infected HCC resulted in caspase-dependent triggering of antitumoural immunity. In therapeutic settings in immunocompetent, but not in immunodeficient or CD8-depleted mice, virotherapy-induced antitumoural immunity efficiently inhibited outgrowth of non-infected lung metastases. Immunotherapeutic efficacy could be significantly improved by bortezomib in experiments with low viral doses.

Conclusion Proteasome inhibition during virotherapy disrupts the UPR, leading to enhanced ER stress-induced apoptosis, improved local oncolysis and antitumoural immunity. The results suggest that combining intratumoural virotherapy with adjuvant systemic therapies, which specifically support the function of the virotherapy as an antitumoural vaccine, is a promising immunotherapeutic strategy against HCC.

  • ER stress
  • apoptosis
  • eIF2α
  • oncolytic virotherapy
  • adenoviral vectors
  • cancer vaccines
  • cell death
  • hepatocellular carcinoma
  • immunotherapy

Statistics from

Significance of this study

What is already known about this subject?

  • The proteasome inhibitor bortezomib is a potent antitumoural agent clinically approved for haematopoietic cancers.

  • Bortezomib can promote apoptosis in HCC cells in vitro by stabilising cell cycle inhibitors and affecting the PI3K/Akt pathway.

  • Tumour-selectively replicating (oncolytic) adenoviruses can effectively lyse solid tumours and can induce systemic antitumoural immune responses by as yet unclear mechanisms.

What are the new findings?

  • Proteasome inhibition and oncolytic adenovirotherapy interfere with ER stress response pathways in HCC, resulting in an increased apoptotic phenotype of virus-induced cell death.

  • Bortezomib enhances oncolytic efficacy in vivo, supports immunogenicity of adenoviral-infected HCC cells and reduces antiviral immune responses in vivo.

  • In therapeutic HCC models, virotherapy induces a strong CD8+ T cell-dependent antitumoural immunity that is therapeutically effective against both primary tumour and distant lung metastases.

  • In the context of the dominant immunogenic effect of virotherapy, additional application of bortezomib in therapeutic models of HCC revealed either supportive or neutral effects, depending on the administered viral dose.

How might it impact on clinical practice in the foreseeable future?

  • Multimodal approaches aiming at provocation of a potent and sustained antitumoural immune response emerge as a very promising strategy for curative treatment of solid tumours. We provide the first description of a multimodal, virotherapy-based treatment against metastatic HCC including a systemic cancer treatment that specifically supports the function of oncolytic virotherapy as an antitumoural vaccine. Our results demonstrate that immunogenic cell death during systemic cancer treatment is capable of inhibiting metastatic growth of HCC, providing an innovative strategy for future clinical trials.


Induction of antitumoural immunity is a challenging aspect in oncolytic treatment of solid tumours by conditionally replicating adenoviruses. However, it is still a matter of debate how molecular mechanisms of cell death induction determine immune responses to intracellular antigens.1–3 Initially, necrotic cell death was considered to be immunogenic whereas apoptosis was associated with tolerance induction.4 5 This has been questioned by findings that necrotic cell death can impair tumour immunity and, vice versa, apoptosis may induce antitumoural immunity under certain conditions.6 7 It is apparent now that the immune response against cellular antigens is influenced by a complex signalling network predominantly determined by the initial cause of cell death.8 9 Immunogenic cell death is thereby characterised, for example, by exposure of ‘eat me’ signals on the cell surface and expression of endogenous ‘danger’ signals such as damage-associated molecular patterns (DAMPs) or pathogen-associated molecular patterns (PAMPs).10

Since elimination of infected cells is a major target of the first-line defence against viral infections, molecular patterns of virus-mediated cell death are believed to be evolutionarily conserved cross-links to innate and adaptive immune responses.11 12 On the other hand, viruses have developed numerous means to inhibit molecular mechanisms of immunogenic cell death to either delay or inhibit the induction of antiviral immune responses.13 14 In contrast to studies using chemotherapy,15 16 the impact of cell death phenotype during tumour-specific viral replication on induction of antitumour immunity has not yet been addressed. In particular, it is unclear whether immunogenic cell death during virotherapy can support cross-presentation of tumour antigens, thus provoking an antitumoural immune response.

We have reported that infection by human telomerase reverse transcriptase promoter-regulated adenovirus (hTert-Ad) sensitises hepatocellular carcinoma (HCC) against chemotherapy-induced apoptosis,17 and it has been shown by Zou et al that Epstein–Barr virus (EBV)-infected lymphomas are sensitised for apoptosis by bortezomib, an inhibitor of the 26S proteasome involved in endoplasmic reticulum (ER) stress induction.18 Sustained ER stress following an abnormal unfolded protein response (UPR) has been implicated in innate immune responses and the bypass of immune tolerance mechanisms in colitis19–21 and other autoimmune diseases.22 23 These observations suggest that UPR and ER stress, as a result of excessive synthesis of viral proteins in tumour cells, may also be relevant for immune responses in tumour-specific adenoviral infection. Although it is tempting to speculate that irrecoverable levels of ER stress due to impaired UPR in a dying tumour cell could be involved in triggering of antitumoural immunity, there are currently no studies investigating a possible link between ER stress-induced apoptosis and induction of immune responses against tumour antigens.

Therefore, we investigated molecular mechanisms of ER stress in HCC cells following application of the proteasome inhibitor bortezomib and the conditionally replicating adenovirus hTert-Ad for selective treatment of telomerase-positive tumours. Both agents not only triggered complementary UPR pathways but also negatively interfered with protective UPR factors, as shown by downregulation of phosphorylation of the translation initiation factor eIF2α and elimination of Grp78/BiP, leading to enhanced apoptosis of HCC cells in vitro. Investigations on the therapeutic potential of ER stress-induced apoptosis during oncolytic treatment revealed improved oncolysis in HCC xenotransplant models in vivo. Remarkably, ER stress-induced apoptosis due to the impaired UPR is linked to an enhanced antitumoural immune response in immunocompetent mouse models of HCC, thus contributing to effective elimination of non-infected HCC lung metastases.

