|Home | About | Journals | Submit | Contact Us | Français|
Glioblastoma multiforme is a deadly primary brain cancer. Because the tumor kills due to recurrences, we tested the hypothesis that a new treatment would lead to immunological memory in a rat model of recurrent glioblastoma multiforme.
We developed a combined treatment using an adenovirus (Ad) expressing fms-like tyrosine kinase-3 ligand (Flt3L), which induces the infiltration of immune cells into the tumor microenvironment, and an Ad expressing herpes simplex virus-1–thymidine kinase (TK), which kills proliferating tumor cells in the presence of ganciclovir.
This treatment induced immunological memory that led to rejection of a second glioblastoma multiforme implanted in the contralateral hemisphere and of an extracranial glioblastoma multiforme implanted intradermally. Rechallenged long-term survivors exhibited anti-glioblastoma multiforme–specific T cells and displayed specific delayed-type hypersensitivity. Using depleting antibodies, we showed that rejection of the second tumor was dependent on CD8+ T cells. Circulating anti-glioma antibodies were observed when glioblastoma multiforme cells were implanted intradermally in naïve rats or in long-term survivors. However, rats bearing intracranial glioblastoma multiforme only exhibited circulating antitumoral antibodies upon treatment with Ad-Flt3L + Ad-TK. This combined treatment induced tumor regression and release of the chromatin-binding protein high mobility group box 1 in two further intracranial glioblastoma multiforme models, that is, Fisher rats bearing intracranial 9L and F98 glioblastoma multiforme cells.
Treatment with Ad-Flt3L + Ad-TK triggered systemic anti–glioblastoma multiforme cellular and humoral immune responses, and anti–glioblastoma multiforme immunological memory. Release of the chromatin-binding protein high mobility group box 1 could be used as a noninvasive biomarker of therapeutic efficacy for glioblastoma multiforme. The robust treatment efficacy lends further support to its implementation in a phase I clinical trial.
Gliomas represent the most common primary brain tumors, and ~50% of these are the most aggressive type, that is, glioblastoma multiforme. The available multimodality clinical treatment for glioblastoma multiforme, that is, surgery, radiotherapy, and chemotherapy, results in an improved median survival from 6 to 18 months (1, 2). This has led to the development of various novel adjuvant treatments such as gene therapy and immunotherapy. Although numerous clinical trials have been conducted in patients with glioblastoma multiforme, the observed outcomes have not yet provided a major therapeutic breakthrough (3–7). Hence, continued research efforts are needed to develop and implement more efficacious and safer therapeutic strategies (8).
We previously reported the efficacy of a combined adenovirus (Ad)–mediated conditional cytotoxic/immune stimulatory gene therapy in preclinical rodent glioblastoma multiforme models (9–13). This gene therapy is composed of two trans-genes: the conditional cytotoxic gene herpes simplex virus type 1–thymidine kinase (TK) and the immune stimulatory transgene fms-like tyrosine kinase-3 ligand (Flt3L); both are delivered simultaneously into the growing glioblastoma multiforme tumor mass using adenoviral vectors (Ad-Flt3L + Ad-TK). This combination therapy results in a powerful therapeutic efficacy that rescues ~70% of the treated rats bearing large intracranial glioblastoma multiforme (10, 14). The therapeutic efficacy of this treatment relies on an immune response mediated by antigen-presenting cells, that is, dendritic cells and macrophages recruited by Flt3L into the glioblastoma multiforme microenvironment and CD4+ and CD8+ T cells (10, 12). Moreover, this combined therapy inhibits the progression of a second glioblastoma multiforme implanted in the contralateral hemisphere at the same time as the primary treated tumor in a rat model of multifocal glioblastoma multiforme (13). In this report, we show that CD8+ T cells mediate the immunological memory induced by Ad-Flt3L + Ad-TK (+ganciclovir) in an intracranial rat model of recurrent glioblastoma multiforme.
It is thought that due to the short survival of patients with glioblastoma multiforme, distant metastases from this tumor outside the cranium are rare. Although rare, systemic metastases of glioblastoma multiforme have been reported in bone marrow, lung, lymph nodes, spinal cord, peritoneum, and bone (15–20). Metastatic glioblastoma multiforme occurs mostly in young patients (<40 years) and is not correlated with overall median survival (17). Molecular genetic analysis of metastatic and primary glioblastoma multiforme suggested that systemic metastases of glioblastoma multiforme emerge from subclones within the primary tumor mass (17). In this study, we determined that the successful treatment of the primary intracranial glioblastoma multiforme mass with Ad-Flt3L + Ad-TK stimulates an immune response strong enough that can inhibit the growth of a extracranial glioblastoma multiforme mass growing outside of the cranial cavity.
We recently reported that the efficacy of the combined conditional cytotoxic/immune stimulatory gene therapy depends on the activation of toll-like receptor-2 (TLR2) on dendritic cells by high mobility group box 1 protein (HMGB1) released by dying tumor cells (12). TLR2 activation by HMGB1 is necessary to induce a therapeutically effective immune response against glioblastoma multiforme. In the present work, we aimed to test the hypothesis that serum levels of HMGB1 could be used as a noninvasive biomarker of therapeutic efficacy in glioblastoma multiforme. Thus, we measured the systemic levels of HMGB1 in several syngeneic glioblastoma multiforme models undergoing gene therapy. We determined the concentration of HMGB1 in serum from F98, 9L, and CNS-1 brain tumor-bearing rats 5 days after the treatment, as well as in naive rats and long-term survivor implanted with CNS-1 tumors in the flank. We found that the systemic levels of HMGB1 were elevated by treatment with Ad-Flt3L + Ad-TK in all glioblastoma multiforme models studied, suggesting that serum HMGB1 levels are a useful biomarker to monitor tumor regression in response to therapeutic intervention.
In summary, our results show that the combined conditional cytotoxic/immune-stimulatory gene therapy induces long-term cellular and humoral immunity against intracranial or extracranial glioblastoma multiforme. These findings bear clinical relevance for primary and recurrent glioblastoma multiforme. Importantly, in this report, we also show that the combined conditional cytotoxic/immune stimulatory gene therapy induces tumor regression and release of the chromatin-binding protein, that is, HMGB1 in several syngeneic glioblastoma multiforme models, which supports its use as a noninvasive bio-marker of glioblastoma multiforme progression.
