(2011) 18, 765–777 & 2011 Macmillan Publishers Limited All rights reserved 0969-7128/11 www.nature.com/gt

ORIGINAL ARTICLE Cancer targeting Gene-Viro-Therapy of liver carcinoma by dual-regulated armed with TRAIL gene

X Cao1, M Yang2, R-C Wei3, Y Zeng4, J-F Gu1, W-D Huang1, D-Q Yang1, H-L Li1,5, M Ding1,NWei1, K-J Zhang1,BXu6, X-R Liu1, Q-J Qian7,8 and X-Y Liu1,7

Liver cancer is a common and aggressive malignancy, but available treatment approaches remain suboptimal. Cancer targeting Gene-Viro-Therapy (CTGVT) has shown excellent anti-tumor effects in a preclinical study. CTGVT takes advantage of both gene therapy and by incorporating an anti-tumor gene into an vector. Potent anti-tumor activity is achieved by virus replication and exogenous expression of the anti-tumor gene. A dual-regulated oncolytic adenoviral vector designated AdÁAFPÁE1AÁE1B (D55) (AdÁAFPÁD55 for short thereafter) was constructed by replacing the native viral E1A promoter with the simian virus 40 enhancer/a-fetoprotein (AFP) composite promoter (AFPep) based on an E1B-55K-deleted construct, ZD55. AdÁAFPÁD55 showed specific replication and cytotoxicity in AFP-positive hepatoma cells. It also showed enhanced safety in normal cells when compared with the mono-regulated vector ZD55. To improve the anti-hepatoma activities of the virus, the tumor necrosis factor-related -inducing ligand (TRAIL) gene was introduced into AdÁAFPÁD55. AdÁAFPÁD55-TRAIL exhibited remarkable anti-tumor activities in vitro and in vivo. Treatment with AdÁAFPÁD55-TRAIL can induce both autophagy owing to the AdÁAFPÁD55 vector and caspase-dependent apoptosis owing to the TRAIL . Therefore, AdÁAFPÁD55-TRAIL could be a potential anti-hepatoma agent with anti-tumor activities due to AFP-specific replication and TRAIL-induced apoptosis. Gene Therapy (2011) 18, 765–777; doi:10.1038/gt.2011.16; published online 17 March 2011

Keywords: cancer targeting Gene-Viro-Therapy; dual-regulated oncolytic adenovirus; tumor necrosis factor-related apoptosis-inducing ligand; autophagy; apoptosis

INTRODUCTION used first in our laboratory.3 The selective replication of ONYX-015 Although major progress has been made in the field of cancer therapy (similar to ZD55) in cancer cells was first thought to be dependent on in the past decades, liver cancer remains one of the most refractory deficiency.4 In 2004, O’Shea et al.5 attributed ONYX-015’s cancer- malignancies. According to the SEER (Surveillance, Epidemiology and specific replication to the late viral RNA export ability of E1B 55K. End Results) Cancer Statistics Review 1975–2006 of the USA National However, ZD55 is a mono-regulated oncolytic vector without the Cancer Institute, the 5-year relative survival rate of patients with ability to target a specific cancer and has some liver toxicity.6 To primary cancer originating from the liver and intrahepatic bile duct achieve liver cancer targeting and improved safety, a dual-regulated was 13.0% from 1999 to 2005 in the United States, which was second oncolytic adenovirus can be developed by using a liver cancer-specific only to that of pancreatic cancer patients (5.7%). Hepatocellular promoter to drive expression of the replication-essential gene E1A in carcinoma (HCC), which is the most common type of human liver the E1B-55K-deleted construct. a-Fetoprotein (AFP) is preferentially cancer and the fifth most frequent worldwide, represents expressed in over 70% of human liver cancer patients through the third leading cause of cancer-related death.1 Novel therapeutic transcriptional activation.7,8 Therefore, the AFP promoter is a good methods with high effectiveness and safety are urgently needed. candidate for this purpose. Some dual-regulated oncolytic adenoviral The Cancer targeting Gene-Viro-Therapy (CTGVT) approach vectors targeting HCC have been explored, including YKL-1001 and developed by our laboratory is one of the most promising strategies AdAFPep/Rep.9,10 In AdÁAFPÁE1AÁE1B (D55) (AdÁAFPÁD55 for short for the treatment of cancer.2 It combines gene therapy and virotherapy thereafter), which is used in this study, E1A expression was under the by utilizing an oncolytic adenoviral vector containing an anti-tumor control of the AFPep composite promoter within the E1B-55K-deleted gene. It has greater anti-tumor effects than gene therapy or virotherapy construct ZD55. alone. Different strategies were applied to construct the oncolytic The anti-tumor activities of adenovirus have been widely studied, virus. The ZD55 vector based on the deletion of the E1B-55K gene was but the mechanism by which adenovirus induces cancer death

1Laboratory of Molecular Cell Biology, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China; 2HIV and AIDS Malignancy Branch, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, MD, USA; 3School of Life Sciences, Guangxi Normal University, Guilin, China; 4Laboratory of Molecular Virology, Institut Pasteur of Shanghai, Chinese Academy of Sciences, Shanghai, China; 5Department of Horticulture and Garden, the Chongqing Engineering Research Center for Floriculture, Southwestern University, Chongqing, China; 6Department of General Surgery, Shanghai 10th People’s Hospital, Tongji University School of Medicine, Shanghai, China; 7Xinyuan Institute of Medicine and Biotechnology, Zhejiang Sci-Tech University, Hangzhou, China and 8Laboratory of Gene and Virus Therapy, Eastern Hepatobiliary Surgery Hospital, Second Military Medical University, Shanghai, China Correspondence: Dr X-Y Liu, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, 320 Yueyang Road, Shanghai 200031, China. E-mail: [email protected] Received 20 July 2010; revised 23 January 2011; accepted 24 January 2011; published online 17 March 2011 Cancer targeting Gene-Viro-Therapy of hepatoma XCaoet al 766

remains elusive. In 2004, Abou El Hassan et al.11 reported that conditionally replicating adenoviruses killed tumor cells via a basic apoptotic machinery-independent mechanism that resembles necrosis-like programmed cell death. More recently, this kind of cell death was identified as autophagic cell death or programmed cell death II.12–14 The oncolytic virus has limited anti-tumor activities, but it can target and replicate several hundred- to several thousand-fold in cancer cells. If we insert a therapeutic gene into a replication- competent oncolytic , the gene copy number can increase by several hundred- to several thousand-fold. Many members of our laboratory have shown the strong anti-tumor effect of the CTGVT strategy.3,15,16 In this study, the tumor necrosis factor-related apoptosis- inducing ligand (TRAIL) gene was incorporated into the oncolytic adenoviral vector AdÁAFPÁD55. TRAIL was first identified and cloned by Wiley et al. in 1995.17 Its receptors and signaling-transduction pathways have been extensively studied in the past decade.18 As TRAIL can induce apoptosis in many types of malignancy without severe damage to normal cells when incorporated into the oncolytic virus, it has been widely used in the field of cancer gene therapy.19 The bystander effect was also observed for the TRAIL-induced anti- tumor activities, although the underlying mechanisms responsible for this event are not yet clear.20–22

RESULTS Construction and characterization of AFP-targeting oncolytic adenoviruses AFP secretion was assayed in different hepatoma cell lines. As shown in Table 1, AFP secretion in conditional medium was highest for the HepG2 cell line, up to 1174±17.33 ng mlÀ1. The HuH-7 and PLC/ PRF/5 cell lines secreted AFP at moderate levels of 424.87±20.36 and 315.20±2.40 ng mlÀ1, respectively. The secretion of AFP by other hepatoma cell lines was low or undetectable. Transcriptional activity of the AFPep composite promoter was measured in different cell lines by use of the luciferase assay. The transcriptional activity of the AFPep composite promoter in different cell lines was consistent with the observed secretion levels of AFP and addition of the simian virus 40 (SV40) enhancer did not alter specificity of the AFP promoter (Figure 1a). It is worth noting that the AFPep composite promoter showed very high transcriptional activity in HuH-7 cells, exceeding that of the pGL3-Control. By contrast, transcriptional activity of the AFPep composite promoter was almost undetectable in the immorta- lized epithelial cell line BEAS-2B. Transcriptional activity could not be measured in the HepG2 and PLC/PRF/5 cell lines owing to their low transfection efficiency.

