Engineering the Brain Tumor Microenvironment Enhances the Efficacy of Dendritic Cell Vaccination: Implications for Clinical Trial Design

Glioblastoma multiforme (GBM) is the most common primary brain tumor in adults and exhibits a dismal prognosis. Despite therapeutic advances, that is, surgical resection followed by radiotherapy and chemotherapy, the prognosis for GBM is still poor with a median survival of 15 to 18 months postdiagnosis (1). Because GBM diffusely infiltrates into the surrounding brain parenchyma, relapse is almost inevitable even after surgical resection.

Immunotherapy strategies, including vaccination with tumor-associated peptides, autologous tumor cells, or dendritic cells (DC), have been intensively studied in human clinical trials for cancer (2–12). One of the most widely used approaches in cancer immunotherapy is DC vaccination, in which patients' DCs are cultured and expanded ex vivo, loaded with specific tumor antigens or autologous tumor lysates and then systemically administered to the patient. Recently, sipuleucel-T (Provenge) became the first antigen-specific, cell-based immunotherapy approach to receive Food and Drug Administration (FDA) approval (13). Sipuleucel-T showed an overall survival benefit to men with castration-resistant prostate cancer in 3 double-blind, placebo-controlled, multicenter trials (14, 15).

Several clinical trials for GBM using DC vaccination approaches have been implemented including vaccination with DCs pulsed with EGFRvIII, a tumor antigen which is expressed in about 30% of human GBM patients (2, 11), or vaccination with DCs loaded ex vivo with autologous tumor lysates (5–10, 12). These therapeutic approaches are designed to facilitate the presentation of brain tumor antigens to naive T cells, thereby inducing the proliferation of brain tumor antigen–specific cytotoxic T cells. In these studies, DC vaccination induced increased antitumor cellular and humoral immune responses against brain tumors, exhibiting a high safety profile (3, 5–10, 12).

We have developed a combined cytotoxic/immunostimulatory gene therapy using intratumoral injection of adenoviral vectors expressing fms-like tyrosine kinase 3 ligand (Ad-Flt3L) and thymidine kinase (Ad-TK) followed by systemic administration of ganciclovir (GCV; ref. 16). Intratumoral expression of Flt3L induces the migration, differentiation, and expansion of antigen-presenting cells, that is, bone marrow–derived DCs (BMDC), within the tumor microenvironment in mice (17, 18) and rats (16, 19, 20). Ad-TK is a conditional cytotoxic strategy which induces the death of actively dividing tumor cells in the presence of GCV, releasing endogenous brain tumor antigens. Ad-TK/GCV-treated tumor cells also release a potent innate adjuvant, that is, high-mobility group B1 protein (HMGB1), a DNA-binding protein constitutively expressed in nucleus of eukaryotic cells (18, 21). We showed that HMGB1 is released from dying tumor cells acting as an endogenous adjuvant to stimulate Toll-like receptor 2 (TLR2) signaling on BMDCs (18). In a large, intracranial GBM model, Ad-Flt3L + Ad-TK immunogene therapy induced long-term survival and immunologic memory that can eliminate recurrent and multifocal brain tumors (19, 22, 23). We also recently showed that Ad-Flt3L + Ad-TK–mediated gene therapy induces memory T cells capable of recognizing brain tumor neoantigens (24). On the basis of these results, gene therapy using Ad-Flt3L and Ad-TK/GCV was recently cleared by the FDA for an upcoming phase I clinical trial for GBM (BB-IND 14574; NIH/OBA Protocol 0907-990; and OSU Protocol 10089).

Herein, we aimed to test the hypothesis that in situ (intratumoral) gene therapy mediated by expression of Flt3L and HSV1-TK would enhance the therapeutic efficacy of systemic DC vaccination, leading to the design of novel phase I trials for GBM. Recent reports indicate that vaccination efficacy depends on the type of cell death used to prepare the tumor lysates, that is, apoptotic tumor lysates are more immunogenic and elicit greater antitumor effects than necrotic tumor lysates (25, 26). As such, we compared the immunogenicity of tumor lysates prepared by treatment with Ad-TK/GCV, the chemotherapeutic agent temozolomide (TMZ), or freeze/thawing cycles (FTC). Our results show that vaccination with cell lysates prepared using Ad-TK/GCV or TMZ displayed equivalent inhibition of brain tumor progression and were more efficacious than necrotic lysates (FTC). We also compared the in vitro and in vivo phenotypic characteristics of DCs, generated by Flt3L + IL-6 treatment or by granulocyte-macrophage colony-stimulating factor (GM-CSF) and IL-4 treatment, that is, activation markers, cytokine release, and phagocytic ability.

