Introduction

Abstract Introduction Materials and Methods Results Discussion References

In vivo gene therapy, in which a potentially therapeutic gene is transferred to target cells in vivo, has the potential to aid in the local control of malignancy by creating high concentrations of cellular transgene products in tumor cells. The present study evaluated the hypothesis that tumor growth is suppressed by delivering to tumor cells genes encoding a cytotoxic protein, theoretically achieving a high concentration of the toxic product in the local milieu while avoiding systemic toxicity. As an example of a gene that might be used in this paradigm, we evaluated human perforin complementary DNA (cDNA) and delivered it with a replication-deficient adenovirus (Ad) vector. Perforin is a cytolytic pore-forming protein produced by natural killer (NK) cells and cytotoxic T lymphocytes (CTL), and is used by these cells to induce cytotoxicity of their target cells (1). The perforin protein binds to cell membranes in the presence of Ca²⁺ and polymerizes to construct lethal, doughnut-like channels with internal diameters of 5 to 20 nm (5, 6). The perforin-dependent cytotoxic pathway is used by NK cells and CTL to suppress infection and eliminate tumor cells (1, 7).

To confer control over the potentially toxic perforin transgene, we constructed a replication-deficient Ad vector containing a glucocorticoid-inducible promoter consisting of five glucocorticoid response elements (GRE) (10) and controlling perforin cDNA (AdGRE.PFP; Figure 1). The study data demonstrated that glucocorticoids will induce the expression of perforin cDNA in cells infected with the AdGRE.PFP vector, leading to suppression of tumor cell growth in vitro and in vivo.

View larger version (14K): [in this window] [in a new window]   Figure 1.   Schematic representation of the expression cassette containing a chimeric GRE promoter driving human perforin cDNA (GRE.PFP). The chimeric promoter represents five GRE (from the rat tyrosine aminotransferase gene) inserted upstream of the Ad2MLP (10). The human perforin cDNA is followed by the SV40 polyA stop signal.

	Materials and Methods

Abstract Introduction Materials and Methods Results Discussion References

Construction of Ad Vectors

The AdGRE.PFP E1, E3 Ad vector was constructed as previously described (11, 12). We prepared the plasmids used for recombination with the adenovirus 5 (Ad5) backbone by inserting the expression cassette containing the chimeric glucocorticoid-responsive promoter (referred to as the "GRE promoter") described by Mader and colleagues (10), utilizing five GREs from the rat tyrosine aminotransferase gene (13) in tandem with insertion of the adenovirus 2 major late promoter (Ad2MLP) TATA box/ initiation site and human perforin (PFP) cDNA (gift of K. Okumura, Juntendo University, Tokyo, Japan [14]), into pCMV.SV2⁺ (15) after removing the cytomegalovirus (CMV) promoter/enhancer from this plasmid (Figure 1). The AdGRE.PFP vector was generated by cotransfecting the plasmid and pJM17 Ad5-based backbone (16) into the 293 embryonic kidney cell line (HEK 293, CRL1573; American Type Culture Collection [ATCC], Rockville, MD) grown in improved Eagle's minimum essential medium (Biofluids, Rockville, MD) containing 10% fetal bovine serum (FBS), 2 mM glutamine, 50 U/ml penicillin G, and 50 µg/ml streptomycin. The AdGRE.CAT vector, constructed and produced in a similar fashion is identical to the AdGRE.PFP vector, but substitutes the chloramphenicol acetyltransferase (CAT) reporter gene (17) for perforin cDNA. The AdNull vector, containing the CMV promoter/enhancer, but no transgene in the expression cassette, was used as a negative control (18, 19). All Ad vectors were purified by cesium chloride density-gradient ultracentrifugation as previously described, and the titers of the virus stocks were determined by plaque-forming assay on 293 cells (11, 12, 20).