Materials and methods

Cell lines and plasmids

Huh-7 cells were obtained from the JCRB (Osaka, Japan). Hepa1-6 and BNL cells (the epoxide-transformed subtype BNL 1ME A.7R.1 was used in this study) were obtained from the American Type Culture Collection (ATCC; Rockville, Maryland, USA). All cells were maintained in Dulbecco's modified Eagle's medium (DMEM) + Glutamax with 10% heat-inactivated fetal bovine serum (FBS; Life Technologies, Gaithersburg, Maryland, USA), 100 U/ml penicillin and 100 μg/ml streptomycin (Seromed, Berlin, Germany). Stable cell lines were generated by retroviral transduction of CReP (constitutive repressor of eIF2α phosphorylation), HA-IRES-EGFP (haemagglutinin–internal ribosome entry site–enhanced green fluorescent protein) or OVA-IRES-EGFP (ovalbumin–IRES–EFGP) and subsequent selection by neomycin treatment or fluorescence-activated cell sorting (FACS). CReP was amplified by PCR from Huh-7-derived cDNA using the primers AAAAGCTTATGGCATCCAGTGACCTGGAAA (forward) and AAGCGGCCGCCAACTCAACATTGCTTGAG (reverse).

Adenovirus preparation

The tumour-specific adenovirus hTert-Ad and the replication-defective adenovirus Ad-LacZ were prepared and stored as described before.17 Infectious titres were determined using a Rapid Titre Kit (BD-Biosciences, San Jose, CA, USA). Before application in mice, the virus was dialysed against a solution containing 10 mM Tris pH 8.0, 1 mM MgCl2 and 140 mM NaCl.

Adenovirus pretreatment with CARex-VP22

To allow effective infection of BNL cells, normally refractory to adenoviral transduction, adenoviruses were pretreated with recombinant, purified CARex-VP22 as described previously.24 As an example, 20 μg of CARex-VP22 was used for pretreatment of 1×109 infectious units (ifu) of hTert-Ad in serum-free medium. The mixture was incubated for 30 min and then directly added to BNL cells at a multiplicity of infection (MOI) of 100. Infection was carried out for 4 h. For intratumoural injection, pretreatment of hTert-Ad with CARex-VP22 was performed in a volume of 100 μl.

Western blot analysis

A total of 5×106 cells were treated with hTert-Ad (MOI 50) and/or bortezomib at the concentrations indicated in the figures. For analyses, cells were collected and lysed in RIPA buffer. A 10–40 μg aliquot of protein was separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) and blotted onto HyBond (Millipore, Billerica, MA, USA). Membranes were blocked by shaking at room temperature in Tris-buffered saline (TBS)/Tween-20 (1% v/v) containing 5% (w/v) skim milk powder for 3 h. In general, blots were developed by shaking overnight at 4°C in TBS/Tween containing 2% (m/v) skim milk powder and primary antibodies at concentrations recommended by the manufacturers. The antibodies, manufacturers and clone/catalogue numbers used are listed in Supplementary table 1. Blots were then subjected to secondary antibody detection (1:10 000 dilutions in TBS/Tween) for 1–2 h. Bands were visualised using the enhanced chemiluminescence (ECL) detection system (Amersham, Freiburg, Germany).

Detection of apoptosis

Suspended cells for a final confluency of 70–90% were treated first with hTert-Ad/bortezomib (MOI/concentrations as indicated in the figures). Induction of apoptosis was analysed after 24 h in cell lysates by caspase-3 activity assay (Clontech, Mountain View, CA, USA) according to the manufacturer's protocol. Results were normalised against the protein concentration (BioRad protein assay; BioRad, Munich, Germany).

Apoptotic cells were further identified by fluorescence microscopy of 4',6-diamidino-2-phenylindole (DAPI)-stained nuclei. For this purpose, treated cells were seeded onto glass coverslips in 6 cm dishes. After 30 h, the supernatant was carefully replaced by fixation solution (1.25% glutaraldehyde in phosphate-buffered saline (PBS)). After 15 min, cells were washed twice (0.02% Tween-20 in PBS) and permeabilised with 0.5% Triton-X 100 in PBS for 5 min. After a further wash step, cells were covered with DAPI (Merck, Darmstadt, Germany) in PBS (0.2 μg/ml) for 5 min. After washing twice, the coverslips were mounted on microscopic slides using Vectashield mounting medium (Vector Labs, Burlingame, CA, USA).

Apoptotic cells were also identified and quantitated by Annexin V/propidium iodide staining. Ad-infected/bortezomib-treated cells were harvested after 32 h and stained with the Annexin V-APC Apoptosis Detection Kit (eBioscience, San Diego, CA, USA) following the manufacturer's protocol. Stained cells were analysed by FACS.

DNA extraction and real-time quantitative PCR (qPCR)

Primary tumours and lung metastases were established in Balb/c mice as described below. Mice received a single dose of 1×109 ifu of hTert-Ad intratumourally and were harvested after 3 days to determine the vector distribution. DNA from livers, lung and subcutaneous tumours was isolated using a QIAamp DNA Mini Kit (Qiagen, Hilden, Germany). A 100 ng aliquot of DNA was subjected to real-time qPCR in an ABI7300 cycler using qPCR Mastermix Plus (Eurogentec, Cologne, Germany). Values were normalised against genomic 18S DNA using the 18S Genomic Control Kit (Eurogentec). The hexon gene was detected as described previously.25

FACS analysis

Cells were detached in cell suspension buffer (Gibco, Paisley, UK), collected by centrifugation and washed with PBS containing 0.1% bovine serum albumin (BSA). After resuspension, 0.5 μg of labelling antibody was added per 5×105 cells and incubated on ice for 3 h. Cells were washed twice and acquired using a FACS Canto-II (BD-Biosciences). Data were analysed using FACSDiva-V5.0.2 (BD-Biosciences) and FlowJo-V7.2.5 software (TreeStar Software, Ashland, Oregon, USA). All labelling antibodies were purchased from eBioscience.