Figure 1 Construction and characterization of adenoviruses. (a) Relative activity of the AFPep composite promoter in various cell lines. The activity of pGL3- Table 1 AFP secretion in cell culture Basicwassetat1.Relativeactivitiesofthevariousconstructsincomparisonto pGL3-Basic are presented as the mean±s.d. (n¼3). (b) Schematic diagram of Cell AFP secretion (ng per day per 105cell) the viruses. All viruses were created using the backbone of wild-type Ad5. The E1B-55K gene was deleted in ZD55. The native promoter of E1A was replaced HepG2 1174.00±17.33 by the AFPep composite promoter in AdÁAFPÁD55andisbasedonZD55.The HuH-7 424.87±20.36 TRAIL expression cassette was inserted into AdÁAFPÁD55 to construct PLC/PRF/5 315.20±2.40 AdÁAFPÁD55-TRAIL. ITR, inverted terminal repeat; c, packaging signal. (c) Hep3B 38.20±1.29 Characterization of AdÁAFPÁD55 and AdÁAFPÁD55-TRAIL expression by western HuH-6 6.62±0.33 blot analysis. Expression of E1A by mock or infection with AdÁAFPÁD55 or AdÁAFPÁD55-TRAIL at an MOI of 10 in HuH-7 cells from 1 to 4 BEL-7402 0.67±0.04 days post-infection was evaluated. Actin was evaluated as a loading control. BEL-7404 N.D. (d) Characterization of AdÁAFPÁD55-TRAIL by -linked immunosorbent SMMC7721 N.D. assay. TRAIL concentrations in the supernatant of HuH-7 cell cultures after Abbreviations: AFP, a-fetoprotein; N.D., not detectable. infection with AdÁAFPÁD55-TRAIL at an MOI of 10 from 1 to 6 days post- The value represents the mean±s.d. of three independent experiments. infection were determined. Data are presented as the mean±s.d. (n¼3).

Gene Therapy Cancer targeting Gene-Viro-Therapy of hepatoma XCaoet al 767

The oncolytic adenoviruses constructed in this work are illustrated The pro-apoptotic TRAIL gene was introduced into the AFP- in Figure 1b. To characterize these viruses, the AFP-positive cell line targeting oncolytic adenoviral vector to form the dual-regulated HuH-7 was infected with AdÁAFPÁD55 or AdÁAFPÁD55-TRAIL. E1A gene-viral system, AdÁAFPÁD55-TRAIL. AdÁAFPÁD55-TRAIL showed expression was assayed by western blot (Figure 1c) and the TRAIL concentration in the conditional medium was measured by enzyme- linked immunosorbent assay (Figure 1d).

Selective cytotoxicity of Ad . AFP . D55 and Ad . AFP . D55-TRAIL in AFP-positive hepatoma cell lines and improved safety in the normal cell line To test infectability of the new constructs in various cell lines, all cell lines were infected with AdÁAFPÁD55 harboring enhanced green fluorescent protein (EGFP)genes(thatis,AdÁAFPÁD55-EGFP) at a multiplicity of infection (MOI) as indicated. All cell lines can be infected by AdÁAFPÁD55-EGFP, although the infection efficiencies may vary between different cell lines (Supplementary Figure 1). The HuH-7 and BEL-7404 cell lines showed the highest infection efficiencies, with 76.01 and 78.79% GFP-positive cells, respectively, infected at an MOI of 10. The differences in replication capabilities between the mono- regulated oncolytic adenovirus ZD55 and the dual-regulated oncolytic adenoviruses AdÁAFPÁD55 and AdÁAFPÁD55-TRAIL in the various cell lines were assayed using the standard tissue culture infectious dose 50 method. AdÁAFPÁD55 and AdÁAFPÁD55-TRAIL can selectively repli- cate in AFP-positive hepatoma cells, and their replication abilities were strongly attenuated in AFP-negative cells, including the HCC cell line BEL-7404. However, the mono-regulated oncolytic adenovirus ZD55 can replicate efficiently in all cancer cells (Figure 2a). The dual-regulated oncolytic adenoviral vector AdÁAFPÁD55 can induce cytopathic effects as efficiently as the mono-regulated oncolytic adenoviral vector ZD55 in AFP-positive hepatoma cell lines (HuH-7, HepG2 and PLC/PRF/5), whereas its oncolytic ability was more strongly attenuated in AFP-negative lines (Bcap-37, SW-620 and BEL-7404) (Figure 2b). The safety of AdÁAFPÁD55 in the immortalized epithelial cell line BEAS-2B was also improved compared with ZD55; it caused little damage to BEAS-2B cells even at an MOI of 200 (Figure 2c). The safety of AdÁAFPÁD55 for normal cells was further supported by the result from normal fibroblast cell line MRC-5 and HFL1 (Supplementary Figure 3). What is more, the enhanced safety of AdÁAFPÁD55 compared with ZD55 was also evaluated in vivo by intravenous injection in nude mice. The serum level of alanine aminotransferase (ALT) and aspartate aminotransferase (AST) in AdÁAFPÁD55-treated mice were reduced compared with the ZD55-treated ones (Po0.05) (Figure 2d). It indicated that AdÁAFPÁD55 was safer and better tolerated than ZD55 when delivered systematically in nude mice.

Figure 2 AdÁAFPÁD55 showed AFP-targeting ability and enhanced safety compared with mono-regulated oncolytic vector ZD55. (a) Selective replication of AdÁAFPÁD55 and AdÁAFPÁD55-TRAIL in AFP-positive cell lines. Progeny virus yields are presented as the mean±s.d. of three separate experiments. (b) AFP-targeting ability of the dual-regulated oncolytic vector AdÁAFPÁD55 compared with that of the mono-regulated ZD55. AFP-positive and -negative cells were infected with ZD55 and AdÁAFPÁD55, respectively, at the indicated MOIs. Cytopathic effect was observed by crystal violet staining at 5 days after infection. (c) Improved safety of AdÁAFPÁD55 compared with ZD55 in the immortalized epithelial cell line BEAS-2B was shown by the crystal violet assay. (d) ZD55, AdÁAFPÁD55 or PBS was intravenously injected into nude mice, respectively, and serum enzyme level of ALT and AST was measured at 24 hs post-injection. The elevation in serum enzyme level of AdÁAFPÁD55-treated mice was attenuated compared with ZD55-treated ones (n¼3, *Po0.05).

Gene Therapy Cancer targeting Gene-Viro-Therapy of hepatoma XCaoet al 768

enhanced cytotoxicity compared with AdÁAFPÁD55 in the AFP-posi- tive hepatoma cell lines (HuH-7 and HepG2), which was shown by crystal violet staining and the 3-(4,5-dimethylthiazol-2-yl)-2,5-di- phenyltetrazolium bromide (MTT) assay (Figures 3a and b). However, no severe side effect was induced by AdÁAFPÁD55 and AdÁAFPÁD55- TRAIL infection at an MOI of 10 in the immortalized epithelial cell line BEAS-2B (Figures 3a and b) and normal fibroblast cell line MRC-5 or HFL1 (Supplementary Figure 4) assayed by the same methods. It is intriguing that AdÁAFPÁD55 and AdÁAFPÁD55-TRAIL also showed some cell-killing effect in AFP-negative hepatoma cell lines BEL-7404 at an MOI of 10, although not effective as that in HuH-7 and HepG2 (Supplementary Figure 2).