We showed the therapeutic efficacy of Flt3L + IL-6 induced DCs loaded with Ad-TK/GCV-generated tumor lysates, in rats bearing established brain tumors. Flt3L is a powerful cytokine that is required for normal development and expansion of DCs in vitro and in vivo (27). Our results show that therapeutic efficacy and induction of antitumor immunity in Lewis rats bearing orthotopic, syngeneic brain tumors was significantly enhanced when DC vaccination was combined with Ad-TK/GCV + Ad-Flt3L gene therapy. These data support the implementation of novel phase I clinical trials for GBM to assess the safety and efficacy of this novel dual therapeutic modality.

We used first-generation, E1/E3-deleted Ad-Flt3L or Ad-TK (described in Supplementary Materials and Methods).

CNS1 rat glioma cells were treated with TK/GCV, TMZ, or FTC; described in Supplementary Materials and Methods.

Bone marrow was isolated from the femurs of naive, adult Lewis rats as described previously and in Supplementary Materials and Methods. Bone marrow precursors were cultured with either Flt3L and interleukin (IL) 6 or GM-CSF and IL-4. Loosely adherent cells were used for subsequent subcutaneous injections and functional assays as described in Supplementary Materials and Methods.

All procedures involving live animals were conducted in accordance with protocols approved by the Cedars-Sinai Medical Center's Institutional Animal Care and Use Committee (IACUC). Five thousand CNS1 glioma cells in 3 μL Dulbecco's modified Eagle's media (DMEM) were stereotactically implanted in the right striatum of syngeneic Lewis rats as described in Supplementary Materials and Methods. Four or 10 days after tumor implantation, rats received a stereotactic, intratumoral injection of Ad vectors; described in Supplementary Materials and Methods. For survival studies, animals were treated with saline (as a control), 3 × 10⁸ plaque-forming units (pfu) of Ad-Flt3L, 1 × 10⁸ pfu of Ad-TK, or in combination with Ad-Flt3L and Ad-TK.

Lysates prepared from 6 × 10⁶ tumor cells were mixed with CpG₂₀₀₆ and injected subcutaneously in the flank 3 times at 7-day intervals as indicated and described in Supplementary Materials and Methods. DC vaccination is described in detail in Supplementary Materials and Methods.

CNS1 cells treated with Ad-TK/GCV, TMZ, or FTC were fixed with 2.5% glutaraldehyde, dehydrated through graded acetone, and embedded in Eponate. Sections (60 nm) were counterstained with uranyl acetate in methanol and in Reynold's lead citrate and viewed with a Jeol 100CX transmission electron microscope; described in Supplementary Materials and Methods.

Acridine orange was used to detect and quantify the development of acidic vesicular organelles. Briefly, acridine orange moves freely across biological membranes and fluoresces green when uncharged, whereas its protonated form accumulates in acidic compartments and fluoresces bright red. The intensity of red fluorescence is proportional to the degree of acidity. CNS1 tumor cells were stained with acridine orange (1 μg/mL) and analyzed by fluorescence microscopy or flow cytometry as described in Supplementary Materials and Methods. Bafilomycin A1, an inhibitor of vacuolar H⁺-ATPase (V-ATPase) that interferes with the fusion of autophagosome and lysosome, was added to the cells 30 minutes before addition of acridine orange.

Analysis of cell lysates by Western blotting was conducted as described in Supplementary Materials and Methods. Labeling was conducted using anti-LC-3B antibody, or α-tubulin as a control, followed by horseradish peroxidase (HRP)-conjugated secondary antibody and visualization with enhanced chemiluminescence (ECL).

CNS1 tumor cell lysates were stained with fluorescein isothiocyanate (FITC)-conjugated Annexin V and propidium iodide (PI) and analyzed by flow cytometry as described by us previously (18, 21) and in Supplementary Materials and Methods.

