Eradication of Glioblastoma, and Breast and Colon Carcinoma Xenografts by Hsp70 Depletion¹

The major stress-inducible Hsp70³ (also known as Hsp72 or iHsp70) is an antiapoptotic chaperone protein expressed abundantly and preferentially in human tumors and tumor cell lines (1, 2, 3). The role of Hsp70 in tumorigenesis is supported by experimental data showing that it enhances the tumorigenic potential of rodent cells in vivo (4, 5, 6, 7). Furthermore, its high expression in various human tumors correlates with therapy resistance and poor prognosis (1). The tumorigenic potential of Hsp70 has been suggested to depend on its ability to transform cells and/or its antiapoptotic function (4, 6). The latter hypothesis is supported by data showing that depletion of Hsp70 by antisense technology induces massive apoptosis-like death in tumorigenic cells but not in nontumorigenic epithelial cells or embryonic fibroblasts (8, 9). Despite the apoptosis-like morphology as judged by electron microscopy, the antisense Hsp70-induced death of breast cancer cells is independent of known caspases. Furthermore, Bcl-2 and Bcl-X_L, which protect tumor cells from most forms of programmed cell death, fail to rescue cancer cells from death induced by Hsp70 depletion. Because Bcl-2/Bcl-X_L expression as well as other alterations resulting in defective caspase activation are common in cancer cells and because many cancer cells can even survive the activation of caspases, Hsp70 depletion may provide a new option to combat tumors resistant to therapies dependent on classical mitochondrion- and caspase-mediated death pathways (10, 11, 12).

All of the cells were propagated in RPMI 1640 with glutamax (Life Technologies, Inc.) supplemented with 10% FCS (Biological Industries) and antibiotics. The medium for MCF-10A cells was additionally supplemented with 100 ng/ml of choleratoxin, 20 ng/ml of epidermal growth factor, 500 ng/ml of hydrocortisone, and 100 ng/ml of insulin.

Ad.asHsp70 and Ad.β-gal are adenoviral shuttle vectors carrying bases 475–796 of the published human Hsp70 sequence in antisense orientation or the β-gal gene, respectively (9). Ad.ΔE1 is an empty adenoviral vector lacking the insert. The infections were performed as described (9).

Before implantation, the tumor cells were washed twice with PBS, counted, and resuspended in PBS. MDA-MB-468 cells (1 × 10⁷ cells in 100 μl PBS) were inoculated into a mammary fat pad of female FOX CHASE SCID mice, and LoVo-36 cells (3 × 10⁶ cells in 100 μl PBS) were inoculated s.c. into the right flank of athymic BALB/cA nu/nu (nude) mice. Breast tumors were treated intratumorally on days 7, 10, and 12, and s.c. colon carcinomas on days 4 and 6 after implantation with 3 × 10⁸ pfu and 5.4 × 10⁹ pfu of virus in 100 μl PBS, respectively. For intracranial implantation, athymic CD1 nu/nu mice (Charles River, Sulzfeld, Germany) were anesthetized and placed in a stereotactic fixation device (Stoelting). A burr hole was drilled into the skull 2 mm lateral to the bregma. The needle of a Hamilton syringe was introduced to a depth of 3 mm. Five × 10⁴ U373MG cells in 4 μl of PBS were injected into the right striatum. At day 7 after implantation, indicated amounts of virus in 8 μl of 10% glycerol in PBS or vehicle alone were injected intralesionally. The mice of the first set of experiments were observed daily and sacrificed by an overdose of anesthetic when the first one developed adverse neurological symptoms. In the second set of experiments, Kaplan-Meier curves were generated. All of the animal work was carried out in accordance with the NIH guidelines.

The MTT reduction assay was used to analyze the survival of the cells (13). Apoptotic nuclear changes were detected by a TUNEL kit (Roche). zVAD-fmk was from Bachem.

The primary antibodies used included mouse monoclonal antibodies against Hsp70 (2H9), Hsc70 (N69; both kindly provided by Boris Margulis, Russian Academy of Sciences, St. Petersburg, Russia), GAPDH (Biogenesis), and ED1 (Serotec). Immunodetection of proteins (20 μg/lane) separated by 8% SDS-PAGE and transferred to nitrocellulose was performed using ECL Western blotting reagents (Amersham) indicated primary antibodies and appropriate peroxidaseconjugated secondary antibodies from Dako.