The first-generation, replication-defective recombinant Ad type 5 vectors used in the study expressed soluble human Flt3L (9, 10, 21) or TK (9, 10) under the transcriptional control of human cytomegalovirus intermediate early promoter embedded within the E1 region (22). The construction of these vectors has been described in detail previously (9, 10, 13). The vectors were scaled up by infecting human embryonic kidney HEK 293 cells with a multiplicity of infection of 3 infectious units per cell of the vector seed stock. The cells were harvested 48 h later and lysed with 5% deoxycholate and DNase I, and the Ad vectors were purified by ultracentrifugation over two cesium chloride step gradients (22). The titers of vectors were estimated in triplicate by end point–dilution cytopathic effect assay (22), and the values were 3.28 × 1011 infectious units per milliliter for Ad-TK and 4.10 × 1010 infectious units per milliliter for Ad-Flt3L. All viral stocks used in this study were replication competent Ad and lipopolysaccharide free (22). The two vectors used to treat the glioblastoma multiforme rat models in the present study possessed a viral particles–versus–infectious units ratio of 18:1.
Glioblastoma multiforme CNS-1 cells (4,500; 3 μL) were implanted in the striatum of Lewis rats (220–250 g; Harlan), as described below (coordinates, 5 mm from the dura and 1 mm anterior and 3.2 mm lateral to the bregma), and treated with Ads 9 d later. Ad-Flt3L + Ad-TK (108 infectious units each in 3 μL) were delivered in three locations (1 μL each) within the tumor (−5.5, 5.0, and −4.5 mm from dura).
Ad-Flt3L + Ad-TK–treated brain tumor survivors were implanted in the contralateral striatum with 4,500 glioblastoma multiforme CNS-1 cells 60 d after primary brain tumor implantation. The day before the second tumor implantation, rats were depleted of specific immune cell populations. Phagocytic antigen-presenting cells and macrophages were depleted by i.p. injection of liposome encapsulated clodronate (Cl2MBP; 12 mg in 2 mL/rat; ref. 23). Clodronate (a gift from Dr. Nico Van Rooijen) was encapsulated in liposomes, as previously described (10, 23). CD4+ and CD8+ T cells were depleted by i.p. injection of 1 mg anti-CD2 antibody (OX-34) and 1 mg anti-CD8 antibody (OX-8), respectively. Hybridoma cell lines for CD8+ cell depletion (OX-8; European Collection of Animal Cell Cultures) and CD4+ cell depletion (OX-34; European Collection of Animal Cell Cultures) were produced by Bioexpress Cell Culture Services.
Glioblastoma multiforme F98 cells (50,000; 3 μL) were implanted in the striatum (coordinates, 6 mm from the dura and 1 mm anterior and 3.2 mm lateral to the bregma) of Fisher rats (Harlan) and treated with Ads 7 d later. Ad-Flt3L + Ad-TK (108 infectious units each in 3 μL) were delivered in three locations (1 μL each) within the tumor (−6.5, −5.5, and −4.5 mm from the dura).
Glioblastoma multiforme 9L cells (500,000; 3 μL) were implanted intracranially in the striatum (coordinates, 5 mm from the dura and 1 mm anterior and 3.2 mm lateral to the bregma) of Fisher rats (Harlan) and treated with gene therapy Ads 9 d later. Ad-Flt3L + Ad-TK (108 infectious units each in 3 μL) were delivered in three locations (1 μL each) within the tumor (−5.5, 5.0, and −4.5 mm from dura).
Rats were kept in controlled conditions of light (12 h light-dark cycles) and temperature (20–25°C) and fed standard laboratory chow and water ad libitum. All animal procedures were carried out in accordance with the NIH guide for the care and use of laboratory animals and approved by the Cedars-Sinai Institutional Animal Care and Use Committee.
Brain tumor implantation was done as previously described (24). Briefly, using a 10 μL Hamilton syringe fitted with a 26-gauge needle, glioblastoma multiforme cell suspensions were injected stereotaxically at the coordinates indicated above. A small pocket was created before depositing the cells by holding the needle 0.5 mm below the stated coordinates for 1 min before moving up to the stated coordinates and injecting the cells slowly over a period of 3 min. The needle was left in place for an additional 5 min before being slowly withdrawn. At the times indicated above, saline or Ad-Flt3L + Ad-TK (108 infectious units each) was injected into the growing tumor mass using the previously drilled burr hole. As controls, a group of rats bearing intracranial CNS-1 tumors were injected intratumorally with an empty Ad (2 × 108 infectious units), as described above. Twenty-four hours after treatment, rats were injected with ganciclovir (25 mg/kg i.p.; Roche Laboratories) twice daily for 10 consecutive days.
CNS-1 cells (3 × 106) in 50 μL DMEM (Cell-Gro) were injected intradermally in the flank of naïve Lewis rats or Ad-Flt3L + Ad-TK–treated survivors after 60 d of CNS-1 tumor implantation in the brain. The length (L) and the width (W) of the skin lesion were measured daily with slide calipers. Tumor volume (V) was calculated using the longest measurement as length and using: V (cubic millimeters) = (L × W2)/2 (25). Optimization of this model was done by injecting 5 × 104, 1 × 106, or 3 × 106 CNS-1 cells in 50 μL DMEM intradermally in the flanks of naïve rats (data not shown).
Animals were monitored daily and euthanized at the first signs of moribund behavior or at predetermined time points for collection of splenocytes for enzyme-linked immunosorbent spot, for detection of HMGB1 or anti–CNS-1 antibodies in serum, or for analysis of brain and flank tumor pathology. Animals were euthanized according to the guidelines of the Institutional Animal Care and Use Committee at Cedars-Sinai Medical Center, by terminal perfusion with Tyrode’s solution (132 mmol/L NaCl, 1.8 mmol/L CaCl2, 0.32 mmol/L NaH2PO4, 5.56 mmol/L glucose, 11.6 mmol/L NaHCO3, and 2.68 mmol/L KCl) followed by perfusion with 4% paraformaldehyde under deep anesthesia. Brains were removed and further fixed in 4% paraformaldehyde for 3 d.