DCs were incubated with CpG₂₀₀₆ or lipopolysaccharide (LPS) and then assessed for either (i) cytokine production by ELISAs specific for IL-10, IFN-γ, TNF-α, and IL-12 or (ii) expression of cell surface markers OX-62, CD161a, CD3, CD4, CD45, CD45R, CD11c, CD80, CD86, and MHCII by flow cytometry; described in Supplementary Materials and Methods.

CNS1 tumor cells stained with Qtracker 655 were treated with the above killing treatments (Ad-TK/GCV, TMZ, or FTC). Dying, labeled tumor cells were incubated Flt3L + IL-6–conditioned DCs labeled with a CD161a-PE antibody. Phagocytosis of tumor cells was assessed by flow cytometry as the percentage of Qtracker 655/CD161a-phycoerythrin (PE)-labeled cells compared with CD161a-PE–labeled cells. Alternatively, microscopy analysis was used to assess the phagocytosis of tumor cell remnants; Qtracker 655 stained tumor cells were cultured with DCs, stained with CD161a followed by an Alexa Fluor 488–tagged secondary antibody. Nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI) and samples were analyzed by confocal microscopy, see Supplementary Materials and Methods.

Cytotoxicity of lymphocytes against tumor cells was assessed by flow cytometry as previously described (28) and in Supplementary Materials and Methods

Serum levels of anti-CNS1 antibodies were measured by flow cytometry, described previously (29) and in Supplementary Materials and Methods.

Kaplan–Meier survival curves were analyzed using the Mantel log-rank test using Prism GraphPad software (version 3.03). The flow cytometry and ELISA data were analyzed by 1-way ANOVA followed by Tukey's posttest using NCSS statistical and power analysis software. Differences between groups were considered significant at P < 0.05. All experiments described in our article were conducted twice to confirm the findings.

Tumor cell death mechanisms have been implicated in the degree of immunogenicity induced by tumor lysates (25, 26). Thus, we characterized the mechanisms of tumor cell death induced by Ad-TK/GCV, TMZ, and FTC in vitro, which we used to prepare tumor cell lysates for vaccination. Electron microscopy analysis indicated that CNS1 cells treated with Ad-TK/GCV exhibited chromatin condensation and cytoplasmic vacuolization, which are typical features of apoptotic cell death (Fig. 1A). FTC-treated CNS1 cells showed typical features of necrosis including cytoplasmic swelling, plasma membrane, and cytoplasmic organelle destruction (Fig. 1A). TMZ induces autophagy in CNS1 rat brain tumor cells, displaying an increased number of autophagic vacuoles in the cytoplasm (black arrow), including autophagosomes, double membrane vacuoles containing organelles (Fig. 1A; white arrow) and autolysosomes vesicular organelles containing digested residual materials (Fig. 1A; white arrowheads). Autophagy was also observed in Ad-TK/GCV-treated tumor cells (Supplementary Fig. S1).

Autophagy is an evolutionarily conserved intracellular degradative process, where cytoplasmic organelles are sequestered into lytic compartments; characterized by the formation of acidic vesicular organelles called autolysosomes (30). To quantify the extent of autophagy induction in CNS1 brain tumor cells, we determined the percentage of cells with accumulated acidic vesicular organelles using acridine orange staining (Fig. 1B and C). Acridine orange moves freely across biological membranes and fluoresces green when in an uncharged environment, whereas its protonated form accumulates in acidic compartments and fluoresces bright red (31, 32). The intensity of red fluorescence is proportional to the degree of acidity. As one of the most efficient triggers of autophagy is nutrient starvation (30, 32), serum-starved cells were used as a positive control. Fluorescence microscopy analysis shows accumulation of red fluorescence in the cytoplasm in CNS1 tumor cells treated either with TMZ or Ad-TK/GCV, indicating the presence of acidic compartments (Fig. 1B). Flow cytometry analysis shows that both TMZ and Ad-TK/GCV increased red fluorescence intensity from 9.4% (basal levels) to 35.9% and 36.0%, respectively, comparable with serum-starved cells (42.3%). Increased red florescence in CNS1 cells treated with TMZ or Ad-TK/GCV was abolished by treatment with bafilomycin A1, an inhibitor of the fusion of autophagosomes and acidic lysosomes (ref. 33; Fig. 1C). Freeze–thaw lysates did not acquire acridine orange staining. These findings indicate that CNS1 cells treated with TMZ or Ad-TK/GCV exhibit acidic vesicular organelle formation associated with autophagy. Autophagy induction was also assessed by Western blotting of LC3, a specific marker for autophagy (32; Fig. 1D). When cells undergo autophagy, LC3-I is incorporated into autophagosomes and converted to LC3-II (lapidated form). However, the increase in LC3-II expression can be associated with increased formation of autophagosomes, or an impaired autophagic degradation (34). To differentiate these 2 possibilities, LC3-II expression was assessed in the presence of bafilomycin A1, which blocks autophagic degradation (33). As shown in Figure 1D, CNS1 cells treated with TMZ or Ad-TK/GCV exhibited further accumulation of LC3-II in the presence of bafilomycin A1, thus supporting that the increased LC3-II expression is a consequence of increased levels of autophagosome formation rather than impaired degradation.

