We tested whether modulation of the CNS-tumor microenvironment by delivery of IFN-α-transduced dendritic cells (DCs: DC-IFN-α) would enhance the therapeutic efficacy of peripheral vaccinations with cytokine-gene transduced tumor cells. Mice bearing intracranial GL261 glioma or MCA205 sarcoma received peripheral immunizations with corresponding irradiated tumor cells engineered to express IL-4 or GM-CSFs, respectively, as well as intratumoral delivery of DC-IFN-α. This regimen prolonged survival of the animals and induced tumor-specific CTLs that expressed TRAIL, which in concert with perforin and Fas ligand (FasL) was involved in the tumor-specific CTL activity of these cells. The in vivo antitumor activity associated with this approach was abrogated by administration of neutralizing mAbs against TRAIL or FasL and was not observed in perforin−/−, IFN-γ−/−, or FasL−/− mice. Transduction of the tumor cells with antiapoptotic protein cellular FLIP rendered the gene-modified cells resistant to TRAIL- or FasL-mediated apoptosis and to CTL killing activity in vitro. Furthermore, the combination therapeutic regimen was ineffective in an intracranial cellular FLIP-transduced MCA205 brain tumor model. These results suggest that the combination of intratumoral delivery of DC-IFN-α and peripheral immunization with cytokine-gene transduced tumor cells may be an effective therapy for brain tumors that are sensitive to apoptotic signaling pathways.
Cancer vaccines using tumor-Ag loaded dendritic cells (DCs)4 (1) or cytokine-gene transduced tumor cells (2, 3, 4) appear to represent promising therapeutic approaches. Although the CNS has often been described as an immunologically privileged site (5, 6), recent studies by our group and others have shown that peripheral vaccinations can induce effective anti-CNS tumor immune responses (7, 8, 9, 10, 11, 12). Based on our previous results using peripheral vaccination with syngeneic glioma cells that had been genetically engineered to secrete IL-4 (8, 9, 13), we have initiated a clinical trial using autologous glioma cells as a vaccine (10, 14, 15). However, the CNS and CNS-localized tumors still represent sites that are suboptimal for the infiltration and function of APCs, such as DCs (16) and antitumor effector cells (17), suggesting a further necessity for modifying the CNS and CNS tumor microenvironment to achieve a better clinical response.
As the CNS and cancers may use similar molecular mechanisms to suppress inflammatory immunity as a “self-defense,” it is not surprising that a recent study has reported that plasmacytoid DCs infiltrating head and neck cancer exhibit deficiencies in producing IFN-α and IL-23 in response to CpG oligonucleotides (18). Delivery of exogenous IFNs in the CNS and brain tumor immunological environment resulted in a remarkable up-regulation of MHC class II on tumor-infiltrating APCs (19) and enhanced recruitment of Ag-specific T cells (20), providing a rationale for site-specific modulation of the CNS microenvironment by cytokine-based immunotherapy.
Type I IFNs, including IFN-α, -β, -δ, and -κ (21, 22), mediate several functions, including the inhibition of cellular proliferation and the generation and cytotoxic activities of activated NK cells and CTLs, and modulate MHC molecule expression (23). Furthermore, IFN-α plays an important role in the generation and maturation of the DCs (24) and in the up-regulation of DC-expressed TRAIL expression, thereby allowing DCs to mediate more effective direct tumoricidal activity (25). Type I IFNs also enhance the recruitment of Ag-experienced type 1 T cells into tumor sites by promoting the expression of ICAM-1 on the proximal microvascular endothelium (26, 27, 28). More recent studies have demonstrated that IFN-α promotes cross-priming by increasing the expression of the peptide transporter TAP-1 in DCs (29), with the critical role of type I IFNs in cross-priming of Ag-specific CTLs directly demonstrated in IFN-αβR-deficient mouse models (30).
With regard to the molecular basis of the CTL function, death receptor ligands such as TNF-α, CD95 (Fas ligand (FasL)), or TRAIL may be used as the mediators of cytotoxic activity (31). TRAIL acts as a tumor suppressor molecule in vivo by inducing apoptotic death of tumor cells (32, 33, 34). However, the degree of sensitivity of tumor cells to death receptor ligand-mediated apoptosis could significantly influence their sensitivity to immunotherapeutic modalities. An inhibitor of apoptosis cellular FLIP (cFLIP) interacts with Fas-associated death domain protein and the protease FLICE and potently inhibit apoptosis induced by all known human death receptors (35). Indeed, cFLIP regulates the susceptibility of glioma cells to FasL (36) and TRAIL (37).
In the present study, we addressed the hypothesis that intratumoral (i.t.) delivery of IFN-α by ex vivo-activated DCs enhances the efficacy of peripheral vaccinations with cytokine-gene transduced syngeneic tumor cells. This combinational therapy appears to enhance the antitumor cytolytic activity of vaccine-induced effector cells mediated via TRAIL, FasL, and perforin. Ectopic expression of cFLIP in the tumor cells abrogated the efficacy of the tumor-specific CTLs and the combinational therapy in vivo. We believe that such combinational therapeutic approaches may prove efficacious in the setting of established CNS tumors that are not refractory to these lytic pathways.