DTH was done in long-term survivors 120 d after rechallenge. CNS-1 cell suspensions were prepared in PBS and then irradiated (30 Gy). Irradiated CNS-1 cells (1 × 106; 50 μL) were injected intradermally into the pinna of right ear of each rat, and the left pinna received 50 μL of saline. Baseline measurements of the thickness of pinna were recorded with slide calipers, and the measurement was repeated after 4, 24, and 48 h.
Following perfusion with Tyrode’s solution and 4% paraformaldehyde, brains were fixed in 4% paraformaldehyde for 3 additional days. Free-floating immunohistochemistry was done in serial coronal sections (50 μm). Sections were treated with 0.3% hydrogen peroxide to inactivate endogenous peroxidase and blocked with 10% horse serum, followed by incubation with the primary antibodies diluted in TBS (pH 7.4) containing 1% horse serum and 0.5% Triton X-100. Sections were incubated for 48 h with the following antibodies: anti-tyrosine hydroxylase (TH) in rabbit (1:5,000; Calbiochem 657012), anti-myelin basic protein (MBP) in mouse (1:1,000; Chemicon MAB1580), and anti-rat CD68 in mouse (clone ED1 to identify macrophages/activated microglia; 1:1,000; Serotec MCA341R). Then, the sections were incubated for 4 h with biotin-conjugated anti-rabbit goat IgG or anti-mouse rabbit IgG (1:800; DAKO E0432 and E0464, respectively), followed by 4 h additional incubation with avidin-biotin complex (Vectastain Elite ABC Kit, Vector Laboratories, Inc.). Nickel-enhanced 0.02% 3,3′-diaminobenzidine in sodium acetate was used as the chromogen. Finally, the sections were mounted onto gelatin-coated slides, dehydrated, and cover slipped using DPX mountant for histology (Sigma-Aldrich).
Nissl staining was done on brain sections to delineate the glioblastoma multiforme tumor mass. The sections were mounted and incubated in cresyl violet (0.1%; Sigma) for 15 min. They were washed in destain solution (70% ethanol, 10% acetic acid) for 1 min and then dehydrated (100% ethanol and xylene) and mounted.
The skin lesions were fixed in 4% paraformaldehyde and embedded in paraffin, and 4-μm serial sections were processed. Following deparaffinization, hematoxylin-eosin staining was done to evaluate tissue histology. Trichrome staining was done to look for the presence of collagen tissue within the lesion and to furnish supplementary evidence that the skin lesions were actually tumor masses and not merely inflammatory lesions. In addition, after antigen retrieval by heated citrate buffer (10 mmol/L citric acid; pH 6.0) followed by trypsin or proteinase K treatment, immunocytochemistry was done using the following antibodies: mouse anti-vimentin (1:1,000; Sigma V6630), mouse anti-human von Willebrand Factor (1:50; DAKO A0082), and anti-rat CD68 in mouse, as described above. Sections were evaluated and photographed with a Zeiss Axioplan microscope.
The glioblastoma multiforme cell lines (5,000 cells per well) were seeded in 12-well plates in triplicate and maintained in culture in DMEM supplemented with 10% fetal bovine serum (FBS; Omega Scientific) and 1% penicillin-streptomycin (CellGro). The growing cells were checked under the microscope daily, and the culture media were changed every 3 d. For counting the cells, wells in triplicate were trypsinized and the cells were harvested. The number of cells per well was then counted by mixing the cell suspension with trypan blue stain 0.4% (Gibco; Invitrogen Co.) to exclude dead cells.
The tumor volumes at 3, 6, or 9 d following implantation of the corresponding glioblastoma multiforme cells in the striatum (n = 3 per cell line) were estimated using unbiased stereological techniques. The postfixed brain was cut into consecutive 30-μm-thick coronal brain sections. A random selection of one twelfth of the brain sections was made, and the sections were processed for Nissl staining. With a Zeiss Axioplan 2 microscope controlled by Ludl electronic MAC 5000 XY stage control (Ludl Electronics Products Ltd.) and Axioplan Z-axis control (Carl Zeiss, Inc.), the tumor mass was visualized under low magnification (original magnification, ×1.25). All of the sections that had tumor mass as revealed by Nissl staining were sampled for tumor volume estimation (6–10 sections per animal). Stereo Investigator software version 8.00.0 (Microbrightfield, Inc.) was used to estimate the tumor volume using the Cavalieri estimator, with grid spacing set at 250 μm; the tumor mass was measured on all sections using point counting (26).
Confirmation of immune cell depletion was done in naïve Lewis rats injected i.p. with clodronate, OX-34, or OX-8 antibodies, as described above. Seven days after depletions, spleens were analyzed by flow cytometry. Splenocytes were harvested, and RBCs were removed by incubating in 3 mL ammonium chloride-potassium carbonate solution (0.15 mmol/L NH4Cl, 10 mmol/L KHCO3, and 0.1 mmol/L sodium EDTA at pH 7.2) for 3 min. Splenocytes were washed in RPMI media (containing 10% FBS, 1% penicillin-streptomycin, 1% L-glutamine); 1 × 106 splenocytes were resuspended in FACS buffer (PBS with 1% FBS and 0.1% sodium azide) containing anti–CD3-fluorescein isothiocyanate, anti–CD4-PE-Cy5, and anti–CD8-PE antibodies (BD Biosciences 554857, 554839, 557354). Macrophages were detected by CD68-FITC (MCA341F; Serotec) after fixation with 2% paraformaldehyde. Cells were analyzed on a FACScan flow cytometer (Beckman Coulter).
Serum samples were collected 7 d after CNS-1 tumor implantation in the brain or the flank. Fixed CNS-1 cells (50,000; 4% paraformaldehyde for 15 min on ice) were incubated in 50 μL of sample serum, naïve rat serum (isotype control), or FACS buffer (PBS with 1% FBS and 0.1% sodium azide) for 30 min on ice. After washing, cells were incubated with 50 μL of FITC–goat anti-rat immunoglobulin (1:200; Jackson Labs 112-096-003) for 30 min on ice. After washing, cells were analyzed on a FACScan flow cytometer (Beckman Coulter).
