TNF-related apoptosis-inducing ligand (TRAIL) is a member of the TNF superfamily of cytokines that induces apoptosis in a variety of cancer cells. The results presented in this study demonstrate that introduction of the human TRAIL gene into TRAIL-sensitive tumor cells using an adenoviral vector leads to the rapid production and expression of TRAIL protein, and subsequent death of the tumor cells. Tumor cell death was mediated by an apoptotic mechanism, as evidenced by the activation of caspase-8, cleavage of poly(ADP-ribose) polymerase, binding of annexin V, and inhibition by caspase inhibitor zVAD-fmk. These results define a novel method of using TRAIL as an antitumor therapeutic, and suggest the potential use for an adenovirus-encoding TRAIL as a method of gene therapy for numerous cancer types in vivo.
Members of the TNF superfamily influence a variety of immunological functions, including cellular activation, proliferation, and death, upon interaction with a corresponding superfamily of receptors (1, 2, 3). Whereas interest in the apoptosis-inducing molecules TNF and Fas ligand has been peaked due to their participation in events such as autoimmune disorders, activation-induced cell death, immune privilege, and tumor evasion from the immune system (4, 5, 6, 7, 8), another death-inducing family member, TNF-related apoptosis-inducing ligand (TRAIL)3 or Apo-2 ligand, is generating excitement because of its apparent unique ability to induce apoptosis in a wide range of transformed cell lines, but not in normal tissues (9, 10). To date, four homologous, but distinct, human TRAIL (hTRAIL) receptors have been identified, with two (DR4 (11 ; hereafter referred to as TRAIL-R1) and DR5/TRAIL-R2 (12, 13, 14, 15)) having the ability to initiate the apoptosis-signaling cascade after ligation, and two (TRID/DcR1/TRAIL-R3 (12, 13, 15, 16) and TRAIL-R4/DcR2/TRUNDD (17, 18, 19)) lacking this ability. Because they lack the ability to directly signal cell death, TRAIL-R3 and TRAIL-R4 have been hypothesized as being protective receptors, either by acting as decoy receptors (11, 12, 18, 19) or via transduction of an antiapoptotic signal (17).
Given the tumor cell selectivity of TRAIL’s cytotoxicity from results obtained in vitro, recent studies have examined the safety and antitumor activity of soluble rTRAIL in vivo (20, 21, 22). TRAIL was found to be well tolerated when multiple doses were given to normal animals, and no observable histological or functional changes were observed in any of the tissues or organs examined. These results were dramatically different from those seen in animals given other apoptosis-inducing molecules, as injection of recombinant Fas ligand or anti-Fas mAb into animals rapidly induced massive hepatocyte degeneration, necrosis, and hemorrhage, and ultimately death (20, 23, 24). Moreover, multiple injections of soluble TRAIL into mice beginning the day after tumor implantation significantly suppressed the growth of the tumors, with many animals being tumor free (20, 21, 22). One potential drawback to these findings was that large amounts of soluble TRAIL were required to inhibit tumor formation. This may be due to the pharmacokinetic profile of soluble TRAIL that indicated that after i.v. injection the majority of the protein is cleared within 5 h (20). Increasing the in vivo t1/2 of soluble rTRAIL or developing an alternative means of delivery may increase the relative tumoricidal activity of TRAIL such that larger, more established tumors could be eradicated as efficiently as smaller tumors. The identification of alternate methods to deliver TRAIL to the tumor site, however, is also critical for the further development and testing of the antitumor activity of TRAIL in vivo. The results presented in this work describe the production of an adenoviral vector engineered to carry the gene for hTRAIL. Shortly after cell infection, TRAIL protein was detected, rapidly leading to the induction of apoptosis in TRAIL-sensitive tumor cells in vitro. Moreover, these results demonstrate the potential of using adenoviral-mediated delivery and local expression of TRAIL to destroy tumors in vivo.
Materials and Methods
Reagents and Abs
The tripeptide caspase inhibitor, z-VAD-fmk, was obtained from Enzyme Systems Products (Livermore, CA). A stock solution of the inhibitor was prepared in DMSO and stored at 4°C. Brefeldin A (BFA) was purchased from Epicentre Technologies (Madison, WI), with a stock solution prepared in 100% EtOH and stored at −20°C. Abs against caspase-8 (provided by Dr. Marcus Peter, University of Chicago), poly(ADP-ribose) polymerase (PARP; PharMingen, San Diego, CA), and hTRAIL (PeproTech, Rocky Hill, NJ) were used for Western blotting according to manufacturer’s instruction. Soluble rhTRAIL was purchased from PeproTech and used at the indicated concentrations.
