Heat-inducible TNF-������������ gene therapy combined withhyperthermia using magnetic nanoparticles as anovel tumor-targeted therapy

Akira Ito, Masashige Shinkai, Hiroyuki Honda, and Takeshi Kobayashi

Department of Biotechnology, Graduate School of Engineering, Nagoya University, Nagoya 464-8603, Japan.

Heat- induced therapeutic gene expression is highly desired for gene therapy to minimize side effects. Furthermore, if the gene

expression is triggered by heat stress, combined therapeutic effects of hyperthermia and gene therapy may be possible. We combined

TNF-� gene therapy driven by the stress - inducible promoter, gadd 153, with hyperthermia using magnetite cationic liposomes

(MCLs). In nude mice, MCLs induced cell death throughout much of the tumor area on heating under an alternating magnetic field.

This heat stress also resulted in a 3- fold increase in TNF-� gene expression driven by the gadd 153 promoter as compared with that

of nonheated tumor. TNF-� gene expression was also observed in the peripheral area where the hyperthermic effect was not enough

to cause cell death. The combined treatment strongly arrested tumor growth in nude mice over a 30-day period, suggesting potential

for cancer treatment. Cancer Gene Therapy (2001) 8, 649–654

Key words: Gene therapy; hyperthermia; stress - inducible promoter; gadd 153; magnetite; TNF-������������.

Control of therapeutic gene expression in tumors isimportant for gene therapy. Using an inducible

promoter to control therapeutic gene expression makesspatial and temporal targeting of therapy possible.1–3 Thisis critical for the use of potent, but cytotoxic, cytokinesthat are normally of limited practical utility due tosystemic toxicity concerns. An example of a cytokinewith potential therapeutic application is tumor necrosisfactor (TNF)-�, which exerts cytotoxic effects on a widerange of tumor cells by inducing apoptosis4 and which alsoactivates the immune system.5 However, its clinical use inhumans has been greatly limited by systemic toxicity.6 Inaddition to minimizing side effects, if therapeutic geneexpression can be induced by an exogenous stimulus suchas radiation or heat shock, a combined therapeutic effectcan be expected.

Hyperthermia is a promising approach for cancertherapy7 and has often been used in multimodalitystrategies as it can enhance the tumor killing effect ofchemotherapy,8 radiotherapy,9 and immunotherapy.10 Wehave investigated hyperthermia produced by means ofsubmicron magnetic particles. Magnetic particles generateheat under an alternating magnetic field (AMF) by

hysteresis loss.11 To promote adsorption and accumulationof magnetic particles in tumor cells, we developedmagnetite cationic liposomes (MCLs).12,13 When theMCLs were administrated to tumor tissue by localinjection and AMF was applied in an in vivo study,complete tumor regression was observed in 87.5% of therats.14

The gadd (growth arrest and DNA damage) 153 gene,a member of the gadd transcription factor family, wasoriginally identified as a DNA damaging reagent inhamster cell lines.15–17 Ropp et al18 reported that heatand oxidative stress could be related to this effect. Weconstructed a plasmid with the TNF-� gene under thecontrol of the gadd 153 promoter and demonstrated acombined therapeutic effect with heat stress in an in vitrostudy.19 Moreover, we showed that the gadd 153promoter was further activated by TNF-� gene expres-sion, causing an extremely high cytotoxic effect.20 Thismeans that the gadd 153-TNF-� system is effective notonly in killing tumor cells in heated regions of the tissue,but also in the surrounding areas.

In the present study, we have combined TNF-� genetherapy driven by a heat - inducible promoter with hyper-thermia produced by irradiation of MCLs. The MCLs anda plasmid containing the human TNF-� gene undercontrol of the gadd 153 promoter were injected into solidU251-SP human tumors formed subcutaneously in nudemice, and AMF was applied. This novel strategy has twotargeting functions. One is based on the magnetites,which are specifically heated under an AMF, and theother is based on the inducibility of the gadd 153promoter. We investigated the feasibility of this strategyin vivo.

Received May 30, 2001.

Address correspondence and reprint requests to Dr. Masashige Shinkai,

Ph.D., Department of Biotechnology, Graduate School of Engineering,

Nagoya University, Nagoya 464 - 8603, Japan. E-mail address:

shinkai@nubio.nagoya-u.ac.jp

This study was partially funded by a Grant - in -Aid for Scientific

Research on Priority Areas (Nos. 10145104 and 11227202) and a Grant -

in -Aid for JSPS Fellows (No. 12004238) from the Ministry of Education,

Science, Sports and Culture of Japan.