ELISpot and ELISAs were performed as described.26 The peptides HA (IYSTVASSL, H2-Kd), LacZ (TPHPARIGL, H2-Ld), OVA (SIINFEKL, H-2Kb), LacZ (DAPIYTNV, H2-Kb) and hexon (KYSPSNVKI, H2-Kd) were obtained from Proimmune, Oxford, UK.

Animal experiments

Six- to-eight-week old, female NMRI-nu/nu mice, C57Bl6 and Balb/c mice were obtained from Charles River Laboratories, Wilmington, MA, USA. Clone-4 and OT-1 mice were obtained from the animal facility of Hannover Medical School.

Induction of antitumoural immunity

For the determination of antitumour immunity after single vaccination/adoptive T cell transfer, cells were infected with CARex-VP22-coated (for BNL cells expressing the HA antigen (BNL-HA)) or uncoated virus (for Hepa1-6 cells expressing ovalbumin (Hepa1-6-OVA)) at an MOI of 100. Cells were then incubated for 4 h with bortezomib (10 nM) and/or zVAD-fmk (5 μM). Treated and untreated cells were collected and mixed at a ratio of 1:4. Mitomycin C (5 μg/ml) was added for 30 min to inhibit cell proliferation. A total of 1×107 cells/mouse were injected intraperitoneally into Balb/c mice or C57Bl6 mice. At day 3, CD8 cells were isolated from OT-1 or Clone-4 donor mice using the CD8+ T cell isolation kit (Miltenyi, Bergisch-Gladbach, Germany) and 1×104 CD8 cells/mouse were adoptively transferred by intravenous injection. Mice were sacrificed at day 20, and splenocytes were isolated and subjected to ELISpot analysis.

To monitor intrinsic priming of antitumoural T cells, mice were injected intraperitoneally with treated cells as described above. Fourteen days later, mice received a booster injection of treated cells. After 7 days, mice were sacrificed, and splenocytes and peripheral blood lymphocytes were isolated and subjected to pentamer staining using αCD8-FITC (fluorescein isothiocyanate), αCD19-PE (phycoerthyrin)/Cy7, αCD3-APC (allophycocyanin) and T cell receptor (TCR)-specific pentamers (obtained from Proimmune: PE-labelled, H2-Kd-restricted IYSTVASSL-pentamer for staining of HA-specific T cells; PE-labelled, H-2Kb-restricted SIINFEKL-pentamer for OVA). Pentamer+ cells were quantitated in the CD8+/CD3+/CD19 gate.

Determination of antiviral immunity

For the determination of antiviral immunity, Balb/c mice were injected with 1×108 ifu of hTert-Ad intravenously. After 3 weeks mice were harvested, and serum, bone marrow and splenocytes were prepared. The humoral response was detected in serum by ELISA. Bone marrow cells were investigated for the frequency of pre-B cells (labelling antibodies: αCD25-APC, αCD138-PE, αCD19-PE/Cy7, αB220-FITC). Splenocytes were subjected to ELISpot.

Tumour growth in xenotransplant models

Tumours were established by subcutaneous injection of 1×107 Huh-7 or BNL cells into the flanks of nude mice. Subcutaneous tumour nodules were grown to a size of ∼200 mm3 prior to the first treatment. For adenoviral infection, subcutaneous tumours were injected with 1×109 ifu of hTert-Ad in 100 μl of serum-free medium. Huh-7 tumours received a weekly infection, BNL tumours were treated twice weekly and adenoviruses were pretreated with CARex-VP22 to achieve effective infection (see above). Bortezomib was administered at 0.5 μg/g body weight weekly. Tumour volume was determined using the equation: V(tumour)=(length × width2)/2.

Tumour growth in lung metastases models

Primary tumours were established by subcutaneous injection of 1×107 BNL-HA cells into the flanks of Balb/c mice. Lung metastases were induced by two consecutive intravenous injections of 5×105 cells. Therapeutic treatment of primary tumours was performed according to the treatment schedule presented in the figures. Blood samples during the trial were drawn from the mandibular vein. CD8 depletion was performed by an initial intraperitoneal injection of 200 μg of antimouse CD8a monoclonal antibody (mAb) Ly-2 (eBioscience) on day 8, followed by 100 μg on days 9, 14 and 20 to maintain the depletion throughout the trial. Once the control group reached a critical tumour size, the volume of primary tumours was determined as described above and mice were sacrificed. Lymphocytes were isolated from peripheral blood and subjected to pentamer staining. Lungs were inflated with 4% paraformaldehyde in PBS; paraffin sections were prepared and stained with H&E following standard protocols. Tumour area of lung metastases was evaluated in 10 lung sections/mouse using Cell*-Software (Olympus, Hamburg, Germany).

Statistical analysis

The results of two treatment groups were compared for statistical significance by unpaired, two-tailed t test using GraphPad Prism V3.02 software (GraphPad Software, La Jolla, CA, USA). A p value <0.05 was considered as statistically significant.