The human prostate carcinoma cell line (PC-3) was obtained from Dr. Michael Cohen (University of Iowa, Iowa City, IA). The human melanoma cell lines (WM 164 and WM 793) were obtained from Dr. Meenhard Herlyn (Wistar Institute, Philadelphia, PA). The human mammary adenocarcinoma cell line (MDA 231) was obtained from Dr. David Lynch (Immunex, Seattle, WA). The human bladder cancer cell line (RT-4) was obtained from Dr. Scott Crist (University of Iowa). PC-3, RT-4, WM 164, and WM 793 were cultured in DMEM supplemented with 10% FBS, penicillin, streptomycin, sodium pyruvate, nonessential amino acids, and HEPES (hereafter referred to as complete DMEM). MDA 231 was cultured in RPMI 1640 supplemented with 10% FBS, penicillin, streptomycin, sodium pyruvate, nonessential amino acids, and HEPES (hereafter referred to as complete RPMI). Normal human prostate epithelial cells (PrEC) were obtained from Clonetics (San Diego, CA) and cultured as directed.
Production of adenovirus encoding the hTRAIL gene
The cDNA for hTRAIL was obtained from Dr. Hideo Yagita (Juntendo University, Tokyo, Japan) (25). A replication-deficient adenovirus encoding the hTRAIL gene (Ad5-TRAIL) expressed from the CMV promotor was generated using standard methods by the University of Iowa Gene Transfer Vector Core (26). Briefly, the entire coding sequence of hTRAIL was cloned into the XhoI and NotI sites of pAd5CMVK-NpA. The resultant plasmid and adenovirus backbone sequences (Ad5) (27) that had the E1 (E1A and E1B) genes deleted were transfected into human embryonic kidney 293 cells, and viral particles were isolated and amplified for analysis of TRAIL expression. Recombinant adenoviruses encoding nuclear-targeted bacterial β-galactosidase (Ad5-βgal) or green fluorescent protein (Ad5-GFP) were used as virus controls. Recombinant adenoviruses were screened for replication competent virus by A549 plaque assay, and virus titer was determined by plaque assay on 293 cells. Purified viruses were stored in PBS with 3% sucrose and kept at −80°C until use.
Cells were cultured in complete medium, and permitted to adhere for at least 6 h before adding adenovirus. Before infection, cells were washed with PBS, and then the vectors were added at the indicated number of PFU/cell in culture medium supplemented, as described above, but with only 2% FBS. After 4 h, cells were washed with PBS and incubated in complete medium for the remainder of the assay.
Reported gene assays for measuring adenoviral infection
Analysis of adenoviral infection efficiency and transferred gene expression was performed using Ad5-GFP and Ad5-βgal, respectively. Cells infected with Ad5-GFP were analyzed by flow cytometry on a FACScan (Becton Dickinson, San Jose, CA) at various time points after infection to determine infection efficiency. Cells infected with Ad5-βgal were assayed for β-galactosidase activity with the Galacto-Light Plus chemiluminescent reporter gene assay system (Tropix, Bedford, MA) to determine the level of transferred gene expression.
In vitro killing of human tumor cell lines with adenoviral vectors
Tumor cell sensitivity to Ad5-βgal, Ad5-GFP, or Ad5-TRAIL was assayed using the following procedure. Cells were added to 96-well plates (2 × 104 cells/well) in complete medium, and then allowed to adhere for at least 6 h before infection with the various adenoviral vectors, as described above. As a positive control, soluble rhTRAIL was added to the target cells at the indicated concentrations. In some experiments, z-VAD-fmk (20 μM) TRAIL-R2:Fc (20 μg/ml; Immunex), Fas:Fc (20 μg/ml; PharMingen), or brefeldin A (5 μg/ml) was added to the medium during and after infection for the remainder of the assay. Cell death was determined after 20 h by crystal violet staining, as described (28). Results are presented as percent cell death: (1 − (OD cells treated per OD cells not treated)) × 100. For analysis of apoptosis, tumor cell targets were incubated as described above and apoptotic cell death was measured by flow cytometry using FITC-conjugated annexin V (R&D Systems, Minneapolis, MN) and propidium iodide (Sigma, St. Louis, MO), as described (29, 30).
Cells were lysed in PBS containing 1% Nonidet P-40, 0.35 mg/ml PMSF, 9.5 μg/ml leupeptin, and 13.7 μg/ml pepstatin A. The lysed cells were centrifuged at 14,000 × g to remove cellular debris, and protein concentrations of the lysates were determined by the colorimetric bicinchoninic acid analysis (Pierce, Rockford, IL). Equal amounts of protein were separated by SDS-PAGE, transferred to nitrocellulose membrane (Novex, San Diego, CA), and blocked with 5% nonfat dry milk in PBS Tween-20 (0.05% v/v) overnight. The membrane was incubated with the anti-caspase-8, anti-PARP, or anti-hTRAIL mAb (diluted according to manufacturer’s instructions) for 1 h. After washing, the membrane was incubated with an anti-mouse or anti-rabbit HRP Ab (diluted 1/1000; Amersham, Arlington Heights, IL) for 1 h. Following several washes, the blots were developed by chemiluminescence according to the manufacturer’s protocol (Renaissance chemiluminescence reagent; DuPont NEN, Boston, MA).