Cell Culture

The lung carcinoma cell line A549, obtained from ATCC (CCL185), was grown in minimal essential medium (GIBCO BRL, Grand Island, NY) with 10% FBS, 2 mM glutamine, 50 U/ml penicillin G, and 50 µg/ml streptomycin.

In Vitro Expression of Perforin cDNA

To evaluate the upregulation of expression of perforin cDNA in the GRE.PFP cassette transferred to A549 cells by the AdGRE.PFP vector, we infected A549 cells (90 min, 37°C) in serum-free medium with AdGRE.PFP (multiplicity of infection [MOI] 5 or 25). After 24 h, 25 nM dexamethasone was added or not added to the culture medium. To assess the upregulation of perforin at the mRNA level, total RNA (10 µg) was isolated through guanidine isothiocyanate phenol-chloroform extraction (21), separated on a 1% agarose gel containing 2.2 M formaldehyde, transferred to a nylon membrane (Schleicher & Schuell, Keene, NH), and hybridized with a ³²P-labeled perforin cDNA probe and with a ³²P-labeled glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA probe (22) as an internal control. Both probes were prepared by random priming, and the Northern blots were evaluated by autoradiography, with quantification by densitometry using a PhosphorImager (Molecular Dynamics, Sunnyvale, CA).

In Vitro Assessment of Cytotoxicity Induced by AdGRE.PFP

A549 cells (4 × 10³) were seeded in 96-well plates (Falcon 3072; Becton Dickinson, Lincoln Park, NJ), infected with AdGRE.PFP (10 moi) or AdNull (10 MOI) for 90 min, and then incubated alone or with dexamethasone (25 nM) for 4 d to induce the GRE promoter. Cytotoxicity to the AdGRE.PFP-infected cells was measured before and 1 to 4 d after vector infection, using a measure of cell viability (Alamar Blue; BioSource International, Camarillo, CA) according to the protocol provided by the manufacturer.

In Vivo Evaluation of AdGRE.PFP Suppression of Tumor Growth

To demonstrate that an Ad vector could transfer genes to tumor masses of A549 cells in vivo, and that a transgene in the expression cassette controlled by the GRE promoter could be upregulated by systemic administration of corticosteroid, we injected A549 cells (10⁷) subcutaneously into the flanks of athymic Balb/c nu/nu mice, and injected the AdGRE.CAT vector (5 × 10⁸ pfu in 50 µl phosphate-buffered saline [PBS]) into the tumor. The animals then received intraperitoneal injections of dexamethasone (50 µg) or PBS on three consecutive days. The tumors were then assessed for CAT activity (23). The CAT activity was standardized to total protein concentration.

To determine whether locally administered AdGRE.PFP activated by systemic administration of dexamethasone could suppress the growth of tumors in vivo, we injected A549 cells (10⁷) subcutaneously into the flanks of athymic Balb/c nu/nu mice. After 7 d, randomized tumors of comparable size were injected with AdGRE.PFP (5 × 10⁸ pfu in 50 µl PBS), AdNull (5 × 10⁸ pfu in 50 µl PBS), or PBS alone (50 µl) intratumorally. The animals then received intraperitoneal injections of dexamethasone (50 µg) on three consecutive days starting at Days 1 and 11. Tumor size was measured in a blinded fashion with calipers before and 9 to 29 d after vector administration, and was calculated as the product of width × length (24).

Statistical Analysis

Results are expressed as means ± SEM. Statistical comparisons were made with the unpaired two-tailed Student's t test.

	Results

Abstract Introduction Materials and Methods Results Discussion References

In Vitro Evaluation of the AdGRE.PFP Vector

Evaluation of the AdGRE.PFP vector in vitro confirmed that the GRE.PFP expression cassette functioned in vitro as expected (Figure 2). In this regard, a 2.0-kb mRNA transcript of human perforin cDNA was found in A549 cells infected with AdGRE.PFP (MOI 5 or 25) when the latter were incubated with 25 nM dexamethasone (Figure 2, lanes 3 and 4), but no perforin mRNA was detected without the addition of dexamethasone (lanes 1 and 2). Uninfected cells showed no expression of perforin transcripts with or without dexamethasone incubation (Figure 2, lanes 5 and 6). The mRNA level for GAPDH, used as an internal control, did not change with the addition of dexamethasone in cells either infected or uninfected with the AdGRE.PFP vector.

