A novel recombinant protein of IP10-EGFRvIIIscFv and CD8⁺ cytotoxic T lymphocytes synergistically inhibits the growth of implanted glioma in mice

Abstract

The epidermal growth factor receptor (EGFR) mutant of EGFRvIII is highly expressed on glioma cells and has been thought to be an excellent target molecule for immunotherapy. IP-10 is a potent chemokine and can recruit CXCR3⁺ T cells, including CD8⁺ T cells that are important for the control of tumor growth. This study is aimed at investigating the therapeutic efficacy of a novel fusion protein of IP10-EGFRvIIIscFv (IP10-scFv) in combination with glioma lysate-pulsed DCs-activated CD8⁺ cytotoxic T lymphocytes (CTLs) in a mouse model of glioma. A plasmid of pET-IP10-scFv was generated by linking mouse IP-10 gene with the DNA fragment for anti-EGFRvIIIscFv, a (Gly₄Ser)₃ flexible linker and a His-tag. The recombinant IP10-scFv in E. coli was purified by affinity chromatography and characterized for its anti-EGFRvIII immunoreactivity and chemotactic activity. C57BL/6 mice were inoculated with mouse glioma GL261 cells in the brain and treated intracranially with IP10-scFv and/or intravenously with CTL for evaluating the therapeutic effect. The glioma-specific immune responses were examined. The IP10-scFv retained anti-EGFRvIII immunoreactivity and IP-10-like chemotactic activity. Treatment with both IP10-scFv and CTL synergistically inhibited the growth of glioma and prolonged the survival of tumor-bearing mice, accompanied by increasing the numbers of brain-infiltrating lymphocytes (BILs) and the frequency of CXCR3⁺CD8⁺ T cells, enhancing glioma-specific IFN-γ responses and cytotoxicity, and promoting glioma cell apoptosis in mice. Our novel data indicate that IP10-scFv and CTL have synergistic therapeutic effects on inhibiting the growth of mouse glioma in vivo.

Introduction

Glioblastoma multiforme (GBM) is the most common primary neoplasm in the brain and represents over 50 % of all tumors in the brain [1]. Although standard multi-modal therapies, including surgical resection and radiotherapy plus temozolomide, are available for the intervention of patients with GBM, the prognosis of patients with GBM is poor. Therefore, the development of effective and safe therapies for the intervention of patients with GBM will be of great significance [2].

T cell-mediated tumor-specific immunotherapy is highly interesting. Therapeutic strategies by recruiting and activating immunocompetent cells, such as CD4⁺, CD8⁺ T cells, natural killer (NK), and natural killer T (NKT) cells, in the tumor and surrounding tissues have been demonstrated to be effective in animal models of tumors, including GBM [3]. Several chemokines have been shown to elicit strong antitumor responses [4]. The IFN-induced protein of 10 kDa (IP-10) can bind to its receptor, CXCR3, which is expressed on activated T cells. IP-10 can promote the migration of activated CD4⁺ and CD8⁺ T cells, NK and NKT cells [5]. Dendritic cells are potent antigen-presenting cells (APCs) and can induce T cell activation and antigen-specific T cell responses. Hence, the therapeutic application of dendritic cells and IP-10 may enhance lymphocyte chemotaxis and specific immune responses. Indeed, our previous study has shown that treatment with IP-10, together with glioma lysate-loaded dendritic cells, inhibits the growth of implanted tumors in vivo [6]. However, there are a few activated T cells infiltrated in the tumors. Given that the numbers of activated T cells infiltrated in the tumors are crucial for tumor-specific T cell immunity, a new approach to recruit more effector T cells may enhance tumor-specific T cell responses and control the progression of tumors.

Aberrant activation of the gene for epidermal growth factor receptor (EGFR) and its arrangement is commonly detected in human glioma, which results in the expression of the mutant EGFRvIII protein [1]. The EGFRvIII is encoded by an 801 base-pair fragment, corresponding to the mRNA exons of 2–7 with a deletion at position 7–273 in the extracellular domain of EGFR [7]. The EGFRvIII is expressed specifically on the surface of glioma cells in 30–50 % of malignant gliomas [8]. Thus, EGFRvIII represents an excellent target for immunotherapy of gliomas [9].