Surface expression of TRAIL was determined by measuring the binding of the anti-hTRAIL mAb M181 (mouse IgG1; Immunex). Briefly, cells were incubated with 10 μg/ml M181 or MOPC-21 (nonspecific mouse IgG1 isotype control; Sigma) in 3% BSA in PBS (PBSA) for 30 min on ice. Following three washes with PBS, cells were incubated with a PE-conjugated, Fc-specific goat anti-mouse F(ab′)2 (Jackson ImmunoResearch, West Grove, PA) for 30 min on ice. Finally, after three washes in PBS, the cells were analyzed on a FACScan (Becton Dickinson).
Cell-mediated killing of human tumor cells
WM 164 cells were infected with 1000 PFU/cell Ad5-TRAIL for 4 h as above, incubated in complete medium for 12 h, washed, and resuspended in complete medium. PC-3 tumor cells were labeled with 100 μCi of 51Cr for 1 h at 37°C, washed three times, and resuspended in complete medium. To determine TRAIL-induced death, 51Cr-labeled tumor cells (104/well) were incubated with varying numbers of Ad5-TRAIL/WM 164 effector cells for 8 h. As a positive control, soluble TRAIL was added to the target cells at the indicated concentrations. In some cultures, TRAIL-R2:Fc or Fas:Fc (20 μg/ml) was added to the Ad5-TRAIL/WM 164 effector cells 15 min before adding tumor cell targets. Assays were performed in round-bottom 96-well plates, and the percent specific lysis was calculated as: 100 × (experimental cpm − spontaneous cpm)/(total cpm − spontaneous cpm). Spontaneous and total release were determined in the presence of either medium alone or 1% Nonidet P-40, respectively. The presence of TRAIL-R2:Fc or Fas:Fc during the assay had no effect on the level of spontaneous release of 51Cr by the target cells.
Production of TRAIL-encoding adenovirus
The cDNA encoding full-length hTRAIL was inserted into the E1 region of a replication-deficient Ad5 construct under the control of the CMV immediate early promoter (Fig. 1,A). This plasmid was transfected into human embryonic kidney 293 cells to propagate the virus. Ad5-TRAIL-infected 293 or uninfected 293 cells were lysed, and the cellular proteins were separated by nonreducing SDS-PAGE to assay for TRAIL expression by Western blotting. Amino acid sequence analysis of the TRAIL cDNA predicts a weight of 32.5 kDa for TRAIL monomers (9, 10). As demonstrated in Fig. 1 B, prominent bands migrating at 32–35 kDa and 55–58 kDa are clearly evident, which correspond to TRAIL monomers and dimers, respectively. In contrast, no corresponding bands were present in the uninfected 293 cell lysate. Ab specificity was demonstrated by using rTRAIL comprised of the entire extracellular domain of 169 aa, which yields a monomer of 19.6 kDa. Higher order forms of the rTRAIL migrating at approximately 40 kDa (dimer) and 60 kDa (trimer) were also present. The differences in monomer and dimer sizes reflect the fact that one is only the extracellular domain (rTRAIL) while the other is full length (Ad-TRAIL-derived). The lack of detectable higher order multimers in the 293/Ad-TRAIL lysate may be due to the method of cellular lysis and only analyzing the solubilized proteins not associated with the cell membrane. It is possible that full-length TRAIL trimers require support from the cell membrane to remain in this form, which would be absent from the soluble component of the cell lysate containing mostly monomeric and dimeric forms. Thus, these results demonstrate that the adenoviral-mediated gene transfer of hTRAIL results in transgene expression in human cells.
