Cure of established GL261 mouse gliomas after combined immunotherapy with GM-CSF and IFNγ is mediated by both CD8+ and CD4+ T-cells

Authors

  • Karin Enell Smith,

    Corresponding author
    1. Department of Clinical Sciences, Glioma Immunotherapy Group, The Rausing Laboratory, Division of Neurosurgery, BMC D14, Lund University, SE-221 84 Lund, Sweden
    • Glioma Immunotherapy Group, The Rausing Laboratory, BMC D14, SE-221 84 Lund, Sweden
    Search for more papers by this author
    • Fax: +46 46 2224606.

  • Sara Fritzell,

    1. Department of Clinical Sciences, Glioma Immunotherapy Group, The Rausing Laboratory, Division of Neurosurgery, BMC D14, Lund University, SE-221 84 Lund, Sweden
    Search for more papers by this author
  • Wiaam Badn,

    1. Department of Clinical Sciences, Glioma Immunotherapy Group, The Rausing Laboratory, Division of Neurosurgery, BMC D14, Lund University, SE-221 84 Lund, Sweden
    Search for more papers by this author
  • Sofia Eberstål,

    1. Department of Clinical Sciences, Glioma Immunotherapy Group, The Rausing Laboratory, Division of Neurosurgery, BMC D14, Lund University, SE-221 84 Lund, Sweden
    Search for more papers by this author
  • Shorena Janelidze,

    1. Department of Clinical Sciences, Glioma Immunotherapy Group, The Rausing Laboratory, Division of Neurosurgery, BMC D14, Lund University, SE-221 84 Lund, Sweden
    Search for more papers by this author
  • Edward Visse,

    1. Department of Clinical Sciences, Glioma Immunotherapy Group, The Rausing Laboratory, Division of Neurosurgery, BMC D14, Lund University, SE-221 84 Lund, Sweden
    Search for more papers by this author
  • Anna Darabi,

    1. Department of Clinical Sciences, Glioma Immunotherapy Group, The Rausing Laboratory, Division of Neurosurgery, BMC D14, Lund University, SE-221 84 Lund, Sweden
    Search for more papers by this author
  • Peter Siesjö

    1. Department of Clinical Sciences, Glioma Immunotherapy Group, The Rausing Laboratory, Division of Neurosurgery, BMC D14, Lund University, SE-221 84 Lund, Sweden
    Search for more papers by this author

Abstract

We were the first to demonstrate that combined immunotherapy with GM-CSF producing GL261 cells and recombinant IFNγ of preestablished GL261 gliomas could cure 90% of immunized mice. To extend these findings and to uncover the underlying mechanisms, the ensuing experiments were undertaken. We hypothesized that immunizations combining both GM-CSF and IFNγ systemically would increase the number of immature myeloid cells, which then would mature and differentiate into dendritic cells (DCs) and macrophages, thereby augmenting tumor antigen presentation and T-cell activation. Indeed, the combined therapy induced a systemic increase of both immature and mature myeloid cells but also an increase in T regulatory cells (T-regs). Cytotoxic anti-tumor responses, mirrored by an increase in Granzyme B-positive cells as well as IFNγ-producing T-cells, were augmented after immunizations with GM-CSF and IFNγ. We also show that the combined therapy induced a long-term memory with rejection of intracerebral (i.c.) rechallenges. Depletion of T-cells showed that both CD4+ and CD8+ T-cells were essential for the combined GM-CSF and IFNγ effect. Finally, when immunizations were delayed until day 5 after tumor inoculation, only mice receiving immunotherapy with both GM-CSF and IFNγ survived. We conclude that the addition of recombinant IFNγ to immunizations with GM-CSF producing tumor cells increased the number of activated tumoricidal T-cells, which could eradicate established intracerebral tumors. These results clearly demonstrate that the combination of cytokines in immunotherapy of brain tumors have synergistic effects that have implications for clinical immunotherapy of human malignant brain tumors. © 2008 Wiley-Liss, Inc.

Glioblastoma multiforme (GBM) is the most abundant and most aggressive primary brain tumor in adults.1 Because of the infiltrative growth of the tumor, surgery will never be radical, but it only reduces the tumor burden. GBM is also highly resistant to both radiation and chemotherapy, which nevertheless are used in the treatment of human gliomas.1 Patients suffering from GBM rarely survive more than 1–2 years after diagnosis.1 Therefore, it is of great importance to find new treatment modalities. The murine brain tumor model, GL261, was induced by methylcholantrene (MCA) and established as a cell line.2–4 GL261 is a very rapidly growing and aggressive tumor, and mice with i.c. tumor die within 30 days. Morphologically, our GL261 tumor resembles a human malignant glioma, displaying a similar cell structure and growth pattern including; necrosis; proliferative capacity and; infiltrative growth (Personal communication with Dr. E. Englund, Dept. of Neuropathology),3 in contrast to others.5 Hence, GL261 is an attractive experimental model for the development and evaluation of new treatments including immunotherapy.