Ad . AFP . D55 and Ad . AFP . D55-TRAIL infection may cause autophagy Infection with ZD55 or AdÁAFPÁD55 at an MOI of 10 for 48 h enhanced the intracellular ATP level in HuH-7 cells (Figure 4A). A decrease in intracellular ATP levels is frequently considered a biochemical marker of necrosis. A drop in the ATP level also indicates decreased or impaired mitochondrial function, and it is often asso- ciated with mitochondrial membrane potential dissipation during apoptosis. Our results suggest that neither necrosis nor apoptosis is the principal cause of cell death induced by ZD55 or AdÁAFPÁD55. To investigate the mechanism of cell death induced by AdÁAFPÁD55 infection, a plasmid encoding the EGFP-microtubule-associated protein light chain 3 (LC3) was transfected into HuH-7 cells before adenoviral infection. LC3 is first cleaved at the carboxy terminus immediately following synthesis to yield a cytosolic form LC3-I. During autophagy, LC3-I is converted to LC3-II through lipidation by an ubiquitin-like system that allows for LC3 to become associated with autophagic vesicles. The distribution of EGFP-LC3 from diffused (LC3-I) in the cytosol to punctuate (LC3-II) indicates that autophagy occurs. EGFP-LC3 diffused into the cytosol in the mock group, whereas numerous punctate structures expressing EGFP were easily seen after infection of HuH-7 cells with ZD55, AdÁAFPÁD55 or AdÁAFPÁD55-TRAIL (Figure 4B). To quantify the cell population that underwent autophagy after different treatments, cells with more than five EGFP-LC3-positive punctae (each greater than 1 mm in diameter) were counted using the Operetta Compact High Content Screening System. In AFP-positive HuH-7 cells, AdÁAFPÁD55 and AdÁAFPÁD55-TRAIL infection caused an apparent increase in EGFP-dotted cells compared with that of mock treatment (Po0.01; t-test), up to 12.11 and 20.46%, respectively. The presence of these structures was inhibited by the addition of 3-methylademine, a widely used autophagy inhibitor. There were no significant differences in the immortalized epithelial BEAS-2B cell line among the different treatments (Figure 4C). Ultrastructures of HuH-7 cells under different treatments were observed by transmission electron microscopy. Both AdÁAFPÁD55- Figure 3 Introduction of the TRAIL gene in AdÁAFPÁD55 showed stronger and AdÁAFPÁD55-TRAIL-infected cells displayed a large quantity of cytotoxicity in AFP-positive hepatoma cells without severe side effects to vacuolated structures in the cytoplasm, which is a hallmark of immortalized epithelial cells. (a)AdÁAFPÁD55-TRAIL showed enhanced autophagy. Furthermore, apoptotic bodies were observed around the tumoricidal activity in AFP-positive hepatoma cells without obvious damage AdÁAFPÁD55-TRAIL-infected HuH-7 cells (Figure 4D, panel a). To to the immortalized epithelial cell line BEAS-2B as shown by the crystal validate the occurrence of autophagy in a more convincing manner, violet assay. (b) Cell viability following mock infection or infection with AdÁAFPÁD55 or AdÁAFPÁD55-TRAIL at the indicated MOIs was measured by the translocation of LC3 to autophagosomes was shown by post- the MTT assay at 5 days post-infection. The results are presented as the embedding immunogold staining (Figure 4D, panel b). Briefly, 10-nm mean±s.d. (n¼4; *Po0.05; **Po0.01) and are expressed as the colloidal gold particles were located on the membrane of autophago- percentage relative to mock-treated control cells. somes after infection with the oncolytic adenovirus, whereas cytoplasmic localization was observed in the mock-treated cells. LC3-II is considerably degraded by lysosomal hydrolases after the The conversion of LC3-I to LC3-II with or without chloroquine formation of autolysosomes, the amount of LC3-II seemed to be only incubation was monitored by western blotting. As endogenous moderately upregulated after infection with either AdÁAFPÁD55 or

Gene Therapy Cancer targeting Gene-Viro-Therapy of hepatoma XCaoet al 769

Figure 4 Both AdÁAFPÁD55 and AdÁAFPÁD55-TRAIL infection caused prominent autophagy in the HuH-7 cell line. (A) Intracellular ATP levels in HuH-7 cells 48 h after mock infection or infection with AdÁAFPÁD55 or AdÁAFPÁD55-TRAIL at an MOI of 10 were assayed by a luciferase-based method (n¼3; *Po0.05; **Po0.01). (B) Translocation of the EGFP-LC3 fusion protein after adenovirus infection. A plasmid encoding EGFP-LC3 was transfected into HuH-7 cells, followed by mock infection or infection with AdÁAFPÁD55 or AdÁAFPÁD55-TRAIL at an MOI of 10 after 6 h. At 48 h post-infection, EGFP distribution in HuH- 7 cells was observed by Leica SP2 confocal microscope. Representative confocal images of EGFP-LC3 are shown. Scale bar¼50 mm. (C) Quantitation of EGFP-LC3-positive cells with more than five EGFP-LC3 punctate dots (greater than 1 mm in diameter) in HuH-7 or BEAS-2B cells following mock infection or infection with AdÁAFPÁD55 or AdÁAFPÁD55-TRAIL at an MOI of 10 with or without 1 mM 3-methylademine pre-treatment (n¼3; **Po0.01; NS, P40.05, not significant). (D) Ultrastructure of HuH-7 cells after mock infection or infection with AdÁAFPÁD55 or AdÁAFPÁD55-TRAIL at an MOI of 10 for 3 days. Panel a: Representative transmission electron micrographs of HuH-7 cells after different treatments. Scale bar¼1 mm. Panel b: Localization of LC3 by post- embedding immunogold staining. Arrows indicate the 10 nm anti-LC3 colloidal gold particles and arrowheads indicate autophagosomes. The 10 nm colloidal gold particles indicated by arrows are magnified. Scale bar¼500 nm for original field and 200 nm for magnified field. (E) At 90 h after mock infection or infection with AdÁAFPÁD55 or AdÁAFPÁD55-TRAIL at an MOI of 10, HuH-7 cells were incubated with or without chloroquine (50 mM) for 6 h. LC3-II and p62 levels were tested by western blot and quantified relative to glyceraldehyde 3-phosphate dehydrogenase.

Gene Therapy Cancer targeting Gene-Viro-Therapy of hepatoma XCaoet al 770

AdÁAFPÁD55-TRAIL without chloroquine incubation. Chloroquine to exclude mitochondria-derived contaminant in the cytosol part. The increases the pH of lysosome, thereby preventing the activity of extrinsic and intrinsic apoptosis pathways were activated after lysosomal acid proteases and causing autophagosomes to accumulate. AdÁAFPÁD55-TRAIL infection of the HuH-7 cell line. Strikingly, LC3-II accumulated at an apparently high level after both adenoviral infection at an MOI of 10 for 4 days compared with mock- Anti-hepatoma effect of Ad . AFP . D55-TRAIL in nude mice infected group with chloroquine incubation for 6 h (Figure 4E). A hepatoma xenograft model was established by inoculating AFP- Autophagy induced by oncolytic adenovirus infection was also positive HuH-7 cells into the right back of nude mice. When tumor shown by the degradation of p62/SQTM1, which is the substrate of volume reached 80–150 mm3, phosphate-buffered saline (PBS), autophagy process (Figure 4E). AdÁAFPÁD55 and AdÁAFPÁD55-TRAIL were administered by intratu- moral injection in each group. Tumor volumes were monitored weekly Ad . AFP . D55-TRAIL-induced apoptosis of HuH-7 cells after the first injection. The tumors grew rapidly and exceeded a mean After different treatments, the nuclear morphology of the HuH-7 and of 1200 mm3 within a month in the PBS group. Tumor growth in the BEAS-2B cell lines was observed under a confocal microscope after AdÁAFPÁD55 and AdÁAFPÁD55-TRAIL groups was significantly inhib- Hoechst 33258 staining. The infection of HuH-7 cells with AdÁAFPÁ ited. It is worth noting that nearly 74% of the tumor burden was D55-TRAIL can induce nuclear fragmentation and chromatin con- inhibited in the AdÁAFPÁD55-TRAIL group compared with the PBS densation, which are characteristic of apoptosis. However, mock group at the end of the fourth week, with a mean volume of 328 mm3 infection and infection with AdÁAFPÁD55 did not cause abnormal (Figure 6a). The serum liver levels of ALT, AST and alkaline phos- nuclear morphology of the HuH-7 cells. Infection with either AdÁAFPÁ phatase (ALP) were assayed in each group (Figure 6b). There were no D55 or AdÁAFPÁD55-TRAIL did not cause damage to the immorta- significant differences between each treatment, indicating that the lized epithelial cell line BEAS-2B (Figure 5a). Similar results were intratumoral administration of AdÁAFPÁD55 and AdÁAFPÁD55-TRAIL confirmed by the DNA fragmentation assay. The DNA ladder was caused little hepatotoxicity. clearly seen in AdÁAFPÁD55-TRAIL-treated HuH-7 cells (Figure 5b). To confirm that the tumoricidal activities were mediated by the To quantitate the proportion of cells that underwent apoptosis, dual-regulated oncolytic virus, expression of the viral E1A samples of HuH-7 hepatoma cells and BEAS-2B immortalized epithe- and Hexon and the therapeutic gene TRAIL were measured by lial cells under different treatments were stained by propidium iodide immunohistochemistry in tumor specimens harvested at 7 days and analyzed by flow cytometry. The number of HuH-7 cells in the after the first injection of virus. The expression of E1A and Hexon sub-G1 phase was elevated after AdÁAFPÁD55-TRAIL infection com- can be observed in both AdÁAFPÁD55- and AdÁAFPÁD55-TRAIL- pared with HuH-7 cells after mock and AdÁAFPÁD55 infection. No treated samples, and expression of TRAIL was only seen in the significant differences were seen in the immortalized epithelial cell line AdÁAFPÁD55-TRAIL-treated sample (Figure 6c). Furthermore, the BEAS-2B after each treatment (Figure 5c). viral replication kinetics of AdÁAFPÁD55-TRAIL in the tumor were To address whether the apoptosis induced by AdÁAFPÁD55-TRAIL measured by TaqMan real-time PCR using the adenoviral shuttle was caspase dependent, the expression of TRAIL and activated caspase- plasmid pXC1 to generate the standard curve (Figure 6d, left panel). 3 in the HuH-7 cell line after adenoviral infection was analyzed by Viral replication of AdÁAFPÁD55-TRAIL peaked at 5 days after the first immunofluorescence. Cleaved caspase-3 was expressed in the injection, with a mean of 4.52Â106 copies per nanogram of total DNA AdÁAFPÁD55-TRAIL treatment group, and it seemed to correlate with (Figure 6d, right panel). Similar replication kinetics were observed for TRAIL expression as neither infection with AdÁAFPÁD55 nor mock AdÁAFPÁD55 (data not shown). infection induced caspase-3 activation (Figure 5d). Expression of the key proteins involved in the extrinsic and intrinsic apoptosis pathways Ad . AFP . D55-TRAIL induced autophagy and apoptosis in vivo was evaluated by western blotting. The expression levels of procaspase- To validate the mechanisms of cell death induced by AdÁAFPÁD55- 3, procaspase-8 and procaspase-9 were reduced, and cleaved caspase-3, TRAIL in vivo, autophagy and apoptosis were assayed in HuH-7 cleaved caspase-8 and cleaved caspase-9 was detected after tumor specimens using different methods. Hexon and LC3 proteins in AdÁAFPÁD55-TRAIL infection at an MOI of 10 for 3 or 4 days. The the ultrathin sections of HuH-7 tumors were stained by the post- poly-(ADP-ribose) polymerase substrate of caspase-3 was also cleaved embedding immunogold cytochemical reaction. As shown in with AdÁAFPÁD55-TRAIL treatment (Figure 5e). Cytoplasmic and Figure 7a, the 10-nm colloidal gold particles staining LC3 were located mitochondrial proteins were extracted from the HuH-7 cell line after in the membrane of autophagosomes in the AdÁAFPÁD55- and different treatments. Translocation of cytochrome c from the mitochon- AdÁAFPÁD55-TRAIL-treated samples, whereas their distribution was dria to the cytoplasm after AdÁAFPÁD55-TRAIL infection of HuH-7 dispersed throughout the cytosol in the PBS-treated sample. The cells was validated by western blotting (Figure 5f). The mitochondrial phenomenon mentioned above indicates that infection with the outer membrane protein voltage-dependent anion channel was applied oncolytic adenovirus might cause autophagic cell death in vivo.

