AAV-encoded expression of TRAIL in experimental human colorectal cancer leads to tumor regression

Gene Therapy (2004) 11, 534–543 & 2004 Nature Publishing Group All rights reserved 0969-7128/04 $25.00 www.nature.com/gt

RESEARCH ARTICLE

AAV-encoded expression of TRAIL in experimental human colorectal cancer leads to tumor regression A Mohr1, G Henderson2, L Dudus3, I Herr4, T Kuerschner5, K-M Debatin1, H Weiher6, KJ Fisher3 and RM Zwacka2 1

University Children’s Hospital, Germany; 2Division of Gene Therapy and Interdisciplinary Clinical Research Center, University of Ulm, Ulm, Germany; 3Department of Pathology and Laboratory Medicine, Tulane University Medical Center, New Orleans, LA, USA; 4 German Cancer Research Center, Molecular Oncology/Pediatrics, Heidelberg, Germany; 5Center for Molecular Biology, Heidelberg, Germany; 6Institute for Diabetes Research, Munich, Germany

Gene transfer vectors based on the adeno-associated virus (AAV) are used for various experimental and clinical therapeutic approaches. In the present study, we demonstrate the utility of rAAV as a tumoricidal agent in human colorectal cancer. We constructed an rAAV vector that expresses tumor necrosis factor (TNF)-related apoptosisinducing ligand (TRAIL/Apo2L) and used it to transduce human colorectal cancer cells. TRAIL belongs to the TNF superfamily of cytokines that are involved in various immune responses and apoptotic processes. It has been shown to induce cell death specifically in cancer cells. Transduction with AAV.TRAIL gave rise to rapid expression of TRAIL,

followed by induction of apoptosis, which could be inhibited by the caspase inhibitor z-VAD.fmk, in several human colon cancer cell lines. The apoptotic mechanism included activation of caspase-3, as well as cytochrome c release from mitochondria. The outgrowth of human colorectal tumors grown in mice was completely blocked by transduction with AAV.TRAIL in vitro, while in vivo transduction significantly inhibited the growth of established tumors. AAV vectors could provide a safe method of gene delivery and offer a novel method of using TRAIL as a therapeutic protein. Gene Therapy (2004) 11, 534–543. doi:10.1038/sj.gt.3302154

Keywords: AAV; TRAIL; colon cancer; apoptosis

Introduction In economically developed countries, colorectal cancer is a major public health problem. In the USA, the annual incidence varied between 55 and 65 cases per 100 000 in the period from 1973 to 1999 (http://seer.cancer.gov/ faststats/html/inc_colorect.html). Current treatment modalities for colon cancer metastases include systemic or loco-regional chemotherapy or hepatic resection if the tumor is localized. However, liver resection is possible in less than 10% of patients and postoperative tumor recurrence is frequent. Hence, there is a clear need for novel approaches to treat such established disease. The use of recombinant, therapeutic proteins or antibodies for different diseases, including cancer, has been a great success in the past two decades. However, in the treatment of malignancies, in particular, a number of problems in terms of efficacy, specificity, delivery and duration will be difficult to overcome using only recombinant proteins. The delivery of genetic expression cassettes encoding therapeutic proteins by means of viral gene transfer vectors is a potential alternative. Despite a number of setbacks, gene therapy approaches to treat serious diseases, including cancer, still hold a lot of Correspondence: RM Zwacka, Division of Gene Therapy and Interdisciplinary Clinical Research Center, University of Ulm, Helmholtzstr. 8/1, 89081 Ulm, Germany Received 25 April 2003; accepted 25 August 2003

promise. Safe and efficient vectors are the prerequisite for future progress of the field. Recombinant adeno-associated viral vectors (rAAV) have been proven to be safe and efficacious in experimental and clinical gene therapy approaches.1,2 AAV is a nonpathogenic human parvovirus that has been studied for more than three decades, and the development of recombinant AAV type 2-based gene therapy vectors started about 15 years ago.3,4 Transient cotransfection of the required helper functions instead of using helper viruses has allowed contamination-free, high titer production of rAAV.5,6 Recombinant AAV has been shown to direct long-term transgene expression without giving rise to host toxicities or serious cellular immune responses to various tissues, including certain tumor types.1,7 Enger et al8 found that it better disseminates through tumors in vivo and consequently transduces cells more efficiently than the larger adenovirus. In the light of long-term expression achieved with rAAV, it is not only suited to deliver acutely acting cellular suicide genes, but could also be used to produce tumoricidal proteins systemically or locally that could act on residual disease after surgery and/or chemotherapy. Tumor necrosis factor (TNF)-related apoptosis-inducing ligand (TRAIL) has been shown to induce apoptosis in a wide range of human cancer cells, but not in normal tissues.9 We and others have shown that human colon cancer cells are sensitive to TRAIL.10,11 TRAIL binds to a type of receptor, of which four homologues have been

AAV-encoded expression of TRAIL A Mohr et al

535

identified so far. While TRAIL-R1 (DR-4) and TRAIL-R2 (DR-5) are able to initiate apoptotic processes upon binding of TRAIL, TRAIL-R3 (DcR1/TRID) and TRAILR4 (DcR4/TRUNDD) lack this ability.12 Therefore, the latter two receptors have been hypothesized to be decoy receptors that are not able to mediate apoptotic signals. However, no correlation between the expression of the two decoy receptors and TRAIL resistance has been found so far.13 The tumor selectivity and safety of rTRAIL in vivo have been examined in various studies.14,15 It was generally found to be well tolerated even when multiple doses were administered to animals15 and, in contrast to injection of recombinant FasL or antiFas mAb, which causes massive hepatocyte apoptosis, hemorrhage and death, no histological or functional abnormalities were detectable in any of the tissues examined following rTRAIL administration. Furthermore, injection of rTRAIL into tumor-bearing mice resulted in complete tumor regression in many cases.15,16 The recently reported toxicity of rTRAIL at high concentrations in human primary hepatocytes17 has been attributed to problems in the preparation of the respective rTRAILs used in these particular studies.18 Therefore, rTRAIL remains a promising antitumor agent. The use of rAAV vector-expressing TRAIL appears to be a specifically safe gene therapy approach, combining the good safety characteristics of AAV and the tumor selectivity of TRAIL. Hence, we set out to develop such a vector and to test it in different in vitro and animal models of human colorectal cancer.