We have recently shown a synergistic effect between GM-CSF producing tumor cells and recombinant IFNγ on survival after therapeutic immunization of mice with i.c. gliomas.6 GM-CSF secreting cancer vaccines have shown promising results, both in experimental7–9 as well as in clinical trials.10, 11 The immunological effect induced systemically by the peripheral administration of GM-CSF is due to an increase in maturation and migration of myeloid progenitor cells from the bone marrow to the site of immunization as well as to local draining lymph nodes.12, 13 This will enhance the presentation of tumor antigens from the vaccine preparation for the T-cells and overall improve the induction of an efficient anti-tumor immune response.12, 14 However, the induction of an efficient anti-tumor response is highly dependent on localization, concentration, and timing of the administration of GM-CSF, since a high and protracted concentration of GM-CSF has been shown to induce immunosuppression through the induction of CD11b+/Gr-1+ myeloid-derived suppressor cells (MDSCs).15, 16 It has been postulated that the window of action for the induction of an effective anti-tumor response, using GM-CSF transduced tumor cells, ranges between 35 ng/106 tumor cells/24 hr17 up to 1500 μg/106 tumor cells/24 hr.13, 16 Above this dose GM-CSF loses its positive anti-tumor efficacy and induces immunosuppression through the induction of CD11b+/Gr-1+ MDSCs. CD11b+/Gr-1+ MDSCs comprise cells of a myeloid origin and are commonly definded by their expression of CD11b and Gr-1. Increased numbers of CD11b+/Gr-1+ MDSCs have been found to accumulate in blood, bone marrow, and secondary lymphoid organs in several mouse tumor models, whereas very few are found in either blood or spleen of tumor-free mice.18, 19 MDSCs are well known to promote their immunosuppressive function through the inhibition of T-cell functions as well as the induction of CD4+CD25+ T regulatory cells (T-regs).15, 20

The combination of GM-CSF-based cancer vaccines with other immunostimulatory factors is under development in order to enhance the anti-tumor response induced by GM-CSF. Some of these factors include CD40L21 or TLR-ligands e.g. CpG22 that stimulate activation of dendritic cell (DC) or anti-cytotoxic T-cell (CTL)A4 that can inhibit the induction of immune suppression.23 The importance of both IFNγ and GM-CSF in immunosurveillance has been demonstrated by the fact that double knock-outs, for both GM-CSF and IFNγ, spontaneously developed tumors.24 IFNγ is a multifunctional cytokine. Some of its effects important for tumor eradication are; induction of MHC I and II expression on tumor cells, inhibition of tumor cell proliferation, enhanced capacity of phagocytosis and antigen presentation of antigen presenting cells (APCs) and anti-angiogenic effects.25, 26 IFNγ is also known to play a crucial role in the induction of a Th1 response, resulting in a tumor-specific CD8+ CTL response.25 We have earlier shown a 37–70% survival in two rat glioma models (N32 and N29) after immunization with IFNγ transduced tumor cells (Janeldize et al. unpublished data).27 On the basis of these results, we have initiated a clinical trial, where patients with GBM receive peripheral immunizations with autologous tumor cells infected by adenoviruses to produce IFNγ (Salford et al. unpublished data).

In our previously published study, tumor-bearing animals were immunized the day after tumor inoculation.6 However, in this work immunizations were delayed until day 5 after tumor inoculation in order to confirm the effect on more established tumors in which will resemble treatment of human tumors. To address the mechanism of this immunotherapy, we also investigated the long-term memory and cellular changes induced by these immunizations.

Abbreviations

APC, antigen presenting cell; CTL, cytotoxic T-cell; DC, dendritic cell; GBM, Glioblastoma multiforme; GL-GM, GM-CFS transduced GL261 cell; GL-wt, GL261 wild-type cell; i.c., intracerebral; MCA, methylcholantrene; MDSC, myeloid-derived suppressor cell; MLTC, mixed lymphocyte–tumor culture; SPC, splenocyte; T-reg, T-regulatory cell.

Material and methods

Animals and cell lines

Male C57Bl/6 were purchased from Scanbur BK AB Sollentuna, Sweden and maintained under specific pathogen-free conditions at the Biomedical Center Animal Facility, University of Lund. Mice 8 weeks of age were used in all experiments. All animal procedures were performed according to the practices of the Swedish Board of Animal Research and were approved by the Committee of Animal Ethics in Lund-Malmö, Sweden. The GL261 mouse glioma cell line (GL-wt) of C57Bl/6 origin was a kind gift from Dr. G Safrany, Hungary. The transduction of the GM-CSF producing GL261 cells (GL-GM) has been described earlier.6

The cell culture medium (R10) used in all cell cultures was RPMI 1640 supplemented with 2 mM L-Glutamine, 1 mM sodium pyruvate, 10 mM HEPES, 50 μg/ml gentamicin (Invitrogen AB, Sweden), and 10% FBS (Fetal Bovine Serum, Biochrom AB, Berlin, Germany). For immunizations FBS and gentamicin were excluded from the injection medium (R0). β-mercaptoethanol, 50 μM, was added to R10 when used in spleen cell cultures.