Bortezomib and oncolytic viral infection activate complementary pathways of ER stress and interfere in the UPR

First, we investigated the possible implications of proteasome inhibition by bortezomib on ER stress-induced apoptosis during adenoviral oncolysis in HCC cells. To determine a suitable bortezomib dose for these investigations, dose-dependent induction of ER stress and apoptosis were determined in Huh-7 cells. As demonstrated by caspase-3 activation, bortezomib induced apoptosis in these cells (figure 1A). Dose-dependent amplification of ER stress markers, such as Grp78/BiP and GADD153/CHOP, indicated that increased ER stress was involved in bortezomib-dependent apoptosis (figure 1B). A bortezomib concentration of 10 nM was used for subsequent investigations, since this concentration was sufficient to inhibit cell growth without visible cell toxicity for at least 48 h. Furthermore, this concentration did not significantly interfere with replication of hTert-Ad in Huh-7 cells, as shown by investigations on adenoviral protein expression, DNA synthesis, particle formation and release of virus progeny (Supplementary figure 1).

Figure 1

Treatment of human hepatocellular carcinoma (HCC) cells with the oncolytic adenovirus hTert-Ad (human telomerase reverse transcriptase promoter-regulated adenovirus) and the proteasome inhibitor bortezomib leads to increased endoplasmic reticulum (ER) stress. (A) Huh-7 cells were treated with different doses of bortezomib. Cells were lysed after 24 h and caspase-3 activity was determined. (B) Protein expression levels were analysed by western blot after 40 h of treatment. (C) Huh-7 cells were treated with 10 nM bortezomib and analysed by western blot at the indicated time points. (D) Huh-7 cells after treatment with bortezomib (10 nM) and/or hTert-Ad (multiplicity of infection (MOI) 50) were analysed by western blot after 24 and 40 h. A short summary of crucial pathways involved in the regulation of the ER stress response is given in (E).

Additionally, ER stress pathways were investigated after proteasome inhibition in a time-dependent manner (figure 1C). Upregulation of IRE1α, Grp78/BiP, GADD153/CHOP and also increased phosphorylation of eIF2α could be observed, while neither the ATF-6 nor the PERK/ATF-4 pathway was triggered. Interestingly, splicing of XBP-1 was inhibited during a short period (8–12 h) as indicated by the occurrence of XBP-1u, which has been considered to play a proapoptotic role in the UPR.27

Next, we investigated ER stress pathways for possible interference of proteasome inhibition and infection with the tumour-selective, oncolytic adenovirus hTert-Ad. In contrast to proteasome inhibition alone, viral infection strongly induced the UPR pathways PERK/ATF-4 and ATF-6, but not IRE1α (figure 1D, see also figure 1E for a schematic overview of major ER stress pathways). Additionally, JNK and c-Jun were activated after adenoviral infection, but not by bortezomib alone. Proteasome inhibition did not interfere with virus-induced PERK/ATF-4, ATF-6, JNK and c-jun activity. Albeit that hTert-Ad seemed to inhibit bortezomib-mediated upregulation of IRE1α, impairment of XBP-1-splicing was unchanged after combined treatment.

In contrast to bortezomib, hTert-Ad infection significantly decreased the levels of Grp78/BiP, an essential protective chaperone for the recovery of ER-stressed cells. Unexpectedly, additional proteasome inhibition did not retard the decrease of Grp78/BiP levels but led to an almost complete elimination of Grp78/BiP. We further observed that combined bortezomib/hTert-Ad interfered in the phosphorylation of eIF2α. Blocking translation by phosphorylation of eIF2α is known as a first-line response in ER-stressed cells allowing them to process excessively synthesised protein.28 Although proteasome inhibition and viral infection each induced eIF2α phosphorylation, combined treatment abrogated this crucial UPR response. Our observations described above were further confirmed in other HCC cells such as Hep3B (data not shown). In summary, the results suggest that bortezomib and oncolytic viral infection not only trigger complementary pathways of ER stress but can also negatively interfere with recovery functions of the UPR as demonstrated by the inhibition of eIF2α phosphorylation and by elimination of Grp78/BiP.

Proteasome inhibition during adenoviral infection enhances apoptosis

Our results described above point to increased ER stress in HCC cells following viral oncolysis and proteasome inhibition. Our observations further indicate critical alterations in the UPR following this treatment that may promote apoptosis. To test this, Huh-7 cells were morphologically examined. Compared with single application, combined hTert-Ad/bortezomib treatment strongly increased the cytopathic morphology of Huh-7 cells (figure 2A) associated with an elevated number of apoptotic cells as demonstrated by DAPI (figure 2B) and Annexin V staining (figure 2C). Cell death following viral infection alone showed relatively weak signs of apoptosis consistent with other studies describing that oncolytic adenoviral infection of human cells mediates an apoptotic machinery-independent, necrosis-like programmed cell death.29 Interestingly, our data therefore suggest that proteasome inhibition is able to confer an apoptotic signature to cells infected by an oncolytic adenovirus.

Figure 2

hTert-Ad (human telomerase reverse transcriptase promoter-regulated adenovirus) infection of human hepatocellular carcinoma (HCC) cells combined with bortezomib leads to increased apoptosis. Huh-7 cells were treated with 10 nM bortezomib and/or hTert-Ad (multiplicity of infection (MOI) 50). Treated cells were investigated by bright-field microscopy (A), by fluorescence microscopy of 4',6-diamidino-2-phenylindole (DAPI)-stained nuclei (B) and by fluorescence-activated cell sorting (FACS) analyses after staining with Annexin V and propidium iodide to allow the assessment of apoptotic cell death (C). (D) Maternal Huh-7 and transgenic Huh-7 cells stably expressing enhanced green fluorescent protein (EGFP) or constitutive repressor of eIF2α phosphorylation (CReP), respectively, were treated as described above. At 24 h following treatment, cells were lysed and caspase-3 activity was determined. (E) Huh-7 cells were treated as described above. At the indicated time points, protein levels were analysed by western blot.