D2001 Nature Publishing Group 0929-1903/01/$17.00/+0www.nature.com/cgt

Cancer Gene Therapy, Vol 8, No 9, 2001: pp 649–654 649

MATERIALS AND METHODS

Cell line and establishment of tumors in nude mice

U251-SP human glioma cells were grown in minimumessential medium, supplemented with 10% fetal calf serum,10 mM nonessential amino acids, 0.1 �g/mL streptomycinsulfate, and 100 U/mL potassium penicillin G. The cellswere incubated at 378C in a humidified atmosphere of 5%CO2 and 95% air.

A total of 3�107 U251-SP cells were transplanted alongthe right flank of 4-week-old female athymic mice (JapanSLC, Hanamatsu, Japan). Tumor-bearing mice wererandomized for the studies when tumors reached 8�8 mmin size. These mice were housed in a special pathogen-freeanimal facility.

Plasmid construction and transfection

The plasmid JymCAT0 including the gadd 153 promoter waskindly provided by Prof. Nikki Holbrook.15 The plasmidSKhTNF� encoding human TNF-� gene was kindlyprovided by Dr. Hirohumi Hamada. In the plasmidpGadTNF,19 the human TNF-� gene is under the controlof the gadd 153 promoter.

A total of 20 �g of pGadTNF was introduced into 8�8mm tumors by the lipofection method, which uses N - (� -trimethylammonioacetyl ) -didodecyl-D-glutamate chloride(Sogo Pharmaceutical, Tokyo, Japan), dilauroylphosphati-dylcholine, and dioleoylphosphatidylethanolamine (SigmaChemical, St. Louis, MO) in a 1:2:2 molar ratio as describedpreviously.19

Preparation of MCLs and hyperthermia treatment

Magnetite (Fe3O4; average particle size: 10 nm), used as thecore of the MCLs, was the kind gift of Toda Kogyo(Hiroshima, Japan). The MCLs were prepared with colloidalmagnetite and a lipid mixture consisting of N -(� -trimethylammonioacetyl ) -didodecyl-D-glutamate chloride,dilauroylphosphatidylcholine, and dioleoylphosphatidyl -ethanolamine in a 1:2:2 molar ratio as previously de-scribed.12 All MCLs concentrations are expressed as the netmagnetite concentration.

One day after pGadTNF transfection, 0.2 mL of MCLssolution (net magnetite weight: 3 mg) was injected at thecenter of the tumor using a 25-gauge needle and an infusionpump (SP100i; World Precision Instruments, Sarasota, FL)for 30 minutes. One day later, mice were subjected tohyperthermia treatment. A magnetic field was created byusing a horizontal coil ( inner diameter: 7 cm; length: 7 cm)with a transistor inverter (LTG-100-05; Dai - ichi HighFrequency, Tokyo, Japan).12 Anesthetized mice were laidinside the coil such that the tumor region was at the center.The magnetic field frequency and intensity were 118 kHzand 30.6 kA/m (384 Oe), respectively. Tumor and rectaltemperatures were measured by optical fiber probe (FX-9020; Anritsu Meter, Tokyo, Japan).

TNF-� ELISA

For the TNF-� ELISA, tumor samples were homogenized inice-cold tris (hydroxymethyl ) aminomethane buffer (pH

7.6) containing the protease inhibitor phenylmethylsulfonylfluoride (1 mM), leupeptin (1 �g/mL), N - tosyl -L-phenyl-alanyl chloromethyl ketone (1 �g/mL), N� - tosyl -L- lysylchloromethyl ketone (1 �g/mL), and pepstatin (1 �g/mL).The homogenates were centrifuged at 14,000 rpm at 48C for10 minutes, 100 �L of the supernatants was subjected to thehumanTNF-�ELISA (Endogen,Woburn,MA) in duplicate,and the ELISA was performed according to the manufactur-er’s instructions. The standards were calibrated using WHOreference lot 87/650. One picogram of standards wasequivalent to one WHO picogram. The total protein contentof the tumor homogenates was determined by using the DCprotein assay reagent (Bio-Rad Laboratories, Hercules, CA).

To determine the human TNF-� concentration in theserum of the mice, 50 �L of serum was subjected to theultra-sensitive TNF-� ELISA (Biosource International,Camarillo, CA).