Materials and Methods
C57BL/6 mice (H-2b), C57BL/6-background Thy1.1 (B6.PL-Thy1a/CyJ), perforin-deficient (C57BL/6-Pfptm1sdz), IFN-γ-deficient (B6129S7-ifng), and FasL-deficient (B6Smn, C3H-Faslgld) mice (6–8 wk old) were purchased from The Jackson Laboratory. IL-12/IL-23 p40-deficient animals were maintained as described previously (38). Animals were handled under aseptic conditions in microisolator cages within the Central Animal Facility at the University of Pittsburgh per an Institutional Animal Care and Use Committee-approved protocol and in accordance with recommendations for the proper care and use of laboratory animals.
Tumor cell lines
Mouse (H-2b) glioma GL261 and fibrosarcoma MCA205 were maintained in RPMI 1640 supplemented with 10% FBS, 2 mM l-glutamine, 100 IU/ml penicillin, and 100 μg/ml streptomycin. MCA205 cells stably transduced with retroviral vectors DFG-cFLIP-neo that encode cFLIP and neomycin-resistance (neo-R) (MCA205-cFLIP) or backbone-MFG-neo (MCA205-neo) were obtained from Dr. P. Robbins (University of Pittsburgh, Pittsburgh, PA). Transduced cell cultures were maintained in the presence of G418 (400 μg/ml) in the culture medium.
Generation of DCs in vitro from bone marrow (BM)
C57BL/6 mouse-derived BM cells were generated as described previously (7). Briefly, cells were cultured in complete medium supplemented with 1000 U/ml recombinant murine GM-CSF and recombinant murine IL (mIL)-4 (Schering-Plough) at 37°C in a humidified, 5% CO2 incubator for 7 days. DCs were then isolated at the interface of 14.5% (w/v) metrizamide (Sigma-Aldrich) in complete medium discontinuous gradients by centrifugation. DCs typically represented >90% of the harvested population of cells based on morphology and expression of the CD11b, CD11c, CD40, CD54, CD80, CD86, and class I and class II MHC Ags as monitored by flow cytometry (data not shown).
The mock adenoviral vector (Ad) Ad-ψ5 and the Ad-encoding mouse cytokine cDNAs were produced as previously reported (39) and provided by the University of Pittsburgh Cancer Institute’s Vector Core Facility.
Adenoviral transduction of DCs and tumor cells
Five million (day 7 cultured) DCs or tumor cells were transduced with Ads at a multiplicity of infection (MOI) of 50, as reported previously (39). After 48 h, adenoviral-transduced cells were harvested and analyzed for phenotype and function. Culture supernatants were also collected for measurement of mouse cytokine production using species-specific ELISA kits (eBioscience).
Tumor challenge and i.t. delivery of DC-IFN-α
To create brain tumor-bearing animals, each animal received an intracranial (i.c.) injection of GL261 (1 × 105/mouse) or MCA205 (1 × 105/mouse) cells as described previously (7). Briefly, using a 10-μl Hamilton syringe, 1 × 105 tumor cells suspended in 2 μl of PBS were stereotactically injected through an entry site at the bregma 2 mm to the right of the sagittal suture and 3 mm below the surface of the skull of anesthetized mice using a Kopf stereotactic frame (Kopf Instruments). Some animals bearing i.c. tumors also received an i.t. injection with 1 × 105 adenovirally transduced DCs in the same location on 4 days after the i.c. tumor challenge. The animals were monitored daily after treatment for manifestation of any pathologic signs associated with elevated i.c. pressure, such as hemiparesis, loss of appetite, or any altered grooming habits. Affected animals were euthanized by CO2 inhalation. Representative animals in each treatment cohort were euthanized at selected time points to obtain tissues (brain tumors, lymphoid organs) for immunological analyses.
Peripheral tumor cell vaccines
Mice bearing i.c. GL261 glioma were injected s.c. with 2 × 106 irradiated (50 Gy) GL261-IL4 or GL261-ψ5 as tumor cell vaccines on days 3, 5, and 7 after the i.c. tumor inoculation. Mice that received 2 × 106 irradiated (50 Gy) MCA205-GM-CSF or MCA205-ψ5 s.c. were challenged i.c. with MCA205 cells (1 × 105/mouse).
CTL activity assay
Spleens or cervical lymph nodes were resected, and single-cell suspensions (2 × 106 cells/ml) were restimulated in vitro with irradiated (50 Gy) MCA205 or GL261 cells (5 × 104 cells/ml) in the presence of 10 U/ml human IL-2 (Chiron) and 50 μM 2-ME (Sigma-Aldrich) in 24-well plates (Corning Glass) for 5 days. Specific CTL activity was determined in 8 or 20 h 51Cr release assays against relevant, control, and irrelevant tumor cells, as described previously (40). Assays were performed in triplicate wells, with spontaneous release of all assays never exceeding 25% of maximum release.