To investigate whether decreased eIF2α phosphorylation following bortezomib/hTert-Ad promotes apoptosis, we used Huh-7 cells stably expressing the catalytic domain of CReP.30 Expression of CReP in Huh-7 cells led to strongly elevated caspase-3 activity after bortezomib/hTert-Ad, whereas this effect was modest after single treatments (figure 2D). These results demonstrate that decreased eIF2α phosphorylation plays an important role in bortezomib-induced apoptosis of adenoviral-infected tumour cells. However, this does not exclude that other mechanisms, such as the elimination of Grp78/BiP, may also contribute to increased apoptosis after bortezomib/hTert-Ad.

Downregulation of phospho-Akt has been previously described as a molecular determinant of bortezomib-mediated apoptosis in HCC cells.31 To investigate whether UPR impairment is involved in the execution of apoptosis in virus-infected cells, we therefore determined the level of Akt/protein kinase B phosphorylation. We found that hTert-Ad-dependent Akt phosphorylation was not significantly reduced by simultaneous proteasome inhibition (figure 2E), suggesting that inhibition of Akt phosphorylation does not play a major role in this treatment model. We further investigated expression of proapoptotic and antiapoptotic mediators involved in regulation of intrinsic apoptosis. Proteasome inhibition alone did not influence the levels of the investigated proteins except a significant stabilisation of Mcl-1, consistent with recent reports.32 After combined treatment, we also found no significant alterations in the protein levels of Bax, Bak, Bcl-xS, Bcl-2 and Bcl-xL. Though proteasome inhibition was able to increase Mcl-1 levels significantly at early stages of adenoviral infection, bortezomib did not prevent adenovirus-induced elimination of Mcl-1 during the late phase of infection (figure 2E).

It has been reported that ER stress can be counterbalanced by induction of autophagy.33 However, we could not find any signs of increased autophagy after combined treatment (Supplementary figure 2).

Altogether, our observations indicate that co-application of proteasome inhibition and oncolytic infection results in increased HCC cell killing by primarily apoptotic mechanisms due to impaired UPR and enhanced ER stress. The apoptotic signature of viral-infected cells could have significant implications for the balance of the antitumoural and antiviral immune response in vivo.

Proteasome inhibition improves therapeutic efficacy of oncolytic adenovirus in human HCC xenotransplants

To evaluate the therapeutic potential of hTert-Ad/bortezomib combination in vivo, we first determined a bortezomib dose to avoid toxicity-related weight loss in treated animals (figure 3A). To investigate oncolysis in vivo, mice with subcutaneous Huh-7 tumours were treated intravenously with bortezomib and/or by intratumoural hTert-Ad injections. Consistent with in vitro observations, combined treatment of HCC xenografts led to a significantly stronger inhibition of tumour growth compared with single treatments (figure 3B). Since ineffective removal of dead tumour cells may lead to misinterpretation of therapeutic efficacy, treated tumours were histologically examined (figure 3C). Confirming the in vitro results, oncolytic efficacy in tumour xenografts was strongly enhanced after combination of bortezomib and hTert-Ad, as illustrated by almost complete tissue destruction.

Figure 3

Application of hTert-Ad (human telomerase reverse transcriptase promoter-regulated adenovirus) in combination with bortezomib leads to effective growth inhibition of human hepatocellular carcinoma (HCC) xenografts in vivo. Six-week-old nude mice were treated with bortezomib at a weekly dose of 0.5 or 1 μg/g body weight. Any increase or decrease of body weight was followed up for 2 weeks (A). Huh-7 xenografts were established on the flanks of nude mice by subcutaneous injection of 1×107 tumour cells. Tumours were treated weekly by intratumoural injection of 1×109 ifu of hTert-Ad (injection marked by arrows) and/or intravenous injections of bortezomib at a dose of 0.5 μg/g body weight. Tumour growth curves are shown in (B) with n=6/group. The experiment was performed twice with comparable results. At the end of treatment, H&E-stained tumour sections were investigated for the extent of tissue destruction (C).

Proteasome inhibition during tumour-specific viral replication leads to reduced antiviral but enhanced antitumoural immune response

Since differences in local tumour destruction following intratumoural injections may influence immune responses, we used a standardised approach for investigating antitumoural and antiviral immune responses in vivo. For this purpose, cells BNL-HA were treated with bortezomib and/or hTert-Ad. To investigate the contribution of ER stress-induced apoptosis to the antitumoural immune response during oncolytic infection, apoptosis was prevented by a pan-caspase inhibitor. Tumour cells were treated in vitro as described in the Materials and methods section and intraperitoneally injected into syngeneic mice, and HA-specific CD8+ T cells were adoptively transferred. Compared with the group that received untreated cells, bortezomib or viral infection significantly enhanced the HA-specific antitumoural immune response, that was further triggered by proteasome inhibition during viral infection (figure 4A). However, this effect was abrogated after caspase inhibition, indicating that improved antitumoural immunity was mediated by ER stress-induced apoptosis during combined treatment. To investigate intrinsic priming of T cells, mice were double vaccinated but did not receive an adoptive T cell transfer. Determination of HA-specific T cells in spleen or peripheral blood (figure 4B) showed an increased antitumoural response following combined treatment but not after proteasome inhibition alone. The observable immune response in the combination group was again effectively abrogated by caspase inhibition, confirming that apoptotic cell death of viral-infected tumour cells decisively contributes to the immunogenicity of this vaccine. Since monitoring of HA-specific responses suggested low immunogenicity of this particular epitope in our model, our results were additionally confirmed in similar experiments using Hepa1-6-OVA hepatoma cells (Supplementary figure 3).