Human tumor cells are susceptible to adenovirus infection
One of the advantages of using an adenoviral vector lies in the ability to infect epithelial cell populations. Group C adenovirus, such as Ad5, requires the interaction between the viral fiber capsid protein to the coxsackievirus and adenovirus receptor, or CAR, and the viral penton base binding to certain integrins (e.g., αvβ3 and αvβ5) for entry into the cell by receptor-mediated endocytosis (31, 32, 33). Therefore, it was critical to determine whether a panel of human tumor cell lines (MDA 231, mammary adenocarcinoma; PC-3, prostate carcinoma; RT-4, bladder papilloma; WM 164, melanoma; and WM 793, melanoma) would be receptive to adenoviral infection before examining the effects of Ad5-TRAIL infection. The tumor cells were infected with either an adenovirus carrying the enhanced green fluorescent protein gene (Ad5-GFP) or the β-galactosidase gene (Ad5-βgal) for 4 h, and then analyzed 20 h later. When infected with 1000 PFU/cell Ad5-GFP, all of the tumor cell lines demonstrated a high percentage of infectivity, ranging from 84.7 to 99.1%, as measured by flow cytometry (Fig. 2,A). In addition, normal PrEC were also found to be highly susceptible to Ad5-GFP infection (95.3%). Infection with Ad5-βgal revealed that all of the cell types produced protein from the transferred gene in a PFU/cell-dependent manner; however, there were greater differences in β-galactosidase activity between the different cell types than seen when examining GFP production (Fig. 2 B). Thus, these results indicate that adenoviral-mediated transfer of the β-galactosidase and GFP reporter genes into the cells of interest resulted in efficient gene transcription and translation into protein, suggesting that infection with Ad5-TRAIL should produce TRAIL protein in a similar percentage of cells.
Production of TRAIL following Ad5-TRAIL infection leads to tumor cell death
With the demonstration that the human tumor cell line panel was adequately infected with adenovirus, subsequent experiments were performed to examine the consequences of Ad5-TRAIL infection. The tumor cells were infected with either Ad5-βgal or Ad5-TRAIL for 4 h, and then cultured for an additional 20 h before determining the amount of cell death. As indicated in Fig. 3,A, minimal cell death of PC-3 cells was observed upon infection with Ad5-βgal. In contrast, a significant increase in cell death was seen with Ad5-TRAIL infection. Moreover, the level of cell death induced by Ad5-TRAIL infection was comparable with that of soluble TRAIL-induced death. This cytotoxic activity was seen with other TRAIL-sensitive tumor cell targets, but not with the TRAIL-resistant melanoma cell line WM 164 or normal PrEC (Table I). Analysis of TRAIL protein production by Western blot revealed detectable levels in PC-3 cell lysates by 1 h postinfection, with levels increasing over the entire time course (Fig. 3 B). In contrast, lysates from uninfected PC-3 cells or PC-3 cells examined 20 h after Ad5-βgal infection had no detectable TRAIL protein present. Thus, these results demonstrate that tumor cells infected with Ad5-TRAIL produce TRAIL protein that, presumably, leads to their death.
|Target Cell .||Ad5-βgala .||Ad5-TRAILa .||TRAILb .|
|MDA 231 (breast)||2.4 ± 1.4||44.7 ± 5.6||89.9 ± 0.9|
|PC-3 (prostate)||1.4 ± 1.8||85.5 ± 5.2||90.1 ± 8.7|
|RT4 (bladder)||1.3 ± 1.0||82.4 ± 6.1||89.4 ± 5.0|
|WM 164 (melanoma)||3.8 ± 2.1||1.2 ± 0.8||5.2 ± 2.7|
|WM 793 (melanoma)||2.5 ± 2.8||56.7 ± 9.7||86.6 ± 7.6|
|Normal prostate epithelium||13.1 ± 4.1||10.0 ± 2.6||14.3 ± 1.9|
|Target Cell .||Ad5-βgala .||Ad5-TRAILa .||TRAILb .|
|MDA 231 (breast)||2.4 ± 1.4||44.7 ± 5.6||89.9 ± 0.9|
|PC-3 (prostate)||1.4 ± 1.8||85.5 ± 5.2||90.1 ± 8.7|
|RT4 (bladder)||1.3 ± 1.0||82.4 ± 6.1||89.4 ± 5.0|
|WM 164 (melanoma)||3.8 ± 2.1||1.2 ± 0.8||5.2 ± 2.7|
|WM 793 (melanoma)||2.5 ± 2.8||56.7 ± 9.7||86.6 ± 7.6|
|Normal prostate epithelium||13.1 ± 4.1||10.0 ± 2.6||14.3 ± 1.9|
Mean percent specific lysis (±SD) at 1000 PFU/cell.
Mean percent specific lysis (±SD) with 1 μg/ml soluble TRAIL.