View larger version (41K): [in this window] [in a new window]   Figure 2.   Dexamethasone (Dex) stimulated expression of human perforin mRNA transcripts in A549 cells following infection with the AdGRE.PFP vector. Shown are Northern blot analyses of total RNA (10 µg per lane) isolated from A549 cells infected with Ad vectors. Cells were infected with AdGRE.PFP (MOI 5 or 25); after 24 h the cells were incubated without or with dexamethasone (25 nM) for an additional 24 h. The RNA was hybridized with a perforin cDNA probe (top) or, as a control, a GAPDH (bottom) probe. Lane 1: AdGRE.PFP (5 MOI) alone. Lane 2: AdGRE.PFP (25 MOI) alone. Lane 3: AdGRE.PFP (5 MOI) + dexamethasone. Lane 4: AdGRE.PFP (25 MOI) + dexamethasone. Lane 5: control, uninfected. Lane 6: control, uninfected but with dexamethasone. The sizes of the perforin mRNA (2.0 kb) and GAPDH mRNA (1.3 kb) are indicated.

In Vitro Cytotoxicity Function

In accord with the knowledge that the infection of A549 cells with AdGRE.PFP induced by dexamethasone resulted in expression of perforin transcripts, the growth of A549 cells infected with AdGRE.PFP was suppressed when the cells were exposed to dexamethasone (Figure 3). The growth of A549 cells infected with the control AdNull vector was similar to that of the naive cells whether the cells were incubated with or without dexamethasone (% survival on Day 4; P > 0.4, all comparisons). Cells treated with AdGRE.PFP with no added dexamethasone had a small decrease in growth compared with naive cells (% survival on Day 4, naive 100 ± 1%, versus AdGRE.PFP-infected, 91 ± 1%; P < 0.002), but this was minor compared with the suppression of growth of cells treated with AdGRE.-PFP with dexamethasone (Day 4, naive 100 ± 1%, versus AdGRE.PFP-infected + dexamethasone, 37 ± 1%; P < 0.0000001). Dexamethasone had no effect by itself on cell growth (Day 4, naive 100 ± 1%, versus naive + dexamethasone, 101 ± 1%; P > 0.3).

View larger version (17K): [in this window] [in a new window]   Figure 3.   In vitro evaluation of the effect of perforin expression in A549 cells. The human perforin cDNA driven by the GRE promoter was transferred to A549 cells in vitro with the AdGRE.PFP vector (10 MOI). After 90 min the cells were incubated with dexamethasone (Dex; 25 nM) for an additional 4 d. Cells were infected with AdNull (10 MOI) ± dexamethasone (25 nM) or given no treatment ± dexamethasone ("naive," 25 nM). The number and percentage of viable cells were assessed with the Alamar blue assay. Shown are data for AdGRE.PFP without dexamethasone (open squares); AdGRE.PFP with dexamethasone (closed squares); AdNull without dexamethasone (open triangles); AdNull with dexamethasone (closed triangles); naive cells without dexamethasone (open circles); naive cells with dexamethasone (closed circles). The data are presented as means ± SE of three independent experiments.

In Vivo Function

Infection of subcutaneous A549 tumors with the AdGRE.-CAT vector and the concomitant systemic administration of corticosteroids demonstrated that the chimeric GRE could be upregulated in vivo as it was in vitro. In this regard, CAT activity in the tumor mass was 32 ± 6 dpm/µg protein, but in animals receiving corticosteroids the CAT activity in the tumor mass was 8-fold greater, at 258 ± 64 dpm/µg protein (P < 0.03).