Brain tumor inoculation

Brain tumors were induced by inoculation of 5 × 103 GL-wt tumor cells in 5 μl R0 into the right caudate nucleus in mice anesthetized with Isoflurane. The head of the mouse was fixed in a stereotactic frame and the needle of a Hamilton syringe was inserted after drilling of a small hole into the skull ∼1.5 mm to the right and 1.0 mm in front of bregma, 2.75 mm deep into the brain. The needle was left in the brain for 5 min and was slowly withdrawn to diminish any backflow through the insertion canal. The hole in the skull was sealed with bone wax.

Mice that had survived for 200 days after receiving immunotherapy were rechallenged with a second tumor into the left caudate nucleus. Otherwise, the inoculation was performed as described earlier. Survival of the mice was monitored for additionally 100 days (300 days in total).

All tumor-bearing mice were carefully observed and immediately killed when neurological symptoms appeared. Brains were removed and examined for remnant tumors.

Immunotherapy

For detection of cytokine levels in serum after immunization, nontumor-bearing mice were immunized with either 160 ng of recombinant IFNγ (R&D Systems, Abingdon, United Kingdom) or 2 × 106 irradiated (40 Gy) GL-GM tumor cells. Blood was collected at the indicated time points after immunization and serum was prepared for measurement of systemic concentrations of GM-CSF or IFNγ with a GM-CSF or IFNγ ELISA kit (BD Pharmingen, Stockholm, Sweden).

For characterization of spleen cell populations shortly after one single immunization in nontumor-bearing mice, spleens were collected 48 hr after immunization with GL-wt, GL-wt + recombinant IFNγ, GL-GM, or GL-GM + recombinant IFNγ. The cells used in the two immunization groups; GL-wt + IFNγ and GL-GM + IFNγ, were pretreated with IFNγ (100 ng/ml) (R&D Systems, Abingdon, United Kingdom) during 4 days before immunization. Recombinant IFNγ, 800 ng/ml (160 ng/mouse, 5 μg/kg) was also added into the injection medium at the time of immunization. All mice were immunized intra peritoneally (i.p.) with 2 × 106 irradiated (40 Gy) tumor cells in 200 μl of R0 without FBS and gentamicin.

For characterization of spleen cell populations in tumor-bearing mice and for detection of systemic cytotoxic immune responses after restimulation in vitro, spleens were collected day 18 after tumor inoculation after having received two immunizations day 1 and 15 after tumor inoculation (same immunization groups as described earlier).

For long-term survival after delayed immunization, tumor-bearing mice were immunized (same immunization groups as described earlier), starting day 5 after tumor inoculation and repeated every second week for a total of three times. Survival of the mice was monitored up to 200 days when the experiment was terminated and brains were examined for remnant tumors.

Flow cytometry

At day 18 after tumor inoculation, the spleens were cut into pieces and incubated for 45 min in R10+ with collagenase Type IV (1.6 mg/ml) (Sigma, Stockholm, Sweden) and DNAse (10 mg/ml) (Sigma) in +37°C. Undigested fragments were filtered through a stainless steel sieve and red blood cells were lysed.

At least 1 × 105 splenocytes (SPCs) were used for FACS-staining. All cells were preincubated with anti-CD16/32 (2.4G2, BD Pharmingen) to block unspecific binding to Fc-receptors by the antibodies. After washing, the SPCs were stained for cell surface molecules at +4°C for 30 min. All antibodies used, unless otherwise stated, were purchased from BD Pharmingen. MAbs used were; FITC-CD11b (M1/70), FITC-F4/80 (CI:A3-1), bio-Gr-1 (RB6-8C5), PE-CD11c (HL3), PE-CD4 (GK1.5), FITC-CD8a (53-6.7), APC-CD3 (145-2C11), and PE-CD25 (PC61.5). Biotin-conjugated antibodies were counter stained using APC-conjugated streptavidin. APC-FoxP3 (FJK-16s) staining was performed using a FoxP3 fixation/permeabilization staining kit (eBioscience, San Diego). Isotype-matched mAbs were included in each experiment. 7-AAD was added to all nonfixed/permeabilized cells before FACS analyzation, for gating of live cells. Fluorescence was measured on a FACSCalibur flow cytometer (BD, Heidelberg, Germany) and analysis of the data was performed using CellQuest software (BD, Heidelberg, Germany).