The extent of apoptosis and necrosis in CNS1 brain tumor cells was assessed by flow cytometry following Annexin V and PI staining (Fig. 1E). Our results indicate that Ad-TK/GCV induced mainly apoptosis (69.6%), TMZ induced a small population of necrotic cells (18.3%) and minimal apoptosis (3.9%), and FTC induced mostly late apoptosis/necrosis (88.2%; Fig. 1E). Taken together, these data show that Ad-TK/GCV induces apoptosis that is accompanied with autophagy, TMZ induces mainly autophagy, and FTC induces necrosis in CNS1 brain tumor cells.

To assess whether immunogenicity of tumor cell lysates is dependent on the mechanism of tumor cell death in our model, we compared the therapeutic efficacy of whole tumor cell lysate vaccines generated by treatment of CNS1 cells with Ad-TK/GCV, TMZ, or FTC. In the first paradigm, the animals were vaccinated before tumor cell implantation (Fig. 2A). Lysates were administered subcutaneously 3 times before CNS1 tumor cell implantation into the striatum (Fig. 2A). Vaccination with tumor lysates prepared from any of the 3 methods tested, failed to significantly prolong median survival (Fig. 2B). All tumor lysate vaccines increased the levels of circulating anti-CNS1 antibodies in the serum when compared with saline controls (Fig. 2C). Levels of circulating anti-CNS1 antibodies were significantly higher in rats vaccinated with Ad-TK/GCV- or TMZ-generated cell lysates than in rats treated with FTC-generated cell lysates (Fig. 2C). These findings show that Ad-TK/GCV–treated (apoptotic and autophagic) or TMZ-treated (autophagic) tumor lysates induce more potent antitumor immune responses than FTC lysates (necrotic).

DC differentiation and proliferation can be elicited by either Flt3L or GM-CSF (35–37). Also, recent studies have shown that IL-6 improves expansion of Flt3L-induced DCs (35, 38). To determine whether vaccination with Flt3L-induced DCs elicits antitumor immunity, we compared in vitro and in vivo the characteristics of BMDCs induced by GM-CSF + IL-4 or Flt3L + IL-6 (Fig. 3). Consistent with previous reports (35, 38), expansion of DCs was higher in Flt3L-treated than GM-CSF + IL-4–treated BM cultures; the addition of IL-6 greatly enhanced the proliferation of Flt3L-induced DCs (Supplementary Fig. S2A). Microscopic analysis showed that under both conditions, DCs exhibited long cytoplasmic processes, a typical feature of DCs (Supplementary Fig. S2B). Loosely attached cells conditioned with Flt3L + IL-6 were analyzed by flow cytometry; about 33% of the cells displayed elevated levels of OX-62, expressed by rat conventional DCs (cDC; CD11c⁺CD45^highMHCII⁺) but not plasmacytoid DCs (pDC; CD3⁻ CD4⁺ CD45R⁺), and about 94% displayed elevated levels of the natural killer (NK) marker CD161a on their cell surface, a phenotype characteristic of all rat DC subtypes (39). Flow cytometric analysis of cell surface markers revealed that a majority of Flt3L + IL-6–conditioned DCs displayed a characteristic phenotype of cDCs; a very small population of DCs expressed cell surface markers characteristic of pDCs (ref. 39; Supplementary Fig. S3).