When the AdGRE.PFP vector was administered (with or without dexamethasone), no associated systemic toxicity (reflected by animal deaths) was observed either from the vector infection or from dexamethasone administration, nor were there differences among the groups in general alertness, state of the skin coat, food intake, or weight of the animals, or any changes in the liver, kidney, or lungs.

In accord with our findings in the in vitro studies with AdGRE.PFP and dexamethasone, administration of AdGRE.PFP to tumors of A549 cells in nude mice, together with systemic administration of dexamethasone, resulted in a suppression of tumor growth as compared with that in the control groups (Figure 4). In this regard, tumors treated with AdGRE.PFP without dexamethasone administration, tumors treated with the AdNull control virus with or without dexamethasone, and tumors injected with PBS alone or with dexamethasone alone all showed similar growth curves over the 29 d of evaluation (P > 0.1, all comparisons, Days 9 to 29). In contrast, tumor injected with the AdGRE.PFP vector in conjunction with systemic dexamethasone showed a marked suppression of growth (P < 0.02, Days 9 to 29) compared with those infected with AdGRE.PFP without dexamethasone treatment.

View larger version (22K): [in this window] [in a new window]   Figure 4.   In vivo growth of subcutaneous A549 tumors following local administration of AdGRE.PFP and intraperitoneal administration of dexamethasone (Dex). Tumors were established in Balb/c nu/nu mice by subcutaneous injection of 107 A549 cells. After 7 d the animals received AdGRE.PFP (5 × 108 pfu, intratumor) and intraperitoneal dexamethasone (closed squares) or PBS (open squares); AdNull (5 × 108 pfu, intratumor) and intraperitoneal dexamethasone (closed triangles) or PBS (open triangles); or PBS (intratumor) and intraperitoneal dexamethasone (closed circles) or PBS (open circles). The vectors or PBS were delivered intratumorally in a 50-µl volume on Day 0 (7 d after subcutaneous administration of tumor cells). After 24 h the animals received intraperitoneal injections of dexamethasone (50 µg) on three consecutive days starting at Days 1 and 11. The data are presented as means ± SE of three independent experiments.

	Discussion

Abstract Introduction Materials and Methods Results Discussion References

Perforin, a 534-amino acid glycoprotein with sequence homology to the membrane attack complex-forming complement component C9, is expressed primarily by NK cells and CTL (1, 7, 25, 26). Together with granzymes released by the same cell that produces it, perforin functions in a Ca²⁺-dependent pathway of NK-, and CTL-induced cytolysis of target cells (1, 7, 27). Perforin functions by creating pores in the plasma membrane (1, 30, 31). In analogy to C9, perforin can integrate into cell membranes and aggregate, forming polyperforin pores comprising 12 to 18 monomers of 5-20 nm internal diameter (5, 6). The present study capitalized on the cytotoxic function of perforin to inhibit tumor growth in vivo, through use of a replication-deficient, recombinant adenovirus to transfer perforin cDNA to tumor masses of human A549 lung carcinoma cells growing in the flanks of nude mice concomitantly receiving dexamethasone. Although infection with the AdGRE.PFP vector per se produced no antitumor activity, concomitant systemic administration of dexamethasone induced the chimeric corticosteroid-sensitive promoter in the AdGRE.PFP expression cassette to effectively suppress local tumor growth without systemic toxicity. These observations are of interest on several fronts. The technique investigated represents a new approach to using gene therapy to inhibit tumor growth, and demonstrates that a promoter regulated by a commonly used pharmacologic agent can be administered systemically in an in vivo gene transfer strategy to regulate a potentially toxic transgene in a local region. It also shows that when delivered intracellularly, perforin can be cytotoxic by itself, in the absence of granzymes or other lymphocyte-derived products.