Results AAV-mediated transgene expression in colorectal cancer cells In contrast to adenovirus-based gene therapy vectors that can cause inflammatory responses and adverse side effects in vivo, AAV vectors are nonpathogenic and relatively safe. Initial clinical trials have substantiated the safety of AAV vectors and have even shown some efficacy. In order to test the utility of AAV as a gene therapy vector system for human colon cancer, we have tested several human colorectal cancer cell lines including DLD-1, HRT18, HCT116, SW480 and Lovo cells. They were transduced with varying MOIs (0, 100, 400) of a recombinant AAV vector harboring the full-length cDNA of EGFP under the control of the cytomegalovirus (CMV) promoter/enhancer element (AAV.EGFP).19 As shown in Figure 1a, FACS analyses revealed high EGFP expression in DLD-1, HRT18, HCT116 and SW480 cells. Lovo cells required relatively high doses before EGFP could be detected. Examination of transduced cells lines by fluorescent microscopy confirmed these findings (data not shown). Despite the differences in the efficiency of transgene expression among the cell lines, these results indicate that AAV vectors hold the potential to be used as gene therapy vectors for colorectal cancer by expressing therapeutic transgenes inside the tumor cells. Since it is generally difficult to achieve transduction efficiencies of 100%, we decided to use a transgene that can give rise to bystander effects and possesses an in-built tumor specificity. To this end, a recombinant form of TRAIL (rTRAIL) as well as adenoviral-mediated expression of TRAIL have shown its cell-death-promoting potential

Figure 1 EGFP expression in human colorectal cancer cells from a recombinant AAV.EGFP vector. (a) Several colorectal cancer cell lines were transduced with AAV.EGFP at varying MOIs (0, 100 and 400) expressed as genomes/cell. (b) The pAV.TRAIL.EGFP construct that was used to generate recombinant AAV.TRAIL viruses. It contains the two expression cassettes for TRAIL and EGFP driven by the early CMV promoter/ enhancer element (CMV-P) and the PGK promoter (PGK-P), respectively, flanked by the ITRs. (c) PHH were transduced at an MOI of 400 with AAV.EGFP and AAV.TRAIL giving rise to a transduction rate of not more than around 1%. EGFP was measured 24 h post-transduction by FACS analysis.

Gene Therapy

AAV-encoded expression of TRAIL A Mohr et al

536

and tumor specificity in several recent studies.15 Hence, an rAAV vector-expressing TRAIL from a CMV promoter/enhancer element plus EGFP driven by a PGK promoter (AAV.CMV.TRAIL:PGK.EGFP; short: AAV.TRAIL) was generated and tested in colon cancer cells. The construct is depicted in Figure 1b. Since TRAIL hepatotoxicity has been reported in some cases20 and would be a concern for therapeutic strategies targeting colorectal cancer liver metastases, we analyzed the selectivity of rAAV with respect to colon cancer cells and hepatocytes. We transduced primary human hepatocytes with AAV.EGFP and AA.TRAIL. As shown in Figure 1c no significant EGFP expression could be achieved, indicating that rAAV based on AAV-2 does not effectively transduce human primary hepatocytes in vitro. The inability to transduce hepatocytes while giving rise to substantial transgene expression in colorectal cancer cells could serve as an additional level of specificity in therapeutic approaches targeting liver metastases with rAAV.

AAV-mediated TRAIL expression in colorectal cancer cells Within the AAV.TRAIL vector, TRAIL and EGFP are expressed from two different promoters. EGFP levels were measured by FACS analyses after AAV.TRAIL transduction in DLD-1 and HCT116 cells, and expression was detectable in both cell lines. In line with the AAV.EGFP transduction data, 50% of all cells were EGFP positive at an MOI of 100, and we decided to continue with this rAAV genome/cell ratio. Analysis of transgenic TRAIL expression by Western blot in AAV.TRAILtransduced DLD-1 cells demonstrates TRAIL expression at an MOI of 100 (Figure 2). The detected band of 30 kDa

Figure 2 The recombinant AAV.TRAIL expression vector leads to TRAIL transgene expression in DLD-1 cells. A Western blot analysis shows TRAIL protein expression in AAV.TRAIL transduced DLD-1 cells. Increasing MOIs were used to transduce cells: Lane 1, marker; lane 2, MOI 0; lane 3, MOI 1; lane 4 MOI 10; lane 5, MOI 100. In lane 6, protein extracts from DLD-1 cells transduced with AAV.EGFP (MOI 100) were loaded as negative control. Recombinant TRAIL was loaded as positive control in lane 7 (0.1 ng) and lane 8 (5 ng). The membrane standing form of TRAIL expressed from AAV.TRAIL is indicated by the full arrow at 30 kDa, while the soluble recombinant TRAIL shown by the dashed arrow runs as a 19 kDa protein. Gene Therapy

has the expected size of the membrane form of TRAIL. The rTRAIL that was used as positive control was detectable at levels of 5 ng/lane but not at 0.1 ng/lane, indicating the relatively poor sensitivity of the antiTRAIL antibody and explaining the lack of signal at lower MOIs. The size of the rTRAIL band is the expected 20 kDa since it represents only the ectodomain of TRAIL. AAV-encoded TRAIL expression in colorectal cancer cells is comparable to levels of rTRAIL protein that have been shown to trigger apoptosis in vitro, that is, 5 ng in 1 ml medium. These results demonstrate the feasibility to express TRAIL from an AAV vector to a degree that is capable of inducing cell death.

