Gene therapy strategies for cancer currently aim at targeting gene delivery to the malignant cell. In a mouse model of intracerebral Lewis lung carcinoma (3LL), adenoviral vectors transduce not only 3LL cells but also brain parenchymal cells including endothelial cells, neurons,microglia, and astrocytes∼ in vivo∼. Furthermore,transgene expression persists longer in brain than in tumor. Transfer of IFN-γ into brain parenchymal cells rather than tumor is both necessary and sufficient to generate antitumor therapeutic benefits. Therefore, parenchymal cells represent an effective and necessary target for delivery of genes that render the brain uninhabitable by the tumor.

Current experimental gene therapy strategies for malignant brain tumors aim at targeting delivery of the transgene to the tumor cell. The genes transferred include those that induce a suicide effect,modulate the immune response, arrest the cell cycle, induce apoptosis,or inhibit neovascularization (1, 2, 3, 4, 5, 6, 7, 8, 9, 10). Historically,delivering the suicide gene herpes simplex thymidine kinase into brain tumors was the first gene therapy strategy for brain cancer to show efficacy in animal models, and this therapy has been tested in humans(2, 11). Because of the inability of retroviral vectors to infect quiescent cells, Culver et al.(2)hypothesized that recombinant retroviruses can target suicide gene delivery into rapidly multiplying tumor cells while sparing the background of nondividing neural tissue. Data from this laboratory showed that treatment of mice harboring a poorly immunogenic carcinoma(3LL) implanted in the brain with an adenoviral vector to deliver IFN-γ (AdIFN) at the same coordinates as the tumor elicits significant prolongation of survival times and tumor rejection. In this mouse brain tumor model, the antitumor effects of AdIFN are independent of adaptive cellular immune mechanisms and appear to be mediated by antiangiogenesis (12). Evidence against a cellular immune-mediated response include: (a) absence of a memory immune response on challenge; (b) lack of antitumor effects at sites distal to inoculation of AdIFN; and (c)preservation of the therapeutic benefits in scid, beige,iNOS C57/BL6 knockout mice and mice treated with NG-nitro-l-arginine-methyl ester. High concentrations of IFN-γ do not inhibit tumor growth in vitro, making it unlikely that the antitumor effect of this treatment acts directly on the growth of the tumor cell. However, gene transfer of IFN-γ inhibits neovascularization of the tumor in the avascular s.c. space in a Matrigel assay in vivo, and AdIFN induces apoptosis of endothelial cells in vivo, thus supporting the idea that AdIFN represses tumor growth by inhibiting angiogenesis (12).

The experiments in this study were designed to determine:(a) the identity of the brain parenchymal cells transduced with AdIFN; and (b) whether the antitumor effects are generated by delivery of IFN-γ into the tumor cell versusbrain parenchyma.

Animals and Vectors.

C57/BL6 mice were purchased from the Jackson Laboratories (Bar Harbor,ME). Animal care was in accordance with institutional guidelines. 3LL(Lewis lung carcinoma; C57/BL6) is a generous gift from Dr. I. Fidler(M. D. Anderson Cancer Center, Houston, TX). The generation of adenoviral vectors AdIFN and AdBGAL was described elsewhere(12). The titers of AdIFN and AdBGAL of viral particle:plaque-forming unit ratios are 43 and 81, respectively. Mice were anesthetized and injected intracerebrally as described previously(12). Statistical calculations were performed by the JMP software (SAS Institute, Cary, NC).

Tissue Staining.

Antibodies against mouse IFN-γ (American Type Culture Collection,Manassas, VA) and tomato lectin (Sigma Chemical Co., St. Louis, MO),anti-GFAP (DAKO, Carpenteria, CA), anti-NeuN (Chemicon, Temecula, CA),and anti-factor VIII (DAKO) were used. Biotinylation reactions were performed following the manufacturer’s specifications (Vector Laboratories, Burlingame, CA). Immunofluorescence and immunohistochemistry were performed as described previously(12).