Perforin-Induced Cytotoxicity

The mechanisms by which CTL induce cytotoxicity are complex, involving several pathways, but perforin clearly is a mediator of at least one of these cytotoxic pathways (1, 32, 33). It was originally proposed that perforin was the major mediator in the Ca²⁺-dependent CTL pathway, but it is now recognized that CTL use granzymes in parallel with perforin to evoke cytolysis of target cells (7, 33- 35). Furthermore, there are data showing that CTL can use both the Ca²⁺-dependent perforin/granzyme pathway and the Fas ligand/Fas-dependent pathway to induce target-cell death (1, 3, 4, 34).

In contrast to the requirement for perforin and granzymes, and possibly for perforin, granzymes, and Fas ligand/Fas for CTL-induced cytotoxicity, the data in the present study suggest that perforin can function independently to kill tumor cells if the perforin gene is transferred and expressed within the tumor cells. The mechanism(s) by which perforin gene transfer induces cytotoxicity is not known, but it is unlikely that anti-Ad vector host immune mechanisms play a significant role in it, because the AdGRE.PFP vector induced cytotoxicity both in vitro and in vivo in immunodeficient mice. The role of corticosteroids (used to induce the GRE promoter) working in concert with the perforin to induce cytotoxicity is difficult to assess, but the corticosteroids did not induce cytotoxicity alone or with a control vector, suggesting that they do not play a major role in the cytotoxic process. However, because the effect of the corticosteroids plus the AdGRE.PFP vector appears to be more extensive than would merely be due to the number of tumor cells transfected by the AdGRE.PFP vector, it is theoretically possible that the therapy works in part via a "bystander" effect on neighboring tumor cells.

Cytotoxic Protein Expression by Using an Inducible Promoter

The chimeric GRE promoter, designed by Mader and coworkers (10), takes advantage of the response of tandem GRE elements to the dose-dependent level of corticosteroids to which the cell is exposed. As shown by Mader and coworkers (10) in the context of plasmid in vitro transfection of HeLa cells, and in the present study in the context of an Ad vector transfection in vitro and in vivo, the chimeric GRE promoter produces little cell leakage in the absence of added corticosteroids, but marked upregulation of cytotoxicity when the promoter is exposed to corticosteroids. The present study capitalized on this regulable expression to achieve local expression of a toxic transgene without obvious systemic effects. Importantly, the inducible, chimeric GRE promoter can be upregulated with a commonly used group of pharmacologic agents that can be given for short periods without adverse effects either on individuals who have a wide variety of disorders or on normal individuals (36), and produces levels of upregulation comparable to other inducible promoters used in gene transfer applications (41).

Cytotoxic Proteins for Cancer Therapy

The major challenge to using cytotoxic proteins for the control of tumor growth is to direct a cytotoxic protein to a tumor and to achieve sufficient concentrations of the protein for a sufficient period to suppress tumor growth. To date, this challenge has been approached by using immunotoxin, consisting of tumor-specific monoclonal antibodies to which are linked the toxic protein, such as Diphtheria toxin, Pseudomonas exotoxin, ricin, gelonin, saporin, and pokeweed antiviral protein (51). The major limitations to applying these strategies to the treatment of solid tumors are the limited access of the immunotoxin to the tumor mass, lack of immunotoxin specificity, tumor cell heterogeneity, antigen shedding, breakdown or rapid clearance of the immunotoxin, and dose-limiting side effects (51). The in vivo gene transfer strategy demonstrated in the present study may be able to overcome some of these limitations, because it is not linked to antigen specificity, the Ad vector will potentially express itself for days to weeks in an immunocompetent host, and the expression is local, thus avoiding systemic toxicity (11, 12, 52). Whether in vivo gene transfer with a cytotoxic agent will be effective in helping control solid tumors will depend on the ability of the gene transfer vector to diffuse through the tumor mass and infect the majority of tumor cells. The use of in vivo gene therapy strategies is potentially limited to local control of disease, although such vectors may eventually be used to control systemic metastases by adding targeting ligands to the vector.