AAV.TRAIL induces apoptosis in DLD-1 cells We next investigated whether TRAIL expression in DLD1 cells at an MOI of 100 causes apoptosis and could be used as a possible therapeutic antitumor approach. Cell death following AAV.TRAIL treatment of DLD-1 cells was measured by Annexin-V/propidium iodide (PI) staining followed by FACS analysis. The results of the Annexin-V/PI analysis are shown in Figure 3a. They reveal that, 24 h post-transduction with AAV.TRAIL, 27% of all cells are apoptotic, as measured by Annexin-V positivity. This compares to 2% in medium and AAV.EGFP control cells, respectively. Treatment with rTRAIL gave rise to cell death induction in 26% of all cells. These results indicate that AAV.TRAIL is at least as effective as rTRAIL. The AAV.TRAIL-induced apoptotic mechanism included activation of caspase-3, as well as cytochrome c release from mitochondria (data not shown). Even though we had observed low transduction rates of rAAV in human primary hepatocytes, we wanted to know whether this could still lead to apoptosis in liver cells. However, transduction of primary hepatocytes with AAV.TRAIL as well as treatment with rTRAIL (5 ng/ml) lead to no detectable apoptosis (Figure 3b). These findings indicate the general safety of TRAIL with respect to hepatotoxicity. AAV.TRAIL leads to attenuated growth of human colorectal cancer cells Following transduction at an MOI of 100 the growth of DLD-1 was followed over a period of 5 days. As depicted in Figure 3c, the growth of AAV.TRAIL-transduced cells was retarded by 500% 4 days post-transduction as compared to medium-treated and AAV.EGFP (MOI 100)-transduced cells. This effect was comparable to the result achieved with rTRAIL treatment at 5 ng/ml (Figure 3c). The growth of TRAIL-expressing cells picked up on day 4, probably due to the outgrowth of nontransduced cells that did not get into contact with TRAIL-positive cells. An alternative explanation is the development of TRAIL-resistant cells that start to grow again 72 h post-TRAIL treatment. However, the results demonstrate that AAV.TRAIL is capable of significantly reducing the growth of colorectal cancer cells in vitro. Most likely the reduced growth is a result of TRAILinduced apoptosis as shown above. Specificity of AAV.TRAIL-induced apoptosis In order to test whether the AAV.TRAIL-induced effects were mediated by the prototypical caspase-mediated

AAV-encoded expression of TRAIL A Mohr et al

apoptotic pathways, we examined cell death induction in the presence of the caspase inhibitor z-VAD.fmk. Figure 4a shows the percentage of apoptosis measured by Annexin-V/PI staining after treatment of DLD-1 cells with medium, AAV.EGFP (MOI 100), AAV.TRAIL (MOI 100) and rTRAIL (5 ng/ml) with and without zVAD.fmk. The caspase inhibitor is capable of completely

inhibiting apoptosis after AAV.TRAIL transduction and rTRAIL treatment. Application of DMSO, the carrier of z-VAD.fmk, had no effect on apoptosis. These results indicate that AAV.TRAIL induces the usual caspase cascade leading to cell death. Furthermore, we investigated activation of caspase-3 and cytochrome c release (data not shown). We found that AAV.TRAIL, as well as rTRAIL, elicit release of cytochrome c from mitochondria and cleavage of procaspase-3 to its active form. Cytochrome c is required to recruit Apaf-1 and caspase-9, together forming the apoptosome, resulting in activation of effector caspases, such as caspase-3. In order to confirm that the observed effects of AAV.TRAIL are indeed caused by the expression of TRAIL, we treated transduced cells with a neutralizing TRAIL antibody. In addition, we cotreated cells with rTRAIL and the antibody. The upper panel of Figure 4b depicts the DNA histograms of medium-, AAV.EGFP(MOI 100), AAV.TRAIL- (MOI 100) and rTRAIL-treated DLD-1 cells. Cells were harvested 24 h post-treatment according to the protocol described by Nicoletti et al.21 These data confirm the results obtained with AnnexinV/ PI staining. AAV.TRAIL induces cell death, as measured and expressed here by the ‘sub-G1’ peak. After 24 h, 56% of all cells are dead following transduction with AAV.TRAIL as compared to 78% of rTRAIL-treated and 5% in medium- and AAV.EGFP-treated cells. When the cells were cotreated with a TRAIL neutralizing antibody, apoptosis was completely inhibited in rTRAIL-induced cells, while it was reduced by 30% in AAV.TRAILtransduced cells. The different neutralizing efficacy of the antibody might be due to the limited accessibility of epitopes on the membrane form of TRAIL as compared to soluble rTRAIL. Therefore, we used a derivative DLD1 cell line, designated DLD-TRAIL-R, that was generated by continuous treatment with rTRAIL and selection of surviving cells. The resulting cells were transduced with AAV.TRAIL and treated with rTRAIL, AAV.EGFP or medium as controls. The results shown in the lower panels demonstrate that these cells do not undergo apoptosis following AAV.TRAIL transduction or treatment with rTRAIL. Taken together, these results prove that the observed cell death in AAV.TRAIL-transduced DLD-1 cells is a specific effect of the expressed TRAIL.

537

3 Figure 3 AAV.TRAIL treatment leads to apoptosis and reduced growth of DLD-1 cells. (a) Apoptosis was measured by staining with Annexin-V and PI 24 h post-transduction with AAV.EGFP and AAV.TRAIL, respectively, at an MOI of 100 or treatment with rTRAIL at a concentration of 5 ng/ml. The diagram is the summary of three independent experiments performed in triplicate (mean values7s.e.). Annexin-V positivity is expressed as percentage apoptosis and shows that AAV.TRAIL and rTRAIL give rise to more than 25% apoptotic cells 24 h post-treatment compared to 2% for medium- and AAV.EGFP-treated cells. (b) PHH were transduced with AAV.EGFP and AAV.TRAIL at an MOI of 400, and apoptosis was measured by the Nicoletti staining. However, no TRAIL-induced apoptosis could be detected 24 h post-transduction. Treatment with rTRAIL (5 ng/ ml) also had no effect on the hepatocytes. (c) The growth of DLD-1 cells transduced (MOI 100) with AAV.TRAIL (m) was followed over 5 days and compared to AAV.EGFP (’)-, medium (E)- and rTRAIL (
)treated cells. The growth of TRAIL-expressing cells was static over 3 days before a small number of cells started to grow. Cells that were treated with rTRAIL (5 ng/ml) behaved similar to AAV.TRAIL-transduced cells. Numbers are the mean value of three independent experiments7s.e. *Po0.01 and **Po0.01. Gene Therapy

AAV-encoded expression of TRAIL A Mohr et al

538

Figure 4 AAV.TRAIL induces caspase-dependent apoptosis. (a) Cotreatment with the caspase inhibitor z-VAD.fmk (50 mM) inhibits AAV.TRAIL-induced apoptosis. DLD-1 cells were treated with medium, AAV.EGFP (100 MOI), AAV.TRAIL (100 MOI) and rTRAIL (5 ng/ml) without and with 50 mM zVAD.fmk for 24 h. As control, a set of cells was treated with 2.5 ml DMSO, which had no significant effect on apoptosis. (b) Neutralizing anti-TRAIL antibodies can inhibit AAV.TRAIL-induced apoptosis and a TRAIL-resistant DLD-1 cell population is AAV.TRAIL insensitive. The upper panel shows a Nicoletti FACS analysis of DLD-1 cells following 24 h treatment with medium, AAV.EGFP (MOI 100), AAV.TRAIL (MOI 100) and rTRAIL (5 ng/ml). The depicted marker comprises the ‘sub G1’ cell fraction and is expressed as [%] dead cells. Treatment with AAV.TRAIL and rTRAIL gives rise to 56 and 78% apoptosis, respectively. As shown in the central panel, the effect of rTRAIL can be completely inhibited by cotreatment with 0.5 mg/ml of a neutralizing anti-TRAIL antibody and the apoptosis induction by AAV.TRAIL can be substantially reduced. The lower panel depicts the DNA histograms of TRAILresistant DLD-1 (DLD-TRAIL-R) cells that were generated by four rounds of successive rTRAIL treatment of the surviving cells. Neither transduction with AAV.TRAIL (MOI 100) nor treatment with rTRAIL (5 ng/ml) gives rise to increased apoptosis in these cells as compared to medium- or AAV.EGFP (MOI 100)-treated cells.