Determination of tumor volume

Tumor sizes were measured every 3 days. The volume wasdetermined by the following formula:21

tumor volume ¼ 0:5ðlength�width2Þ

where the unit of length and width is a millimeter.

Preparation of specimens for immunohistochemicalstaining

For immunohistochemical staining, the tumor was resectedand fixed in a 30% formalin solution at 24 hours afterhyperthermia treatment. The tumor tissues were sectionedlongitudinally at a 6-�m thickness. These sections weredeparaffinized and incubated with 10% skim milk at 378Cfor 10 minutes to block background staining. They were thenincubated at 378C for 60 minutes with mouse anti–rat IgG toTNF-� antibody (Santa Cruz Biotechnology, Santa Cruz,CA) and at 378C for 60 minutes with biotinylated goat anti–mouse IgG. They were subsequently incubated at 378C for30 minutes with peroxidase-conjugated streptavidin(DAKO, Kyoto, Japan). Each step was followed by washingwith phosphate buffer (50 mM, pH 7.4). Peroxidase activitywas visualized by treatment at room temperature for 10minutes with 0.02% diaminobenzidine tetrahydrochloridesolution containing 0.005% hydrogen peroxide. Thesesections were also stained with hematoxylin and eosin.

Statistical analysis

Levels of statistical significance in the gene expression andtumor growth experiments were evaluated using the Mann-Whitney rank sum test.22

RESULTS

MCL-induced hyperthermia

One day after MCL injection into the tumor, an AMF wasapplied to the whole body of the mouse. Figure 1 shows thetemperature of the tumor tissue and in the rectum duringirradiation with the AMF. Tumor temperature increased

Cancer Gene Therapy, Vol 8, No 9, 2001

650 ITO, SHINKAI, HONDA, ET AL: HEAT-INDUCIBLE TNF-� GENE THERAPY

rapidly to 468C in 3 minutes and was then maintained for 30minutes by control of the power of the AMF. When the AMFwas turned off, tumor temperature decreased to 368C within10 minutes. In contrast, the temperature in the rectum did notincrease over 388C during the irradiation. This suggests thatthe MCLs injected into the tumor were heated specifically bythe AMF, and that heat generation was controllable byadjusting the power of the AMF generator.

TNF-� gene expression in tumors

Both pGadTNF and MCLs were injected into tumors, andthe concentration of human TNF-� transcript was measuredin tumor homogenates following hyperthermia treatment(Fig 2). TNF-� gene expression driven by the gadd 153promoter was significantly induced by hyperthermia(P<.005): a 3-fold increase in expression was observed inheated tumors compared to nonheated tumors. Withouthyperthermia, the TNF-� concentration was 10–15 pg/mgtotal protein in the tumor homogenates. However, inhyperthermia- treated tumors, which were heated as shownin Figure 1, TNF-� concentration increased up to 45 pg/mgprotein on day 1 (one day after hyperthermia treatment ). Theinduction of TNF-� gene expression peaked on day 1 andgradually decreased until day 5. TNF-� was not detectablein mouse serum sampled one day after treatment or inhomogenates of tumors from control animals (hyperthermiaalone or no treatment).

Therapeutic effect of MCL-induced hyperthermiacombined with TNF-� gene therapy driven bythe gadd 153 promoter

Tumor growth after pGadTNF transfection is shown inFigure 3. A control group of five mice was injected with

empty liposomes and MCLs, but AMF was not applied.Tumor volume increased over 30 days in the control groupand the average tumor volume on day 32 was 6.6±2.2 cm3. Inthe case of the gene therapy group, which was injected withpGadTNF-containing liposomes and MCLs but not irradi-ated, tumor volume was significantly suppressed from day 17after treatment as compared with the control group (P<.05).Average tumor volume on day 32 was 2.9±1.1 cm3 (P<.01).For the hyperthermia group, both empty liposomes andMCLs were injected and AMF was applied. The time courseof tumor volume was similar to that observed for the genetherapy group. Tumor temperature increased up to 468C, asshown in Figure 1. Tumor volume was significantlysuppressed from day 14 after the treatment (P<.05) andwas 2.0±1.1 cm3 on day 32 (P<.005). For the combinedtherapy group, tumor volumes decreased slightly until day 5,were suppressed from day 8 (P<.05), and then were stronglyarrested for 30 days and the average tumor volume on day 32was 0.5±0.4 cm3 (P<.005), which was significantly smallerthan that of the hyperthermia (P<.01) or TNF-� genetreatments (P<.005).