Restimulation of SPCs in vitro

The spleens were prepared from immunized mice as described earlier. The SPCs were restimulated in a mixed lymphocyte–tumor culture (MLTC). The SPCs (2 × 106) were incubated with 1 × 105 irradiated (40 Gy) GL-wt cells, in a 48-well plate for 48 hr. All SPCs analyzed with FACS for detection of intracellular cytokines were treated with Brefeldin A, 10 μg/ml, (Sigma-Aldrich) during the last 6 hr of incubation. After 48 hr, the restimulated SPCs were counted and all supernatants were collected for measurement of IFN-γ with ELISA (BD OptEIA Set: Mouse IFN-γ, BD Pharmingen). The harvested SPCs were analyzed with either FACS or ELIspot.

Detection of intracellular IFN-γ

SPCs harvested from the MLTC were counted and washed with PBS + 1% BSA. Briefly, the staining procedure was as follows. Fc receptors were blocked with CD16/32 (2.4G2, BD Pharmingen), and thereafter cell surface antigens were stained before permeabilization. For fixation and permeabilization, a fixation/permeabilization kit was used (BD Pharmingen). Briefly the cells were incubated in a 4% paraformaldehyde solution for 15 min in room temperature; thereafter the cells were washed and permeabilized and stained with APC conjugated IFN-γ antibodies (XMG1.2). Preincubation with unconjugated IFN-γ antibody was used to demonstrate specificity of staining. FACS analyses were performed as described earlier.

Granzyme B ELISpot

SPCs harvested from the MLTC were counted and immediately transferred to a Granzyme B ELISpot PVDF 96-well plate (R&D Systems, Abingdon, United Kingdom) precoated with mouse Granzyme B antibody. The ELIspot procedures were conducted according to the manufacturer's recommendations.

T-cell depletion

Injection of antibodies for T-cell depletion was initiated 3 days before tumor inoculation, and then two times every week, for a total of 5 weeks. 200 μg of either anti-CD4 (GK1.5) (eBioscience) or anti-CD8 (53-6.7) (eBioscience) mAbs was injected i.p. The efficiency of the depletion was analyzed in parallel on SPCs using FACS after three mAbs injections.

Immunizations with GL-GM+IFNγ, of mouse receiving depletion antibodies as well as the nondepleted control group only receiving immunization, were initiated day 1 after tumor inoculation and then every 2 weeks for a total of three times. Survival of the mice was monitored up to 100 days when the experiment was terminated and brains were examined for remnant tumors.

Statistics

Statistical differences between more than two groups were initially determined by using the nonparametric Kruskal-Wallis test. Kaplan Meier survival curves were compared using a logrank test. The p < 0.05 was being considered statistically significant. The tests were performed using GraphPad Prism®4.0c software (GraphPad software, San Diego, CA).

Results

Increased serum concentrations of both GM-CSF and IFNγ after immunization i.p.

Both IFNγ and GM-CSF are important for the recruitment and activation of immune cells to the immunization site. Being aware of their respectively short half-life in vivo,28–30 we were interested to examine their potential role in the initiation of a systemic anti-tumor immune response after immunization. Therefore, the pharmacokinetics for both cytokines after i.p. immunizations was investigated.

Nontumor-bearing mice were immunized with 2 × 106 irradiated GL-GM cells or 160 ng of recombinant mouse IFNγ and blood was collected after indicated time points. The serum concentration of GM-CSF peaked after 6 hr (975.4 ± 201.0 pg/ml) and was almost undetectable after 48 hr (20.1 ± 12.7 pg/ml) (Fig. 1a). These results indicated that the irradiated GL-GM cells were rapidly eradicated shortly after the time of immunization. The serum concentration of IFNγ peaked between 1 and 2 hr (310.1 ± 67.62 pg/ml and 278.1 ± 108.8 pg/ml) after immunization (Fig. 1b). Thereafter, the concentration declined 4 hr after immunization, and at 6 hr, the serum concentration of IFNγ was very low (17.71 ± 12.38 pg/ml).

Figure 1.

Pharmacokinetics of GM-CSF and IFNγ in serum after i.p. immunizations. Blood was collected from mice after the indicated hours and the concentration of (a) GM-CSF and (b) IFNγ were measured in serum using ELISA. The n was at least three for each time point measured. Each time point shows the mean concentration with SEM.

These pharmacokinetic results showed that both GM-CSF and IFNγ were detectable systemically in serum after immunization and that the injection of recombinant IFNγ displayed a shorter window of action than GM-CSF produced by the tumor cells.

Increase of CD11b+/Gr-1+ MDSCs in the spleen after a single immunization with either GL-GM or GL-GM+IFNγ

GM-CSF is well known to increase myeloid populations in lymphoid organs after systemic administration.28, 31 To further evaluate the direct systemic effect of GM-CSF produced by the GL-GM cells at the immunization site, flow cytometry of SPCs was performed shortly after immunization on nontumor-bearing mice.