Ad5-TRAIL infection induces apoptosis in tumor cell targets
Although crystal violet staining of the tumor cells infected with the adenoviral vectors or treated with rTRAIL as presented in Fig. 3 indicates the amount of cell death, it does not discriminate between apoptotic and necrotic cell death. Previous reports have demonstrated that TRAIL-induced cell death occurs through an apoptotic mechanism characterized by the activation of a cascade of intracellular proteases, or caspases, and the cleavage of numerous intracellular proteins (9, 10, 14, 34, 35, 36). To confirm that the tumor cell death following Ad5-TRAIL infection was mediated through an apoptotic mechanism, caspase activation and cellular protein cleavage were examined. Thus, PC-3 cells were infected with Ad5-TRAIL for 4 h, cell lysates were prepared at various times after infection, and the cellular proteins were separated by SDS-PAGE for Western blot analysis of caspase-8 activation and PARP cleavage. Activation of caspase-8 occurred within 2 h after infection, while PARP cleavage took place by 4 h after infection (Fig. 4,A). By 20 h after infection, the levels of the active p18 subunit of caspase-8 and 85-kDa fragment of PARP had dropped below the level of detection, due to extensive apoptotic destruction. To further demonstrate the importance of caspase activation in the death of Ad5-TRAIL-infected cells, the caspase inhibitor z-VAD-fmk (carbobenzyloxy-Val-Ala-Asp (OMe) fluoromethyl ketone) was added to the culture medium throughout the assay. z-VAD-fmk completely inhibited PC-3 cell death, whereas equal concentrations of the peptide vehicle (DMSO) did not (Fig. 4 B).
A second critical event that takes place during apoptosis is the alteration in plasma membrane composition that appears to serve as a signal for phagocytes to recognize and engulf the apoptotic cells before membrane integrity is compromised. It has been suggested that phosphatidylserine, a phospholipid component of the inner leaflet of the cell membrane that appears in the outer leaflet during apoptosis, serves as the marker for phagocytosis (37, 38). Annexin V has been shown to preferentially bind to phosphatidylserine (39, 40), and can be used to detect the expression of phosphatidylserine on apoptotic cells (29, 30). Thus, PC-3 cells infected with Ad5-βgal or Ad5-TRAIL were analyzed for annexin V binding. Upon staining the PC-3 cells 6 h after infection or incubation with soluble hTRAIL (100 ng/ml), only those cells infected with Ad5-TRAIL or incubated with soluble hTRAIL were positive for FITC-annexin V binding (Fig. 4 C), providing further evidence that the death of the Ad5-TRAIL-infected tumor cells was through an apoptotic mechanism. Morphological changes, such as membrane blebbing and the release of apoptotic bodies, were also observed in cells infected with Ad5-TRAIL using light microscopy (data not shown).
Ad5-TRAIL-induced death can be inhibited with BFA, but not TRAIL receptor:Fc
The results obtained to date demonstrated that Ad5-TRAIL infection leads to TRAIL protein production and the subsequent induction of apoptotic cell death. However, it was important to also demonstrate the cell death to be a TRAIL-dependent phenomenon with the expression of TRAIL on the surface of the infected cells. Thus, the TRAIL-resistant (both soluble TRAIL and Ad5-TRAIL) human melanoma WM164 was infected with Ad5-βgal or Ad5-TRAIL as in previous experiments, and then analyzed for TRAIL expression by flow cytometry after 8 h. TRAIL-resistant WM 164 cells were used in this study because the cell death that occurred in the TRAIL-sensitive PC-3 cells following Ad5-TRAIL infection made them difficult to analyze accurately as nonspecific staining increased as the cells became apoptotic. Whereas no TRAIL was detectable on uninfected or Ad5-βgal-infected WM 164 cells, those cells infected with Ad5-TRAIL did express TRAIL on the cell surface (Fig. 5,A). Interestingly, treatment of the Ad5-TRAIL-infected WM 164 cells with BFA resulted in the inhibition of TRAIL expression at the cell surface. BFA blocks the anterograde migration of proteins through the Golgi complex, and thus prevents their expression on the cell surface. The BFA treatment did not, however, inhibit the production of TRAIL protein, as both treated and untreated Ad5-TRAIL-infected WM 164 cell lysates contained TRAIL protein as determined by Western blotting (Fig. 5 B).