Sustained cell death-inducing activity of AAV-encoded TRAIL In order to analyze possible advantages of AAV.TRAIL over rTRAIL we investigated the mode of action of AAVencoded TRAIL in more detail. Even though the use of TRAIL, as a tumor-specific inducer of apoptosis, provides a good degree of safety, we wanted to know whether the TRAIL produced in transduced tumor cells would remain in the plasma membrane or would be processed and shed into the extracellular space. First, we transduced DLD-1 cells with AAV.TRAIL (MOI 100) and harvested the medium after 72 h. After carefully removing DLD-1 cells and cellular debris by centrifugation, we applied the medium to a fresh batch of DLD-1 cells. No apoptosis could be detected visually or by Annexin-V/PI staining up to 72 h post-treatment (data not shown), Gene Therapy

indicating that AAV.TRAIL-transduced DLD-1 cells did not shed TRAIL. This finding is in line with the results obtained by Kagawa et al,22 who showed a bystander effect of adenoviral vector-encoded TRAIL in DLD-1 cells due to membrane expression of TRAIL. However, they could also not find soluble TRAIL in the medium of TRAIL-expressing cell cultures and the cell killing effect was not transferable with the medium of TRAILexpressing cell cultures. In addition, DLD-1 cells were transduced with AAV.TRAIL (MOI 100) and cotreated with Brefeldin-A (BFA). BFA inhibits the antereograde transport of proteins through the Golgi complex, and thus prevents expression of membrane proteins on the cell surface. Cotreatment with BFA significantly inhibited AAV.TRAIL-induced cell death as measured by AnnexinV binding, while it was ineffective against the effects of

AAV-encoded expression of TRAIL A Mohr et al

rTRAIL (data not shown). Griffith et al23 have demonstrated similar effects using BFA in cells transduced with a recombinant adenovirus expressing full-length TRAIL. These results demonstrate that AAV-encoded TRAIL is retained in the plasma membrane. This mode of expression will localize the effects of AAV.TRAIL to the site of transduction potentially providing an additional level of safety. In order to analyze possible advantages of sustained expression of TRAIL as mediated by AAV.TRAIL transduction in comparison to treatment with rTRAIL, we analyzed cell death kinetics over 72 h. We measured Annexin-V binding in combination with PI staining at different time points after transduction. Figure 5a depicts the degree of Annexin-V staining at 24, 48 and 72 h posttransduction. Annexin-V binds to phosphatidylserine that only appears on the outside of the plasma membrane during apoptosis. The percentage of Annexin-V-positive cells is a good measure for early apoptotic processes. In AAV.TRAIL-treated cells, 28% bind Annexin-V as compared to 3% in AAV.EGFP- and mediumtreated cells at 24 h. Subsequently, at 48 and 72 h the percentage of Annexin-V-binding cells drops to below 20% in AAV.TRAIL-transduced cells, while it does not change significantly in the control cells. This could be due to dynamic processes that see apoptotic cells developing necrotic features over time and becoming accessible to PI and leading to double positivity for Annexin-V and PI as shown in Figure 5b. While the cells treated with rTRAIL show a rapid burst of apoptosis during the first 24 h of treatment, which translates into the same percentage of Annexin-V/PI double-positive cells at the later time points, AAV.TRAIL appears to cause a more gradual but continuous wave of apoptosis followed by necrosis. This kind of action profile could be advantageous over the treatment with bolus injections of rTRAIL. Therefore, we believed that AAV.TRAIL warranted further testing in animals that carry engrafted human colon cancers.

AAV.TRAIL reduces the growth of human colorectal tumors in mice In vitro transduction for 6 h of 5
105 DLD-1 cells with AAV.TRAIL (MOI 100) completely blocked the outgrowth of tumors when these cells were subsequently injected into BALB/c nu/nu mice. The photograph in Figure 6a shows mice 4 weeks after tumor cell injections. In contrast to AAV.TRAIL-transduced cells, AAV.EGFPtreated DLD-1 cells develop tumors. The growth kinetics of these tumors are depicted in Figure 6b. We then wanted to know whether AAV.TRAIL also has antitumor activity when applied in vivo. Therefore, we injected 2x106 DLD-1 cells into BALB/c nu/nu mice. When the resulting tumors had reached an average volume of 500 mm3, usually after 1 week, 2
1010 AAV.TRAIL genomes in 200 ml HEPES-buffered saline (HBS) were injected into the tumors. In parallel, tumors were injected with AAV.EGFP (2
1010 genomes) and HBS, respectively. Figure 6c shows EGFP expression in frozen sections of a tumor 3 weeks after AAV.EGFP in vivo transduction. Approximately 10% of the tumor was EGFP-positive at this time point. Figure 6d shows the growth of the tumors over a period of 3 weeks following rAAV injections. During this period, AAV.EGFP- and

539

Figure 5 AAV.TRAIL treatment leads to sustained induction of apoptosis. (a) Annexin-V positivity: DLD-1 cells were treated with medium, AAV.EGFP (MOI 100), AAV.TRAIL (MOI 100) or rTRAIL 5 ng/ml. The percentages of Annexin-V-positive cells are depicted at 24 h (&), 48 h ( ), and 72 h (’) post-treatment. (b) Annexin-V/PI double positivity: The percentage of double-positive cells at 24 h (&), 48 h ( ), and 72 h (’), post-treatment are shown. AAV. TRAIL gives rise to a gradual increase in double-positive cells indicative of apoptotic cells turning necrotic. Recombinant TRAIL in contrast leads to a burst of apoptosis/ necrosis that declines over time. The more long-term activity of AAV.TRAIL could be advantageous in a therapeutic context.