Trypsinized glioblastoma multiforme cells were incubated with FITC-conjugated mouse anti-rat major histocompatibility complex class II (BD Pharmingen 554919) or FITC-conjugated mouse IgG1κ isotype control, peridin chlorophyll protein-conjugated mouse anti-rat MHCII (BD Pharmingen 557016), or Per.CP-conjugated Ig1κ isotype control antibodies (4°C for 30 min). Cells were analyzed on a FACScan flow cytometer (Beckman Coulter). WinMDI 2.9 software (J. Trotter, Scripps Research Institute, La Jolla, CA) was used to determine fluorescence intensity.
ELISPOT assay was carried out using the rat IFN-γ development module (R&D Systems), according to the manufacturer’s instructions. Splenocytes were collected as described above, resuspended in 90% FBS + 10% dimethyl sulfoxide and frozen in liquid nitrogen. Frozen splenocytes were thawed rapidly and resuspended in RPMI-10 (RPMI medium supplemented with 10% FBS, 50 units/mL penicillin, 50 μg/mL streptomycin) and incubated at 37°C in 5% CO2 for 48 h in the presence of 10 U/mL recombinant human IL-2. One million cells per well were then plated in Imobilon-P membrane–lined 96-well plates (Millipore), precoated with mouse anti-rat IFN antibody. Stimulants were added as follows: (a) Concanavalin A at 1.6 μg/mL and (b) cell lysates, 100 μg/mL protein equivalent of cell lysate prepared from CNS-1, DSL6A, or L2 cells. Each well was supplemented with a further 10 U/mL of IL-2, and the plate was incubated as above for a further 48 h before discarding the cells and detecting IFN spots with a biotinylated secondary antibody, alkaline phosphatase–conjugated streptavidin, and a chromagenic substrate mixture of 5-bromo-4-chloro-3-indolyl phosphate/nitroblue tetrazolium (R&D Systems). Spots were counted manually under a dissection microscope.
HMGB1 expression was determined in rat serum using a specific anti-HMGB1 ELISA (IBL International), following the manufacturers protocol. Briefly, 100 μL sample diluent was added to each well. Next, 10 μL of rat serum was added to each well and incubated at 37°C for 24 h. Wells were washed five times with wash buffer and incubated for 2 h at 25°C with 100 μL peroxidase-conjugate solution. Wells were washed a further five times in wash buffer and incubated for 30 min at room temperature with substrate solution. The reaction was stopped by adding 100 μL stop solution to each well, and the absorbance was read at 450 nm (the background was subtracted by measuring absorbance at 570 nm).
RNA was extracted from rat glioma cell lines or mouse melanoma cell line B16 using the RNeasy Plus Mini kit (Qiagen). RNA quality and concentration was measured using DU640 Spectrophotometer (Beckman Coulter, Inc.). A total of 1 μg of RNA, 25 ng of Random primer, 20 units of Recombinant RNasin RNase Inhibitor, and 200 units of M-MLV Reverse transcriptase (Promega) were used in one reverse transcription reaction. The RT reaction was run at 37°C for 1 h. cDNA (1 μL), 800 nmol of both reverse and forward primers, and Go-Taq Flexi DNA polymerase (Promega) were used for PCR. PCR was done using Gene Amp PCR system 9700 (AB Applied Biosystem). The thermal cycles used for Trp-2 amplification were 10 min at 95°C, followed by 35 cycles of 95°C for 50 s, 56°C for 50 s, and 72°C for 50 s, ending with 10 min at 72°C. The thermal cycles used for GAPDH amplification were 10 min at 95°C, followed by 30 cycles of 95°C for 50 s, 58°C for 50 s, and 72°C for 50 s, ending with 10 min at 72°C. The reaction mixture was run on 0.8% w/v agarose gel with ethidium bromide. Primers, corresponding annealing temperature, and fragment length of PCR product size are listed in a table in Supplementary Fig. S1.
The survival curves were analyzed using the log-rank test using Prism GraphPad software (version 3.03). In vitro and in vivo tumor growth rates were determined by nonlinear regression analysis using Prism GraphPad software. NCSS statistical and power analysis software was used for the statistical analysis for flank tumor size, in vitro cell growth, and DTH data using randomization test. The flow cytometry, ELISPOT, and ELISA data were analyzed by Student’s t test or ANOVA. When data failed normality or Levene’s test for variance homogeneity, they were analyzed by Kruskal-Wallis multiple comparison followed by Dunn’s test. Differences between groups were considered significant at P < 0.05.
We have previously showed that gene therapy using Ad-Flt3L + Ad-TK induces brain tumor regression and immunological memory in rodent models of glioblastoma multiforme (12–14). Here, we characterize the phenotype of the immune cells that mediate the effector phase of the combined gene therapy approach using the recurrent rat glioblastoma multiforme model in Lewis rats. We first determined whether Ad-Flt3L + Ad-TK induces clonal expansion of tumor antigen–specific T lymphocytes and also the specificity of this response. Thus, we collected splenocytes from rats bearing intracranial CNS-1 tumors 7 days after the treatment with saline or Ad-Flt3L + Ad-TK and exposed them to cell extracts from CNS-1 cells or, as controls, DSL6A cells (syngeneic pancreatic carcinoma) or L2 cells (syngeneic rat yolk sac carcinoma cells). An increase in the frequency of T-cell precursors that released IFN-γ in response to exposure to specific tumor antigen was only detected in splenocytes from Ad-Flt3L + Ad-TK–treated rats that were challenged with CNS-1 cell extracts (Fig. 1A). No response was observed when splenocytes were exposed to DSL6A or L2 cell extracts or when the spenocytes originated from rats treated with saline. In addition, when Ad-Flt3L + Ad-TK (+ganciclovir)–treated long-term survivors were rechallenged in the contralateral brain hemisphere with a second CNS-1 tumor (60 days after primary tumor implantation), we observed an increase in the number of tumor antigen–specific T-cell precursors that release IFN-γ (Fig. 1B) when compared with tumor-bearing naïve rats, used as controls. Cellular immunity against CNS-1 cells in rechallenged long-term survivors was also assessed using the DTH reaction (Fig. 1C). DTH test was done 120 days after intracranial rechallenge of Ad-Flt3L + Ad-TK–treated long-term survivors. Naïve rats, used as controls, did not exhibit a positive DTH reaction over the 48-hour period of observation (Fig. 1C). However, rechallenged Ad-Flt3L + Ad-TK (+ganciclovir)–treated long-term survivors developed a strong DTH reaction, detected at 24 and 48 h after CNS-1 injection in the ear (Fig. 1C).