Figure 5 AdÁAFPÁD55-TRAIL induced caspase-dependent apoptosis in an AFP-positive hepatoma cell line. (a) HuH-7 and BEAS-2B cells were stained with Hoechst 33258 at 2 days after mock infection or infection with AdÁAFPÁD55 or AdÁAFPÁD55-TRAIL at an MOI of 10. Arrows indicate apoptotic cells. Scale bar¼50 mm. (b) DNA fragmentation analysis. Genomic DNA was harvested from HuH-7 or BEAS-2B cells at 72 h after mock infection or infection with AdÁAFPÁD55 or AdÁAFPÁD55-TRAIL at an MOI of 10 and separated on a 2% agarose gel. (c) Flow cytometric analysis of propidium iodide-stained cells was performed at 48 h after mock infection or infection with AdÁAFPÁD55 or AdÁAFPÁD55-TRAIL at an MOI of 10. The percentages of sub-G1 cells are shown. (d) TRAIL and cleaved caspase-3 expression levels were determined by fluorescent immunostaining at 48 h after mock infection or infection with AdÁAFPÁD55 or AdÁAFPÁD55-TRAIL at an MOI of 10 in HuH-7 cells. Scale bar¼50 mm. (e) HuH-7 cells were mock infected or infected with AdÁAFPÁD55 or AdÁAFPÁD55-TRAIL at an MOI of 10 for 3 or 4 days. Key proteins involved in both extrinsic and intrinsic apoptotic pathways were assayed by western blotting. (f) Mitochondrial and cytosolic proteins were prepared of HuH-7 cells after mock infection or infection with AdÁAFPÁD55 or AdÁAFPÁD55-TRAIL at an MOI of 10 for 3 days. Cytochrome c release from the mitochondria to the cytoplasm was tested by western blot.

Gene Therapy Cancer targeting Gene-Viro-Therapy of hepatoma XCaoet al 771

Gene Therapy Cancer targeting Gene-Viro-Therapy of hepatoma XCaoet al 772

TUNEL-positive cells was only seen in the AdÁAFPÁD55-TRAIL- treated samples (Figure 7b). Expression levels of TRAIL and cleaved caspase-3 were measured by immunofluorescence. As shown in Figure 7c, the apoptosis induced by AdÁAFPÁD55-TRAIL in vivo was caspase-3 dependent and correlated with the expression of TRAIL.

DISCUSSION To achieve targeting of HCC and improved safety in normal cells, a dual-regulated oncolytic vector designated as AdÁAFPÁD55 and based on the ZD55 vector was constructed using the AFPep composite promoter to drive E1A expression. The AFPep composite promoter consists of a 0.3-kb AFP promoter and a 235-bp SV40 enhancer. The 0.3-kb AFP promoter contains a glucocorticoid response element and two binding sites for hepatocyte nuclear factor-1, a hepatocyte-specific transcriptional factor. It also contains a TATA box, but it lacks the typical CCAAT sequence. Although a previous study showed that the 0.3-kb AFP promoter had sufficient activity to stimulate the tran- scription of downstream genes, its efficiency was limited.23,24 For this reason, the AFP enhancer is often inserted upstream of the promoter to increase its activity. However, the AFP enhancer spans a long sequence. The AFP enhancer utilized by Minoru et al.10 was 0.7 kb. To save space for the exogenous gene expression cassette, the shorter SV40 enhancer was used. The SV40 enhancer is comprised of a 72- repeat. Its efficiency at increasing promoter activity was previously shown by many groups.25,26 Introduction of the SV40 enhancer upstream of the 0.3-kb AFP promoter did not change the transcriptional specificity of the AFP promoter (Figure 1a). Transcrip- tional activities of the AFPep composite promoter may not be as powerful as the original E1A promoter of the wild-type adenovirus as the replication ability of AdÁAFPÁD55 was weaker than that of ZD55, even in AFP-positive cells (Figure 2a). However, the cytotoxicity of AdÁAFPÁD55 in AFP-positive cells was comparable to that of ZD55 (Figure 2b). Obvious autophagy was observed during both AdÁAFPÁD55 and AdÁAFPÁD55-TRAIL infection and was illustrated by EGFP-LC3 translocation, autophagosome formation and LC3-II induction (Figures 4B–E). Autophagy is a conserved catabolic process for the autophagosomic-lysosomal degradation of long-lived proteins, orga- nelles and bulk cytoplasm.27,28 Although autophagy can act to promote cell survival during starvation, excessive autophagy may result in type II programmed cell death.29,30 Jiang et al.31 speculated that autophagy may facilitate material recycling, energy production and viral release during adenovirus infection. Our data support this hypothesis to some extent. Elevated intracellular ATP levels were observed during oncolytic adenoviral infection (Figure 4A), which suggest that adenoviral infection may increase ATP levels to provide Figure 6 Anti-hepatoma activity of AdÁAFPÁD55-TRAIL in the HuH-7 xenograft model. (a) Female BALB/c nude mice were inoculated subcutaneously with sufficient energy for the increased anabolism in infected cells. Cyto- HuH-7 cells (5Â106). Tumor size was recorded and tumor volume was plasmic structure and nuclear and cellular membranes were destroyed calculated. Arrows indicate virus injection. Points indicate mean values (n¼5) late in the adenoviral infection cycle by excessive autophagy and bars indicate standard deviation (*Po0.05; **Po0.01). (b) The levels of (Figure 4D), which facilitated the release of virions. serum ALT, AST and ALP are presented (NS, P40.05, not Introduction of the TRAIL gene in the dual-regulated oncolytic significant). (c) Representative micrographs of immunohistochemical staining vector AdÁAFPÁD55 resulted in the induction of apoptosis. Extrinsic for E1A, Hexon and TRAIL in tumor tissue. Arrows indicate cells that stained positively for the indicated protein. Scale bar¼100 mm. (d) Analysis of the viral and intrinsic apoptotic pathways were involved in cell death by the replication kinetics of AdÁAFPÁD55-TRAIL in HuH-7 tumors by TaqMan real- AdÁAFPÁD55-TRAIL virus as shown by western blotting (Figure 5e). time PCR. Plasmid pXC1 was used as a template to generate the standard Interestingly, the percentage of autophagy is higher in TRAIL-expres- curve (left panel). The primers and probe for E1A were used to detect the sing cells than that in AdÁAFPÁD55-infected cells as shown in genomic copy number of adenovirus in the tumor tissue. Figure 4C. The roles of TRAIL in autophagy have been investigated extensively, but still elusive. TRAIL-induced autophagy was found in The terminal deoxynucleotidyl transferase-mediated dUTP-biotin lumen formation of MCF-10A human mammary epithelial cells nick-end labeling (TUNEL) assay was used to detect apoptosis in vitro,32 apoptosis-defective tumor cells33 and untransformed in paraffin sections of HuH-7 tumors in situ. A large area of human epithelial cells.34 Activation of TRAIL receptor 2 by a human