HBS-treated tumors grew to double the size as compared to AAV.TRAIL-injected tumors. Hence, AAV.TRAIL has tumoricidal activity in vivo. Its effects could potentially be further enhanced by multiple injections and combination treatment with 5-fluorouracil (5-FU). Owing to the advantageous properties of AAV as a gene transfer vector and the specific antitumor effect of TRAIL, gene therapy treatment of colorectal cancer using AAV.TRAIL is a putative safe and efficacious approach.

Discussion In the present study, we have explored a therapeutic approach using an AAV vector-expressing TRAIL to treat human colorectal tumors. The AAV-encoded TRAIL is expressed within transduced colon tumor cells and is capable of inducing apoptosis and growth retardation in vitro. In vivo delivery of AAV.TRAIL to human colorectal tumors in mice leads to a marked suppression of their growth. This is the first study that demonstrates the potential therapeutic utility of rAAV in colorectal tumors. The results are comparable to the effects reported with rTRAIL. The potential advantages of such Gene Therapy

AAV-encoded expression of TRAIL A Mohr et al

540

gene therapy approaches are the sustained expression and antitumor activity of TRAIL, as well as the expression in its natural context as a membrane protein. To this end, we have shown that AAV.TRAIL causes a continuous wave of cell death as compared to rTRAIL that only leads to a short-term burst of apoptosis. The former mechanism might be of therapeutic benefit, since it could potentially inhibit the generation of TRAILresistant cells due to the persistent presence of AAV-

encoded TRAIL. While the presented vector system has utility for at least orthotopic delivery, further improvements to the targeting mechanism of the vector and/or the mode of TRAIL expression have to be made to be able to use it for disseminated tumor disease. We found that AAV-encoded TRAIL is expressed as a membrane protein, and we could not detect significant processing and release of a soluble form of TRAIL. Cell death was inflicted on adjacent cells by TRAIL-expressing cells, thereby restricting the killing zone to the immediate surroundings of AAV.TRAIL-transduced cells and avoiding systemic side effects. In contrast to rTRAIL, AAVencoded TRAIL cannot be eliminated or diluted out by binding to decoy or scavenging receptors. However, the membrane-bound form of TRAIL is locally restricted in its tumor-killing activity. The expression of a secreted and soluble form of TRAIL from an rAAV vector might be a possibility to engineer the advantages of systemic administration of rTRAIL into a gene therapy system. Beyond the good safety profile of rAAV vectors, demonstrated by the lack of toxicity on primary human hepatocytes, the use of TRAIL as the therapeutic transgene provides an additional layer of specificity and thereby safety to the chosen experimental gene therapy system. Several studies have demonstrated the tumor-specific activity and safety of TRAIL.15 In accordance with these reports, we also did not detect any side effects such as weight loss or apathy in AAV.TRAILtreated animals. Somewhat surprisingly, we found high transduction rates with rAAV vectors in most of the colorectal cancer cells investigated. This observation is in contrast to studies in ovarian cancer cells that could not find significant transgene expression in three different cell lines.24 The reasons for this discrepancy can be manifold and include differences in viral entry mechanisms and/ or factors (e.g. FKBP5225) that are involved in second (leading) strand synthesis of the AAV genome, which is believed to be the transduction rate determining factor.26 While the present AAV.TRAIL vector has been shown to be a safe and efficacious experimental gene therapy 3 Figure 6 AAV.TRAIL gives rise to regression of human colorectal tumors in nude mice. (a) 5
105 DLD-1 cells were transduced in vitro with HBS, AAV.EGFP (MOI 100) or AAV.TRAIL (MOI 100) for 6 h and then transplanted into BALB/c nu/nu mice. The photograph shows mice 4 weeks after tumor cell injections. AAV.TRAIL transduction completely blocks tumor growth. (b) The increases in tumor volume of AAV.EGFP (n ¼ 7) and HBS (n ¼ 4) as compared to AAV.TRAIL (n ¼ 3) in vitro treated tumors are shown. The fold increase of the tumors after the initial measurement after 1 week is depicted in the graph. The AAV.EGFP and HBS tumor sizes increased by a factor of 6.3 and 7.2, respectively, while the AAV.TRAIL tumors did not grow at all. (c) The left panel shows a section of a tumor that was injected with HBS and removed after 4 weeks. The right panel depicts a section of an AAV.EGFP-injected tumor. EGFP signals could be detected in 10% of the tissue, indicating successful in vivo transduction with rAAV (
40 objective). (d) DLD-1 (2
106) cells were injected into BALB/c nu/nu mice. After 1 week when tumors had reached a volume of 500 mm3, 2
1010 AAV.TRAIL (E) genomes in 200 ml HBS were slowly injected into the tumor mass. Tumors that were injected with 2
1010 genomes AAV.EGFP (’) and HBS (m), respectively, were used as controls. The graph depicts the growth of the tumors over a period of 4 weeks. AAV.TRAIL-treated tumors grew at a significantly reduced rate. The number of animals per group was as follows: AAV.TRAIL (n ¼ 4), AAV.EGFP (n ¼ 8) and HBS (n ¼ 7). Numbers represent mean values7s.e. *Po0.05 and **Po0.05.

Gene Therapy

AAV-encoded expression of TRAIL A Mohr et al

system, further improvements, in light of its potential use against hepatic metastasis of colorectal cancer, can be made. Vectors based on AAV type 5 instead of type 227 or rhesus monkey AAV-828 have recently been found to facilitate substantially higher transduction rates in liver. Such vectors might be better suited for transduction of liver and tumor tissue. Owing to its lack of systemic toxicity, TRAIL expressed on hepatocytes would only act on metastatic cells growing in the vicinity. Combination therapies of AAV.TRAIL and 5-FU might further increase the efficacy of the proposed approach. 5FU is used in chemotherapeutic regimens for hepatic metastases, and synergistic effects of rTRAIL and 5-FU have been demonstrated.29,30 In conclusion, this study is the first to show that longterm expression of TRAIL in colorectal cancers can be achieved by rAAV-mediated in situ transduction with a single intratumoral injection of the vector. The apoptosisinducing activity of the AAV-encoded TRAIL is restricted to the neighboring tumor cells (data not shown), thereby achieving a restricted bystander effect without losing the advantage of gene therapy approaches to direct transgene expression to specific tissues in order to avoid side effects.