Maturation of DCs, with upregulation of MHCII or costimulatory molecules such as CD86 and CD80, is critical to induce the clonal expansion of antigen-specific T-cell populations (40). Thus, we evaluated the maturation status of BMDCs incubated with Flt3L + IL-6 or GM-CSF + IL-4. Flt3L + IL-6–generated DCs expressed CD86 and CD80 (27.3% and 55.1%, respectively); GM-CSF + IL-4–generated DCs expressed higher levels of CD86 and CD80 (80.7% and 78.7%, respectively; Fig. 3A). MHCII expression was similar in both Flt3L + IL-6- and GM-CSF + IL-4–cultured DCs (75.6% and 83.3%, respectively; Fig. 3A). To induce further maturation of DCs, cells were further incubated for 48 hours without cytokines in the presence or absence of activation agents CpG oligodendronucleotide (CpG-ODN) or LPS. Deprivation of cytokines induced upregulation of CD86, CD80, and MHCII (Fig. 3A). Stimulation by CpG-ODN or LPS further increased the expression of those surface molecules in Flt3L + IL-6–cultured DCs. Treatment with CpG-ODN or LPS had little effects on the expression of maturation markers on GM-CSF + IL-4–induced DCs (Fig. 3A). These findings indicate that Flt3L + IL-6 conditioning maintains BMDCs in a more immature state (lower expression of maturation makers) when compared with GM-CSF + IL-4. The balance between the induction of T helper cell type 1 (Th1) and Th2 CD4⁺ T cells determines the efficiency and the type of immune response. Secretion of Th1-polarizing cytokines such as IL-12 and IFN-γ produced by DCs, which promote a shift toward Th1 dominant balance, has been implicated in potent antitumor immunity (40). Thus, we compared the cytokine secretion profile between GM-CSF + IL-4- and Flt3L + IL-6–conditioned DCs by ELISA (Fig. 3B). Activation of DCs by CpG-ODN or LPS promoted production of both Th1-polarizing cytokines, TNF-α, IL-12, and IFN-γ, and the Th2-polarizing cytokine, IL-10 (ref. 40; Fig. 3B). Although there was no difference in the secretion pattern of TNF-α, both IL-12 and IFN-γ were more robustly produced by Flt3L + IL-6–generated DCs as compared with GM-CSF + IL-4–generated DCs (Fig. 3B). Despite evidence of upregulated production of Th1-polarizing cytokines IL-12 and IFN-γ, we also observed elevated levels of IL-10 production in Flt3L + IL-6–generated DCs. The ratio between IL-12 and IL-10 represents the balance between Th1- and Th2-polarizing cytokine production by DCs (20). Flt3L + IL-6–generated DCs displayed a high Th1 balance when stimulated with either CpG or LPS; however, GM-CSF + IL-4–generated DCs only displayed a high Th1 balance when stimulated with LPS but not CpG (Fig. 3B).

Functionally, DCs are characterized by their capacity to present antigen to T cells and induce clonal expansion of antigen-specific T cells. Thus, we assessed the capacity of Flt3L + IL-6–conditioned DCs loaded with Ad-TK/GCV–derived tumor lysates to induce T-cell proliferation. Splenocytes from tumor-bearing rats were cocultured with tumor-loaded BMDCs, unloaded BMDCs, or no BMDCs (Supplementary Fig. S4A). T-cell proliferation was only induced when T lymphocytes were cocultured with tumor-loaded DCs (Supplementary Fig. S4B). T cells stimulated by CNS1-loaded DCs effectively killed CNS1 tumor cells in vitro in a dose-dependent manner, when compared with lymphocytes from a naive rat as a control (Supplementary Fig. S4C). Thus, these data show that tumor-loaded BMDCs have the capacity to induce the proliferation of tumor-specific cytotoxic T cells.

While most DC vaccination clinical trials rely on GM-CSF to induce the proliferation and conditioning of DCs, recent studies have shown that Flt3L and IL-6 improves expansion of DCs (38). As such, we assessed the in vivo therapeutic efficacy of systemic vaccination with DCs generated using Flt3L + IL-6 or GM-CSF + IL-4 loaded with Ad-TK/GCV–derived tumor lysates. DCs were delivered 3 times at 7-day intervals before tumor implantation (days −14, −7, and 0). Vaccinated rats were then challenged with CNS1 cells in the striatum at day 0 (Fig. 3C). Rats vaccinated with Flt3L + IL-6–generated DCs completely inhibited tumor growth and exhibited 100% long-term survival, whereas only 60% of rats vaccinated with DCs generated by GM-CSF + IL-4 exhibited long-term survival (Fig. 3D). These data suggest that DCs generated with Flt3L + IL-6 result in vaccines that induce more robust antitumor immune responses than DC vaccines prepared with GM-CSF + IL-4.