Gene Therapy Cancer targeting Gene-Viro-Therapy of hepatoma XCaoet al 773

Figure 7 Validation of the mechanisms of anti-hepatoma activity induced by AdÁAFPÁD55-TRAIL in vivo.(a)AdÁAFPÁD55 and AdÁAFPÁD55-TRAIL infection induced autophagy in HuH-7 tumor tissue. Hexon and LC3 were detected by the post-embedding immunogold staining assay. LC3 was translocated to the membrane of autophagosomes after adenovirus infection. Arrows indicate the colloidal gold particles and arrowheads indicate autophagosomes. Scale bar¼500 nm. (b) Tumor sections derived from the in vivo study were subjected to an in situ cell apoptosis detection kit to detect apoptotic cells. The TUNEL-positive cells were only observed in the AdÁAFPÁD55-TRAIL-treated group. Scale bar¼50 mm. (c) Tumor sections derived from the in vivo study were subjected to an immunofluorescence assay to evaluate the expression of TRAIL and cleaved caspase-3. Scale bar¼22 mm. single-chain fragment variable HW1 can induce autophagic One important concern of hepatoma treatment is the resistance of cell death in both TRAIL-sensitive and -resistant cancer cells.35 HCC to drugs, including TRAIL. Previous reports showed that E1A Although the concurrence of autophagic and apoptotic cell death expression enhanced TRAIL sensitivity in human cancer cells and was reported from a combination of adenoviral virotherapy and normal cell lines.37,38 Hence, incorporation of TRAIL gene into temozolomide in malignant glioma,36 the results pre- oncolytic adenovirus with E1A expression may make some TRAIL- sented here are the first to show that these two distinct programmed resistant cancer cells sensitive to our gene-viral agent. As the E1A cell death pathways can be simultaneously induced by a single gene- expression was under the control of AFPep composite promoter in viral agent in hepatoma. AdÁAFPÁD55, E1A expression was dependent on AFP status of cell line

Gene Therapy Cancer targeting Gene-Viro-Therapy of hepatoma XCaoet al 774

infected and could not express in normal cell line, leaving ligated by treatment with T4 DNA ligase (TaKaRa, Shiga, Japan) to form a AdÁAFPÁD55-TRAIL-infected normal cells spared. This explanation novel shuttle plasmid designated pZXC2-ZD55, which lacks the E1A promoter was supported by cytotoxic effect of AdÁAFPÁD55-TRAIL in TRAIL- and E1B-55K. To obtain the AFPep composite promoter, PCR was performed resistant cell line HepG2 and immortalized epithelial cell line BEAS-2B with high-fidelity polymerase KOD-plus (ToYoBo, Osaka, Japan) and the (Figure 3). Although the volume of HuH-7 xenograft tumor was plasmid pDRIVE03-SV40enh/AFP(h) v04 was used as a template. The primers used for amplification were as follows: sense primer, 5¢-GCACTCGAGGGCCT diminished by AdÁAFPÁD55-TRAIL in this study, it was not elimi- GAAATAACCTCTG-3¢; antisense primer, 5¢-GACTACGTATGTTATTGGCAGT nated. Our previous work verified that the entire inoculated tumor GGTGG-3¢.TheXhoIandSnaBI recognition sites are underlined. The PCR could be completely eradicated by utilization of two genes with product was digested with XhoIandSnaBI and ligated into the XhoI/SnaBI site compensatory or synergetic effects, which is the basis of the Cancer of pZXC2-ZD55 to form plasmid pAFP-D55. The TRAIL expression cassette Targeting Dual Gene-Viro-Therapy strategy.39–41 As SOCS3 is defi- (with hCMV promoter and SV40 polyadenylation signal) was released from the cient in HCC and may sensitize human hepatoma cells to TRAIL- previously constructed plasmid pCA13-TRAIL by digestion with BglII. It was mediated apoptosis,42,43 AdÁAFPÁD55 armed with the SOCS3 gene was inserted into the BglII site of pAFP-D55 to generate the recombinant plasmid constructed. The anti-hepatoma activities of AdÁAFPÁD55-SOCS3 and named pAFP-D55-TRAIL. The plasmid pAFP-D55-EGFP was constructed by AdÁAFPÁD55-TRAIL co-treatment are currently being assessed. the same method.

MATERIALS AND METHODS Generation, purification and titration of recombinant adenovirus Cell culture Oncolytic adenoviruses (AdÁAFPÁD55, AdÁAFPÁD55-TRAIL and AdÁAFPÁD55- The HepG2, Hep3B, BEL7404, BEL-7402, SMMC-7721 (human HCC), Bcap-37 EGFP) were constructed by standard techniques (human breast cancer), SW620 (human ), MRC-5 and HFL1 using the adenoviral shuttle plasmids (pAFP-D55, pAFP-D55-TRAIL and (normal lung fibroblast cell) cell lines were purchased from the Cell Bank of pAFP-D55-EGFP) and the adenovirus packaging plasmid pBHGE3 (Microbix Type Culture Collection of Chinese Academy of Sciences (Shanghai, China). Biosystems Inc.) in HEK293 cells. One microgram of shuttle plasmid was The human HCC cell lines HuH-7 and HuH-6 were obtained from the RIKEN mixed with 0.5 mg of pBHGE3 and then co-transfected into HEK293 cells (six- Cell Bank (Ibaraki, Japan). The highly spontaneous metastatic human HCC cell well format) using the Effectene Transfection Reagent (Qiagen, Hilden, line HCCLM6 was obtained from the Liver Cancer Institute, Zhongshan Germany). The resultant recombinant adenovirus was isolated from a single Hospital, Fudan University (Shanghai, China). The human hepatoma cell line plaque and expanded in HEK293 cells. Genomic DNA harvested from the PLC/PRF/5 was kindly provided by Dr Qi-Jun Qian from Second Military candidate viruses was extracted using the Genomic DNA MiniPreps Kit for Medical University (Shanghai, China). Immortalized epithelial cell line Blood (Generay, Shanghai, China) and evaluated by PCR to verify its structure BEAS-2B, which was isolated from normal human bronchial epithelium and exclude wild-type adenovirus contamination. obtained from autopsy of non-cancerous individuals, was a kind gift from Large-scale purification of adenoviruses was performed by ultracentrifuga- Professor Hong-Bin Ji’s lab in Shang Institute of Biochemistry and Cell Biology. tion with cesium chloride. The titers were determined by plaque assay with The HEK293 cell line was purchased from Microbix Biosystems Inc. (Toronto, HEK293 cells. Ontario, Canada). Cell lines were maintained in Dulbecco’s modified Eagle’s medium (GIBCO BRL, Grand Island, NY, USA) supplemented with 10% fetal Enzyme-linked immunosorbent assay to evaluate TRAIL expression bovine serum (GIBCO BRL). All cell lines were incubated at 37 1Cina in conditional medium humidified air atmosphere with 5% CO2. After AdÁAFPÁD55-TRAIL infection at an MOI of 10, the supernatants of HuH- 7 cells were collected from 1 to 6 days post-infection. The expression of TRAIL AFP measurement in cell culture media was measured with a commercial human TRAIL enzyme- Cells were seeded in 24-well plates and cultured at 37 1C for 24 h. The cultured linked immunosorbent assay kit following the manufacturer’s instructions medium was then replaced by fresh medium without fetal bovine serum. After (Bender MedSystems, Vienna, Austria). 24 h of cultivation, AFP concentrations in the media were measured by the radioimmunity method at Zhongshan Hospital, Fudan University (Shanghai, Virus progeny production assay China). Cells were infected with different viruses at an MOI of 10. After 48 h of infection, the medium and cells were collected and lysed by three freeze–thaw Reporter gene assay to measure the transcriptional activity of the cycles. The products were titrated by the tissue culture infectious dose 50 AFPep composite promoter method using HEK293 cells, and the results were transformed to PFU per cell. The plasmid pDRIVE03-SV40enh/AFP(h) v04 containing the SV40 enhancer These experiments were repeated three times. and human AFP promoter sequence was purchased from InvivoGen Inc. (San Diego, CA, USA). Luciferase reporter plasmids pGL3-Basic (promoter-free) Cytopathic assay and pGL3-Control (with SV40 promoter/enhancer) were purchased from Different cell lines were grown to subconfluence and infected with adenoviruses Promega Corp. (Madison, WI, USA). The plasmid pGL3-AFPep was con- at various MOIs. At 5 days after infection, cells were exposed to 2% crystal structed by inserting the AFPep composite promoter sequence into the pGL3- violet in 20% methanol for 15 min, washed with distilled water (dH O) and Basic plasmid. To measure the transcriptional activity of the AFPep composite 2 documented in photographs. promoter in different cell lines, cells were seeded in 24-well plates and co- transfected with 0.8 mg of reporter plasmid and 0.08 mg of internal control plasmid pCMV-b-galactosidase encoding b-galactosidase using Lipofectamine Cell viability by the colorimetric MTT assay 3 2000 (Invitrogen, Carlsbad, CA, USA). At 48 hours post-transfection, cells were Cells were seeded in 96-well plates at a density of 6Â10 cells per well. When harvested and the firefly luciferase and b-galactosidase activities were assayed cells reached subconfluence, they were infected with adenoviruses at the using the Luciferase reporter assay system (Promega) and the b-galactosidase indicated MOI. The colorimetric MTT assay was performed as described 45 assay kit (Beyotime, Haimen, China), respectively. previously. The percentage of cell viability was calculated as follows: (mean A570 of infected cells)/(mean A570 of uninfected cells)Â100%. Construction of the oncolytic adenoviral shuttle vectors The adenoviral shuttle plasmids pZXC2 (ref. 44) and pZD55 (ref. 3) were ATP assay previously constructed and preserved in our laboratory. The two plasmids At 48 h after ZD55 or AdÁAFPÁD55 infection at an MOI of 10, the intracellular mentioned above were digested with EcoRI and XbaI, respectively. The smaller ATP level of HuH-7 cell line was determined by an ATP Assay Kit according to fragment of the former (1.5 kb) and larger fragment of the latter (7.3 kb) were the manufacturer’s instructions (Beyotime).