Figure 4

hTert-Ad (human telomerase reverse transcriptase promoter-regulated adenovirus)/bortezomib-dependent apoptosis of murine hepatocellular carcinoma (HCC) cells leads to an improved induction of an antitumour response and reduced antiviral immunity. (A) Murine haemagglutinin (HA)-expressing BNL cells were infected in vitro by CARex-VP22-coated hTert-Ad, and treated with bortezomib and/or zVAD-fmk as described in the Materials and methods section. Cells were harvested, growth was arrested by mitomycin C and they were injected intraperitoneally into syngeneic mice. At day 3, HA-specific T cells were adoptively transferred by intravenous injection. At day 20, splenocytes were isolated and antitumour immunity was investigated by ELISpot. (B) For a model of intrinsic priming, BNL-HA tumour cells were injected intraperitoneally into syngeneic mice. Fourteen days later, mice received a booster injection with treated cells. After 7 days, mice were sacrificed, and splenocytes (left) and peripheral blood lymphocytes (right) were isolated and subjected to HA pentamer staining. (C–E) Mice were injected intravenously by 1×108 ifu of hTert-Ad and/or bortezomib (0.5 μg/g body weight). After 20 days, serum was investigated for antiadenoviral immunoglobulin G (IgG)/IgM titres by ELISA (C) and splenocytes were investigated for hexon-specific interferon γ (IFNγ) release by ELISpot (D). The frequency of pre-B cells (CD19+ B220+ CD25+ CD138+) was determined in bone marrow by fluorescence-activated cell sorting (FACS) analysis (E). Representative results of three independent experiments are shown.

A further question is the antiviral immune response that restricts the therapeutic efficacy of oncolytic viruses. We therefore analysed the antiviral humoral and hexon-specific cellular immune response after systemic injection of 1×108 ifu of hTert-Ad. In contrast to the antitumoural response during oncolytic tumour infection, bortezomib did not increase, but rather inhibited the antiviral immune responses (figure 4C,D). Proteasome inhibition may compromise the antiviral immune response since it has been previously reported that bortezomib interferes with lymphocyte development.34 Consistent with this hypothesis, impaired B cell development could be observed after bortezomib, particularly when combined with virus (figure 4D). This result is in agreement with other studies demonstrating that antibody production in bortezomib-treated animals is compromised due to the high susceptibility of B cells to ER stress-induced apoptosis.35 36 Since antiviral antibodies opsonise viral particles for subsequent uptake by antigen-presenting cells, the reduced antiviral cellular immune response could be a consequence of bortezomib-mediated inhibition of antiviral antibody production.

Murine BNL cells allow production of infectious adenovirus and are a suitable model to investigate ER stress-induced apoptosis

Murine cells that are permissive for propagation of oncolytic adenoviruses have been described for different tumour species.26 However, the lack of suitable HCC cells has been a limitation for rational investigations on tumour-specific adenoviruses in animal models of HCC. We identified a transformed subtype of murine BNL cells that allows generation of infectious particles after infection with hTert-Ad (figure 5A). Western blot analyses of ER stress-involved proteins predominantly revealed similar observations compared with human HCC cells (figure 5B). In particular, BNL cells showed a strong activation of phospho-PERK/ATF-4 and ATF-6 pathways following virus or combination treatment. Viral infection led to an even stronger phosphorylation of eIF2α than observable in Huh-7 cells that was completely abrogated by combined treatment, suggesting that impairment of the UPR is also involved in the induction of apotosis in BNL cells. Consistent with the results in Huh-7 cells, reduced levels of Grp78/BiP and Mcl-1 could also be observed after virus infection alone or in the combined treatment group. Caspase-3 assays (figure 5C), DAPI staining (figure 5D) and Annexin V FACS (figure 5E) showed that BNL cells responded to adenoviral infection alone with significant induction of apoptosis which could not be observed in human cells. Consistent with the observations in human HCC cells, proteasome inhibition strongly enhanced ER stress-induced apoptosis of infected BNL cells.

Figure 5

Murine BNL cells are a replication-competent model that closely reflects endoplasmic reticulum (ER) stress-induced apoptosis by hTert-Ad (human telomerase reverse transcriptase promoter-regulated adenovirus)/bortezomib in human hepatocellular carcinoma (HCC) cells. Murine HCC cells (BNL) were infected with CARex-VP22-coated hTert-Ad or Ad-LacZ (replication-defective control) at a multiplicity of infection (MOI) of 10. Murine CMT-64 cells were used as permissive control cells.26 At the time points indicated, cells were lysed and the infectious progeny were titred (A). BNL cells were infected with CARex-VP22-coated hTert-Ad (MOI 100) and treated with bortezomib (10 nM). After 40 h, protein expression was investigated by western blot (B). To measure apoptosis, cells were lysed after 24 h, and caspase-3 activity was determined (C). After 32 h, cells were stained with 4',6-diamidino-2-phenylindole (DAPI) and investigated by fluorescence microscopy (D). After 34 h, cells were stained with Annexin V and propidium iodide, and analysed by fluorescence-activated cell sorting (FACS) (E). BNL xenografts were established on the flanks of nude mice by subcutaneous injection of 1×107 tumour cells. Tumours were treated twice weekly by intratumoural injection of 1×109 ifu of hTert-Ad (marked by arrows) and/or weekly by intravenous injections of bortezomib at a dose of 0.5 μg/g body weight. Tumour growth curves are shown in (F) with n=6/group. Representative results of two independent experiments are shown.

Testing the oncolytic efficacy of hTert-Ad and/or bortezomib in BNL tumours in nude mice (figure 5F) revealed that bortezomib alone was inefficient. Virus alone or the combined treatment were able to retard the growth of this aggressively growing tumour while the combination was significantly more efficient.