Material and methods Cloning of TRAIL cDNA The full-length cDNA was cloned by RT-PCR from Jurkat cell RNA, using primers that were described in Wiley et al:31 50 , GCACGTCGACCAGGATCATGGCTATGATGG; CGTGAGCGGCCGCCAGGTCAGTTAGCCAACT. 30 , The oligos carried additional restriction enzyme sites (50 BamHI; 30 EcoRI) in order to facilitate cloning into pcDNA3 (Invitrogen, Carlsbad, CA, USA). The resulting vector was checked by sequence analysis and functionally tested by Western blot in pcDNA3.TRAIL-transfected 293 cells. Transfections were carried out by a standard Ca3(PO4)2 method. Generation of recombinant AAV.TRAIL vectors Recombinant AAV.TRAIL was generated by cloning the full-length cDNA of TRAIL into the AAV-backbone plasmid pAV.55K(2)-GFP.19 The EcoRI/BamHI TRAIL fragment from pcDNA3.TRAIL.EGFP was ligated into the EcoRI/BglII digested backbone plasmid. The resulting plasmid was called pAV.TRAIL.EGFP. All plasmid manipulations were performed in SURE E. coli cells (Stratagene, La Jolla, CA, USA) to ensure the integrity of the inverted terminal repeats (ITR). The ITR integrity was analyzed by SmaI digests of the resulting clones. One of the clones was expanded and DNA prepared (Qiagen, Hilden, Germany). The AAV.TRAIL vectors were generated according to a protocol described in Liang et al.32 Briefly, 100 plates of subconfluent (70–80%) 293 cells were cotransfected with pAV.TRAIL and a helper plasmid (pAd.Help.Rep/Cap) that provides the necessary adenoviral (E2a, E4 and VA RNA) and AAV functions (Rep and Cap).32 A total of 12.5 mg of pAV.TRAIL.EGFP and 37.5 mg of pAd.Help.Rep/Cap were Ca3(PO4)2-transfected adding a total of 3 ml of the transfection mix onto each 15 cm plate of 293 cells grown in 20 ml of normal growth medium. The following day the medium was changed to 2% fetal calf serum (FCS)-

containing medium. The cells were harvested 72 h posttransfection by scraping cells off into the medium, which was collected in 500 ml centrifugation bottles. The cells were spun down for 30 min at 1500 g in a refrigerated centrifuge. The supernatant was discarded and the cells were resuspended in 50 ml 10 mM Tris (pH 8.0) and divided into two 50 ml conical tubes. The cell suspension was sonicated on ice for 2
1 min and 1
45 s at 15%, each interrupted by periods of at least 1 min on ice using a Fisher Sonic Dismembrator (Fisher, Pittsburgh, PA, USA). Subsequently, RNase-A and DNase-I were added to a final concentration of 0.2 and 0.1 mg/ml, respectively, followed by a 30 min incubation at 371C, during which the tubes were inverted every 5–10 min. After 30 min, Na-deoxycholate from a filtered 10% stock solution was added to a final concentration of 0.5%, and incubated for a further 10 min at 371C, followed by a 10 min incubation on ice. Finally, 0.454 g/ml of CsCl was added, adjusting the CsCl concentration to 1.3 g/ml. The sample was then loaded onto a step gradient in SW28 tubes consisting of 9 ml of 1.6 g/ml CsCl, 9 ml of 1.4 g/ ml CsCl and 19.5 ml sample at 1.3 g/ml. The gradients were centrifuged for 24 h with 25 000 r.p.m. at 41C. Subsequently, fractions of 1 ml were collected using a Beckman fraction collector (BeckmanCoulter, Fullerton, CA, USA). Refractory indexes were determined of all fractions, and 12 fractions bracketing a density of 1.37 g/ ml were tested on 293 cells grown in six- well plates. To this end, 2 ml of each fraction plus wild-type adenovirus to enhance EGFP transgene expression were applied to 293 cells. After 24 h, EGFP expression, indicating the presence of AAV, was visualized under an inverted fluorescent microscope. The positive fractions were pooled adjusted to 1.3 g/ml and applied to a second SW28 CsCl gradient. Subsequently, two SW41 gradients, consisting of 4 ml of 1.6 g/ml, 4 ml of 1.4 g/ml and 4 ml sample at 1.3 g/ml, were run with 35,000 rpm at 41C for 24 h. After each centrifugation, fractions were assayed for AAV on 293 cells and the EGFP-positive fractions were pooled and adjusted to 1.3 g/ml with 10 mM Tris (pH 8.0). Finally, the AAV.TRAIL containing pool was dialyzed against HBS, which was changed three times every 3 hours at 41C. The dialyzed virus was aliquoted and stored at –801C. The rAAV were titered by Southern blot analyses using an EGFP radioactive probe. The number of AAV genomes was determined by comparison against serial dilutions of an EGFP-containing plasmid. MOI is expressed as the number of genomes per cell. The ratio of infectious units per viral was found to be 26 by transducing 293 cells with serial dilutions of AAV.TRAIL in the presence of wild-type adenovirus. Cells were analyzed by FACS for the percentage of EGFP-positive cells 24 h post-transduction. The control vector AAV.EGFP has been described earlier.19

541

Cell lines 293 cells were cultured in DMEM medium supplemented with 10% FCS and 1% penicillin/streptomycin (Biochrom, Berlin, Germany and Invitrogen, Carlsbad, CA, USA). DLD-1, HRT18 and Lovo cells were grown in RPMI 1640 medium (Invitrogen, Carlsbad, CA, USA) containing 10% FCS and 1% penicillin/streptomycin. HCT116 and SW480 human colorectal cancer cells were grown in McCoy‘s 5A and L-15 medium (Invitrogen, Gene Therapy

AAV-encoded expression of TRAIL A Mohr et al

542

Carlsbad, CA, USA), respectively, containing 10% FCS and 1% penicillin/streptomycin. DLD-TRAIL-R cells that are resistant against rTRAIL were generated by four subsequent rounds of treatment of the respective surviving cells with 5 ng/ml rTRAIL (R&D Systems, Minneapolis, MN, USA). The resulting cells were almost completely TRAIL resistant and remained resistant over several weeks.