After counting, no differences were detected in the number of SPCs because of the type of immunization (data not shown). Significantly higher percentages of CD11b+/Gr-1+ MDSCs were detected after immunizations with either GL-GM (2.04% ± 0.15) or GL-GM+IFNγ (1.8% ± 0.19), when compared with immunizations with GL-wt (0.82% ± 0.04; p < 0.01 for both; Fig. 2).

Figure 2.

Increase of CD11b+/Gr-1+ MDSCs in the spleen early after immunization with either GL-GM or GL-GM+IFNγ. Spleens were collected 48 hr after immunization with GL-wt, GL-wt+IFNγ, and either GL-GM or GL-GM+IFNγ. SPCs were stained and analyzed with flow cytometry, showing CD11b+/Gr-1+ MDSCs. Percentages shown are out of total SPCs, live gated using 7-AAD. This experiment was repeated three times, where n = 3 for each group.

Hence, shortly after a single immunization with either GL-GM or GL-GM+IFNγ, the proportion of CD11b+/Gr-1+ MDSCs in the spleen was increased.

Increase of differentiated myeloid cell populations and T-regs in tumor-bearing mice after receiving two immunizations with either GL-GM or GL-GM+IFNγ

To verify if the increase in CD11b+/Gr-1+ MDSCs would be sustained after several immunizations in tumor-bearing mice, SPCs were investigated after the second immunization with GL-wt, GL-wt + recombinant IFNγ, GL-GM, or GL-GM + recombinant IFNγ.

The SPCs were counted before staining; no differences in SPC number were detected after the different immunizations (data not shown). The percentages of CD11c+ DCs were significantly increased after immunization with either GL-GM (3.84% ± 0.19) or GL-GM+IFNγ (3.64% ± 0.13), compared with immunizations with GL-wt (3.01% ± 0.14; p < 0.01 and 0.05, respectively; Fig. 3a). A significant increase in F4/80+ macrophages was detected after immunization with GL-GM (4.44% ± 0.50) and GL-GM+IFNγ (4.15% ± 0.26), compared with immunizations with GL-wt (3.30% ± 0.21; p < 0.05 and 0.05, respectively; Fig. 3b). On the other hand, there were no significant differences between the types of immunizations when analyzing the percentages of CD11b+/Gr-1+ MDSC (Fig. 3c). These results demonstrated that after two immunizations with either GL-GM or GL-GM+IFNγ only an increase of the differentiated myeloid cell population but not of the immature myeloid cell population expressing CD11b and Gr-1 could be detected.

Figure 3.

Increase of differentiated myeloid cell populations and T-regs in tumor-bearing mice after receiving two immunizations with either GL-GM or GL-GM+IFNγ. Spleens were collected at day 18, after tumor-bearing mice had received two immunizations with GL-wt, GL-wt+IFNγ, and either GL-GM or GL-GM+IFNγ. SPCs were stained and analyzed with flow cytometry, showing (a) CD11c+ DCs, (b) F4/80+ macrophages, (c) CD11b+/Gr-1+ MDSCs, (d) CD3+/CD4+ T-cells, (e) CD3+/CD8+ T-cells, and (f) CD4+/CD25+/FoxP3+ T-regs. Percentages shown in (a)–(e) are out of total SPCs, whereas (f) shows the percentage out of CD4+ cells. The SPCs, except for the FoxP3 staining, were live gated with 7-AAD. Experiments (a)–(c) and (f) were repeated three times and (d) and (e) were repeated two times, where n = 3 for each group.

Minor, but nonsignificant alterations were seen among the CD4+ and CD8+ T-cell populations (Fig. 3d,e), indicating that the immunizations did not affect the proportions of T-cells systemically. However, when analyzing the proportion of CD4+CD25+FoxP3+ T-regs, a significant increase was detected after immunization with either GL-GM (12.35% ± 0.37) or GL-GM+IFNγ (13.36% ± 0.26) compared with immunization with GL-wt (10.47% ± 0.27; p < 0.05 and 0.001, respectively; Fig. 3f).

Finally, to exclude the possibility that the tumor itself affected the proportions of myeloid or T-cell populations systemically, SPCs from tumor- and nontumor-bearing mice were compared. Since no differences were detected, any potential suppressive effects on the proportion of SPCs induced by the tumor itself could be excluded (data not shown).

To summarize, mice immunized twice with either GL-GM or GL-GM+IFNγ showed an increase in differentiated myeloid cells (DCs and macrophages) and T-regs compared with mice immunized with GL-wt. At the same time, the proportion of CD11b+/Gr-1+ MDSCs remained unchanged between the different types of immunizations. All systemic effects observed were due to immunization with GL-GM, since there was no indication that the addition of IFNγ was responsible for the differences detected.

Immunizations with GL-GM+IFNγ increased the systemic cytotoxic immune response toward the tumor

Although no changes in the proportion of either CD4+ or CD8+ T-cells were detected systemically, their function could have been affected by the immunizations. Since we detected an increased proportion of APCs after immunization with either GL-GM or GL-GM+IFNγ, this indicated that the antigen presentation and activation of T-cells could be affected. Therefore, the cytotoxic functions of the SPCs were investigated after restimulation in vitro with irradiated tumor cells.