Cryostat sections (16 μm) of tumors were cut, air-dried, and stored at −20°C. For the assessment of tumor volume, sections were stained with H&E and analyzed by MCID software (Imaging Research Inc.). For immunohistochemistry slides were incubated with acetone for 10 min, air-dried, incubated with 1% H₂O₂ for 30 min, washed with PBS, and blocked with BSA (10 mg/10 ml PBS) for 10 min. Primary antibodies were incubated overnight. Bound primary antibody was labeled with biotinylated antimouse IgG antibodies, washed, incubated with avidin-biotin reagent (Vectastain Elite ABC peroxidase; Vector, Wiesbaden, Germany), and developed in diaminobenzidine (Sigma Chemical Co., St. Louis, MO).

Human monocytes were isolated by elutriation, differentiated to macrophages by 8–12-day culture on 48-well microtiter plates, and stained for 1 h with DiI (Molecular Probes; 2.5 mg/ml) to visualize cell bodies (14). Tumor cells infected with Ad.asHsp70 or Ad.β-gal were stained with Fast Blue (Sigma; 1 μg/ml) and added to macrophages at a ratio of 5:1 for 2 h at 37°C. The phagocytosis index was scored by counting ingested cells per phagocyte as described (14). For the analysis of macrophage activation, macrophages were incubated with or without LPS (Sigma) in the presence of cell corpses. TNF concentration in the supernatant was determined by ELISA (Pierce, Rockford, IL).

The significance of the observed effects was evaluated by t test. Kaplan-Meier curves were drawn as detailed (15).

We have earlier shown that partial depletion of Hsp70 by Ad.asHsp70 induces apoptosis-like cell death in MDA-MB-468, MCF-7, BT-549, and SK-BR-3 human breast cancer cell lines (9). To test whether the cytotoxic effect of Ad.asHsp70 was specific for cancers of mammary origin, we infected human cancer cell lines originating from glioblastoma (U373MG and LN-18), colon carcinoma (LoVo-36), prostate carcinoma (PC-3 and DU145), or hepatocellular carcinoma (HUH-7) with Ad.asHsp70 at a multiplicity of infection (250–1500) required to achieve a 100% infection. Akin to breast cancer cells, tumor cells of other origins also responded to Ad.asHsp70 treatment with reduction in Hsp70 protein level, shrinkage, detachment, and a loss of viability 3–5 days after the infection (Fig. 1; data not shown). Similar infections with control viruses resulted in some cases in a slight growth inhibition (maximum 20% over 4 days compared with noninfected logarithmically growing cells) but not in any detectable morphological changes or loss of viability. Importantly, Ad.asHsp70 had no effect on the survival of nontumorigenic breast- or prostate-derived epithelial cells (Fig. 1 c).

To ensure that Hsp70 is not merely needed for tumor cell growth in culture conditions, we next investigated whether Hsp70 expression is also required for the in vivo growth of tumors. MDA-MB-468 breast carcinoma cells left untreated or infected with Ad.β-gal or Ad.asHsp70 were inoculated into the mammary fat pad of nude mice 24 h after the infection. Whereas all of the mice injected with noninfected cells or cells infected with Ad.β-gal developed aggressively growing tumors, none of the mice injected with Ad.asHsp70-infected cells showed progressive tumor growth during the observation period of 2 months (Fig. 2 a).

The dramatic anticancer effects of Ad.asHsp70 observed in vitro may have great clinical relevance. Therefore, we next tested the effect of Ad.asHsp70 on the growth of progressively growing tumors in vivo. For this purpose we chose three tumor xenograft models in immunodeficient mice: MDA-MB-468 breast carcinoma growing in mammary fat pad, s.c. LoVo-36 colon carcinoma, and U373MG glioblastoma inoculated into the striatum. Remarkably, Ad.asHsp70 treatment induced a dramatic reduction in tumor volumes in all of the in vivo model systems (Fig. 2, b and c; Fig. 3,a). Furthermore, 28 days after the inoculation of glioblastoma cells, 83% of control animals presented with neurological symptoms, whereas only 1 of 17 animals treated with Ad.asHsp70 showed mild symptoms (Fig. 3,b). To additionally evaluate the clinical usefulness of Ad.asHsp70 treatment, we investigated the survival of glioblastoma-bearing mice. The median survival of mice treated with a single Ad.asHsp70 injection 1 week after the implantation of tumor cells was twice as long as that of mice treated with the vehicle alone or an empty control virus (Fig. 3 c).