To determine which immune cells are responsible for the immunological memory against glioblastoma multiforme, Ad-Flt3L + Ad-TK–treated rats that survived a primary CNS-1 tumor were rechallenged with a second CNS-1 tumor in the contralateral striatum 60 days after primary tumor implantation and depleted of specific immune cell populations immediately before implantation of the second tumor. Phagocytic antigen-presenting cells were depleted by i.p. injection of liposome encapsulated clodronate, and CD4+ and CD8+ T cells were depleted by i.p. injection of anti-CD2 antibody (OX-34) and anti-CD8 antibody (OX-8), respectively. Depletion of the specific immune cell populations was confirmed in naïve Lewis rats 7 days following antibody administration (Fig. 1D). We found that depletion of antigen-presenting cells and CD4+ cells does not interfere with the rejection of the second glioblastoma multiforme, whereas the depletion of CD8+ T cells abrogates the anti-tumor immunological memory elicited by Ad-Flt3L + Ad-TK. Note that the rechallenged long-term survivors that were not depleted rejected the second tumor and survived long term after rechallenge (Fig. 1D).
In these experiments, we aimed to assess if the combined Ad-Flt3L + Ad-TK (+ganciclovir) gene therapy delivered into large intracranial glioblastoma multiforme was effective at inhibiting the progression of a glioblastoma multiforme implanted in the flank. The experimental paradigm is depicted in Fig. 2A. Lewis rats were implanted intracranially with 4,500 CNS-1 cells in the right striatum, and 9 days later, they were treated by intratumoral injection of either saline, an empty Ad, or Ad-Flt3L + Ad-TK, followed by administration of ganciclovir. Although all the saline- and empty Ad–treated rats succumbed because of tumor burden (median survival, 19 and 16, respectively), the combined treatment rescued ~80% of the rats bearing large intracranial glioblastoma multiforme (Fig. 2B). Sixty days after primary glioblastoma multiforme implantation in the brain, long-term survivors were implanted with CNS-1 cells (3 × 106) in the flank. Naïve rats implanted in the flank with CNS-1 cells served as controls (Fig. 2A). Tumor size was measured regularly in both groups. In the long-term survivors following successful treatment of intracranial glioblastoma multiforme by Ad-Flt3L + Ad-TK, the skin lesions were very small, attaining a maximum size of 14.8 ± 4.1 mm3 (mean ± SE) on day 3 and became undetectable 10 days of post–CNS-1 cell implantation. The skin lesions in naïve control rats grew rapidly to a maximum size of 903.3 ± 58.5 mm3 at 12 days postimplantation (Fig. 2C) and then started to regress spontaneously (not shown). The size and macroscopic appearance of a representative skin lesion 12 days following CNS-1 cell implantation in the flank is shown in Fig. 2A.
We have recently determined that mouse, rat, and human glioblastoma multiforme cells release HMGB1 upon cell death induced by Ad-TK + ganciclovir (12, 14). HMGB1 is an abundant chromatin-binding protein that acts as an endogenous TLR2 and TLR4 agonist when released by dying cells or inflammatory cells (27–29). In our experiments, systemic levels of HMGB1 increased rapidly during tumor cell death in vivo, and this was essential for the efficacy of Ad-Flt3L + Ad-TK in eradicating primary intracranial tumors in rodent models of glioblastoma multiforme (12, 14). We therefore aimed to determine whether this protein would also be released during the regression of an extracranial glioblastoma multiforme lesion. We found that 17 days after flank tumor implantation, the systemic levels of HMGB1 were increased in naïve and Ad-Flt3L + Ad-TK–treated long-term survivors (Fig. 2D).
Seventeen days after intradermal glioblastoma multiforme implantation (77 days after intracranial glioblastoma multiforme implantation), rats were euthanized, and the flank lesions were harvested for histopathologic analysis (Fig. 3). Although Ad-Flt3L + Ad-TK–treated brain tumor survivors did not exhibit tumor remnants in the flank, hematoxylin-eosin–stained extracranial glioblastoma multiforme lesions from naïve rats revealed tumors exhibiting necrosis with surrounding pseudopalisading cells (pseudopalisading necrosis), coagulative necrosis, microvascular hyperplasia, and also areas of hemorrhages (Fig. 3A), features that are all typical of glioblastoma multiforme. Trichrome Masson staining (Fig. 3B) showed the presence of abundant collagen along the tumor periphery but not in the central tumor mass. In addition, immunocytochemistry was done with antibodies against vimentin (to visualize CNS-1 cells that express vimentin), CD68/ED1 (macrophages), and von Willebrand factor (vascular endothelial cells). Microscopic examination uncovered areas of positive immunoreactivity against vimentin (Fig. 3C) and profuse CD68/ED1+ macrophages (Fig. 3D). Positive staining against von Willebrand factor was also evident and revealed characteristic microvascular hyperplasia (Fig. 3E).