Cytokine production from lymphocytes
Lymphocytes obtained from mice vaccinated with GL261-IL4 were restimulated in mixed lymphocyte tumor cell cultures with irradiated GL261 cells for 6 days, as previously described (41), with slight modifications. Briefly, CD4+ and CD8+ cells were negatively selected by paramagnetic beads conjugated with Ab mixtures (MicroBeads; Miltenyi Biotec) and tested for cytokine production by culturing lymphocytes (2 × 105/well) in 96-well flat-bottom plates with 1.0 μg/well of anti-CD3 mAb (145-2C11; R&D Systems) at 37°C for 20 h. Culture supernatants were assayed for cytokine content using specific ELISAs (eBioscience).
Microscopic analyses of the tumor tissues
For immunohistochemical analyses of immune cell infiltration, animals bearing i.c. tumors were euthanized at indicated intervals after tumor inoculation. Whole brains bearing i.c. tumors were fixed for 2 h in 4% paraformaldehyde (in PBS) and then cryoprotected in 30% sucrose in PBS before being shock-frozen in liquid nitrogen-cooled isopentane. Five-micrometer frozen sections were then made on a cryostat. After air-drying, sections were stained with PE-conjugated specific Abs or isotype controls. The sections were then mounted in Vectashield H1000 (Vector Laboratories) and observed using a Nikon Eclipse E800 microscope equipped with a cooled charge-coupled device color camera.
To assess the fate and function of DCs injected into the i.c. tumor site, day 7 BM-derived DCs were transduced with Ad-IFN-α, as indicated above. Forty eight hours later, the DCs were labeled with PKH67-green, and 1 × 105 virally infected DCs were injected into i.c. PKH-26 red-labeled GL261 tumors established in syngeneic C57BL/6 (H-2b) mice. After 1 additional day, whole brains and cervical lymph nodes were harvested and fixed with 4% paraformaldehyde. Sections (1 mm) were imaged for PKH67-green labeled DCs and PKH26-red-labeled GL261 glioma cells using a two-photon microscope comprising a titanium-sapphire ultrafast tunable laser system (Coherent Mira Model 900-F), Olympus Fluoview confocal scanning electronics, an Olympus IX70 inverted system microscope, and custom built input-power attenuation and external photomultiplier detection systems. Single-plane image acquisition used two-photon excitation at 850 nm with Olympus water immersion objectives (×20 UApo 0.7NA, ×60 UplanApo 1.2NA). Emission filters (Chroma Technology) comprised a HQ535/50m filter (green emission) and 565dclp dichroic mirror.
Adoptive transfer of peripheral vaccine-induced lymphocytes into GL261-bearing mice
Thy1.1+ C57BL/6 mice (The Jackson Laboratory) were inoculated with 2 × 106 GL261-IL4 or PBS s.c. on the right flank three times every other day. Five days after the last peripheral vaccination, draining inguinal lymph node-derived cells and splenocytes (SPCs) were harvested and adoptively transferred (5 × 106/mouse) i.v. into Thy1.2+ C57BL/6 mice bearing day 6 i.c GL261 gliomas that had previously received i.t. DC-IFN-α or DC-ψ5 (1 × 105/mouse) injections on day 4 posttumor implantation. Three days after the adoptive transfer, brain-infiltrating lymphocytes (BILs) from the tumor-bearing brain hemisphere and cells in the draining cervical lymph nodes (CLNs) were harvested, stained with FITC-conjugated anti-mCD8 Abs and PE-conjugated anti-mThy1.1 Abs (BD Biosciences), and examined by flow cytometry (Coulter EPICS cytometer, Beckman Coulter). For isolation of BILs, mice were euthanized by CO2 asphyxia, then immediately perfused through the left cardiac ventricle with PBS to remove residual intraventricular blood pools. Brains were removed and BILs were isolated by enzymatic digestion and modified Percoll Gradients (Sigma-Aldrich) as described previously (42). The enzymes used in this procedure do not degrade the principal surface receptors on T cells, allowing ex vivo phenotypic analyses of cells to be performed without in vitro culture.
Survival estimates and median survival times were determined using the method of Kaplan and Meier. Survival data were compared using a log-rank test. Comparative T cell responses were compared by Student’s t test for two samples with unequal variances. Statistical significance was determined at the <0.05 level.
Cytokine production and direct tumoricidal activity mediated by IFN-α-transduced DCs
We evaluated the effects of IFN-α transduction on murine BM-derived DCs with regard to their expression of type 1 cytokines and their direct cytotoxic activity against syngeneic murine tumor cell lines in vitro (Fig. 1).