Next, we examined the Ad.asHsp70-induced reduction in the tumor volume in more detail. To this end, we treated the animals 5 days after the intracerebral implantation of U373MG cells, sacrificed them 2, 5, or 7 days later, and prepared cryostat sections for additional analysis. The treatment of tumors with a single injection of Ad.asHsp70 resulted in a significant reduction in the expression level of Hsp70, an increased number of cells with condensed nuclei, and massive DNA degradation (Fig. 3,d). These changes were evident already 2 days after the treatment (data not shown) and peaked 5 days after treatment (Fig. 3 d). One week after the treatment, an area of apoptotic/necrotic tissue surrounded the injection site (data not shown), and the tumor volumes of the Ad.asHsp70-treated mice were significantly smaller than the tumors of the Ad.ΔE1-treated mice (2 mm³ versus 7 mm³).

To test whether Ad.asHsp70-treated glioblastomas were able to attract microglia, we stained the tumor samples with an anti-ED1 antibody recognizing cells of macrophage linage. Ad.asHsp70-treated tumors (n = 3) contained 578 ± 81 ED1-positive cells/mm², whereas control tumors (n = 3) only had 179 ± 30 ED1 positive cells/mm² as analyzed on day 10 (Fig. 3 d).

Classic caspase-dependent apoptosis is associated with the triggering of an anti-inflammatory response in phagocytosing cells (Fig. 4,a; Ref. 16) and a blunted immune attack toward tumor tissue (16). Therefore, we examined the effect of Ad.asHsp70-treated tumor cells on HMDM. Tumor cells dying after Ad.asHsp70 treatment were effectively phagocytosed by HMDM. The significant phagocytic uptake of cell corpses as well as tumor cell death appeared to be independent of caspases, because it was neither affected by zVAD-fmk nor by forced overexpression of caspase-3 in MCF-CASP3 cells (Fig. 4, b and c; data not shown). It should be noted that contrary to parental or vector-transfected MCF-7 cells, Ad.asHsp70 triggered caspase activation in MCF-CASP3 cells. However, the overexpression of caspase-3 had no significant effect on the susceptibility of MCF-7 cells to Ad.asHsp70-induced death (data not shown). Most importantly, cells killed by Ad.asHsp70 did not reduce the LPS-induced activation of macrophages as analyzed by the ability of macrophages to produce TNF (Fig. 4,a). In fact Ad.asHsp70-treated tumor cells had some stimulatory effect on macrophages even in the absence of LPS (Fig. 4,a). In this respect, the Ad.asHsp70-treated cells differed from classic apoptotic cells, which effectively suppressed the proinflammatory macrophage response (Fig. 4 a).

The data presented above suggest that the depletion of Hsp70 may offer an effective means to combat cancer. In vitro treatment with Ad.asHsp70 was cytotoxic to all of the tumorigenic cell lines tested, pending that they could be effectively infected by adenovirus. More importantly, Ad.asHsp70 treatment resulted in a remarkable reduction in tumor volumes in human brain, breast, and colon cancer xenografts in mice without causing any detectable side effects. Furthermore, Hsp70 depletion resulted in the recruitment of macrophages into the tumor site and caspase-independent engulfment of dying cells. Because Hsp70-depleted cells did not inhibit the activation of macrophages in a way apoptotic cells do, the infiltrating macrophages may enhance the therapeutic effect of Ad.asHsp70 by inciting an inflammatory response and secretion of cytotoxic cytokines such as TNF.

The ineffective delivery of adenoviral vectors into the tumor site appears to be the major limitation for the usage of the Ad.asHsp70 therapy in the treatment of human cancer, which is often a systemic disease. Thus, clinical applications of this approach require additional development of the delivery systems or other means to neutralize the antiapoptotic effect of Hsp70. In the case of tumors that kill their host as a result of local tissue infiltration and invasion, but not distant metastasis, e.g., glioblastoma, locoregional treatment with Ad.asHsp70 may be particularly useful.

We thank Ingrid Fossar Larsen, Birgit Poulsen, Gabriele von Kürthy, and Torill Rignes for excellent technical assistance.