Neuropathology of the Ad-Flt3L + Ad-TK (+ganciclovir)–treated long-term survivors that also rejected the extracranial glioblastoma multiforme implanted in the flank was also assessed (Supplementary Fig. S2). Long-term survivors showed no significant structural alterations in the striatum when compared with the untreated contralateral hemisphere. The brains of these rats did not display residual tumor, as determined by Nissl staining and vimentin immunocytochemistry, but exhibited abundant vimentin+ reactive astrocytes in the corpus callosum, external capsule, and also along the injection tract, typical of overreactive astrocytes (Supplementary Fig. S2). Immune staining of striatal dopaminergic fibers using anti-TH antibodies revealed similar density of TH+ fibers in both hemispheres. The density of myelinized fibers was observed by immune staining of MBP staining. MBP+ oligodendrocytes seemed to be intact in both hemispheres; no signs of demyelinization were observed. Scarce CD8+ T cells and abundant CD68+ macrophages/activated microglia were detected at and around the scar at the site of tumor implantation. MHCII immunoreactive cells were seen not only in the injection site but also in the corpus callosum and external capsule of brain parenchyma (Supplementary Fig. S2).
To evaluate whether humoral immunity was stimulated in response to Ad-Flt3L + Ad-TK (+ganciclovir) treatment during the rejection of the primary glioblastoma multiforme or the second glioblastoma multiforme implanted in the flank, we assessed the presence of circulating anti–CNS-1 antibodies in Lewis rats bearing intracranial and/or flank glioblastoma multiforme cells. Lewis rats were implanted intracranially with CNS-1 cells and 9 days later received intratumoral injection of either saline (Fig. 4A) or Ad-Flt3L + Ad-TK (+ganciclovir; Fig. 4B). All the saline-treated and a set of Ad-Flt3L + Ad-TK (+ganciclovir)–treated rats were euthanized 8 days after the treatment (17 days after intracranial glioblastoma multiforme implantation), and the serum was collected for detection of anti–CNS-1 antibodies. Control naïve rats (Fig. 4C) and a set of Ad-Flt3L + Ad-TK (+ganciclovir)–treated rats that survived to 60 days (Fig. 4D) were implanted with CNS-1 cells in the flank and were euthanized 17 days later (77 days after intracranial glioblastoma multiforme implantation in long-term brain tumor survivors). Serum was collected and used to determine the presence of anti–CNS-1 antibodies. Anti–glioblastoma multiforme cell antibodies were determined by incubating CNS-1 cells with serum from glioblastoma multiforme–bearing rats or from non–tumor bearing naïve rats, used as isotype control. The cells were then incubated with a FITC-tagged anti-rat secondary antibody, and mean fluorescence of CNS-1 cells bound to putative opsonizing antibodies was determined by flow cytometry. We found that, whereas rats with intracranial glioblastoma multiforme that were injected with saline had undetectable levels of anti–CNS-1 antibodies (Fig. 4A), rats treated with Ad-Flt3L + Ad-TK exhibited circulating antibodies against CNS-1 antigens 7 days after the treatment (Fig. 4B). In addition, anti–CNS-1 antibodies were detected in naïve rats (Fig. 4C) and Ad-Flt3L + Ad-TK–treated long-term survivors (Fig. 4D) implanted with glioblastoma multiforme in the flank and euthanized 17 days later.
One important criterion to plan the implementation of novel treatment strategies in phase I clinical trials is to show their efficacy and safety in preclinical animal models. We therefore tested the efficacy of the combined conditional cytotoxic/immune-therapeutic strategy in two further syngeneic brain tumor models. We evaluated therapeutic efficacy in intracranial rat glioblastoma multiforme F98 and 9L orthotopic tumors in Fisher rats, using the CNS-1 model in Lewis rat as reference. In vitro, the proliferation rate of F98 cells (doubling time, 1.033 days) was significantly higher than 9L cells (doubling time, 1.328 days) and CNS-1 cells (doubling time, 1.167 days; Fig. 5A). In all the three cell lines, we found expression of MHCI (Fig. 5B) and lack of expression of MHCII (Supplementary Fig. S3). In addition, we did not detect expression of the tumor antigen Trp-2 in any of the three glioblastoma multiforme cell lines (Supplementary Fig. S1A); this antigen is commonly overexpressed in melanoma cells and has been detected in mouse glioblastoma multiforme cells (12, 30) and glioblastoma multiforme specimens from human patients (31). We characterized the tumor growth rate and the histopathologic characteristics of these glioblastoma multiforme models in vivo at 3, 6, and 9 days after implantation using Nissl staining (Fig. 5C–E). The tumors were already evident at day 3 in all three models; the doubling time was similar, ranging from 1.1 to 1.6 days. Although the tumor size at day 9 after implantation was similar for the three tumor models, F98 (~40 mm3) and 9L tumors (45 mm3) were slightly larger than CNS-1 tumors (~30 mm3). The three models exhibited infiltration of immune cells, as assessed by staining of CD68+ macrophages (Fig. 5C–E) and CD8+ T cells (not shown).
To evaluate therapeutic efficacy in these glioblastoma multiforme models, F98 cells were implanted in the striatum of Fisher rats and treated 7 days later with intratumoral injection of saline, Ad-TK, Ad-Flt3L, or the combination Ad-Flt3L + Ad-TK (Fig. 6A). Rats treated with Ad-TK (+ganciclovir) alone exhibited a median survival of 26 days, which was significantly longer than the 15-day survival of saline-treated rats (median survival ratio, 1.73; 95% confidence interval, 1.4–2.1). Although Ad-Flt3L alone did not improve survival, combination with Ad-TK (+ganciclovir) led to a significant extension of the median survival compared with rats that received single treatment or saline-treated rats (median survival ratio, 2.9; 95% confidence interval, 2.5–3.2). Fisher rats harboring 9L tumors were treated with intratumoral injection of saline, Ad-TK, Ad-Flt3L, or Ad-Flt3L + Ad-TK 9 days after tumor implantation (Fig. 6B). Whereas Ad-TK (+ganciclovir) significantly improved the survival of tumor-bearing rats (median survival ratio, 1.5; 95% confidence interval, 1.16–1.9) and even led to long-term survival in 2 of 10 rats, Ad-Flt3L alone did not improve survival. However, the combined therapy using Ad-Flt3L + Ad-TK induced long-term survival in >50% of the animals treated. Tallying with our previous results, combined therapy using Ad-Flt3L + Ad-TK 9 days after CNS-1 tumor implantation in Lewis rats (used as positive controls) led to >70% long-term survival (Fig. 6C). Images in Fig. 6 show the appearance of the brain in saline-treated moribund rats bearing F98, 9L, or CNS-1 tumors, and the neuropathologic analysis of Ad-Flt3L + Ad-TK (+ganciclovir)–treated long-term survivors 60 days after tumor implantation. Note that the brains of long-term survivors did not exhibit demyelinization or loss of TH expression after tumor rejection induced by Ad-Flt3L + Ad-TK (+ganciclovir) treatment (Fig. 6A–C).