Antibody-based reagents have been applied for anti-tumor therapies, and single-chain variable (scFv) antibodies are more powerful because of their high penetration, specificity, and affinity [10]. The scFv-based reagents have been developed and currently are being tested for the diagnosis and immunotherapy of different types of diseases [11–13]. Given that EGFRvIII is a glioma-specific antigen, we hypothesize that a fusion protein of anti-EGFRvIIIscFv and IP-10 that retains the specificity and affinity of the antibody and the chemotactic activity of IP-10 may specifically target the glioma, where the fusion protein recruits and activates T cells, enhancing glioma-specific T cell immunity and inhibiting glioma progression.

In this study, we first generated a recombinant fusion protein of IP10-EGFRvIIIscFv (IP10-scFv), and after characterizing its binding affinity and chemotactic activity, we employed a mouse model of glioma to test whether treatment with the IP10-scFv and glioma cell lysate-pulsed dendritic cell-activated CD8⁺ cytotoxic T lymphocytes (CTL) could synergistically inhibit the growth of implanted glioma. Furthermore, we determined the potential mechanisms underlying the action of the therapy in vivo. We found that treatment with both IP10-scFv and CTL synergistically inhibited the growth of implanted glioma, accompanied by promoting CTL infiltration and cytotoxicity, and inducing glioma cell apoptosis in vivo. These novel data indicated that combination of chemokine with targeting tumor-specific antigen enhanced T cell immunity against tumors, which may be a promising strategy for the intervention of human malignant tumors in the clinic.

Materials and methods

Cell lines and animals

Mouse glioma cell line, GL261, stably expressing EGFRvIII (GL261-EGFRvIII) and human glioblastoma-astrocytoma epithelial-like cell line, U87, stably expressing EGFRvIII (U87-EGFRvIII) were maintained in our laboratory. These cells were cultured regularly in DMEM medium (Invitrogen, Carlsbad, USA) supplemented with 2 mM l-glutamine, 100 U/ml of penicillin, 100 μg/ml of streptomycin (Sigma), and 10 % fetal bovine serum (FBS, Invitrogen) in a humid atmosphere of 5 % CO₂/95 % air at 37 °C. Mouse fibroblast L929 cell line, Lewis lung carcinoma cell line 3LL, hepatoma cell line Hepa1-6, embryo fibroblast cell line NIH3T3 were obtained from the Union Hospital Central Laboratory and cultured in 10 % FBS RPMI 1640 medium (Invitrogen).

Eight-week-old female C57BL/6 mice (H-2^b) were obtained from the Tongji Medical College, Huazhong University of Science and Technology and housed in a specific pathogen-free facility. The experimental protocol was established, according to the guidelines of NIH Animal Research and Care, and approved by the Ethics Committee of the Union Hospital of Tongji Medical College.

Construction of a plasmid for expressing the IP10-EGFRvIIIscFv gene

The IP-10 gene fragment was cloned into pGEM (TianGen Biotech, China), as described previously [14]. The cDNA (GenBank Accession No. U76382.1) encoding anti-EGFRvIIIscFv was previously obtained using a phage display library [15]. Subsequently, the IP-10 gene fragment, together with a flexible linker of (Gly₄Ser)₃ and helix-histidine tag (His6-tag), was cloned into the up-stream of the scFv to form the His6-tag-IP-10-Gly₄Ser-EGFRvIIIscFv fragment and further cloned into the XhoI and EcoRI sites of the PET30a (Invitrogen) to generate a plasmid of pET-IP10-scFv (Fig. 1a), which was analyzed by DNA sequencing.

Fig. 1.