The cytotoxic response of the SPCs directed toward the tumor cells was measured using Granzyme B ELIspot. Restimulated SPCs from mice immunized with either GL-GM or GL-GM+IFNγ displayed a significantly higher amount of spots positive for Granzyme B, when compared with restimulated SPCs from mice immunized with either GL-wt or GL-wt+IFNγ (p < 0.05 and 0.05, respectively; Fig. 4a). Immunizations with IFNγ did not additively affect the Granzyme B production of the SPCs.

Figure 4.

Immunizations with GL-GM+IFNγ increased the systemic cytotoxic immune response toward the tumor. Spleens were collected at day 18, after tumor-bearing mice had received two immunizations with GL-wt, GL-wt+IFNγ, and either GL-GM or GL-GM+IFNγ. The SPCs were restimulated with irradiated GL-wt cells (black bars) for 48 hr and compared with nonrestimulated SPCs (white bars). (a) The harvested SPCs were transferred to a Granzyme B Elispot 96-well plate or stained for intracellular production of IFNγ in (b) CD4+ or (c) CD8+ T-cells. (d) Supernatants were collected and IFNγ production by the SPCs was measured with an IFNγ-ELISA kit. Each experiment was repeated three times, where n = 3 for each group.

The intracellular production of IFNγ from both CD4+ and CD8+ T-cells was analyzed using flow cytometry. The percentage of CD4+ and CD8+ T-cells producing IFNγ was higher after immunization with either GL-GM or GL-GM+IFNγ, when compared with immunizations with GL-wt (Fig. 4b,c). However, this increase was only significant after immunization with the combined therapy of GL-GM and IFNγ (p < 0.05) when compared with GL-wt immunizations.

The amount of IFNγ produced by the SPCs after restimulation in vitro was analyzed using ELISA. The same trend was observed using ELISA as observed earlier with flow cytometric analysis of intracellular production of IFNγ by T-cells, albeit not statistically significant. Immunization with either GL-GM or GL-GM+IFNγ increased the amount of IFNγ being produced after restimulation in vitro, when compared with immunizations with GL-wt (Fig. 4d).

In conclusion, the cytotoxic functions of the SPCs were significantly enhanced after immunizations with GL-GM+IFNγ, when compared with immunizations using GL-wt. These results demonstrate the synergistic effect of combining GM-CSF and IFNγ for the induction of an efficient cytotoxic anti-tumor immune response.

Immunotherapy with irradiated whole tumor cells induced a long-term memory

To assess whether an adaptive immune response was induced and thus an immunological memory was present, mice that had rejected the i.c. tumors after receiving immunizations were rechallenged with a second i.c. tumor without receiving any additional immunizations. Thus, all mice that had survived for more than 200 days were rechallenged with a second tumor into the opposite brain hemisphere, thereby avoiding involvement of scar tissue possibly remaining from the previous tumor inoculation. After their second tumor challenge, 100% of mice survived for another 100 days (300 days in total) despite not receiving any additional immunizations (Fig. 5). No tumors were detected in the brains of the mice surviving another 100 days (data not shown). This demonstrated that immunotherapy using whole tumor cells induced a long-term memory, which was independent of the presence of GM-CSF and IFNγ at the immunization site during the induction of an immune response.

Figure 5.

Immunotherapy with irradiated whole tumor cells induced a long-term memory. Mice surviving more than 200 days after tumor inoculation followed by immunotherapy were rechallenged with a second i.c. tumor but received no further immunization (n = 12). They were compared with previously nontreated controls receiving their first tumor (n = 6). Survival was monitored for 100 days. Mice were immediately killed when neurological symptoms were detected.

CD4+ and CD8+ T-cells were crucial for survival after peripheral immunizations with GL-GM+IFNγ

Because of a higher cytotoxic systemic activity of the T-cells after immunization with GL-GM+IFNγ the importance of the T-cell subsets for survival was analyzed. To deplete the T-cells, monoclonal antibodies specific for CD4 or CD8 were injected into the mice, starting 3 days before tumor inoculation and thereafter twice every week. The mice being depleted of T-cells, also received immunizations with GL-GM+IFNγ starting day one after tumor inoculation and then every 2 weeks.

The efficiency of the depletion using either CD4 or CD8 monoclonal antibodies was analyzed after the third antibody injection. Both CD4+ as well as CD8+ T-cells were almost completely depleted after three antibody injections (Fig. 6a,b).

Figure 6.