Histological study

Histological observation showed that in tumors injected withempty liposomes andMCLs without hyperthermia treatment,the MCLs injected into the center of tumor remained in thetumor owing to their positive charges (Fig 4A). However, inthe tumors injected with pGadTNF-containing liposomesand MCLs and irradiated, the MCLs spread throughout the

Figure 2. Induction of TNF-� gene expression driven by the gadd153 promoter. Tumors were transfected with pGadTNF and treated

with ( ) or without (6 ) hyperthermia using MCLs. TNF-� gene

expression was determined in tumor homogenates from animalssacrificed at indicated times after hyperthermia treatment using the

human TNF-�–specific ELISA that does not detect murine TNF-�.

TNF-� concentration (pg/mg protein ) represents picogram TNF-� /

milligram of total protein in the tumor homogenates. Each pointrepresents the means±SD of five mice. In nontransfected tumors with

and without hyperthermia, no human TNF-� was detectable. *P<.05;

**P<.01; ***P<.005.

o

Figure 1. MCL- induced hyperthermia. MCLs were injected directly

into subcutaneous tumors of mice. One day after the MCL injection,

the mice were irradiated with an AMF for 30 minutes. Tumor andrectal temperatures were measured by optical fiber probes. : tumor,

6: rectum. Each point represents the means±SD of five mice.o

Cancer Gene Therapy, Vol 8, No 9, 2001

ITO, SHINKAI, HONDA, ET AL: HEAT-INDUCIBLE TNF-� GENE THERAPY 651

tumor and a large necrotic area was also observed as a resultof hyperthermia (Fig 4B). The area of necrotic cell deathwas about 80% of the area of the tumor affected byhyperthermia. Immunohistochemical observation revealedthat the cells in peripheral areas of the tumor survivedhyperthermia and expressed TNF-� strongly (Fig 4D).TNF-� was not observed in tumors that had not beentransfected with pGadTNF, whether they were irradiated(data not shown) or untreated (Fig 4C).

DISCUSSION

We have demonstrated the feasibility of a combined TNF-�gene therapy and hyperthermia strategy. TNF-� has agrowth inhibition effect on some human glioma cells,although other glioma cells are resistant. We previouslydemonstrated that U251-SP cells showed a resistance toTNF-� in vitro.19 However, when pGadTNF transfectionwas combined with hyperthermia, a synergistic cell killingeffect was observed. In the present in vivo study, TNF-�expression alone was not able to promote strong tumorgrowth arrest (Fig 3). However, when the transfected tumorswere heated, TNF-� gene expression was induced (Fig 2)

and a synergistic effect of TNF-� and hyperthermia wasobserved (Fig 3). These results suggest that the combinationtherapy can arrest tumor growth strongly compared withhyperthermia or TNF-� gene therapy alone.

Although the MCLs could induce cell death throughoutmost of a treated tumor by heating under AMF (Fig 4B),tumors did not decrease in size (Fig 3). However, the heatstress increased TNF-� gene expression from the gadd 153promoter even at the periphery of the tumor, enhancing thetherapeutic effect. This TNF-� gene expression areacoincided with the area where the hyperthermic effect wasnot enough to cause cell death. The combined therapypromoted cell killing throughout the tumor and effectivelyarrested tumor growth over 30 days as shown in Figure 3.TNF-� gene expression in the peripheral area can beimportant for the treatment of malignant glioma character-ized by neovascular formation, because TNF-� damages thetumor vasculature.

In TNF-� gene therapy, several groups have reported thatintratumorally injected adenovirus can leak into systemiccirculation and induce abnormally high levels of cytokines,which leads to serious side effects.23,24 Therefore, TNF-�gene therapy in practice is also limited by systemic toxicityin some cases. One of the typical side effects of TNF-� is

Figure 3. U251-SP tumor volumes in athymic mice treated with pGadTNF and hyperthermia using MCLs.6; Control group: tumors were injected

with empty liposomes in PBS on day 0 and with MCLs on day 1, AMF was not applied. ~; Gene therapy group: tumors were transfected with

pGadTNF on day 0 and injected with MCLs on day 1, but without an AMF irradiation. &; Hyperthermia group: tumors were injected with empty

liposomes in PBS on day 0 and with MCLs on day 1 and an AMF was applied on day 2. ; Combined therapy group: tumors were transfected withpGadTNF on day 0 and injected with MCLs on day 1 and an AMF was applied on day 2. Five mice in each group were used for monitoring tumor

growth. The experiment was performed at least twice. Inset figures represent a smaller scale used in the first few days after treatment. *P<.05;