Uptake of tumor cell remnants followed by processing and presentation of tumor antigens to naive T cells by DCs is critical for the induction of adaptive immunity. Thus, we tested whether tumor uptake by DCs depends on the method used for tumor cell killing, that is, autophagy, apoptosis, and necrosis. As shown in Figure 4A, DCs displayed higher levels of phagocytic activity when cocultured with dying tumor cells, as compared with mock-treated cells. Furthermore, Ad-TK/GCV or TMZ treatment increased tumor uptake by DCs when compared with FTC treatment.

We next tested whether the mechanism of tumor cell killing used to prepare the tumor lysates influences the therapeutic efficacy of DC vaccination. Tumor lysate–loaded DCs were administered subcutaneously into Lewis rats 3 times at 7-day intervals before tumor implantation (days −14, −7, and 0). Vaccinated rats were then challenged with CNS1 cells in the striatum at day 0 (Fig. 4C). Rats receiving DCs loaded with Ad-TK/GCV- or TMZ-generated tumor lysates completely inhibited tumor development and animals exhibited 100% long-term survival, whereas only 60% of rats vaccinated with DCs loaded with FTC tumor lysates survived long term (Fig. 4D). Levels of circulating anti-CNS1 antibodies were significantly higher in rats treated with DCs loaded with either Ad-TK/GCV- or TMZ–generated tumor lysates than in rats treated with DCs loaded with FTC tumor lysates (Fig. 4E).

DC vaccination strategies have been shown to induce antitumor immune responses in patients with GBM undergoing clinical trials (3, 5–10). In this report, we wished to test the hypothesis that manipulating the tumor microenvironment via in situ immunogene therapy would enhance therapeutic efficacy of DC vaccination. First, we assessed whether DCs loaded with either Ad-TK/GCV–or TMZ-generated tumor lysates in combination with CpG as an adjuvant could elicit regression of an established intracranial tumor mass in the Lewis rat model. We compared therapeutic efficacy of in situ immunogene therapy to DC vaccination in a small tumor model, where tumor-bearing rats were treated 4 days after tumor implantation (Fig. 5A). In the small brain tumor model, Ad-Flt3L + Ad-TK immunogene therapy induced 100% survival, whereas we observed comparable therapeutic efficacy with either the DC vaccine loaded with TMZ-derived tumor lysates or the DC vaccine loaded with TK/GCV-derived tumor lysates in the small tumor model (Fig. 5B).

We next tested whether in situ immunogene therapy and DC vaccination could be used in combination to potentiate therapeutic efficacy in a rigorous, large tumor model, where tumor-bearing rats were treated 10 days after tumor implantation (Fig. 6A). The combination treatment of in situ immunogene therapy with DC vaccination led to long-term survival in about 90% of tumor-bearing animals. These data show a 40% increase in levels of long-term survival when compared with in situ immunogene therapy alone and an 80% increase in levels of long-term survival when compared with DC vaccination alone. The therapeutic efficacy data are corroborated by serum levels of anti-CNS1 antibodies in animals treated with combination therapy, which were increased by about 2.5-fold when compared with either arm of the therapy alone (Fig. 6C).

DC vaccination approaches for brain tumors have been implemented in human clinical trials (5–10). Several clinical trials for GBM using DC vaccines have shown cellular and humoral antitumor immune responses; and albeit safe, the clinical efficacy of DC vaccination for GBM remains limited (3, 5–10). Our group has shown that in situ Ad-Flt3L/TK–mediated immunogene therapy elicits an influx of DCs, macrophages, CD4⁺ T cells, and CD8⁺ T cells into the brain tumor microenvironment (17) and stimulates effective anti-GBM immune response resulting in tumor regression and long-lasting CD8⁺ T-cell–mediated antitumor immunologic memory in several mouse and rat orthotopic brain tumor models (16, 18, 22–24, 29). On the basis of these data, a phase I clinical trial for GBM was recently cleared by the FDA and is slated to commence in 2011.