Generation and characterization of recombinant IP10-scFv. a Illustration of an inserter for the generation of recombinant plasmid for expressing recombinant IP10-scFv. b Characterization of the purified IP10-scFv. Data are expressed as a representative image from three separate experiments. Lane M: Molecular mass (Kda); Lane 1 The purified IP10-scFv; Lane 2 BSA. c The affinity of IP10-scFv binding to EGFRvIII. Data are expressed as a Scatchard equation that was generated by three separate experiments. V/a (A ₀ − A)/A ₀/[a ₀ − I ₀*(A ₀ − A)/A ₀]; V (A ₀ − A)/A ₀, means the concentrations of free antigen at equilibrium. The related coefficient of equation is 0.989. d The IP10-scFv immunoreactivity. e The chemotactic activity of the purified IP10-scFv. Data are expressed as chemotaxis index of different groups of T cells from three separate experiments. The EGFRscFv displayed immunoreactivity, similar to that of IP10-scFv (data not shown). **p < 0.01 versus the EGFRvIIIscFv

Expression and purification of IP10-scFv fusion proteins

To obtain the purified IP10-scFv fusion proteins, the pET-IP10-scFv plasmid was transformed in E. coli Rosetta (DE3), and the IP10-scFv was purified using the Ni-His-tag Fusion Protein Purification Kit and 200 mM imidazole (pH7.4) for elusion (TOYOBO, Japan), according to the manufacturers’ instruction. The eluted proteins were diluted and dialyzed against 1,000 volumes of refolding buffer (50 mM Tris–HCl pH 8.0, 200 mM NaCl, 1 mM EDTA, 5 % glycerol, and 5 mM DTT) containing gradually decreased concentrations of GuHCl in the presence of 1 mM oxidized glutathione (GSSG) and reduced glutathione. After centrifugation, the purified proteins were characterized by SDS-PAGE on 12 % gel and analyzed by silver staining and Western blot assays using peroxidase-conjugated anti-his-tag monoclonal antibody (Abcam), and chemiluminescent detection (ECL Plus, Amersham Pharmacia).

Competitive ELISA

The affinity of the purified IP10-scFv was measured by a competitive ELISA. A 13-amino acid peptide, LEEKKGNYVVTDHC, that contains an epitode recognized by the anti-EGFRvIIIscFv was synthesized [16] and cross-linked with OVA as an antigen (0.05 μg/well) for coating ELISA plates. After blocking with 3 % BSA, individual wells were added in triplicate with a mixture of 2.94 × 10⁻⁸ M IP10-scFv with various concentrations (2 × 10⁻⁸ to 8 × 10⁻⁷ M) of peptide/OVA that had been incubated for 24 h at room temperature, and the plates were incubated for at least 1 h at room temperature. After washing, the bound IP10-scFv was detected by biotinylated anti-His6 monoclonal antibody and visualized using HRP-conjugated avidin (Peprotech) and substrate of ABTS (Sigma), followed by reading with absorbance at 405 nm. The wells without antigen were used as negative controls. We designated [a ₀], [I ₀], [A ₀], and [A] for as initial concentration of antigen, initial concentration of antibody, OD value without antigen, OD value with additive antigen, respectively. The affinity of IP10-scFv was determined by a Scatchard equation of (A ₀ − A)/A ₀/[a ₀ − I ₀(A ₀ − A)/A ₀] = 1/K _D [1−(A ₀ − A)/A ₀] [17].

Antigen binding assay

U87-EGFRvIII cells were washed with PBS (containing 3 % BSA, pH 7.4) and incubated with 100 ng IP10-scFv or the isotype control antibody at 4 °C for 1 h. Subsequently, the cells were incubated with 1 μg/ml of anti-His6 monoclonal antibody, and after washing, the cells were stained with FITC-conjugated rabbit anti-mouse IgG and analyzed under a fluorescent microscope or by flow cytometry.

Cell migration assays

The ability of IP10-scFv to attract activated T cell migration was determined by transwell migration assays. Briefly, the upper chambers of 24-well transwell plates (5-μm pore size; Corning) were coated with 5 mg/ml of fibronectin (Sigma-Aldrich). Splenic mononuclear cells were isolated from C57BL/6 mice and stimulated with 5 μg/ml of Con A (Sigma) for 48 h. The activated CD8⁺ T cells were purified using mouse CD8⁺ T cell microbeads, according to the manufacturers’ instruction (MACS, German). Subsequently, individual top chambers were loaded with 1 × 10⁵/well of the purified CD8⁺ T cells in 1 % BSA RPMI 1640, and the bottom chambers were filled in triplicate with 50 μl of different concentrations of IP10-scFv, rIP-10 (Peprotech), anti-EGFRvIIIscFv (Lanjing Bisynthesis, China) [15] in 1 % BSA RPMI1640 or PBS for incubation at 37 °C for 3 h. The numbers of CD8⁺ T cells that migrated to the lower chamber were stained with crystal violet and counted. The results were expressed as chemotaxis index = the numbers of cells migrated in the presence of chemokine/the numbers of cells migrated in the absence of chemokine (medium alone) × 100 %.