Gene Therapy Cancer targeting Gene-Viro-Therapy of hepatoma XCaoet al 775

Autophagy assay by EGFP-LC3 translocation rhodamine isothiocyanate-conjugated rabbit anti-mouse IgG+IgM (H+L) To measure autophagy, HuH-7 cells seeded in 24-well plates were transfected (Jackson ImmunoResearch Laboratories). The samples were viewed and images with 0.8 mg pEGFP-LC3 by Lipofectamine 2000 (Invitrogen), followed by mock were obtained with a confocal imaging system (Leica TCS SP2). infection or AdÁAFPÁD55 or AdÁAFPÁD55-TRAIL infection at an MOI of 10 after 6 h oftransfection. At 48 h after post-infection, the cellular localization Subcellular protein fractionation pattern of EGFP-LC3 was photographed using a Leica TCS SP2 confocal HuH-7 cells with total number of 5Â107 were mock infected or infected with microscope (Leica Microsystems, Wetzlar, Germany). To quantify the percen- AdÁAFPÁD55 or AdÁAFPÁD55-TRAIL at an MOI of 10 for 3 days. Mitochondrial tage of EGFP-LC3-positive cells with EGFP-LC3 punctate dots, HuH-7 or and cytosolic proteins were prepared with a Mitochondrial Protein Isolation kit BEAS-2B cells seeded in 96-well plates were treated as above with or without (KeyGen, Nanjing, China). 1mM 3-methylademine (Merck-Calbiochem, Darmstadt, Germany) pre-treat- ment before infection. Cells with more than five EGFP-LC3 punctate dots Western blot analysis m (greater than 1 m in diameter) were counted using an Operetta Compact High For oncolytical adenovirus characterization, HuH-7 cells seeded in six-well Content Screening System (Perkin-Elmer, Waltham, MA, USA). Four random plates were mock infected or infected with AdÁAFPÁD55 or AdÁAFPÁD55-TRAIL fields per well were counted in triplicate well per condition per experiment. at an MOI of 10 for 1–4 days. For LC3-II and p62 detection, HuH-7 cells seeded in six-well plates were mock infected or infected with AdÁAFPÁD55 or Transmission electron microscopy for morphological observation AdÁAFPÁD55-TRAIL at an MOI of 10 for 4 days with or without incubation and post-embedding immunogold staining with 50 mM chloroquine (Sigma) for 6 h before sample collection. For detection The ultrastructure of HuH-7 cells after mock infection or infection with key proteins involved in apoptotic pathway, HuH-7 cells seeded in six-well AdÁAFPÁD55 or AdÁAFPÁD55-TRAIL at an MOI 10 for 3 days was observed plates were mock infected or infected with AdÁAFPÁD55 or AdÁAFPÁD55-TRAIL by transmission electron microscopy using the standard procedure. For at an MOI 10 for 3 or 4 days. Total proteins from different samples were determination of the subcellular localization of LC3, ultrathin sections of extracted with cell buffer for western blot and immunoprecipitation HuH-7 cells or tumors were stained by post-embedding immunogold staining analyses (Beyotime) and quantitated using an Enhanced BCA Protein Assay using the Leica EM IGL Automated Immunolabeling System (Leica Micro- Kit (Beyotime). Western blot analysis was carried out using standard proce- systems). Anti-LC3 antibody was purchased from MBL International Corpora- dures. E1A, cleaved caspase-3, cleaved caspase-8, cleaved caspase-9, glyceralde- tion (Nagoya, Japan) and anti-Hexon antibody was obtained from Millipore hyde 3-phosphate dehydrogenase and voltage-dependent anion channel Corporation (Billerica, MA, USA). Colloidal gold-labeled goat anti-rabbit were purchased from Cell Signaling Inc. (Danvers, MA, USA). immunoglobulin (Ig)G (10 nm) (Sigma, St Louis, MO, USA) was applied to Procaspase-3, procaspase-8 and horseradish peroxidase-conjugated secondary detect anti-LC3 antibody and rabbit anti-goat IgG (10 nm) (Sigma) was applied antibodies were obtained from Santa Cruz Biotechnology (Santa Cruz, CA, to detect anti-Hexon antibody. Images were obtained with a Jeol JEM-1230 USA). The procaspase-9 antibody was manufactured by Thermo Fisher transmission electron microscope (JEOL Ltd, Akishima, Japan). Scientific Inc. (Fremont, CA, USA) and the cytochrome c antibody was purchased from R&D Systems (Minneapolis, MN, USA). LC3 and p62 anti- Hoechst 33258 staining and DNA fragmentation body was purchased from MBL International Corporation (Nagoya, Japan). HuH-7 and BEAS-2B cells were grown on glass coverslips in a six-well plate and The expression levels of LC3-II and p62 were quantified by densitometry using treated with AdÁAFPÁD55 or AdÁAFPÁD55-TRAIL at an MOI of 10 for 48 h or the Image J software and expressed relative to glyceraldehyde 3-phosphate left untreated (mock). Cells were fixed with 4% paraformaldehyde and stained dehydrogenase signals. with 1 mgmlÀ1 Hoechst 33258 (Molecular Probes, Eugene, OR, USA) for 5 min. Cells were then washed twice with PBS and visualized with a Leica TCS SP2 Animal experiments confocal microscope. Four-week-old female athymic nude mice were purchased from the Shanghai For the DNA fragmentation assay, HuH-7 or BEAS-2B cells were seeded in Experimental Animal Center (Shanghai, China). All procedures were per- 10-cm culture dishes (5Â106 cells per dish) and infected with the indicated formed in accordance with the National Institutes of Health Guide for the viruses. At 72 h post-infection, genomic DNA was extracted as previously Care and Use of Laboratory Animals. For the treatment of pre-established described and separated by electrophoresis through 2% agarose gels.46 tumors, 5Â106 HuH-7 cells in 150 ml Dulbecco’s modified Eagle’s medium were subcutaneously injected into the right back of each mouse. When tumors analysis by flow cytometric analysis reached 80–150 mm3, mice were randomly divided into three groups (five mice To quantitate apoptosis using flow cytometry, both adherent and floating cells per group). AdÁAFPÁD55, AdÁAFPÁD55-TRAIL (2Â109 PFUs per mouse) or were harvested at 48 h after mock infection or infection with AdÁAFPÁD55 or PBS was injected into the tumors for 5 consecutive days at 50 mlperinjection. AdÁAFPÁD55-TRAIL at an MOI 10, washed with PBS and fixed in 70% ethanol Tumor volume (mm3) was measured by a vernier caliper weekly and calculated at 4 1C overnight. Fixed samples were washed again with PBS and incubated at as (lengthÂwidth2)/2. 37 1Cwith20mgmlÀ1 RNaseA. After 30 min, cells were washed with PBS and For the toxicology studies, mice in three groups (n¼4 per group) were stained with 500 ml(50mgmlÀ1) propidium iodide (Sigma) for 1 h. Cellular treated by the same method as mentioned above. All mice were killed for serum DNA distribution was analyzed with a FACSCalibur flow cytometer (BD AST, ALTand ALP measurements at 3 days after treatment. The serum levels of Biosciences, Franklin Lakes, NJ, USA). The cell cycle profile was analyzed using AST, ALT and ALP were determined by use of a Hitachi Automatic Analyzer the FlowJo version 5.7.2 software (Tree Star Inc., Ashland, OR, USA). 7080 (Tokoyo, Japan).