Since TRAIL must bind to TRAIL-R1 or TRAIL-R2 (the death domain, death-inducing TRAIL receptors) to initiate the apoptotic process, it was predicted that disruption of this interaction would protect the TRAIL-sensitive tumor cells from Ad5-TRAIL-induced death. To test this, PC-3 cells were infected with Ad5-TRAIL and then cultured for 20 h in medium alone, or medium containing the recombinant soluble receptors for TRAIL (TRAIL-R2:Fc) (14) or Fas (Fas:Fc). Surprisingly, TRAIL-R2:Fc did not inhibit cell death in the Ad5-TRAIL-infected PC-3 cells (Fig. 6,A) at concentrations that completely inhibited cell death induced by soluble TRAIL (Fig. 6,B). It was reasoned that since the PC-3 cells are adherent, nonpolarized cells TRAIL could be expressed on surfaces that were inaccessible to TRAIL-R2:Fc, but still able to engage the TRAIL-R1 or TRAIL-R2 expressed there. Seeing that BFA could inhibit the expression of TRAIL (Fig. 5,A), the same experiment was tried, but in the presence of BFA or its vehicle EtOH. In this setting, the addition of BFA was able to block the cell death resulting from Ad5-TRAIL infection, whereas EtOH did not (Fig. 6,C). This inhibition by BFA did not interfere with the signaling mechanism of TRAIL-R1/R2 as BFA-treated PC-3 cells were as sensitive to soluble TRAIL-induced death as those cultured without BFA (data not shown). Further support for TRAIL-mediated killing following Ad5-TRAIL infection was obtained by using Ad5-TRAIL-infected PrEC or WM 164 cells to induce apoptosis in PC-3 target cells. PrEC or WM 164 was infected with 1000 PFU/cell Ad5-TRAIL for 4 h, and then 12 h later incubated with 51Cr-labeled PC-3 cells at various E:T cell ratios. Whereas uninfected PrEC or WM 164 demonstrated no cytolytic activity against the PC-3 target cells, the Ad5-TRAIL-infected cells displayed comparable activity to soluble TRAIL (Fig. 6,D). This activity was TRAIL specific, as inclusion of soluble TRAIL-R2:Fc, but not Fas:Fc, to the Ad5-TRAIL-infected WM 164 cells blocked target cell lysis (Fig. 6,E). Similar results were also obtained with PrEC (data not shown). Collectively, the results presented in Figs. 5 and 6 demonstrate that the apoptotic death following Ad5-TRAIL infection results from TRAIL expression on the cell surface, where it binds to either TRAIL-R1 or TRAIL-R2.
The data presented in this study describe the generation of an adenoviral vector engineered to carry the gene for hTRAIL. Ad5-TRAIL infection resulted in the rapid transcription and translation of the transferred hTRAIL gene into functional TRAIL protein that, when expressed on the cell surface, induced apoptotic death in TRAIL-sensitive tumor cell targets, but not TRAIL-resistant tumor cells or normal cells. Addition of BFA, a fungal metabolite that specifically blocks protein transport from the endoplasmic reticulum to the Golgi apparatus (and ultimately the cell surface), to the target cells inhibited TRAIL surface expression and subsequent apoptotic death. To our knowledge, this is the first use of TRAIL in a gene transfer/gene therapy setting, which presents a variety of new possibilities for using TRAIL (the gene and/or the protein) as an antitumor agent.
The identification of TRAIL several years ago generated a great deal of interest, when it was determined that it appeared to have the ability to induce apoptosis in a variety of tumor cell lines, but not in normal cells in vitro. Moreover, it was observed that TRAIL mRNA is constitutively expressed in a wide variety of cells and tissues. These were unusual characteristics for a death-inducing molecule in the TNF family, as the expression of TNF, lymphotoxin-α, and Fas ligand is tightly regulated since they can have toxic effects on normal tissues. The tumor-specific activity of TRAIL was extended in vivo with the observation that treatment of SCID and nude mice bearing human tumors with soluble TRAIL significantly inhibited tumor outgrowth without any observable toxic side effects to the host (20, 21, 22). This inhibition of tumor outgrowth, though, required high amounts of TRAIL given over several days shortly after tumor implantation. Pharmacokinetic analysis revealed that soluble TRAIL given to mice i.v. displayed an elimination t1/2 of just under 5 h (20). Given that many normal tissues express mRNA for at least one of the four TRAIL receptors, this suggests that almost all the tissues in the body have the potential to bind and sequester soluble TRAIL and, thus, prevent it from reaching the tumor.
An alternative approach would be to administer TRAIL locally, where it would exist at a greater concentration and have a better chance of significantly inducing tumor cell death. Such localized, intratumoral injections of soluble TRAIL would, however, be limited in that only a relatively small volume could be administered, suggesting that a potentially suboptimal amount of TRAIL protein would be used. In contrast, adenovirus can be produced at high titers, such that small volumes would contain high numbers of infectious adenoviral particles carrying the hTRAIL gene. The use of the CMV promoter to drive the transcription of the transferred hTRAIL gene serves as an additional mechanism to further increase the local concentration of TRAIL protein, because it may not be transcriptionally regulated in the same manner as the TRAIL promoter. Only when the process of apoptosis has disrupted cellular functions sufficiently to affect protein production in the Ad5-TRAIL-infected tumor cell will the generation of TRAIL stop.