Generation of CD8⁺ CTL

The glioma cell lysate-pulsed dendritic cells were prepared, as previously described [14]. To generate glioma-specific CTL, splenic CD8⁺ T cells were purified using a mouse CD8a⁺ T Cell Isolation Kit II (Miltenyi Biotec), and CD8⁺ T cells (10⁷/ml) were stimulated with glioma cell lysate-pulsed dendritic cells (10⁶/ml) in the presence of 20 ng/ml of IL-2 for 4 weeks, as previously described [18]. The CTLs were harvested for infusion.

Tumor challenge

To examine the effect of IP10-scFv on the growth of glioma in vivo, C57BL/6 mice were anesthetized and inoculated intracranially (i.c) with 2 × 10⁴ GL261-EGFRvIII cells in 2 μl PBS by stereotactically injection 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 mice using a Reword stereotactic frame (Reword Instrument, China). The mice were randomized and injected i.c with 0.5 μg IP10-scFv and/or intravenously with 10⁴ CTL on day 7, 14, and 21 post-inoculation, respectively (n = 10 per group). The mice with IP10-scFv alone also received i.v PBS injection, while the mice with CTL alone received i.c PBS injection. Other groups of tumor-bearing mice were injected i.c. with 0.5 μg EGFRvIIIscFv and i.v with PBS or PBS only on the same days as the controls. The some mice from each group were euthanized in a CO₂ container on day 7, 14, 21, or 28 post-inoculation, respectively, and their tumors were dissected out and measured using a Vernier caliper. The tumor volumes were calculated by the formula: 1/2L × W ². Another set of experimental mice were monitored for their survival up to 100 days post-inoculation, and the mice were monitored closely for their death.

Flow cytometry analysis of effector CD8⁺ T cells in the brain

To characterize the brain-infiltrating leukocytes (BILs), some mice from each group were killed on day 14 and 28 post-inoculation for the isolation of infiltrated inflammatory cells, as described previously [19]. Briefly, individual mice were perfused with 20 ml of PBS, and their whole brains were dissected out, weighed, and digested with collagenase (Sigma), as described previously [20]. The single-cell suspension was prepared from individual mice and loaded on 70 % Percoll for centrifugation (Pharmacia). The collected cells were further subjected to density gradient centrifugation using 50–30 % Percoll at 400×g for 20 min. The enriched BILs were characterized for the frequency of CD8⁺ T and CXCR3⁺ T cells by flow cytometry analysis using FITC-anti-CD8 and PE-anti-CXCR3.

Enzyme-linked immunospot (ELISPOT) assay

T cell immunity to glioma was determined by ELISPOT assay [21, 22]. Briefly, splenocytes were isolated from IP10-scFv and CTL, IP10-scFv alone, CTL alone-treated, and EGFRvIIIscFv-treated mice on day 14 post-inoculation. Splenocytes (1 × 10⁵/well) were co-cultured in triplicate with 2 × 10⁵/well of GL261-EGFRvIII cells in 10 % FBS RPMI1640 at 37 °C for 48 h in 96-well ELISPOT plates (U-Cytech, Holland) that had been coated with anti-IFN-γ capture antibody. Splenocytes or GL261-EGFRvIII cells alone in medium were used as negative controls. After washing, the plates were incubated with biotinylated anti-IFN-γ detection antibody and visualized using HRP-streptavidin and AEC. The spot forming cells (SFCs) were counted using the Immunospot analyzer (Cellular Technology, Cleveland, USA).