To identify the cell types transduced with AdIFN in vivo, groups of mice that received intracerebrally 3LL implants were injected 10 days later with AdIFN at the same coordinates as the tumor, and their brains were sectioned 4 days later. Immunofluorescence staining for astrocytes (green) and IFN-γ (red)reveals IFN-γ reactivity predominantly in the brain, outside and surrounding the area of the tumor (Fig. 1). Furthermore, double immunofluorescence staining demonstrates that astrocytes (Fig. 2, a–c), microglia (Fig. 2, d–f), neurons (Fig. 2, g–i),and endothelial cells (Fig. 2, j–l) are all transduced with AdIFN in vivo.

To examine whether selective gene transfer into brain parenchymal cells is sufficient to generate the antitumor effects of AdIFN, mice first received implants of either AdIFN or AdBGAL and were then reinjected 4 days later with wt43LL at the same coordinates and followed for survival. The therapeutic benefits of AdIFN in these experiments were identical to the effects of AdIFN in the treatment of mice with 3LL brain tumors (Fig. 3, a and b); AdIFN generated statistically significant prolongation of survival times as well as tumor rejection in 4 of 10 mice (Fig. 3,b). The latter survived for >85 days, and histological analysis of their brains showed cavity formation (data not shown). To study the biological effects of selective gene transfer of IFN-γ into tumor cells, 3LL cells transduced in vitro to secrete 1.125 μg IFN-γ/106 cells/24 h were implanted in mice, and the mice were followed for survival. The survival benefits in these experiments were modest but statistically significant; however, none of these mice survived longer than 36 days (Fig. 3 c).

To determine the persistence of transgene expression in brain and tumor, groups of mice (n = 6 each)received implants of: (a) 3LL cells transduced with AdBGAL in vitro; or (b) AdBGAL followed 4 days later with wt 3LL cells at the same coordinates. AdBGAL-transduced tumor cells instead of AdIFN transduced tumor cells were used because IFN-γsecreted by the tumor induces microglia to secrete endogenous IFN-γ,thus obscuring the localization of the virus. The brains were sectioned 10 days after the first injection. Although all 3LL cells transduced with AdBGAL in vitro contained β-gal activity at day 0(Fig. 4,a), none of the brains implanted with these tumors stained for β-gal at day 10 (Fig. 4, b and c). Nonetheless, mice first injected with AdBGAL showed prominent β-gal expression in brain parenchyma surrounding the tumor (Fig. 4, d and e).

The results show that AdIFN transduces not only tumor cells but also brain parenchymal cells including endothelial cells, microglia,neurons, and astrocytes in vivo. Although implantation of 3LL cells transduced in vitro into the brain is associated with modest survival benefits (Fig. 3,c), it fails to duplicate the therapeutic effects of AdIFN in treating 3LL brain tumors(Fig. 3,a), making it unlikely that the antitumor effects of AdIFN in the latter model are mediated solely by transduction of the tumor cell. Thus, the data argue that gene transfer into brain parenchyma is necessary to optimize the therapeutic benefits. Furthermore, selective transfer of IFN-γ into brain parenchyma reproduces the survival benefits of AdIFN in treating 3LL brain tumors(Fig. 3,b), suggesting that brain transduction alone is sufficient to generate the therapeutic response (Fig. 3,a). The gains of gene transfer into brain parenchymal cells may stem from the fact that unlike rapidly multiplying tumor cells, they are less likely to lose the transgene and are more likely to produce sustained high amounts of the gene product for longer periods of time (Fig. 4).

Targeting gene transfer into brain parenchymal cells located in proximity of the tumor is an attractive adenoviral-mediated strategy for delivering molecules that are either antiangiogenic, exert immunomodulatory effects on both the tumor and immune cells, inhibit tumor cell growth directly, or induce selective tumor cell apoptosis. Tumor cells constitute an unstable source for transgene production because they multiply rapidly, thus diluting out the episomal transgene(Fig. 4). Furthermore, by killing a large number of malignant cells, a potentially successful therapeutic strategy may fail because of prematurely reducing or eliminating transgene production by the tumor. Because brain parenchymal cells are nondividing or replicate slowly,they make up a reliable and potentially controllable factory for producing a gene product cloned under an inducible promoter (Fig. 4). Furthermore, such a strategy is particularly attractive for gliomas because they tend to recur within a few centimeters of the original tumor site. In summary, the results present a proof of principle that the tumor bed, in this instance, brain parenchymal cells, is an effective and necessary target for delivering genes that render the brain uninhabitable by the tumor.