As demonstrated in Fig. 1,A, at 24 h following the transduction with Ad-IFN-α, DCs produced 3100 ± 330 (pg/ml) of murine IFN-α, with nonstimulated DCs failing to produce detectable levels of IFN-α. Although transduction of DCs with Ad-ψ5 or stimulation of nontransduced DCs with LPS or poly(I:C) induced a slight augmentation in IFN-α production by treated DCs, these levels were 70- to 80-fold lower than those observed for DCs transduced with Ad-IFN-α. Interestingly, transduction of DCs with Ad-IFN-α also induced DC secretion of IL-12p70 and IFN-γ. Although LPS treatment and Ad-ψ5 transduction also induced similar levels of IFN-γ, the induction of IL-12p70 appeared to represent a unique event associated with Ad-IFN-α but not Ad-ψ5 transduction. As human DCs may mediate the direct apoptotic death of cancer cells via multiple cytotoxic TNF family ligands, we also evaluated the direct cytotoxic effects of engineered DCs against syngeneic mouse tumor cells (Fig. 1,B). Although DCs exhibited a higher degree of cytotoxicity than “effector” fibroblasts against GL261 gliomas and MCA205 sarcoma cells (p < 0.01 for both DC-ψ5 vs fibroblast-ψ5 and DC-IFN-α vs fibroblast-IFN-α), transduction of DCs with Ad-IFN-α did not appear to significantly enhance DC-mediated killing to levels greater than those observed for effector DC-ψ5 (p = 0.33 and 0.114 against GL261 and MCA205, respectively). In our flow cytometric analyses, we were unable to detect any dramatic changes in TRAIL and FasL expression by Ad-IFN-α or Ad-ψ5-infected DCs (data not shown). These data suggest that while Ad-IFN-α transduction may not yield a better DC killer cell, it likely results in an APC possessing higher Ag-presentation and type 1 T cell stimulatory capabilities (when compared with nontransduced DC or DC-ψ5) and that the improved stimulatory capacity presented in Fig. 2 may result from augmented DC production of not only IFN-α but also (indirectly) up-regulated secretion of IL-12p70 and IFN-γ.
DC-IFN-α injected intratumorally into established i.c. GL261 glioma induce potent CTL responses in the draining CLN
To evaluate whether i.t. injections with DC-IFN-α influence the induction of tumor-specific T cell responses in the draining CLN, mice bearing day 5 i.c. GL261 gliomas received intratumoral injections with 1 × 105 DC-IFN-α. Control animals received the same number of fibroblasts transduced with Ad.IFN-α or DCs infected transduced with Ad.ψ5. CLN cells (CLNCs) obtained from animals that had received DC-IFN-α therapy exhibited remarkable GL261-specific CTL responses (Fig. 2,A) and IFN-γ production (Fig. 2 B), following a single round of in vitro restimulation. Intratumoral delivery of syngeneic fibroblasts secreting similar levels of ectopic IFN-α (fibroblasts-IFN-α) induced only a modest alteration in CTL and IFN-γ responses to GL261 (vs DC-ψ5 treated or untreated controls), and injection with DC-ψ5 failed to induce either tumor-specific CTL or IFN-γ responses (above background) in this model. These data suggest that the genetic modulation of DCs with Ad-IFN-α dramatically promotes the Ag-presenting capability of DCs in the draining CLNs when applied in the setting of the i.c. microenvironment of GL261 gliomas.
There has been some controversy as to whether induction of T cell responses against brain-derived Ags requires the migration of brain-localized APCs that have captured Ags to the CLNs or whether the drainage of brain-derived Ags to the CLNs is sufficient to support this process (43). To investigate whether DC-IFN-α physically associate with the tumor cells and consequently migrate to the CLNs, we also performed two photon microscopic analyses using fluorescence labeled GL261 glioma cells and DCs (Fig. 2,C). Animals bearing PKH26-red-labeled GL261 tumors received an injection with PKH67-green-labeled DC-IFN-α. At 48 h later, the brain and bilateral CLNs were harvested, fixed, sectioned, and analyzed by two-photon microscopy. The original tumor site in the brain demonstrated the presence of both injected tumor cells and DCs (Fig. 2,CI). At a higher magnification of the same section, in addition to representative PKH67-green-labeled DC-IFN-α and PKH26-red-labeled GL261 cells (Fig. 2,CII; right panels), there were some PKH67-green-labeled cells that appeared to be physically interacting with PKH26-red-labeled cells displaying yellowish fluorescence signals as a result of the colocalization of green and red signals (Fig. 2,CII; left panels with arrows). CLNs demonstrated the presence of PKH67-green-labeled DCs that also express PKH26-red-derived signals (Fig. 2 CIII with arrows), suggesting that the DCs had phagocytosed PKH26-red-labeled GL261 cells or cellular debris, then migrated to the CLNs. These functional (CTL activity) and morphological data indicate that DC-IFN-α are potent APCs that actively acquire tumor Ags within brain tumor sites before migrating to the CLN where they present these Ags to cognate T cells.
Intratumoral DC-IFN-α injection enhances the therapeutic effects of peripheral tumor cell-based vaccines
We have also assessed the efficacy of combinational DC-IFN-α cytokine gene therapy strategies that incorporate peripheral vaccination with gene-modified tumor cells. Syngeneic C57BL/6 mice with established i.c. GL261 tumors received s.c. injections of irradiated GL261 cells that had been engineered ex vivo to secrete murine IL-4 (GL261-IL4) on days 3, 5, and 7 following the i.c. tumor inoculation, as well as i.t. treatments, including the injection of DC-IFN-α (Fig. 3,A). Fig. 3,B provides representative macroscopic pictures of the brains obtained on day 30 following the i.c. tumor inoculation. An animal that had received i.t. DC-IFN-α and s.c. GL261-IL4 (Fig. 3,Ba) had no visible tumor mass around the needle track for tumor and DC inoculations (left; arrow), whereas a control animal (treated with i.t. DC-ψ5 and s.c. GL261-ψ5) had a massive tumor in the right hemisphere (Fig. 3 Bb; with arrows).