Primary hepatocytes Primary human hepatocytes (PHH) were isolated by a two-step collagenase perfusion and differential centrifugation method as previously described.33 PHH were seeded in 12-well culture plates (about 2
105 cells/well) coated with collagen Type I (Sigma, St Louis, MS, USA) and maintained for 36 h in maintenance medium with 5% FCS; then FCS was removed. AAV transduction of human colorectal cancer cells AAV transductions (AAV.EGFP and AAV.TRAIL) were performed in medium containing 2% FCS and 1% penicillin/streptomycin for 24 h, after which cells were harvested for transgene analysis or functional tests. MOIs of 0, 100 and 400 genomes/cell were used. TRAIL expression analyses RT-PCR and Western blot analyses were performed to detect transgenic TRAIL expression. RT-PCR. Total DNase-treated RNA was isolated using the RNAeasy Kit (Qiagen, Hilden, Germany) followed by reverse transcription using a first-strand cDNA synthesis kit (Invitrogen, Carlsbad, CA, USA). Both RNA extraction and RT reaction were carried out according to the manufacturer’s instructions. Subsequently, a PCR was performed using TRAIL-specific primers amplifying a band of 300 bp. Western blot analysis. Whole-cell extracts (100 mg protein) were separated on a denaturing 12.5% SDSPAGE and transferred to a Nitrocellulose membrane (Amersham Pharmacia, Little Chalfont, UK) by electroblotting for 1 h at 100 V. The membrane was blocked in 4% nonfat dry milk solution in PBS, supplemented with 0.3% Tween 20. This solution was used for all other antibody incubations and washing steps. The primary polyclonal antibody against TRAIL (Peprotech, Rocky Hill, NJ, USA) was diluted 1:200 and incubated overnight at 41C, after which the membrane was washed four times. A 1:2000 diluted goat anti-rabbit antibody coupled to horseradish peroxidase (AmershamPharmacia, Little Chalfont, UK) was used as secondary antibody and incubated for 1 h at room temperature. A sheep antihuman-CuZnSOD antibody (The Binding Site, Birmingham, UK), used as a loading control, was diluted 1:2000. A donkey anti-sheep secondary antibody conjugated to horseradish peroxidase (Jackson ImmunoResearch, West Grove, PA, USA) was used to detect the CuZnSOD signal. Following incubations with the secondary antibody, the membrane was washed four times in blocking buffer and finally twice in PBS. Proteins were visualized, using an enhanced chemiluminescence mix (ECL, Amersham, Little Chalfont, UK) on X-ray film (Kodak, Rochester, NY, USA). Gene Therapy

Growth curves Cells were seeded at a concentration of 2.5
104 cells per well in a 24-well plate. After 24 h, cells were first counted and then transduced with AAV.TRAIL in RPMI medium containing 2% FCS and 1% penicillin/streptomycin. AAV.EGFP-, medium- and rTRAIL-treated cells were used as controls. Cells were counted every 24 h over a period of 5 days. The results are the mean results of three independent experiments. Cell death and apoptosis measurements Apoptosis was measured by two different methods. Annexin-V/PI staining: DLD-1 cells were seeded at 2.5
104 cells per well in a 24-well plate. After 24 h, cells were transduced with AAV.TRAIL at an MOI of 100. Controls were AAV.EGFP (MOI 100)-transduced and medium (HBS)- or rTRAIL (5 ng/ml)-treated cells. The following day, (24 h post-transduction), cells were harvested by trypsinization and spun down at 300 g for 7 min at 41C. Subsequently, cells were washed with Sterofundin (Braun, Melsungen, Germany) supplemented with 10 mM HEPES (pH 8.0) and recentrifuged. The cell pellets were resuspended in Annexin-V-APC (Bender MedSystems, Vienna, Austria) staining solution (15 ml Sterofundin/HEPES, 5 ml Annexin-V-APC per sample) and incubated for 15 min at 41C in the dark. Cells were then washed again in Sterofundin/HEPES and centrifuged. Supernatants were discarded and the pellets were resuspended in 50 ml of Sterofundin/HEPES solution containing 4 mg/ml PI (Sigma-Aldrich, Deisenhofen, Germany). Samples were immediately analyzed on a FACS Calibur Cytometer (B,D Pharmingen, San Diego, CA, USA). A minimum of 30,000 events per sample were acquired, stored in listmode files and subsequently analyzed with Cellquest software (BD Pharmingen, San Diego, CA, USA). Nicoletti staining. Apoptosis was also measured according to Nicoletti et al.21 After induction of apoptosis, cells were harvested and washed once with PBS. Cells were resuspended in hypotonic fluorochrome solution containing 50 mg/ml PI, 0.1% sodium citrate and 0.1% Triton-X 100. After incubation at 41C for 16 h, cells were analyzed by flow cytometry. For Nicoletti analyses, 6,000 events were measured. In both methods, specific apoptosis was calculated by subtracting values of basal apoptosis (medium) from values of induced apoptosis. Cotreatments with z-VAD.fmk (Alexis, Montreal, Canada) were performed at a concentration of 50 mM of the caspase inhibitor. The stock concentration was made up in DMSO at a concentration of 20 mM. Brefeldin-A (BFA) was dissolved in ethanol and used at a concentration of 2.5 mg/ml. The neutralizing anti-TRAIL antibody ((Peprotech, Rocky Hill, NJ, USA) was applied at the same time as rTRAIL (5 ng/ml) at a concentration of 0.5 mg/ml. All data were processed for statistical significance using the Student’s t-test. Animal studies In the first animal set, 6-week-old female Balb/c nu/nu mice (Charles River, Sulzfeld, Germany) were injected with 5
105 DLD-1 cells in Hank‘s-buffered saline (Invitrogen, Carlsbad, CA, USA) that were transduced

AAV-encoded expression of TRAIL A Mohr et al

for 6 h in serum-free medium with AAV.TRAIL, AAV.EGFP at an MOI of 100 in vitro. The growth of the tumors was followed over 4 weeks. Three different diameters of each tumor were measured once a week. The tumor volume was calculated using the formula 3,14158/6*(d1*d2*d3). In the second animal set, 2
106 DLD-1 cells were subcutaneously injected into 6-weekold female Balb/c nu/nu mice and grown to a size of 500 mm3. At that point, 2
1010 AAV.TRAIL or AAV.EGFP genomes were directly injected into the tumor. Subsequently, the growth was followed over 3 weeks. HBS-injected tumors were used as controls for AAV effects. The data were analyzed for statistical significance using the Student’s t-test. The animal studies were performed according to the regional and national laws and covered by license number 209.1/211-2531-72/02 from the Bavarian government.