We recently showed that therapeutic efficacy of Ad-Flt3L + Ad-TK (+ganciclovir) depends on the release of HMGB1 from dying tumor cells in a syngeneic mouse model (12) and in the CNS-1 glioblastoma multiforme model in Lewis rats (14). We therefore assessed whether HMGB1 would also be released upon treatment with Ad-Flt3L + Ad-TK in F98 and 9L glioblastoma multiforme model in Fisher rats (Fig. 6D). We found that the systemic levels of HMGB1 were elevated 5 days after the treatment with Ad-Flt3L + Ad-TK in the three rat tumor models studied. This supports the notion that serum HMGB1 levels could be useful to monitor tumor regression in response to the treatment.
Supplementary Table S1 shows the median survival or percentage of long-term survival in the different brain tumor models treated with the combined therapeutic approach described. Note that combination of Ad-Flt3L + Ad-TK lead to at least 60% survival in 9L and CNS-1 tumor models, and thus, the median survival was not reached.
Glioblastoma multiforme grows within the brain; because of the lack of classic lymphatic outflow channels, the secretion of immune-inhibitory molecules by glioblastoma multiforme cells and a paucity of antigen-presenting cells within the brain parenchyma, it is challenging to elicit strong and therapeutically effective anti–glioblastoma multiforme immune responses (32). To elicit an anti–glioblastoma multiforme immune response from the brain tumor microenvironment, we developed a combined gene therapy strategy involving Ad-mediated delivery of Flt3L that recruits antigen-presenting cells into the brain tumor mass (9, 10, 13, 21), coupled with local release of tumor antigen and inflammatory molecules upon Ad-TK + ganciclovir–mediated tumor cell death (9, 10, 12–14). This therapeutic approach induces an immune response against the primary intracranial tumor that is dependent on phagocytic immune cells, that is, macrophages and dendritic cells, as well as CD4 + and CD8+ T cells (10). Considering the high rate of tumor recurrence in glioblastoma multiforme patients, the ability to suppress the growth of recurrent brain tumors growing at the original site or even at locations distant from the primary lesion without further treatment has important clinical implications for brain tumor therapy (19, 20, 33–36). We have previously shown that Ad-Flt3L + Ad-TK treatment induces immunological memory, and most treated rodents that survive the primary tumor are able to reject a second tumor implanted in the contralateral hemisphere without further treatment (10, 12, 14). In this article, we show that long-term survivors that reject the second tumor exhibit a high frequency of tumor antigen–specific T-cell precursors in the spleen and display DTH in response to tumor antigen, suggesting that cellular immunity is responsible for the anti–glioblastoma multiforme immunological memory. Furthermore, using depleting antibodies for specific immune cell types, we found that rejection of the second tumor is dependent on CD8+ memory T cells.
In the present study, we also tested the efficacy of this immune-therapeutic gene therapy strategy to inhibit the progression of glioblastoma multiforme implanted in the flank of long-term glioblastoma multiforme survivors in the syngeneic Lewis rat model. Our results show that the Ad-Flt3L + Ad-TK (+ganciclovir) treatment of intracranial glioblastoma multiforme induces cellular anti–glioblastoma multiforme immunological response that hampers the growth of a second glioblastoma multiforme lesion implanted in the flank. Histopathologic analysis of brain and skin glioblastoma multiforme sections from Ad-Flt3L + Ad-TK (+ganciclovir)–treated long-term survivors revealed the absence of any residual tumor in the brain or the flank. On the other hand, following intradermal implantation of CNS-1 cells in the flank of naïve rats, the tumor mass grew rapidly up to day 12. These flank tumors in naïve rats exhibited histologic features compatible with glioblastoma multiforme, namely, pseudopalisading necrosis, microvascular hyperplasia, and hemorrhages. Although these tumors were still visible to the naked eye 17 days postimplantation, they started regressing spontaneously after day 12. This finding is in line with previous reports showing spontaneous regression of peripherally implanted glioblastoma multiforme without any further treatment (37–39). Glioblastoma multiforme cells may be immunogenic in the periphery, as suggested by the presence of high levels of circulating anti–CNS-1 antibodies. Immunity against extracranial glioblastoma multiforme could be related to the fact that, in response to rapid tumor growth, areas of the tumor become necrotic and/or apoptotic, causing the release of proinflammatory molecules (i.e., HMGB1) and eliciting danger signals, which in turn could trigger the priming of an antitumor immune response (40). In agreement with this, we detected high levels of circulating HMGB1 in the serum of naïve rats implanted with flank glioblastoma multiforme tumors. HMGB1 is a nuclear protein that, when released from dying cells, acts as a cytokine with potent proinflammatory and chemotactic activity (27, 28, 41). HMGB1 levels were also high in the serum of Ad-Flt3L + Ad-TK–treated long-term survivors that rejected the extracranial glioblastoma multiforme tumor. However, because the growth of the flank tumor is abrogated in these animals by the immune response, it is possible that another source of circulating HMGB1 are the immune cells involved in the rejection of the extracranial tumor. In fact, we have recently shown that systemic levels of HMGB1 in Ad-Flt3L + Ad-TK–treated rats are higher than those observed in the rats treated with Ad-TK + ganciclovir alone (14). It has been shown that HMGB1 can also be actively released from immune cells, including macrophages/monocytes (42).
The fact that CNS-1 tumors are rejected from an extracranial site but not from an intracranial implantation site confirms once more that antitumor immune priming does not occur when antigens are injected carefully in the brain parenchyma. Upon implantation into the brain parenchyma, unless dendritic cells are recruited to the brain tumor microenvironment, the brain tumors will continue to grow without being challenged by an immune response. If dendritic cells are recruited to the brain and an immune response is stimulated, then the immune response is able to abrogate intracranial tumor growth. This is not necessary when glioblastoma multiforme cells are implanted in the flank because local infiltrating dendritic cells will be able to take up glioblastoma multiforme antigens that are released because of spontaneous tumor necrosis.