In control animals, i.t. DC-IFN-α injections and s.c. GL261-IL4 vaccinations were replaced by i.t. DC-ψ5 injection and/or s.c. GL261 transduced with backbone Ad-ψ5 (GL261-ψ5) or i.t. injection with syngeneic fibroblasts transduced with Ad-IFN-α (Fig. 3 C). When animals received both i.t. DC-based IFN-α and peripheral vaccinations with GL261-IL4, this combination regimen resulted in a significant improvement in survival (with 8 of 15 mice survived longer than 80 days) in comparison to other regimens for i.t. injections in the animals that received s.c. GL261-IL4 vaccines (p < 0.0001 for i.t. DC-IFN-α + s.c. GL261-IL4 vs i.t. DC-ψ5 + s.c. GL261-IL4; and for i.t. DC-IFN-α + s.c. GL261-IL4 vs i.t. fibroblast-IFN-α + s.c. GL261-IL4). None of the animals that received i.t. injections with DC-ψ5 or fibroblasts secreting similar levels of IFN-α survived longer than 47 or 36 days, respectively, suggesting that injection of i.t. DC-IFN-α was critical in enhancing the effects of peripheral vaccines.
Although i.t. DC-IFN-α delivery exhibited significant therapeutic effects without local IL-4 expression at the peripheral vaccine site (p = 0.0022 for i.t. DC-IFN-α + s.c. GL261-ψ5 vs i.t. DC-ψ5 + s.c. GL261-ψ5 treatment regimens) with 2 of 15 mice surviving longer than 80 days, addition of local transgene-derived IL-4 secretion at the peripheral vaccine site significantly improved the therapeutic efficacy (p = 0.0208 for i.t. DC-IFN-α + s.c. GL-IL4 vs i.t. DC-IFN-α + s.c. GL-ψ5). None of the 15 mice that had received s.c. vaccinations with GL261-IL4 and i.t. DC-ψ5 resulted in the long-term survival (>90 days).
Moreover, 80 days after initial tumor challenge, surviving mice from i.t. DC-IFN-α and s.c. GL261-IL4 therapy were rechallenged i.c. with 1 × 105 parental GL261 tumor cells in the contralateral hemisphere from the primary tumor injection site. All mice (n = 3) exhibited protection and long-term survival until the end of the observation (130 days after initial tumor challenge), with no sign of tumor growth upon necropsy. Control nonimmunized mice receiving GL261 tumor cells died within 30 days after tumor challenge, suggesting that protective immunity to lethal tumor cell challenge was established in tumor-bearing mice treated with i.t. DC-IFN-α and peripheral vaccinations with GL261-IL4.
We also examined the efficacy of i.t. DC-IFN-α delivery can be extended to another i.c. H-2b tumors, namely i.c. MCA205 fibrosarcoma model (44, 45) (Fig. 4). Our pilot experiments with the i.c. MCA205 model revealed that peripheral s.c. vaccinations with MCA205 transduced with Ad-GM-CSF (MCA205-GMCSF) induced more effective antitumor immunity than MCA205-IL4, suggesting that the cytokine of choice for peripheral transduced tumor cell vaccination may vary for each tumor type and/or cell line evaluated (9, 46). In addition, with i.c. MCA205 tumors and s.c. vaccinations with MCA205-GMCSF, we were able to demonstrate prolonged survival only in prophylactic models (data not shown). Therefore, we used a preimmunization paradigm with a s.c. vaccination with MCA205-GMCSF as illustrated in Fig. 4,A. With a control s.c. vaccination with MCA205-ψ5, none of the animals survived for longer than 30 days after the i.c. tumor inoculation, regardless of whether DC-ψ5 or DC-IFN-α were used for i.t. injections. Although s.c. vaccination with the MCA205-GMCSF and i.t. DC-ψ5 regimen resulted in long-term survival (>90 days) in 1 of 10 animals, the combination of i.t. DC-IFN-α delivery and s.c. vaccination with MCA205-GMCSF resulted in the highest percentage of long-term survival in 6–10 animals (p = 0.0262 for i.t. DC-IFN-α plus s.c. MCA205-GMCSF vs i.t. DC-ψ5 plus s.c. MCA205-GMCSF) (Fig. 4 B).
In addition, 80 days after initial tumor challenge, three surviving mice from the i.t. DC-IFN-α plus s.c. MCA205-GMCSF group were rechallenged with 1 × 105 parental MCA205 tumor cells in the contralateral hemisphere from the primary tumor injection site. All mice exhibited protection and long-term survival until the end of the observation (130 days after initial tumor challenge). Control nonimmunized mice receiving MCA205 tumor cells died within 15 days after tumor challenge (data not shown). These results indicate that protective immunity to lethal tumor cell challenge was established in mice treated with i.t. DC-IFN-α delivery and peripheral vaccinations with MCA205-GMCSF.
Taken together, the efficacy of the combination strategy consisting of i.t. DC-IFN-α delivery and s.c. cytokine-tumor vaccine was confirmed in two distinct tumor cell models.