Proteasome inhibition can support virotherapy-mediated elimination of virally uninfected HCC metastases

To investigate whether virotherapy-induced antitumoural immunity is effective in vivo against distant, non-infected HCC metastases, we first established subcutaneous, primary BNL-HA tumours in immunocompetent mice. When tumours became palpable, BNL-HA lung metastases were induced by systemic cell injection 48 and 24 h prior to local treatment of the tumour. Then, a therapeutic scheme (figure 6A) of intratumoural hTert-Ad and/or systemic bortezomib was applied which was well tolerated, as confirmed by regular haemograms (figure 6B).

Figure 6

hTert-Ad (human telomerase reverse transcriptase promoter-regulated adenovirus)/bortezomib-treatment of subcutaneous BNL tumours in a syngeneic model leads to effective, immune-mediated elimination of established lung metastases. Primary tumours and lung metastases of BNL cells expressing the haemagglutinin antigen (BNL-HA) were established in Balb/c mice and treated by intratumoural injection of 1×109 ifu of hTert-Ad and intravenous injections of bortezomib (0.5 μg/g body weight per injection) following the therapeutic scheme in (A). During the trial, blood samples were drawn from the mandibular vein and investigated for haematotoxicity. Total leucocytes (WBC), platelets (PLT) and differential haemograms were determined (B). Lungs of treated mice (n=6/group) were examined macroscopically and the burden of metastases was quantified on histological sections (C and D). (E) One day following initial injection of hTert-Ad, adenoviral DNA was determined in lung tissue by real-time quantitative PCR (n=4/group; LM, lung metastases). Metastases-free mice served as background control. To investigate the contribution of T cell-mediated immunity or immediate cytokine release, a therapeutic experiment according to (A) was performed in athymic nude mice (n=6/group). (F) Lung metastasis was investigated on histological sections (left) and quantified (right).

Examinations of lung tissue revealed that oncolytic virotherapy of the primary tumour resulted in a significantly decreased metastases burden compared with mice receiving systemic bortezomib treatment or untreated controls. This observation suggests that oncolytic infection provides an essential danger signal that provokes a significant antitumoural immune response. Consistent with increased antitumoural immune responses after in vitro treatment of HCC cells, the strongest therapeutic effect was observed in the group treated with proteasome inhibition during oncolytic infection of the primary tumour (figure 6C,D).

To rule out that direct infection may contribute to antimetastatic activity, viral DNA load was examined in relevant target tissues. In contrast to the potent viral infection of the treated primary tumour and a low infection of the liver, no significant infection could be detected in the lungs of treated animals, thus excluding a therapeutic effect by direct viral oncolysis of lung metastases (figure 6E). To confirm finally that inhibition of outgrowth of metastases was a consequence of antitumoural immunity but not of direct infection of metastases or systemic release of antitumoural cytokines after virus application, the experiment was repeated in athymic nude mice (figure 6F). In agreement with our hypothesis, we could not observe significant therapeutic effects in any treatment group.

To further stress the therapeutic efficacy of virotherapy-induced antitumoural immunity, the start of the tumour treatment was delayed in animals with primary tumours and lung metastases (figure 7A). Histological examination of lungs and primary tumours revealed a complete inhibition of outgrowth of lung metastases and a striking reduction of primary tumour burden after viral infection and combined treatment, but not in the bortezomib group. These results underline that viral infection of the primary tumour is an essential mechanism for triggering antitumoural immunity (figure 7B,C). This was further confirmed by increased frequency of antigen-specific T cells in peripheral blood in these two groups (figure 7D). Since the results of our experiment in nude mice (figure 6F) suggested a contribution of T cell-mediated immunity, we tried to identify the responsible cell type by selective depletion of CD8+ cells with Ly-2 mAb. Depletion of CD8+ T cells in the combination group resulted in a complete rescue of tumour growth, providing clear evidence that antitumoural immunity following oncolysis is mediated by the adaptive, CD8-dependent T cell response. Interestingly, we could demonstrate a strong reduction of the large primary tumour burden in the virus alone group or the combination group, with total remissions observable in both groups. These findings, with regard to the low efficacy of tumour treatments in nude mice and after CD8 depletion in immune-competent mice, corroborate the immense therapeutic potential of viral oncolysis as an antitumoural vaccine. In this model, we could not observe a statistically significant difference between virus alone and the combination group, which is most probably explained by the experimental set-up. In this model a significantly higher viral load was required to control the growth of the subcutaneous tumour, resulting in a maximal inhibition of lung metastases that could not be further improved by bortezomib.

Figure 7

Elimination of lung metastases by hTert-Ad (human telomerase reverse transcriptase promoter-regulated adenovirus)/bortezomib treatment of subcutaneous BNL tumours is mediated by CD8+ cytolytic T lymphocytes. Primary tumours and lung metastases of BNL cells expressing the haemagglutinin antigen (BNL-HA) cells were established in Balb/c mice and treated by intratumoural injection of 2×109 ifu of hTert-Ad and intravenous bortezomib following the therapeutic scheme in (A). The contribution of CD8+-dependent tumour immunity was investigated by Ly-2 monoclonal antibody-mediated depletion of CD8+ T cells. At day 27, mice were sacrificed and the burden of lung metastases was quantified on histological sections (B). The volume of the primary tumour was determined (C) and the HA-specific CD8+ T cell frequency was investigated in peripheral blood (D). Representative results from two independent experiments with n=6/group are shown.

On the other hand, attempts to address the control of local tumour growth by increased doses of bortezomib resulted in side effects such as weight loss, and in a significant reduction of antitumoural immune response (data not shown). Together, our findings suggest that well-balanced timing and dosage of both oncolytic virus and adjuvant systemic proteasome inhibition are critical parameters for an optimised treatment.