Indirect immunofluorescence assay Serology assays HuH-7 cells were placed overnight in six-well plates on poly-L-lysine-coated To show enhanced safety of AdÁAFPÁD55 over ZD55 in vivo,ZD55, 9 glass coverslips. At 48 h post-infection with AdÁAFPÁD55 or AdÁAFPÁD55- AdÁAFPÁD55 (2Â10 PFUs per mouse) or PBS was intravenously injected into TRAIL at an MOI 10 or mock treatment, cells were washed twice with PBS, nude mice (n¼3 per group) by a single injection of 250 ml. At 24 h post- fixed with freshly prepared 4% formaldehyde in PBS and incubated with injection, the serum levels of ALT and AST were determined by automated 0.05% Triton X-100 for 5 min at room temperature. For the in vivo study, colorimetric assays with Hitachi Automatic Analyzer 7080. tumors were fixed with 4% formaldehyde, embedded in Jung tissue freezing medium (Leica Microsystems) and 10-mm sections were cut. TRAIL (BioVi- Adenoviral replication kinetics in vivo as determined by TaqMan sion, Palo Alto, CA, USA) or cleaved caspase-3 (Beyotime) antibody was real-time PCR applied to the cells or tissue sections and detected with a tetramethyl To study viral replication kinetics in vivo, tumors injected with AdÁAFPÁD55 or rhodamine isothiocyanate-conjugated goat anti-rabbit IgG (H+L) (Jackson AdÁAFPÁD55-TRAIL were collected at 5, 10, 15 and 20 days after the first ImmunoResearch Laboratories, West Grove, PA, USA) or a tetramethyl injection and quickly frozen in liquid nitrogen. Total tissue DNA was extracted