Induction and characterization of specific CTLs against the GL261 and MCA205 tumor cell lines
We subsequently examined whether specific and systemic CTL responses had been induced in those animals that had rejected their lesions in response to the combinational treatment regimens. As systemic CTL responses against CNS tumors have been demonstrated previously in SPCs on day 11 or thereafter following inoculation of tumor cells in the brain (17, 47), SPCs were isolated from GL261- or MCA205-bearing animals that survived for longer than 90 days following i.t. DC-IFN-α plus peripheral vaccinations with cytokine-transduced corresponding tumors and stimulated in vitro with irradiated tumor cells in the presence of 50 U/ml human IL-2 for 7 days. The responder lymphocytes were analyzed and found to be specific in mediating cytotoxicity against the relevant tumor cells (Fig. 5,A). Expression of TRAIL was high on CTLs raised against GL261 but only barely higher than the background level on anti-MCA205 CTL. Both CTLs also expressed FasL and lymphocyte function-associated Ag (LFA-1) (Fig. 5,B), as well as CD3 and CD8 (data not shown). This phenotype prompted us to investigate whether perforin, TRAIL and/or FasL play key roles in the anti-tumor reactivity of the CTLs. A specific perforin inhibitor concanamycin A (CMA) inhibited the activity of anti-GL261 CTL in the dose-dependent fashion (Fig. 5,C). In addition, neutralizing anti-TRAIL, anti-FasL, and the combination of anti-TRAIL + anti-Fas Abs inhibited the activity of anti-GL261 CTLs (Fig. 5 D). These data suggest that perforin, TRAIL, and FasL each play significant roles in anti-GL261 cytotoxicity of CTLs primed in vivo.
Intratumoral DC-IFN-α delivery enhanced the trafficking of CD8+ T cells to the tumor site and cervical lymph nodes
To evaluate the impact of DC-IFN-α therapy on the trafficking of vaccine-activated lymphocytes into the i.c. tumor site and draining CLNs, we isolated and characterized BILs and CLNCs following the adoptive transfer of SPCs (derived from GL261-IL4-vaccinated Thy1.1+ mice) into recipient (Thy1.2+) mice bearing i.c. GL261 tumors. Fig. 6 depicts the total numbers of BILs (Fig. 6,A) and CLNCs (Fig. 6,D) per mouse, as well as Thy1.1+ (donor derived; Fig. 6, B and D) and total (both donor and recipient derived; Fig. 6, C and F) CD8+ BIL and CLNC/mouse analyzed by flow cytometry. Due to the limited numbers of BILs obtained, only CD8+ subsets were analyzed. Intratumoral DC-IFN-α delivery resulted in the accumulation of larger numbers of total lymphocytes, donor-derived (Thy1.1+) and total (donor and recipient derived) CD8+ T cells when compared with tissues isolated from animals treated with injections of DC-ψ5. Indeed, i.t. delivery of control DC-ψ5 cells did not lead to the recruitment of sufficient BILs to allow for the flow cytometric assessment of Thy1.1+CD8+ cells (Fig. 6 B). A substantial number of recipient-derived CD8+ T cells appear to contribute to the observed increase in total CD8+ BIL and CLN cells. These results suggest that i.t. delivery of DC-IFN-α promotes the recruitment and activation of effector CD8+ T cells within the i.c. tumor site and the draining CLN.
Vaccine-stimulated CD4+ and CD8+ T cells play critical roles in the antitumor effects promoted by combinational cytokine gene therapy
The role of the CD4+ and CD8+ T lymphocytes in the antitumor response was investigated by in vivo depletion experiments. Mice treated with i.t. DC-IFN-α and s.c. GL261-IL4 were injected i.p. with anti-CD4 mAb, anti-CD8 mAb, or control Ab immediately before and during the therapy to specifically deplete each subset of immune cells. We observed that depletion of any of these cell populations abrogated the efficacy of this combinational immunogene therapy, whereas control IgG injections had no effect (Fig. 7), suggesting that each of the CD4+ T cell and CD8+ T cell effectors play important roles in supporting the efficacy of this therapeutic regimen.
Perforin, TRAIL, FasL, and IFN-γ also play critical roles in the in vivo antitumor effect elicited by the i.t. DC-IFN-α plus s.c. GL261-IL4 combination therapy
On the basis of in vitro data demonstrating perforin, TRAIL and FasL contribute to the anti-GL261 CTL activity; we also evaluated the roles of these individual effector molecules in our GL261 combinational treatment model. Administration of anti-TRAIL or anti-FasL mAb alone abrogated the anti-i.c. tumor effectiveness of this strategy, although the combined treatment with anti-TRAIL and anti-FasL mAbs did not further abbreviate survival (Fig. 8,A). To discern the requirement of perforin, FasL, and type 1 cytokines, including IFN-γ and IL-12/IL-23 p40 in the antitumor effect, gene knockout mice deficient in each of these molecules received the same i.c. GL261 inoculation and the combinational immunogene therapy regimen. Fig. 8 B demonstrates that the host-derived perforin, FasL, and IFN-γ were critical to the observed antitumor effects, whereas three of six IL-12/IL-23 p40-deficient mice survived longer than 90 days, suggesting that the antitumor effects elicited by this regimen were less affected by endogenous IL-12/IL-23 production.