Circulating antibodies against CNS-1 cells were observed when glioblastoma multiforme cells were implanted in the flank of naïve or long-term survivors. However, rats bearing intracranial glioblastoma multiforme only exhibited circulating anti–glioblastoma multiforme antibodies upon treatment with Ad-Flt3L + Ad-TK. The presence of antibodies against intracranial brain tumors has been previously reported, and there is a decline of seroreactivity with increased malignancy, which has been related to loss of antigenicity as part of tumor escape mechanisms (43). Serum antibodies against mutant p53 have been detected only occasionally in glioblastoma multiforme patients, and the humoral immune response is known to be rather weak compared with antibody reactivity reported in other cancers (44). The presence of antibodies against brain tumor cells in Ad-Flt3L + Ad-TK rats suggest that, in addition to the CD4+ and CD8+ T cell–mediated cellular immunity (10), Ad-Flt3L + Ad-TK also induces a humoral immune response against the intracranial tumor.
To translate novel therapeutic approaches into clinical trials, it is critical to show their efficacy and safety in relevant preclinical animal models. We evaluated the efficacy of the combined conditional cytotoxic-immunotherapeutic approach in two further syngeneic brain tumor models. We found that treatment with Ad-Flt3L + Ad-TK also induces tumor regression in large intracranial 9L and F98 glioblastoma multiforme cells in Fisher rats. These data suggest that intratumoral administration of Ad-Flt3L + Ad-TK could be effective against a wide range of primary brain tumors. In addition, increased levels of HMGB1 in the serum of rats bearing 9L and F98 tumors after treatment with Ad-Flt3L + Ad-TK suggests that HMGB1 release is a wide spread event upon tumor regression. These results support the use of serum HMGB1 levels as a biomarker of antitumor therapeutic efficacy (12, 14).
Glioblastoma multiforme is an invasive primary brain tumor that recurs in most patients (1, 2, 45); therapies that stimulate the immune system to target and eliminate glioblastoma multiforme cells that migrated far from the main tumor mass have been developed with encouraging results (46, 47). Vaccination of a glioblastoma multiforme patient with dendritic cells pulsed with autologous tumor lysate following surgery, radio-therapy, and temozolomide led to robust CD8+ T-cell response against the tumor, without adverse side effects (46). Type I cytokine responses were also reported in glioblastoma multiforme patients vaccinated with dendritic cells in a phase II clinical trial (48). The epidermal growth factor receptor variant III, a mutated protein that is expressed in 20% to 30% of primary glioblastoma multiforme (49, 50), constitutes a tumor antigen that has been targeted using immunotherapy. Anti-epidermal growth factor receptor variant III antibodies have been detected in the serum of glioblastoma multiforme patients after immunization against this protein (47).
Our therapeutic approach aims at modifying the brain tumor microenvironment to recruit antigen-presenting cells into the brain tumor mass and expose these tumor infiltrating immune cells to tumor antigens and endogenous immune stimulatory molecules, that is, HMGB1. This treatment could be administered at the time of glioblastoma multiforme resection, delivering the combined therapy into the tumor mass or into the brain region surrounding the resection cavity. Finally, release of the chromatin-binding protein HMGB1 during tumor regression could be monitored in serum as a noninvasive biomarker of therapeutic efficacy. This combined gene therapy strategy, together with the potential use of HMGB1 as a biomarker of tumor response to treatment, warrants to be tested in a phase I clinical trial for glioblastoma multiforme.
Glioblastoma multiforme, the most common primary brain tumor, carries a dismal prognosis. Glioblastoma multiforme mostly kills due to recurrences; we therefore tested the hypothesis that a combined conditional cytotoxic/immune stimulatory strategy would lead to cellular and humoral immunological memory against recurrent glioblastoma multiforme lesions. Our results indicate that adenovirus (Ad) expressing fms-like tyrosine kinase-3 ligand + Ad expressing herpes simplex virus-1–thymidine kinase elicits CD8+ T cell–dependent immunological memory that leads to the rejection of a second glioblastoma multiforme implanted in the brain or in the flank. Circulating anti–glioblastoma multiforme antibodies were observed in the long-term survivors. Treatment with Ad expressing fms-like tyrosine kinase-3 ligand + Ad expressing herpes simplex virus-1–thymidine kinase induced tumor regression and release of the chromatin-binding protein high mobility group box 1 in two additional intracranial glioblastoma multiforme models, that is, Fisher rats bearing intracranial 9L and F98 glioblastoma multiforme cells, thus circulating levels of high mobility group box 1 could be used as a noninvasive biomarker of therapeutic efficacy for glioblastoma multiforme. Our results show the efficacy of this combined treatment in alternative glioblastoma multiforme models as a prelude for its clinical implementation.
We thank Drs. S. Melmed, L. Fine, and Mark Greene for the support and academic leadership; Patricia Lin from the flow cytometry core facility and Dr. John Young and his staff from the Department of Comparative Medicine Department, both at Cedars Sinai Medical Center, for their expert assistance; and Ivan B. Di Stefano from the Gene Therapeutics Research Institute at Cedars-Sinai Medical Center for taking the photographs of the flank tumors.
Grant support: NIH/National Institute of Neurologic Disorders and Stroke (NINDS) grants 1R21-NS054143, 1UO1 NS052465, and 1RO1-NS057711, and Medallions Group Endowed Chairs in Gene Therapeutics (M.G. Castro); NIH/NINDS grants 1 RO1 NS 054193, RO1 NS061107, and 1R21 NS047298, and Bram and Elaine Goldsmith (P.R. Lowenstein); the Linda Tallen and David Paul Kane Foundation Annual Fellowship; the Board of Governors at Cedars Sinai Medical Center; NIH/NINDS 1F32NS050303 (G.D. King); and NIH/NINDS 1F32 NS058156 (M. Candolfi).