The combined conditional cytotoxic/immunostimulatory gene therapy approach, utilizes HSV1-TK to kill actively dividing brain tumor cells in the presence of GCV, thereby releasing endogenous brain tumor antigens and also innate immune adjuvants such as HMGB1 (18, 21). HMGB1 belongs to a class of innate immune adjuvants called damage-associated molecular pattern molecules (DAMP) which mediate signaling by binding to a family of receptors called pattern recognition receptors (PRR), thereby promoting innate and adaptive immune responses (41). DCs express a large repertoire of PRRs and several studies have shown that signaling through PRRs leads to DC activation, which is characterized by high levels of MHC-antigen complexes on the DC cell surface, upregulation of costimulatory molecules such as CD80 and CD86, and the production of cytokines such as IL-12 and IFN-α. The production of these cytokines is directly involved in priming Th1-biased immune responses (40). We have previously shown that HMGB1 is released from dying brain tumor cells in response to treatment with Ad-TK/GCV or TMZ, and acts as an endogenous TLR2 agonist to activate bone marrow–derived, brain tumor–infiltrating DCs (18). The other arm of our therapeutic strategy involves expression of Flt3L within the tumor microenvironment. Flt3L recreates a missing immune circuit from the brain, by inducing the expansion and migration of DCs into the brain tumor milieu where they encounter and phagocytose newly released endogenous brain tumor antigens (16, 18, 22–24, 29).

Herein, we tested the hypothesis that combining Ad-Flt3L/TK in situ immunogene therapy with peripheral DC vaccination, would lead to enhanced therapeutic efficacy in a syngeneic brain tumor model. Our data show that the therapeutic combination, that is, in situ immunogene therapy with DC vaccination led to long-term survival in 90% of rats bearing large, syngeneic brain tumors; showing about 40% increase in survival compared with Ad-Flt3L/TK immunogene therapy alone and about 80% increase in survival compared with DC vaccination alone. Our results also showed enhanced antitumor humoral immune response, that is, about 2.5-fold increase in the levels of circulating antitumor antibodies compared with either treatment alone.

The enhanced therapeutic efficacy observed could result from expression of Flt3L within the brain tumor microenvironment which could enhance the trafficking of systemically delivered DC vaccines to the draining lymph nodes, where they present tumor antigens to naive T cells and induce the clonal expansion of tumor-specific CTLs. To this effect, it has been previously shown that Flt3L elicits recruitment of DC populations, including fully differentiated DCs in lymphoid organs (40) and traffic of subcutaneously delivered DC vaccines to the inguinal draining lymph nodes in mice (42). We showed that intracranial delivery of Ad-Flt3L leads to circulating Flt3L in rat serum (19), thus supporting hypothesis that Flt3L could also act systemically and enhance the trafficking of DCs to draining lymph nodes. Increased trafficking of systemically delivered DCs to draining lymph nodes could lead to enhanced priming of antitumor immune responses, inducing increased levels of tumor regression and long-term survival as observed in this study.

An additional explanation for the increased therapeutic efficacy of in situ immunogene therapy in combination with DC vaccination is that DAMPs, that is, HMGB1, are known to enhance the activation status of DCs and facilitate Th1 immune responses (40). We have previously shown that treatment of intracranial brain tumors with TK/GCV results in increased levels of circulating HMGB1 in the sera of mice (18) and rats (21). Therefore, HMGB1 released into the systemic circulation could further activate systemically delivered DCs, thus enhancing their ability to prime an adaptive, antitumor immune response.

When used to treat small, established intracranial tumors, DC vaccination exhibited higher levels of long-term survival (∼30%) when compared with treatment of large intracranial tumors (∼10% long-term survival). These data suggest that the efficiency of a DC vaccination approach is associated with the degree of tumor burden in the host. Furthermore, our data show that DC vaccination is highly efficacious (100% long-term survival) when administered before tumor implantation, suggesting that DC vaccines would be more effective at preventing tumor recurrences after initial surgical debulking, chemotherapy, and radiotherapy.