Anti-MCA205 and anti-GL261 CTL responses, as well as in vivo antitumor effects associated with combinational IFN-α gene therapy, are abrogated in tumors overexpressing cFLIP
Among the various ways that tumor cells escape from immunosurveillance, their resistance against death receptor-induced apoptosis may represent an important survival mechanism (48, 49). The antiapoptotic protein cFLIP protects cells from Fas- and TRAIL-induced apoptosis (50, 51). Indeed, cFLIP expression in glioma cells renders them resistant to FasL (36)- and TRAIL (37)-mediated apoptosis. We next addressed whether cFLIP overexpression by tumor cells would obviate the generation and antitumor efficacy of therapy-induced anti-GL261 and anti-MCA205 CTLs. Both MCA205 and GL261 cells were sensitive to apoptosis induced by murine TRAIL and murine FasL, but ectopic expression of cFLIP dramatically suppressed the TRAIL- or FasL-induced lytic death of both cell lines in vitro (data not shown). As shown in Fig. 9, A and B, tumor cell ectopic expression of cFLIP dramatically suppressed the activity of tumor-specific CTLs that were capable of efficiently lysing mock-transduced relevant target tumor cells. Finally, the role of cFLIP was examined in animals bearing i.c. MCA205 tumors (Fig. 9 C). Syngeneic animals that had been preimmunized s.c. with MCA205-GMCSF received i.c. injection with MCA205-cFLIP or control MCA205-neo and then received i.t. injection with DC-IFN-α. Four of 10 animals that had received MCA205-neo survived longer than 180 days, whereas all animals bearing MCA205-cFLIP (n = 10) died by day 73 (p = 0.0132). These data demonstrate that the efficacy of therapy using i.t. DC-IFN-α and s.c. vaccinations with cytokine-transduced tumor cells relies upon intact intracellular apoptosis-inducing pathways and that the resistance of tumor cells against apoptosis may serve as a dominant limiting variable to effective combinational immunotherapies designed to elicit effector T cells.
Our current study demonstrates that i.t. delivery of IFN-α-transduced DCs remarkably enhances the efficacy of peripheral tumor vaccines for brain tumors that are sensitive to apoptotic signaling pathways.
Adenoviral transduction of DCs with IFN-α induced constitutive IFN-γ and IL-12p70 from the transduced DCs, a finding consistent with reports that human DCs produce IL-12p70 in response to IFN-α stimulation (52). Intratumoral injection of DC-IFN-α induced potent antitumor CTL and IFN-γ responses in the organ-draining CLN and enhanced the therapeutic effects of peripheral vaccinations with cytokine-gene transduced tumor cells, which is consistent with our previous observation that i.t. delivery of DC-IFN-α promotes the cross-presentation of OVA-derived T cell epitopes in the CLNs and promoted therapeutic effects of OVA-specific vaccines in an OVA expression tumor model (53). As peripheral s.c. vaccines with IL-4- or GM-CSF-transduced tumors induced the high levels of tumor-specific IFN-γ production from host SPCs (54), it is presumed that this “prime/boost” approach enhances antitumor immunity by amplifying the pool of peripheral vaccine-induced type 1 antitumor T cells in the CLNs. Also, based on a recent study demonstrating that the site of Ag capture by cross-presenting APCs (i.e., within the CNS tumor site) is the critical factor for in vivo imprinting of brain-tropic homing phenotypes on tumor-specific T cells (42), DC-IFN-α injected into the CNS tumor may preferentially prime antitumor CTLs that possess brain-tropic surface adhesion molecule profiles.
We used IL-4-transduced GL261 glioma and GM-CSF-transduced MCA205 sarcoma cells in the current study because peripheral vaccinations with GL-261-IL4 and MCA205-GMCSF exerted more efficacious therapeutic immunity against corresponding i.c. tumors in comparison to vaccinations with reciprocal cytokine-cell line combinations (i.e., GL261-GMCSF and MCA205-IL4) in our pilot experiments. In this pilot study, both GL261 and MCA205 tumor cells secreted similar levels of IL-4 and GM-CSF following specific adenoviral transduction (unpublished data). Therefore, differential immunity provoked against each of the transduced tumor vaccines is unlikely to be related to technical aspects, such as variance in vector preparation and transduction efficiency. It is possible that IL-4 may promote the expression of a different cohort of homing receptors on effector T cells that dictates a more favorable homing pattern into gliomas, and we are currently evaluating this possibility. Nevertheless, our data do support previous observations by others that the optimally “immunogenic” cytokine for integration into peripheral tumor cell-based vaccines may vary between tumor types and/or cell lines (9, 46).