CD4+ and CD8+ T-cells were important for survival after peripheral immunizations with GL-GM+IFNγ. Injection of antibodies for T-cell depletion was initiated 3 days before tumor inoculation, and continued for 5 weeks, two times a week. (a),(b) The efficiency of the T-cell depletion was confirmed in SPCs using FACS after three injections of antibodies for either (a) CD4 (GK1.5) or (b) CD8 (53-6.7). Dot-plots show either CD4+ or CD8+ T-cells from CD3 gated SPCs for nontreated mice compared with depleted mice. The numbers indicate the percentage of CD4+ or CD8+ T-cells out of CD3+ T-cells. (c),(d). Tumor-bearing mice were immunized with GL-GM+IFNγ three times, every 2 weeks, starting day one after tumor inoculation. The immunization time points are indicated with arrows. Tumor-bearing mice immunized with GL-GM+IFNγ and depleted for either C. CD4+ T-cells (n = 5) or (d) CD8+ T-cells (n = 5) were compared with mice only immunized with GL-GM+IFNγ (n = 5) or nontreated control mice (n = 5). Survival was monitored for a total of 100 days and mice were immediately killed when neurological symptoms were detected.

Both CD4+ as well as CD8+ T-cells turned out to be crucial for survival, since depletion of either CD4+ or CD8+ T-cells abolished the effect of the combined therapy (p = 0.014 and 0.016, respectively; Fig. 6c,d).

Survival after delayed immunization with GL-GM+IFNγ

We have earlier shown that almost 90% of the tumor-bearing mice survived when immunized with GL-GM+IFNγ starting day 1 after tumor inoculation.6 When immunizing with either GL-GM or GL-wt+IFNγ, the survival was 44 and 13%, respectively, demonstrating the synergistic effect of combining both GM-CSF and IFNγ. To evaluate whether the combined therapy with GM-CSF and IFNγ would have any effect on larger established tumors and also to increase the therapeutic window to mimic the clinical situation, we delayed the immunizations until day 5 after tumor inoculation. After having delayed the immunizations until day 5, 33% of the mice receiving immunizations with GL-GM+IFNγ survived more than 100 days (Fig. 7), whereas mice immunized with GL-GM, GL-wt+IFNγ, or GL-wt did not survive. The increased survival after combining GL-GM with IFNγ was significant when compared with mice immunized with GL-GM only (p = 0.0093).

Figure 7.

Survival after delayed immunization with GL-GM+IFNγ. Mice were immunized day 5 after tumor inoculation, with GL-wt (n = 6), GL-wt+IFNγ (n = 6), GL-GM (n = 12), or GL-GM+IFNγ (n = 12) and compared with nontreated control mice (n = 6). The immunization time points are indicated with arrows. Survival was monitored for 200 days. Mice were immediately killed when neurological symptoms were detected.

These results clearly demonstrated the synergistic effect of combining GM-CSF and IFNγ for survival even after delayed immunizations on more established tumors.

Discussion

In this article, we show that by using the combination of GM-CSF and IFNγ, survival of mice with preestablished i.c. tumors, was synergistically enhanced, even if the first immunization was delayed until day 5. This investigation analyzes the different components of the immune response induced by the combined immunotherapy to explain the underlying mechanisms.

To dissect the impact of GM-CSF and IFNγ during the initiation of a systemic immune response, we investigated the in vivo peak concentrations and duration of detectable levels for the two cytokines involved. GM-CSF secreted by the transduced cells in vivo after one immunization i.p. peaked after 6 hr at almost 1 ng/ml. The systemic effect of the secreted GM-CSF could then be followed as an increase in the percentage of myeloid cells in the spleen after a single immunization. The general belief behind using GM-CSF as an adjuvant in immunotherapy is its ability to recruit bone marrow-derived progenitor cells and to enhance their migration and differentiation into APCs at the immunization site.9, 10 A too high and sustained concentration of GM-CSF has been shown by Serafini et al. to recruit and expand the number of CD11b+/Gr-1+ MDSCs at the immunization site, which could inhibit T-cell responses.16 In this study, we confirm a minor and transient increase of CD11b+/Gr-1+ MDSCs, but this does not inhibit the induction of an efficient anti-tumor immune response observed as an increase in T-cell cytotoxicity and prolonged survival. We hypothesize that the minor and transient increase in CD11b+/Gr-1+ MDSCs detected shortly after immunization, could result in a maturation and differentiation of CD11b+/Gr-1+ MDSCs into DCs and macrophages ultimately leading to an enhanced antigen presentation.

The peak concentration of IFNγ was detected already after 1–2 hr after immunization, but declined rapidly. This short but apparently efficient window of action for IFNγ was not reflected by any systemic changes in lymphocyte cell populations. Neither is it probable that it would have induced changes in the transduced tumor cells at the site of immunization.32 IFNγ has been shown to upregulate costimulatory molecules, e.g. B7-1, on APCs.25, 33 Therefore, the synergistic effect seen on cytotoxicity and survival could be due to the fact that IFNγ affected the differentiation and maturation of APCs rather than the recruitment of cells to immunization site.