Acknowledgements This study was supported by the Deutsche Forschungsgemeinschaft (DFG) within their Emmy Noether Program (ZW 60/2-1) and the Scottish Hospital Endowments Research Trust (SHERT) through a travel grant (RMZ). The project was initiated at the University of Edinburgh, Department of Oncology and the initial work was carried out at Tulane Medical School, Department of Pathology.

References 1 Snyder RO. Adeno-associated virus-mediated gene delivery. J Gene Med 1999; 1: 166–175. 2 Monahan PE, Samulski RJ. Adeno-associated virus vectors for gene therapy: more pros than cons? Mol Med Today 2000; 6: 433–440. 3 Hermonat PL, Muzyczka N. Use of adeno-associated virus as a mammalian DNA cloning vector: transduction of neomycin resistance into mammalian tissue culture cells. Proc Natl Acad Sci USA 1984; 81: 6466–6470. 4 Tratschin JD, West MH, Sandbank T, Carter BJ. A human parvovirus, adeno-associated virus, as a eucaryotic vector: transient expression and encapsidation of the procaryotic gene for chloramphenicol acetyltransferase. Mol Cell Biol 1984; 4: 2072–2081. 5 Xiao X, Li J, Samulski RJ. Production of high-titer recombinant adeno-associated virus vectors in the absence of helper adenovirus. J Virol 1998; 72: 2224–2232. 6 Clark KR, Liu X, McGrath JP, Johnson PR. Highly purified recombinant adeno-associated virus vectors are biologically active and free of detectable helper and wild-type viruses. Hum Gene Ther 1999; 10: 1031–1039. 7 Snyder RO et al. Efficient and stable adeno-associated virusmediated transduction in the skeletal muscle of adult immunocompetent mice. Hum Gene Ther 1997; 8: 1891–1900. 8 Enger PO et al. Adeno-associated viral vectors penetrate human solid tumor tissue in vivo more effectively than adenoviral vectors. Hum Gene Ther 2002; 13: 1115–1125. 9 Wajant H, Pfizenmaier K, Scheurich P. TNF-related apoptosis inducing ligand (TRAIL) and its receptors in tumor surveillance and cancer therapy. Apoptosis 2002; 7: 449–459. 10 Gliniak B, Le T. Tumor necrosis factor-related apoptosisinducing ligand’s antitumor activity in vivo is enhanced by the chemotherapeutic agent CPT-11. Cancer Res 1999; 59: 6153–6158.

543 11 Zwacka RM, Stark L, Dunlop MG. NF-kappaB kinetics predetermine TNF-alpha sensitivity of colorectal cancer cells. J Gene Med 2000; 2: 334–343. 12 Schulze-Osthoff K et al. Apoptosis signaling by death receptors. Eur J Biochem 1998; 254: 439–459. 13 Zhang XD et al. Relation of TNF-related apoptosis-inducing ligand (TRAIL) receptor and FLICE-inhibitory protein expression to TRAIL-induced apoptosis of melanoma. Cancer Res 1999; 59: 2747–2753. 14 Ashkenazi A, Dixit VM. Death receptors: signaling and modulation. Science 1998; 281: 1305–1308. 15 Ashkenazi A et al. Safety and antitumor activity of recombinant soluble Apo2 ligand. J Clin Invest 1999; 104: 155–162. 16 Fulda S, Wick W, Weller M, Debatin KM. Smac agonists sensitize for Apo2L/TRAIL- or anticancer drug-induced apoptosis and induce regression of malignant glioma in vivo. Nat Med 2002; 8: 808–815. 17 Jo M et al. Apoptosis induced in normal human hepatocytes by tumor necrosis factor- related apoptosis-inducing ligand. Nat Med 2000; 6: 564–567. 18 Lawrence D et al. Differential hepatocyte toxicity of recombinant APO2L/trail versions. Nat Med 2001; 7: 383–385. 19 Dudus L et al. Persistent transgene product in retina, optic nerve and brain after intraocular injection of RAAV. Vision Res 1999; 39: 2545–2553. 20 Lin T et al. Targeted expression of green fluorescent protein/ tumor necrosis factor- related apoptosis-inducing ligand fusion protein from human telomerase reverse transcriptase promoter elicits antitumor activity without toxic effects on primary human hepatocytes. Cancer Res 2002; 62: 3620–3625. 21 Nicoletti I et al. A rapid and simple method for measuring thymocyte apoptosis by propidium iodide staining and flow cytometry. J Immunol Methods 1991; 139: 271–279. 22 Kagawa S et al. Antitumor activity and bystander effects of the tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) gene. Cancer Res 2001; 61: 3330–3338. 23 Griffith TS et al. Adenoviral-mediated transfer of the TNFrelated apoptosis-inducing ligand/Apo-2 ligand gene induces tumor cell apoptosis. J Immunol 2000; 165: 2886–2894. 24 Vermeij J et al. Transduction of ovarian cancer cells: a recombinant Adeno-associated viral vector compared to an adenoviral vector. Br J Cancer 2001; 85: 1592–1599. 25 Qing K et al. Role of tyrosine phosphorylation of a cellular protein in adeno-associated virus 2-mediated transgene expression. Proc Natl Acad Sci USA 1997; 94: 10879–10884. 26 Fisher KJ et al. Transduction with recombinant adeno-associated virus for gene therapy is limited by leading-strand synthesis. J Virol 1996; 70: 520–532. 27 Mingozzi F et al. Improved hepatic gene transfer by using an adeno-associated virus serotype 5 vector. J Virol 2002; 76: 10497–10502. 28 Gao GP et al. Novel adeno-associated viruses from rhesus monkeys as vectors for human gene therapy. Proc Natl Acad Sci USA 2002; 99: 11854–11859. 29 Keane MM et al. Chemotherapy augments TRAIL-induced apoptosis in breast cell lines. Cancer Res 1999; 59: 734–741. 30 Beltinger C, Fulda S, Walczak H, Debatin KM. TRAIL enhances thymidine kinase/ganciclovir gene therapy of neuroblastoma cells. Cancer Gene Ther 2002; 9: 372–381. 31 Wiley SR et al. Identification characterization of a new member of the TNF family that induces apoptosis. Immunity (1995); 3: 673–682. 32 Liang FQ et al. AAV-mediated delivery of ciliary neurotrophic factor prolongs photoreceptor survival in the Rhodopsin knockout mouse. Mol Ther 2001; 3: 241–248. 33 Protzer U et al. Interferon gene transfer by a Hepatitis B virus vector efficiently suppresses wild-type virus infection. Proc Natl Acad Sci USA 1999; 96: 10818–10823. Gene Therapy