Another important implication from the current study is that our combinational cytokine gene therapy induced tumor-specific CTL responses that used TRAIL, FasL, and perforin pathways to mediate tumor cell death in vitro and in vivo. With regard to the roles of TRAIL in antitumor immunity, it has been well established that TRAIL plays a major role in NK cell-mediated antitumor immunosurveillance (33). Tumor-specific CD4+ CTLs use TRAIL to exert their specific cytotoxicity against TRAIL-sensitive tumors (55, 56, 57). IFN-α induced TRAIL expression on mitogen-activated CD4+ and CD8+ human T cells, and their ability to mediate the cytolysis of target cells involves the TRAIL pathway (58). A clear role for TRAIL in the T cell-mediated immune defense against the tumor was demonstrated in the graft vs tumor model (59). In the report, type I IFNs stimulated TRAIL expression on T cells in vitro; however, it was not known whether administration of type I IFNs in vivo could have further enhanced the graft vs tumor effects. In our study, we hypothesized that local i.t. delivery of DC-IFN-α might up-regulate surface expression of TRAIL on tumor-infiltrating effector T cells. Although we were able to demonstrate TRAIL expression and its role in tumor-specific CTLs, we could only detect a slight enhancement of TRAIL expression on CTLs by coincubation of the CTLs with DC-IFN-α in vitro (data not shown). Nevertheless, our results demonstrate a critical involvement of TRAIL in the anti-CNS tumor CTL response. Although expression of TRAIL on the tumor-specific CTLs diminished within 6 h after a withdrawal from the stimulation with relevant tumor Ags and IL-2, continued stimulation with IFN-α maintained augmented levels of TRAIL expression (data not shown). It has been reported recently that tumor-infiltrating plasmacytoid DCs exhibit reduced production of IFN-α and IL-23 in response to CpG triggering of TLR9 (18). Hence, the tumor microenvironment may suppress the in situ production of these cytokines that would otherwise be postulated to promote type 1 antitumor T cell responses. Therefore, the local expression of IFN-α at the CNS tumor site may contribute to maintain TRAIL expression on activated infiltrating CTLs, thereby prolonging and accentuating their antitumor effectiveness. Models incorporating the administration of neutralizing anti-FasL mAb or the use of perforin-, IFN-γ- or FasL-deficient mice further corroborated the critical therapeutic role(s) of each of these molecules in vivo.
Our adoptive transfer experiments demonstrated that i.t. DC-IFN-α delivery promoted recruitment of CD8+ T cells in the i.c. tumor site. Indeed, it was noteworthy that DC-IFN-α delivery attracted not only donor-derived (peripherally immunized) but also recipient-derived (peripherally nonimmunized) CD8+ T cells. This is consistent with our observation in earlier parts of the current study that i.t. delivery of DC-IFN-α without peripheral vaccination induces significant levels anti-GL261 CTL responses in the CLNs. Although the current model has not assessed the Ag-specific nature of the infiltrating T cells, when taken together with our previous study showing i.t. DC-IFN-α therapy promotes the activation of brain tumor-specific CTLs (60), these cumulative results suggest that i.t. delivery of DC-IFN-α promotes the recruitment and activation of effector CD8+ T cells within the i.c. tumor site and the draining CLN.
Our data from mice depleted of CD4+ or CD8+ T cells, as well that derived from IFN-γ-deficient mice that failed to reject the i.c. tumors, supports our notion that type 1 T cell responses are critical to the success of our in vivo therapy model. With regard to the roles of CD4+ T cells, although regulatory CD4+CD25+ T cells down-regulated the IFN-α-induced antitumor immunity (61), CD4+ Th cells appear critical in our model, which is consistent with similar findings by others in alternate tumor models (62, 63). However, our results using IL-12/IL-23-deficient mice suggest that biologically functional IL-12 and IL-23 may not be necessary for the therapeutic effectiveness of this approach.
Lastly, cFLIP transduction rendered the tumor cells resistant against TRAIL-, FasL-, and CTL-mediated cytotoxicity in vitro, and ectopic expression of cFLIP in the CNS-implanted MCA205 tumors abrogated the therapeutic effect by our combinational therapy. This may limit the ultimate clinical use of such therapies to those tumors that are not refractory to TRAIL-, FasL- and perforin-mediated lysis. Defection of death receptor signaling appears to be one of the major mechanisms of “tumor escape from immunosurveillance” (64), and therefore, preclinical screening for the expression of cFLIP in the tumor tissue might be done to see what patients might get the most benefit from the vaccine therapy.
On the basis of our current study, novel clinical studies that evaluate the safety and efficacy of cytokine-gene modified DCs, such as DC-IFN-α, delivered in the CNS tumors in combination with peripheral vaccines tumors appear to be warranted.
We thank Hideo Yagita (Juntendo University) for providing mAbs (N2B2 and MLF-3).
The authors have no financial conflict of interest.
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.
This work was supported by NIH/NINDS P01 NS40923 (to. I.F.P., S.C.W., H.O.) and a grant from the Copeland Fund of the Pittsburgh Foundation (to H.O.).
Abbreviations used in this paper: DC, dendritic cell; FasL, Fas ligand; cFLIP, cellular FLIP; i.t., intratumoral; BM, bone marrow; mIL, murine IL; Ad, adenoviral vector; i.c., intracranial; BIL, brain-infiltrating lymphocyte; CLN, cervical lymph node; CLNC, CLN cell; SPC, splenocyte; CMA, concanamycin A.