We are indebted to Roger Rosenberg for support and helpful discussions.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

      
1

Supported by NIH Grants R01-CA81367 and R29-CA78825 and by the 1998 Scholar Award from the Children’s Brain Tumor Foundation of The Southwest.

                  
4

The abbreviations used are: wt, wild-type;β-gal, β-galactosidase; o.m., original magnification; GFAP, glial fibrillary acidic protein.

1
Fathallah-Shaykh H. M., McIntire D. D. Brain tumors in the elderly: undefeated and gaining grounds.
Arch. Neurol.
,
55
:
905
-906,  
1998
.
2
Culver K. W., Ram Z., Wallbridge S., Ishii H., Oldfield E. H., Blaese R. M. In vivo gene transfer with retroviral vector-producing cells for treatment of experimental brain tumors.
Science (Washington DC)
,
256
:
1550
-1552,  
1992
.
3
Folkman J. Antiangiogenic gene therapy.
Proc. Natl. Acad. Sci. USA
,
95
:
9064
-9066,  
1998
.
4
Saleh M., Stacker S. A., Wilks A. F. Inhibition of growth of C6 glioma cells in vivo by expression of antisense vascular endothelial growth factor sequence.
Cancer Res.
,
56
:
393
-401,  
1996
.
5
Millauer B., Shawver L. K., Plate K. H., Risau W., Ullrich A. Glioblastoma growth inhibited in vivo by a dominant-negative Flk-1 mutant.
Nature (Lond.)
,
367
:
576
-579,  
1994
.
6
Fueyo J., Gomez-Manzano C., Yung W. K. A., Liu T. J., Alemany R., McDonnell T. J., Shi X., Rao J. S., Levin V. A., Kyritsis A. P. Overexpression of E2F-1 in glioma triggers apoptosis and suppresses tumor growth in vitro and in vivo.
Nat. Med.
,
4
:
685
-690,  
1998
.
7
Chen J., Willinghan T., Margraf L. R., Schreiber-Agus N., Dephinho R. A., Nisen P. D. Effects of the MYC oncogene antagonist, MAD, on proliferation, cell cycling and the malignant phenotype of human malignant brain tumour cell.
Nat. Med.
,
1
:
638
-643,  
1995
.
8
Kondo S., Kondo Y., Li G., Silverman R. H., Cowell J. K. Targeted therapy of human malignant glioma in a mouse model by 2-5A antisense directed against telomerase RNA.
Oncogene
,
16
:
3323
-3330,  
1998
.
9
Hiraga S., Ohnishi T., Izumoto S., Miyahara E., Kanemura Y., Matsumura H., Arita N. Telomerase activity and alterations in telomere length in human brain tumors.
Cancer Res.
,
58
:
2117
-2125,  
1998
.
10
Yu J. S., Sena-Esteves M., Paulus W., Breakefield X. O., Reeves S. A. Retroviral delivery and tetracycline-dependent expression of IL-1 β-converting enzyme (ICE) in a rat glioma model provides controlled induction of apoptotic death in tumor cells.
Cancer Res.
,
56
:
5423
-5427,  
1996
.
11
Ram Z., Culver K. W., Oshiro E. M., Viola J. J., DeVroom H. L., Otto E., Long Z., Chiang Y., McGarrity G. J., Muul L. M., Katz D., Blaese M., Oldfield E. H. Therapy of malignant brain tumors by intratumoral implantation of retrovirus-producing cells.
Nat. Med.
,
3
:
1354
-1361,  
1997
.
12
Fathallah-Shaykh H. M., Zhao L. J., Kafrouni A. I., Smith G. M., Forman J. Gene transfer of IFN-γ into established brain tumors represses growth by antiangiogenesis.
J Immunol.
,
1
:
1
-2,  
2000
.