Gene Therapy Cancer targeting Gene-Viro-Therapy of hepatoma XCaoet al 776

using the Genomic DNA MiniPrep Kit for Cells/Tissue (Generay). Viral DNA 9 Kim J, Lee B, Kim JS, Yun CO, Kim JH, Lee YJ et al. Antitumoral effects of recombinant content was measured by TaqMan real-time PCR as described previously.47 The adenovirus YKL-1001, conditionally replicating in alpha-fetoprotein-producing human primers and probe used for TaqMan real-time PCR are listed below: liver cancer cells. Cancer Lett 2002; 180: 23–32. 10 Takahashi M, Sato T, Sagawa T, Lu Y, Sato Y, Iyama S et al. E1B-55K-deleted E1-adenovirus forward, 5¢-GGGTGAGGAGTTTGTGTTAGATTATG-3¢; adenovirus expressing E1A-13S by AFP-enhancer/promoter is capable of highly specific E1-adenovirus reverse, 5¢-TCCTCCGGTGATAATGACAAGA-3¢; E1-adenovirus replication in AFP-producing hepatocellular carcinoma and eradication of established probe, TET-5¢-AGCACCCCGGGCACGGTTG-3¢-TAMRA (detecting Ad5 gene tumor. Mol Ther 2002; 5:627–634. E1A). The plasmid pXC1 (Microbix Biosystems Inc.) containing the E1 region 11 Abou El Hassan MA, van der Meulen-Muileman I, Abbas S, Kruyt FA. Conditionally replicating adenoviruses kill tumor cells via a basic apoptotic machinery-independent of adenovirus was applied to generate the standard curve. mechanism that resembles necrosis-like programmed cell death. JVirol2004; 78: 12243–12251. 12 Ito H, Aoki H, Kuhnel F, Kondo Y, Kubicka S, Wirth T et al. Autophagic cell death of Immunohistochemical analysis malignant glioma cells induced by a conditionally replicating adenovirus. J Natl Cancer Tumor tissues were fixed in 4% formaldehyde, dehydrated with an ethanol Inst 2006; 98: 625–636. gradient and embedded in paraffin. Tissue sections were dewaxed and 13 Jiang H, Gomez-Manzano C, Aoki H, Alonso MM, Kondo S, McCormick F et al. rehydrated according to a standard protocol. The sections were washed with Examination of the therapeutic potential of Delta-24-RGD in brain tumor stem cells: role of autophagic cell death. JNatlCancerInst2007; 99: 1410–1414. PBS, treated with 3% H2O2 and blocked with blocking solution. These steps 14 Ulasov IV, Tyler MA, Zhu ZB, Han Y, He TC, Lesniak MS. Oncolytic adenoviral vectors were followed by overnight incubation with primary antibody at the proper which employ the survivin promoter induce glioma oncolysis via a process of beclin- dilution. The following procedures were performed with the rabbit or goat SP dependent autophagy. Int J Oncol 2009; 34: 729–742. (horseradish peroxidase) Staining System (Westang Biotechnology, Shanghai, 15 Yang M, Cao X, Yu MC, Gu JF, Shen ZH, Ding M et al. Potent antitumor efficacy of ST13 for colorectal cancer mediated by oncolytic adenovirus via mitochondrial apoptotic cell China) according to the manufacturer’s instructions. Hematoxylin was used as death. Hum Gene Ther 2008; 19:343–353. a counterstain. All sections were examined with bright field microscopy. 16 Zhang KJ, Wang YG, Cao X, Zhong SY, Wei RC, Wu YM et al. Potent antitumor effect of interleukin-24 gene in the survivin promoter and retinoblastoma double-regulated oncolytic adenovirus. Hum Gene Ther 2009; 20: 818–830. TUNEL assay 17 Wiley SR, Schooley K, Smolak PJ, Din WS, Huang CP, Nicholl JK et al. Identification Apoptotic cells in tumor tissue sections were assessed using an in situ apoptosis and characterization of a new member of the TNF family that induces apoptosis. detection kit (KeyGen) according to the manufacturer’s instructions. Sections Immunity 1995; 3:673–682. 18 Johnstone RW, Frew AJ, Smyth MJ. The TRAIL apoptotic pathway in cancer onset, were counterstained with hematoxylin. progression and therapy. Nat Rev Cancer 2008; 8:782–798. 19 Qiu S, Ruan H, Pei Z, Hu B, Lan P, Wang J et al. Combination of Targeting Gene- ViroTherapy with 5-FU enhances antitumor efficacy in malignant colorectal carcinoma. Statistical analysis JInterferonCytokineRes2004; 24:219–230. Statistical significance was assessed with the Student’s t-test and one-way 20 Kagawa S, He C, Gu J, Koch P, Rha SJ, Roth JA et al. Antitumor activity and bystander analysis of variance by GraphPad Prism 5.0 (GraphPad Software Inc., La Jolla, effects of the tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) gene. CA, USA). Results were considered statistically significant when Po0.05. Cancer Res 2001; 61: 3330–3338. 21 Lee J, Hampl M, Albert P, Fine HA. Antitumor activity and prolonged expression from a TRAIL-expressing adenoviral vector. Neoplasia 2002; 4: 312–323. 22 Seol JY, Park KH, Hwang CI, Park WY, Yoo CG, Kim YW et al. Adenovirus-TRAIL can CONFLICT OF INTEREST overcome TRAIL resistance and induce a bystander effect. Cancer Gene Ther 2003; The authors declare no conflict of interest. 10: 540–548. 23 Tamaoki T. Human alpha-fetoprotein transcriptional regulatory sequences. Application to gene therapy. Adv Exp Med Biol 2000; 465:47–56. ACKNOWLEDGEMENTS 24 Ohguchi S, Nakatsukasa H, Higashi T, Ashida K, Nouso K, Ishizaki M et al. Expression We are grateful to Professor Mu-Jun Zhao for providing the EGFP-LC3 of alpha-fetoprotein and albumin genes in human hepatocellular carcinomas: limita- plasmid, Professor Hong-Bin Ji for providing the BEAS-2B cell line, Professor tions in the application of the genes for targeting human hepatocellular carcinoma in gene therapy. Hepatology 1998; 27: 599–607. You-Cheng Xu for critical reading the manuscript, Dr Han Di for help with the 25 Wasylyk B, Wasylyk C, Chambon P. Short and long range activation by the SV40 molecular cloning and Ms Lan-Ying Sun for help with the cell culture. This enhancer. Nucleic Acids Res 1984; 12: 5589–5608. study was supported by grants from the National Nature Science Foundation of 26 Ondek B, Shepard A, Herr W. Discrete elements within the SV40 enhancer China (No. 30623003), the Science and Technology Commission of Shanghai region display different cell-specific enhancer activities. EMBO J 1987; 6: 1017–1025. Municipality (No. 06DZ22032), the National Basic Research Program of China 27 Baehrecke EH. How death shapes life during development. Nat Rev Mol Cell Biol (973 Program) (No. 2004 CB51804), the Hi-Tech Research Development 2002; 3: 779–787. Program of China (863 Program) (No. 2007AA 021006), the Key Project of the 28 Maiuri MC, Zalckvar E, Kimchi A, Kroemer G. Self-eating and self-killing: crosstalk Chinese Academy of Science (No. KSCX2-YW-R -09, R-04) and the Zhejiang between autophagy and apoptosis. Nat Rev Mol Cell Biol 2007; 8: 741–752. 29 Berry DL, Baehrecke EH. Autophagy functions in programmed cell death. Autophagy Sci-Tech University Grant 0616033. 2008; 4: 359–360. 30 Debnath J, Baehrecke EH, Kroemer G. Does autophagy contribute to cell death? Autophagy 2005; 1:66–74. 31 Jiang H, White EJ, Gomez-Manzano C, Fueyo J. Adenovirus’s last trick: you say lysis, we say autophagy. Autophagy 2008; 4: 118–120. 1 Parkin DM, Bray F, Ferlay J, Pisani P. Estimating the world cancer burden: Globocan 32 Mills KR, Reginato M, Debnath J, Queenan B, Brugge JS. Tumor necrosis 2000. Int J Cancer 2001; 94: 153–156. factor-related apoptosis-inducing ligand (TRAIL) is required for induction of 2 Liu XY, Gu JF. Targeting gene-virotherapy of cancer. Cell Res 2006; 16:25–30. autophagy during lumen formation in vitro. Proc Natl Acad Sci USA 2004; 101: 3 Zhang ZL, Zou WG, Luo CX, Li BH, Wang JH, Sun LY et al. An armed oncolytic 3438–3443. adenovirus system, ZD55-gene, demonstrating potent antitumoral efficacy. Cell Res 33 Han J, Hou W, Goldstein LA, Lu C, Stolz DB, Yin XM et al. Involvement of protective 2003; 13: 481–489. autophagy in TRAIL resistance of apoptosis-defective tumor cells. JBiolChem2008; 4 Bischoff JR, Kirn DH, Williams A, Heise C, Horn S, Muna M et al. An adenovirus 283: 19665–19677. that replicates selectively in p53-deficient human tumor cells. Science 1996; 274: 34 Herrero-Martin G, Hoyer-Hansen M, Garcia-Garcia C, Fumarola C, Farkas T, 373–376. Lopez-Rivas A et al. TAK1 activates AMPK-dependent cytoprotective autophagy in 5 O’Shea CC, Johnson L, Bagus B, Choi S, Nicholas C, Shen A et al. Late viral RNA TRAIL-treated epithelial cells. EMBO J 2009; 28:677–685. export, rather than p53 inactivation, determines ONYX-015 tumor selectivity. Cancer 35 Park KJ, Lee SH, Kim TI, Lee HW, Lee CH, Kim EH et al. A human scFv antibody Cell 2004; 6: 611–623. against TRAIL receptor 2 induces autophagic cell death in both TRAIL-sensitive and 6 Wei N, Fan JK, Gu JF, He LF, Tang WH, Cao X et al. A double-regulated oncolytic TRAIL-resistant cancer cells. Cancer Res 2007; 67: 7327–7334. adenovirus with improved safety for adenocarcinoma therapy. Biochem Biophys Res 36 Ulasov IV, Sonabend AM, Nandi S, Khramtsov A, Han Y, Lesniak MS. Combination of Commun 2009; 388: 234–239. adenoviral virotherapy and temozolomide chemotherapy eradicates malignant 7 Camper SA, Tilghman SM. The activation and silencing of gene transcription in the glioma through autophagic and apoptotic cell death in vivo. Br J Cancer 2009; 100: liver. Biotechnology 1991; 16:81–87. 1154–1164. 8 Nakabayashi H, Hashimoto T, Miyao Y, Tjong KK, Chan J, Tamaoki T. A position- 37 Hu B, Zhu H, Qiu S, Su Y, Ling W, Xiao W et al. Enhanced TRAIL sensitivity by E1A dependent silencer plays a major role in repressing alpha-fetoprotein expression in expression in human cancer and normal cell lines: inhibition by adenovirus E1B19K human hepatoma. Mol Cell Biol 1991; 11: 5885–5893. and E3 proteins. Biochem Biophys Res Commun 2004; 325: 1153–1162.

Gene Therapy Cancer targeting Gene-Viro-Therapy of hepatoma XCaoet al 777

38 Shao R, Lee DF, Wen Y, Ding Y, Xia W, Ping B et al. E1A sensitizes cancer cells to 43 Kusaba M, Nakao K, Goto T, Nishimura D, Kawashimo H, Shibata H et al. Abrogation of TRAIL-induced apoptosis through enhancement of caspase activation. Mol Cancer Res constitutive STAT3 activity sensitizes human hepatoma cells to TRAIL-mediated 2005; 3:219–226. apoptosis. JHepatol2007; 47: 546–555. 39 Pei Z, Chu L, Zou W, Zhang Z, Qiu S, Qi R et al. An oncolytic adenoviral vector of Smac 44 Zou W, Luo C, Zhang Z, Liu J, Gu J, Pei Z et al. A novel oncolytic adenovirus increases antitumor activity of TRAIL against HCC in human cells and in mice. targeting to telomerase activity in tumor cells with potent. Oncogene 2004; 23: Hepatology 2004; 39: 1371–1381. 457–464. 40ZhangY,GuJ,ZhaoL,HeL,QianW,WangJet al. Complete elimination of colorectal 45 Mosmann T. Rapid colorimetric assay for cellular growth and survival: tumor xenograft by combined manganese superoxide dismutase with tumor necrosis application to proliferation and cytotoxicity assays. J Immunol Methods 1983; 65: factor-related apoptosis-inducing ligand gene virotherapy. Cancer Res 2006; 66: 55–63. 4291–4298. 46 Herrmann M, Lorenz HM, Voll R, Grunke M, Woith W, Kalden JR. A rapid and simple 41 Zhao L, Dong A, Gu J, Liu Z, Zhang Y, Zhang W et al. The antitumor activity of TRAIL method for the isolation of apoptotic DNA fragments. Nucleic Acids Res 1994; 22: and IL-24 with replicating oncolytic adenovirus in colorectal cancer. Cancer Gene Ther 5506–5507. 2006; 13: 1011–1022. 47 Puntel M, Curtin JF, Zirger JM, Muhammad AK, Xiong W, Liu C et al. Quantification of 42 Niwa Y, Kanda H, Shikauchi Y, Saiura A, Matsubara K, Kitagawa T et al. Methylation high-capacity helper-dependent adenoviral vector genomes in vitro and in vivo,using silencing of SOCS-3 promotes cell growth and migration by enhancing JAK/STAT and quantitative TaqMan real-time polymerase chain reaction. Hum Gene Ther 2006; 17: FAK signalings in human hepatocellular carcinoma. Oncogene 2005; 24: 6406–6417. 531–544.

Supplementary Information accompanies the paper on Gene Therapy website (http://www.nature.com/gt)

Gene Therapy