The concept of gene therapy has recently developed into a viable method of treating malignant transformation and cancer progression. Whereas some therapies have focused on replacing the absent critical functional genes in the target cells to restore a normal phenotype, other approaches have been based on introducing genes that encode immunostimulatory molecules to activate the immune system against the tumor. Many of these studies have employed the use of replication-deficient adenoviral vectors derived from Ad5 to transfer the gene of interest into the target tumor cells. For example, adenoviral vectors encoding CD80, IFN-β, IL-2, IL-7, and IL-12 have all demonstrated the ability to stimulate antitumor responses after administration (41, 42, 43, 44, 45, 46). The combination of adenovirus-mediated delivery of the herpes simplex virus thymidine kinase gene and ganciclovir therapy has proved efficacious in treating prostate cancer (47, 48). Also, adenoviral vectors expressing Fas ligand have been tested in the treatment of prostate cancer models, experimental glioma, and renal carcinoma (49, 50, 51). Even with these promising observations, immunogenicity remains a potential problem with adenoviral-based vectors. Abs present in the patient may quickly neutralize the adenovirus before it can deliver its genetic load, as there is widespread immunity to a variety of adenovirus serotypes in humans. Recent results from a phase I clinical trial, however, reported the safety of using adenoviral vectors as a gene delivery vehicle in humans and demonstrated successful transgene expression even in the presence of preexisting immunity to adenovirus (43). Intratumoral administration of Ad5-TRAIL may provide the virus with an appropriate environment for infection of the tumor and surrounding tissue, which will lead to gene expression and sufficient TRAIL protein production to induce tumor cell death. Such therapeutic studies with Ad5-TRAIL are in progress.
It was surprising that there were differences in tumor cell death following Ad5-TRAIL infection as compared with soluble TRAIL-induced death. The relative activity of Ad5-TRAIL is dependent upon its ability to infect a target cell. Adenovirus infection requires the expression of CAR (coxsackievirus and adenovirus receptor) and the expression of certain integrins, such as αvβ3 and αvβ5 (31, 32, 33). All of the tumor cell lines and the normal PrEC were highly susceptible to adenovirus infection, suggesting that each cell type differentially regulated the translation of TRAIL mRNA into protein. Additional regulation may result in the transport of any TRAIL that is produced from the cytoplasm to the cell surface at different rates. The surface expression of TRAIL is required for apoptosis induction, as the inhibition of protein transport by BFA inhibited cell death, but not the production of TRAIL protein, following Ad5-TRAIL infection.
Whereas surface expression of TRAIL is essential for the observed tumoricidal activity of Ad5-TRAIL, the sensitivity of the cell to TRAIL-induced apoptosis is also an important component of this phenomenon. This was best demonstrated by the fact that the melanoma cell line WM 164 and the normal PrEC, which were resistant to soluble TRAIL-mediated apoptosis, were also resistant to effects of Ad5-TRAIL infection. The identification of two TRAIL receptors that are capable of initiating the apoptotic machinery (TRAIL-R1 and TRAIL-R2) and two that are not (TRAIL-R3 and TRAIL-R4) led to the initial hypothesis that the expression of TRAIL-R3 and/or TRAIL-R4 conferred resistance to TRAIL-induced death (11, 12, 18, 19). However, this hypothesis was formulated by examining TRAIL receptor mRNA expression in several normal tissues and tumor cell lines, and from experiments in which TRAIL-R3 or TRAIL-R4 were overexpressed in transfected cells. Most of the tumor cell lines used in this study express TRAIL-R3 and/or TRAIL-R4, yet were sensitive to TRAIL (soluble or Ad5-TRAIL-derived)-mediated death (36, 52), so it is unlikely that expression of either TRAIL-R3 or TRAIL-R4 plays a role in determining their resistance to TRAIL. One possible explanation for the resistance of the WM 164 cells to Ad5-TRAIL may come from a component of the cell death machinery called FLIP (Fas-associated death domain-like IL-1-converting enzyme inhibitory protein). FLIP is believed to inhibit the death receptor signaling machinery at its most proximal point by preventing the interaction of caspase-8 and/or Fas-associated death domain protein to the death domains of cross-linked death receptors, and thus inhibit any downstream apoptotic signaling events (53). Intracellular levels of FLIP are high in the TRAIL-resistant melanoma cell line WM 164 (36), and high FLIP levels have also been shown to correlate with resistance to TRAIL-mediated apoptosis in primary vs transformed keratinocytes (54). While FLIP may have a protective function in the WM 164 tumor cell line, it is likely to be one of several intracellular proteins that cooperate with other molecules (both intracellular and at the cell surface) to regulate TRAIL-induced death in tumor cell lines.