**P<.01; ***P<.005.

o

Cancer Gene Therapy, Vol 8, No 9, 2001

652 ITO, SHINKAI, HONDA, ET AL: HEAT-INDUCIBLE TNF-� GENE THERAPY

hepatocellular degeneration. In the present study, however,no TNF-� was detectable in the systemic circulation. Thismeans that there is no measurable systemic release of thecytokine, thus limiting the side effect of TNF-�. Moreover,no liver damage was detected in any of the mice on day 32after transfection with pGadTNF (data not shown). Mizunoet al 25 reported that conjugation of antibody to cationicliposomes enhanced gene expression in targeted cells. In thepresent study, we have used MCLs that are designed fordirect injection into tumors, but we have already succeededin the preparation of an antibody-conjugated magneto-liposome for more specific targeting and demonstrated itsaccumulation in tumors in vivo.26 In either case, magnetitesare heated only under AMF irradiation. Therefore, eventhough MCLs can leak into systemic circulation, heat is onlygenerated where the AMF is applied. In our strategy,therapeutic gene expression only occurs where the AMF isapplied.

This combined treatment arrested tumor growth stronglyover 30 days and there was no measurable systemic release

of TNF-�. We conclude that a powerful and safe treatmentfor tumors is possible using this novel strategy.

ACKNOWLEDGMENTS

We are grateful to Nikki Holbrook (National Institute onAging, Baltimore, MD) for providing the JymCAT0 plasmidincluding the gadd 153 promoter and Hirohumi Hamada(Cancer Chemotherapy Center, Japanese Foundation forCancer Research) for providing the plasmid SKhTNF-�including the hTNF-� gene.

REFERENCES

1. Gerner EW, Hersh EM, Pennington M, et al. Heat - induciblevectors for use in gene therapy. Int J Hyperthermia.2000;16:171–181.

Figure 4. Representative sections of U251-SP tumor following treatment with pGadTNF transfection and MCL- induced hyperthermia. Tumor

specimens were prepared at 24 hours after hyperthermia treatment. A and B show hematoxylin and eosin stains of the central part of the tumor.

Arrows indicate MCLs. A: Tumor injected with empty liposomes in PBS and MCLs without hyperthermia treatment. B: Tumor transfected with

pGadTNF and injected MCLs with hyperthermia treatment. C and D show immunohistochemical staining of tumor peripheral areas with anti–human TNF-� antibody. Brown-stained areas indicate TNF-�. C: Tumor injected with empty liposomes in PBS and MCLs without hyperthermia

treatment. D: Tumor injected with pGadTNF and MCLs with hyperthermia treatment. Staining corresponding to human TNF-� expression was

only observed in specimens from pGadTNF- transfected tumors treated with hyperthermia.

Cancer Gene Therapy, Vol 8, No 9, 2001

ITO, SHINKAI, HONDA, ET AL: HEAT-INDUCIBLE TNF-� GENE THERAPY 653

2.Walther W, Stein U, Fichtner I, et al. mdr1 promoter-driventumor necrosis factor-� expression for a chemotherapy-controllable combined in vivo gene therapy and chemotherapyof tumors. Cancer Gene Ther. 2000;7:893–900.

3. Manome Y, Kunieda T, Wen PK, et al. Transgene expression inmalignant glioma using a replication -defective adenoviralvector containing the EGR-1 promoter: activation by ionizingradiation or uptake of radioactive iododeoxyuridine. Hum GeneTher. 1998;9:1409–1417.

4. Bellomo G, Perotti M, Taddei F, et al. Tumor necrosis factor �induces apoptosis in mammary adenocarcinoma cells by anincrease in intranuclear free Ca2+ concentration and DNAfragmentation. Cancer Res. 1992;52:1342–1346.

5. Palladino MA, Shalaby MR, Kramer SM, et al. Characteriza-tion of the antitumor activities of human tumor necrosis factor-� and the comparison with other cytokines: induction of tumorspecific immunity. J Immunol. 1987;138:4023–4032.

6. Spriggs DR, Yates SW. Cancer chemotherapy: experience withTNF administration in humans. In: Beutler B, ed. TumorNecrosis Factors: the Molecules and Their Emerging Roles inMedicine. New York: Raven Press; 1992:383–406.