When analyzing the systemic immune response in the spleen of tumor-bearing mice, we could detect an increasing proportion of FoxP3+ T-regs as well as CD11b+/Gr-1+ MDSCs, after immunizations with either GL-GM or GL-GM and IFNγ. Growing tumors can inhibit the induction of efficient immune responses, for example by interfering with the differentiation, function and activation of APCs, which will lead to an impaired T-cell activation. Human gliomas have been shown to produce immunosuppressive factors including TGFβ, IL-10, GM-CSF, and prostaglandin E234–37 which all have been reported to trigger the induction of MDSCs. In both human as well as in mouse gliomas it has been shown that FoxP3+ T-regs accumulate in the growing tumor as well as peripherally in blood.38–40 We have not been able to measure GM-CSF from the GL261 cells, but we have seen low levels of TGFβ after cell culture in vitro (data not shown). However, we could not detect any differences in either T-regs or CD11b+/Gr-1+ MDSCs, when comparing tumor-bearing mice with nontumor-bearing mice (data not shown). More likely, this increase in T-regs after immunization with GM-CSF could have been induced by the CD11b+/Gr-1+ MDSCs.

Despite the induction of an increased proportion of CD11b+/Gr-1+ MDSCs and T-regs after immunization with GM-CSF and IFNγ, we detected an increased systemic cytotoxic anti-tumor response, including a higher production of Granzyme B of SPCs and IFNγ secreted by the T-cells. Hence, contrary to most previous reports we can correlate a systemically increased proportion of CD11b+/Gr-1+ MDSCs and T-regs with an increased cytotoxic anti-tumor response resulting in enhanced survival. An explanation for this is that the induction of T-regs and CD11b+/Gr-1+ MDSCs, seen after immunization with GL-GM and IFNγ, could be a negative side effect induced by the immunotherapy. Indeed Alderson et al. recently reported an increase on T-regs after successful immunotherapy of an experimental mouse tumor.41 Whether blocking of immunosuppression as by depletion of CD25+ cells38, 40, 42, 43 or injection of anti-CTLA444 would further strengthen the effect of the combined therapy awaits elucidation.

The systemic immune responses induced by the therapy with GM-CSF and IFNγ might not wholly reflect or represent what takes place within the tumor tissue. At the tumor site several cytotoxic cells have been addressed for their importance in tumor eradication, including CD8+ T-cells, NK-cells, and macrophages.6, 26, 45–47 CD4+ T-cells have been shown to be critical for the development of CD8+ effector T-cells and induction of a long-lasting anti-tumor memory.48 CD8+ effector T-cells are of great importance for tumor eradication, whereas a long-lasting anti-tumor memory inhibits tumor progression and formation of new tumors. We clearly demonstrate that a long-lasting anti-tumor memory was induced, since mice previously immunized after their first encounter with the tumor, could survive a tumor rechallenge without any additional immunizations. Additionally, the rejection of tumors in /rechallenge experiments implies that the induction of the anti-tumor T-cell response was initiated by resident brain APCs.49

Induction of anti-tumor immunity with immunotherapy using GM-CSF has been shown to be dependent of CD4+ and CD8+ T-cells.12 The cytotoxic functions of the T-cells, analyzed systemically in the spleen, were significantly enhanced after immunization with the combined therapy of GM-CSF and IFNγ. To evaluate the significance of T-cells for tumor eradication after peripheral immunization with both GM-CSF and IFNγ, either CD4+ or CD8+ T-cells were depleted during the whole period of immunization. T-cell depletion started before tumor inoculation, thereby affecting the initiation phase of the induction of an immune response. There was a profound negative impact on survival if either the CD4+ or CD8+ T-cells were depleted, since the effect of immunizing with GM-CSF and IFNγ was entirely abolished.

Experimental immunotherapies of tumors often use models that do not resemble the clinical situation in various aspects. The conclusions that can be deducted from experiments with preimmunizations and subsequent tumor establishment are dubious as the treatment might well target the establishment and initial angiogenesis of the tumor implantation. Therefore, by delaying the initiation of immunization until day 5 after tumor inoculation, this would more resemble a situation where the tumor is established and also increase the therapeutic window in order to mimic the clinical situation. This experiment again proved that the combination of both GM-CSF and IFNγ was essential for survival, whereas neither monotherapy alone had effect on survival.6 Therefore, we believe that immunotherapy combining both GM-CSF and IFNγ, has a great potential for further investigations in the combination with conventional therapies for the treatment of glioma.

In conclusion, by combining both GM-CSF and IFNγ, we have established a robust immunotherapy in the mouse glioma model, GL261, illustrated by the survival induced after delayed immunization as well as the induction of a long-term memory. Although the combined therapy affects the recruitment and possibly the function of APCs at the immunization site, T-cells are crucial for the effect. Because of the existing experience of both IFNγ and GM-CSF in clinical use the translation of the combined therapy into clinical trials would be feasible in a relatively short time perspective.

Ancillary