The participation of TRAIL-R3 and TRAIL-R4 in regulating TRAIL sensitivity may be greater, however, in normal cells/tissues or primary tumors than in established tumor cell lines. The treatment of normal human umbilical vein or microvascular endothelial cells with phospholipase C (to strip the GPI-linked TRAIL-R3 from the surface) and cycloheximide (to prevent the reexpression of any TRAIL-R3) sensitized these cells to TRAIL (13). This would suggest that TRAIL-R3 is a key regulator of the sensitivity of normal cells to TRAIL-induced death, but the addition of cycloheximide may inhibit the production of some other protein (perhaps FLIP) critical for TRAIL resistance. Furthermore, it is not known how much TRAIL-R3 or TRAIL-R4 is needed to inhibit the formation of a competent TRAIL-R1 or TRAIL-R2 signaling complex. RT-PCR analysis of the normal PrEC detected mRNA species for all four TRAIL receptors (T.S.G., unpublished observation), making it possible for TRAIL-R3 and/or TRAIL-R4 to enter into the TRAIL-ligated receptor complex on the PrEC, and thus making it incapable of initiating apoptosis. As with the tumor cell lines, it is difficult to determine at this time what is the exact mechanism that regulates TRAIL sensitivity and resistance in normal cells and tissues. Additional studies will be required to determine whether regulation is determined at the surface (i.e., TRAIL-R3, TRAIL-R4 expression), within the cell (i.e., FLIP, Bcl-2 family member), or both.
The observed suicide-like death of the Ad5-TRAIL-infected tumor cells is not the only mechanism by which tumor cells may die with this kind of gene transfer therapy in vivo. The intralesional injection of Ad5-TRAIL would likely result in the infection of both cancerous and normal cells surrounding the injection site. While the normal PrEC tested in the report were not killed when infected with Ad5-TRAIL, they still produced TRAIL protein from the transferred gene, as evidenced by the fact that they could then be used to kill PC-3 cells in a TRAIL-dependent manner (Fig. 6, D and E). This suggests that it would not be imperative for Ad5-TRAIL to infect the tumor cell, as infection in either the tumor itself or the surrounding normal tissue would lead to the localized production of TRAIL. In addition, the apoptotic death resulting from Ad5-TRAIL infection may help initiate a T cell-mediated immune response against any remaining tumor cells. Recent reports have demonstrated that immature dendritic cells can engulf apoptotic bodies and present Ags derived from these cell fragments in an MHC class I-restricted fashion upon maturation, resulting in the CTL activity (55, 56). Likewise, the combination of Ad5-TRAIL with an immunostimulatory cytokine (i.e., IL-12, IFN-γ) may result in the initiation of a tumor-specific immune response against any remaining tumor cells. Conversely, one potential disadvantage to using the Ad5-TRAIL gene therapy would be the chance of inducing TRAIL expression on a normally TRAIL-resistant tumor cell, giving it another means to evade or repel cell-mediated attempts of tumor rejection. Indeed, mammary adenocarcinoma cells engineered to express TRAIL demonstrated enhanced growth in vivo compared with the parental tumor cells (57). This situation is similar to that reported for a number of Fas ligand-expressing tumors (8, 58, 59, 60). Careful evaluation of the primary tumor will be necessary before Ad5-TRAIL injection to minimize this possibility.
The development of alternate or adjuvant forms of cancer therapy is crucial, due to the increasing rates of many cancers throughout the world. For example, prostate cancer is one of the most prevalent cancers among U.S. males, with annual death rates currently estimated at over 40,000 (61). Current treatment for localized prostate cancer is limited to surgery or radiation therapy, whereas androgen ablation is generally accepted as the best method for treating metastatic prostate cancer. Unfortunately, a significant number of patients with advanced prostate cancer fail to demonstrate any initial positive response to androgen ablation therapy. Moreover, prostatic cells often lose their dependency on androgen during cancer progression, and androgen ablation becomes ineffective, leading to tumor progression and death within 3 yr. The development of an adenovirus expressing the gene for hTRAIL is the first of what will likely be numerous gene therapy vectors to treat cancer in the future.
We thank Drs. Michael Cohen, David Lubaroff, Lyse Norian, and Oskar Rokhlin for careful reading of the manuscript. We also acknowledge Maria Scheel for her help with virus production, and Linda Buckner for secretarial assistance.
This work was supported by Grant IN-122U from the American Cancer Society, administered through The University of Iowa Cancer Center.
Abbreviations used in this paper: TRAIL, TNF-related apoptosis-inducing ligand; Ad5, adenovirus 5; BFA, brefeldin A; EtOH, ethanol; FLIP, Fas-associated death domain-like IL-1-converting enzyme inhibitory protein; GFP, green fluorescent protein; hTRAIL, human TRAIL; PARP, poly(ADP-ribose) polymerase; PrEC, prostate epithelial cells; Ad5-βgal, recombinant adenovirus encoding nuclear-targeted bacterial β-galactosidase; Ad5-GFP, recombinant adenovirus encoding nuclear-targeted green fluorescent protein.