7. Kobayashi T. Hyperthermia for brain tumors. Jpn J Hyperther-mic Oncol. 1993;9:245–249.

8. Raaphorst G, Yang H, Wilkins D, et al. Cisplatin,hyperthermia and radiation treatment in human cisplatin -sensitive and resistant glioma cell lines. Int J Hyperthermia.1996;12:801–812.

9. Montes H, Hynynen K. A system for the simultaneous deliveryof intraoperative radiation and ultrasound hyperthermia. Int JHyperthermia. 1995;11:109–119.

10. Watanabe N, Niitsu Y, Umeno H, et al. Synergistic cytotoxicand antitumor effects of recombinant human tumor necrosisfactor and hyperthermia. Cancer Res. 1988;48:650–653.

11. Shinkai M, Matsui M, Kobayashi T. Heat properties ofmagnetoliposomes for local hyperthermia. Jpn J HyperthermicOncol. 1994;10:168–177.

12. Shinkai M, Yanase M, Honda H, et al. Intracellularhyperthermia for cancer using magnetite cationic liposomes:in vitro study. Jpn J Cancer Res. 1996;87:1179–1183.

13. Yanase M, Shinkai M, Honda H, et al. Intracellularhyperthermia for cancer using magnetite cationic liposomes:ex vivo study. Jpn J Cancer Res. 1997;88:630–632.

14. Yanase M, Shinkai M, Honda H, et al. Intracellular

hyperthermia for cancer using magnetite cationic liposomes:an in vivo study. Jpn J Cancer Res. 1998;89:463–469.

15. Luethy JD, Fargnoli J, Park JS, et al. Isolation andcharacterization of the hamster gadd153 gene. J Biol Chem.1990;265:16521–16526.

16. Luethy JD, Holbrook NK. Activation of the gadd153 promoterby genotoxic agents: a rapid and specific response to DNAdamage. Cancer Res. 1992;52:5–10.

17. Delmastro DA, Li J, Vaisman A, et al. DNA damage inducible -gene expression following platinum treatment in humanovarian carcinoma cell lines. Cancer Chemother Pharmacol.1997;39:245–253.

18. Ropp M, Courgeon AM, Calvayrac R, et al. The possible roleof the superoxide ion in the induction of heat - shock andspecific proteins in aerobic Drosophila cells during return tonormoxia after a period of anaerobiosis. Can J Biochem CellBiol. 1983;61:456–461.

19. Ito A, Shinkai M, Bouhon IA, et al. Induction of TNF-� geneexpression by heat inducible promoter gadd 153. Jpn JHyperthermic Oncol. 2000;16:91–98.

20. Ito A, Shinkai M, Bouhon IA, et al. Bystander-killing effectand cyclic induction of TNF-� gene under heat induciblepromoter gadd153. J Biosci Bioeng. 2000;90:437–441.

21. Yoshida J, Mizuno M, Yagi K. Secretion of human � -interferon into the cystic fluid of glioma transfected with theinterferon gene. J Clin Biochem Nutr. 1991;11:123–128.

22. Montgomery D. Design and Analysis of Experiments, 3rd edn.New York: John Wiley and Sons; 1991.

23. Bramson JL, Hitt M, Gauldie J, et al. Pre -existing immunity toadenovirus does not prevent tumor regression followingintratumoral administration of a vector expressing IL-12 butinhibits virus dissemination. Gene Ther. 1997;4:1069–1076.

24. Zhang R, Straus FH, DeGroot LJ. Effective genetic therapy ofestablished medullary thyroid carcinomas with murine inter-leukin -2: dissemination and cytotoxicity studies in a rat tumormodel. Endocrinology. 1999;140:2152–2158.

25. Mizuno M, Yoshida J, Sugita K, et al. Growth inhibition ofglioma cells transfected with the human � - interferon gene byliposomes coupled with a monoclonal antibody. Cancer Res.1990;50:7826–7829.

26. Le B, Shinkai M, Kitade T, et al. Preparation of tumor-specificmagnetoliposomes and their application for hyperthermia. JChem Eng Jpn. 2000;33:821–826.

Cancer Gene Therapy, Vol 8, No 9, 2001

654 ITO, SHINKAI, HONDA, ET AL: HEAT-INDUCIBLE TNF-� GENE THERAPY