Antitumoural immune responses dictate the long-term therapeutic success of cancer treatment.16 Consequently, advanced treatments must eliminate cancer cells in a manner that effectively provokes antitumoural immunity to prevent tumour recurrence and progression. During cancer treatment effective antitumoural immunity can only be elicited when both ‘eat me’ and ‘danger’ signals are effectively emitted by the dying cancer cell and perceived by antigen-presenting cells in a fine-regulated spatiotemporal pattern.37 Improved understanding of molecular mechanisms that determine immunogenic cell death thus may facilitate the integration of tumour vaccination into effective multimodal cancer treatment strategies. Recently, it has been shown that whether a dying cell is infected or not dictates whether the immune response is ameliorated by T regulatory cells (Tregs) or supported by induction of T helper 17 (Th17) cells, emphasising the importance of danger signals in eliciting an immune response.3 We hypothesised that the mode of tumour cell death after oncolytic virus infection may also have implications in the antitumoural immune response, since viruses are evolutionarily selected to prevent immunogenic cell death.

Induction of ER stress is a hallmark of viral infections due to the synthesis of large amounts of viral proteins that are required for successful production of progeny virus. While viruses try to dampen ER stress by triggering a UPR,38 it has been shown that bortezomib-induced ER stress mediates apoptosis in cancer cells as well as in antibody-secreting plasma cells by disruption of the cell-protective UPR.27 35 36 39 We could show that ER stress in hTert-Ad-infected tumour cells is counterbalanced by PERK/ATF-4 and ATF-6 pathway activation. In contrast to viral infection, application of bortezomib resulted in caspase-3 activation and induction of apoptosis in human tumour cells. Consistent with the results of Lee et al,27 ER stress-mediated apoptosis in HCC after bortezomib treatment may be partially explained by transient expression of the proapoptotic XBP-1u (figure 1C,D). However, we did not detect a further increase of XBP-1u in bortezomib-treated HCC cells during viral infection. In contrast, we showed that bortezomib and oncolytic viral infection activate complementary pathways of ER stress and interfere with the UPR by inhibiting phosphorylation of eIF2α and by increased elimination of the protective chaperone Grp78/BiP. Thus, disruption of the UPR and enhanced apoptosis in virus-infected cells after bortezomib treatment could be explained by inhibition of eIF2α phosphorylation and Grp78/BiP rather than by alteration of XBP-1 splicing.

Although bortezomib interferes with adenovirus-induced UPR, net production of infectious adenoviral particles was unchanged during low-dose proteasome inhibition in HCC, providing an important prerequisite for simultaneous therapeutic application of hTert-Ad and bortezomib. Actually, tumour growth inhibition in HCC xenotransplant mouse models was significantly improved after combined treatment. Remarkably, in immunocompetent HCC mouse models, UPR impairment during virotherapy resulted in a significantly enhanced antitumoural immune response. This enhancement correlated with caspase activation, indicating that ER stress-mediated apoptosis following proteasome inhibition and viral oncolysis was actually responsible for successful triggering of immunogenic cell death. The assumption that ER stress-mediated apoptosis allows for effective cross-presentation of cellular antigens was recently supported by the observation of doxorubicin-induced immunogenic cell death in tumour cells due to ER stress-mediated apoptosis.40 Furthermore, it has been reported by other groups that bortezomib treatment of tumour cells can trigger CD8+ T cell antitumoural immune responses.41–43 Particularly in the context of viral oncolysis we found that bortezomib efficiently triggered a CD8+ T cell response. However, systemic application of bortezomib alone failed to inhibit HCC lung metastases in our therapeutic settings, which may be explained by insufficient apoptosis in primary tumours in response to bortezomib (figures 5F, 6C,D and 7C).

It has been shown that viral infection not only leads to a strong immune response against viral antigens, but also to MyD88- and TLR-dependent cross-priming of cellular antigens,11 12 providing a rationale for induction of antitumoural immunity by tumour-specific viruses. The broad triggering of PAMPs by the oncolytic adenovirus hTert-Ad may thus explain the stronger antitumoural immune response and the improved inhibition of lung metastases by virotherapy compared with proteasome inhibition alone. Furthermore, our data strongly suggest that the mode of virus-induced cell death influences the immune response against intracellular antigens of infected cells. In one therapeutic model, ER stress-induced apoptosis of tumour cells due to impaired UPR during oncolytic infection significantly enhanced the antitumour immune response, thus resulting in improved elimination of HCC lung metastases compared with single treatments. In a more challenging therapeutic tumour model involving larger tumour sizes and increased virus doses, we could not observe an additional therapeutic benefit after combined application of proteasome inhibition and oncolytic adenovirus. This finding has to be regarded in the light of an extremely efficient antitumoural immune response raised by the virus alone in both models which could be a phenomenon restricted to application of human adenoviruses in murine models. We could observe that human adenovirus significantly induced apoptosis in murine but not in human tumour cells, suggesting that mechanisms that allow human adenovirus to hide from the immune response in their natural host might be compromised in mice. Though the extent of antitumoural immunity could therefore be less pronounced in humans, it is interesting to speculate whether combining oncolysis and proteasome inhibition could even lead to increased synergism when transferred to human patients.

Together, our results indicate that ER stress-induced apoptosis during oncolytic adenoviral infection is an essential immunological danger signal that can be utilised for immunotherapeutic strategies against HCC. However, our results also suggest that timing and dosage of oncolytic adenovirus and adjuvant proteasome inhibition have to be carefully optimised to release the full potential of this complementary treatment in creating an effective and well-tolerated antitumoural vaccine.


Supplementary materials


  • BB, BM, FK and SK contributed equally to this work.

  • Funding Deutsche Forschungsgemeinschaft, Mildred-Scheel-Foundation and Wilhelm-Sander-Foundation.

  • Competing interests None.

  • Ethics approval All experiments involving mice were carried out according to German legal requirements (TierSchG) and in accordance with the NIH criteria for the care and use of laboratory animals.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.