Effective uptake and loading of tumor-associated antigens onto MHC complexes of DCs and expansion of DC subgroups that can efficiently prime naive T cells play a critical role in the effectiveness of DC vaccination. Therefore, it is critically important to optimize the preparation of tumor cell antigens and DCs. In this study, we compared the immunogenicity and levels of phagocytosis of apoptotic, autophagic, and necrotic tumor cells lysates. In line with previous evidence (26, 43, 44), our data indicate that triggering autophagy and/or apoptosis to generate tumor cell lysates increases the immunogenicity of tumor cells and enhances the delivery of tumor-associated antigens to DCs. Furthermore, we have previously reported that treatment of GBM cells with Ad-TK/GCV or TMZ results in the release of the endogenous TLR2 ligand HMBG1 from GBM cells (18). Autophagic GBM cells have also been shown to release HMGB1, without causing lysis of the cell membrane and classical necrosis (45). Previous reports showed that DCs loaded with purified autophagosomes from autophagic tumor cells induced tumor-specific immune responses (26), suggesting that autophagosomes contain a wide range of tumor-associated antigens and immune adjuvants, that is, HMGB1 (46). There is experimental evidence that suggests that intratumoral delivery of a recombinant cytotoxin composed of Pseudomonas exotoxin fused to IL-13 into mice bearing human xenografts causes cell death not only by apoptosis but also by necrosis. Furthermore, Kawakami and colleagues showed that delivery of the IL-13-PE cytotoxin induced phagocytes may play a role in cytotoxin-mediated tumor regression (47). In addition, it has been previously shown that the fusion of the recombinant Pseudomonas exotoxin to a model tumor antigen may enhance vaccine potency (48).

GM-CSF combined with IL-4 has been previously used to generate DCs from both murine and human bone marrow progenitor cells (30). Flt3L has been introduced as an alternative means to generate DCs (35, 37, 49) and recent reports showed that Flt3L combined with IL-6 enhances the expansion of Th1-polarizing DCs, a requirement for the efficient induction of antitumor immune responses (50). Thus, we compared the characteristics of GM-CSF + IL-4–and Flt3L + IL-6–generated DCs to establish the optimal parameters for ex vivo expansion of DCs to be used in the vaccination paradigms described. Our results showed that ex vivo conditioning with Flt3L resulted in higher levels of DCs when compared with GM-CSF + IL-4; these results are in line with results previously obtained with canine DC cultures (49). Levels of Th1 polarizing cytokines such as IL-12 and IFN-γ were higher when DCs were cultured with Flt3L + IL-6. These data are in line with evidence that Flt3L induces DCs that preferentially secrete Th1-polarizing cytokines, when compared with GM-CSF cultured DCs (35). Consistent with these results, inhibition of tumor progression and therapeutic efficacy of DC vaccination was higher when we used Flt3L + IL-6–generated DCs. Addition of IL-6 in combination with Flt3L further enhanced the proliferation of BMDCs. Flt3L + IL-6–induced DCs displayed decreased levels of cell surface markers CD80, CD86, and MHCII, and exhibited high phagocytic capacity consistent with an immature phenotype.

The advantage of combining immunogene therapy (Ad-Flt3L/Ad-TK) with DC vaccination is that gene therapy can be administered into the tumor bed/cavity at the time of surgical resection to immediately initiate antibrain tumor immune responses while the autologous DC vaccine with autologous tumor lysate is being prepared ex vivo. After the DC vaccine preparation is completed, it can be administered using the appropriate vaccination regime to enhance the antitumor immunity and therapeutic efficacy. In summary, the data presented show that immunogene therapy not only elicits tumor regression and antitumor immunity but also potentiates the therapeutic efficacy elicited by DC vaccination and support the implementation of novel phase I clinical trials to assess the safety and efficacy of this combined therapeutic strategy.

No potential conflicts of interest were disclosed.

Our work was supported by NIH/National Institute of Neurological Disorders & Stroke (NIH/NINDS) grants 1R21-NS054143; 1UO1-NS052465; and 1RO1-NS057711 to M.G. Castro; NIH/NINDS grants 1RO1-NS 054193 and 1RO1-NS061107 to P.R. Lowenstein; and F32NS0503034 to G.D. King. The Bram and Elaine Goldsmith and the Medallions Group Endowed Chairs in Gene Therapeutics to P.R. Lowenstein and M.G. Castro, respectively, The Linda Tallen & David Paul Kane Foundation Annual Fellowship, The Drown Foundation, and the Board of Governors at CSMC.

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.