Cytotoxic assay

The cytotoxicity of CTL against glioma GL261-EGFRvIII cells was determined by ⁵¹Cr-release assay with minor modification. Briefly, the BILs were isolated from the IP10-scFv and CTL, IP10-scFv, CTL, and EGFRvIIIscFv-treated mice on day 14 post-inoculation. The BILs (1 × 10⁵/well) were stimulated with GL261-EGFRvIII cells in 10 % FBS RPMI1640 for 5 days, and the activated cells were used as effector cells. Simultaneously, GL261-EGFRvIII, 3LL, Hepa1-6, and NIH3T3 target cells (10⁶) were labeled with 40 μCi ⁵¹Cr-sodium chromate (Amersham Pharmacia Biotech) at 37 °C for 2 h, and after washing, the target cells were incubated in triplicate with activated effectors at different ratios of 40:1, 20:1, 10:1, and 5:1 in 96-well U-bottom microtiter plates (Nunc, Roskilde, Denmark) for 4 h, respectively. The target or effector cells alone were used as the controls. The target cells were treated with 1 % triton x-100 and used as the maximum release of ⁵¹Cr. The plates were centrifuged at 600×g, and the supernatant was counted using a gamma counter. The percentage of glioma-specific cytotoxicity was calculated by the formula: {[CPM (experimental) − CPM (spontaneous)]/[CPM(maximum) − CPM (spontaneous)]} × 100 %. Similarly, the cytotoxicity of CTL against different types of cells was tested by the same protocol.

Apoptosis assay

The GL261-EGFRvIII tumors in the brains of different groups of mice were dissected on day 21 post-inoculation, and the tumor cells were prepared using the Tumor Dissociation Kit, according to the manufacturers’ instruction (Miltenyi). Subsequently, the cells were stained with 5 μl of APC-Annexin V and 10 μl of PI for 10 min. The apoptotic cells were characterized by flow cytometry.

Statistical analysis

Data are expressed as the mean ± SD. The difference between different groups was analyzed by one-way analysis of variance and post hoc Bonferroni correction, and the difference between two groups was analyzed by Student’s t test using SPSS version 16.0 software (SPSS, USA). The survival of individual groups of mice was analyzed by the Kaplan–Meier and log rank Mantel–Cox methods. A p value of <0.05 was considered statistically significant.

Result

The recombinant IP10-scFv retains its immunoreactivity and chemotactic activity in vitro

To generate a plasmid for expressing recombinant IP10-scFv, a DNA fragment of mouse IP-10 gene was linked with the DNA fragment for anti-EGFRvIIIscFv, together with a (Gly₄Ser)₃ flexible linker and a His-tag, at the same coding frame to generate the plasmid pET-IP10-scFv (Fig. 1a). As shown in Fig. 1b, the purified IP10-scFv was near 40 kDa and had a purity of greater than 93 %, as determined by densimetric scanning (Fig. 1b).

Next, we characterized the immunoreactivity of IP10-scFv by a competitive ELISA and found that interaction of IP10-scFv with increased amounts of EGFRvIII antigen peptide reduced the binding of IP10-scFv to the antigen coated. The rewritten Scatchard equation for the binding of the IP10-scFv to EGFRvIII antigen was plotted in Fig. 1c. The affinity constant K _D value of the IP10-scFv was 1.32 ± 0.05 × 10⁻⁸ M, which was lower than that of anti-EGFRvIII mAb (1.3–25 × 10⁻⁹ M) [23], but was similar to that of MR1 immunotoxin (K _D = 1.1–2.2 × 10⁻⁸ M) [15]. Immunofluorescent and flow cytometry analysis indicated that the IP10-scFv bound to the surface of U87-EGFRvIII cells (Fig. 1d).

The chemotactic activity of IP10-scFv was tested for the migration of activated CD8⁺ T cells by transwell migration assays. While there was no detectable chemotactic activity of anti-EGFRvIIIscFv, the IP10-scFv stimulated the migration of CD8⁺ T cells, similar to that of rIP-10 (Fig. 1e). Collectively, these data clearly indicate that the purified recombinant IP10-scFv retains immunoreactivity of scFv against its EGFRvIII antigen and the chemotactic activity of IP-10.

IP10-scFv and CTL synergistically inhibit the growth of glioma in mice

We tested the effect of IP10-scFv on the growth of glioma in vivo. As shown in Fig. 2a, treatment with CTL or IP10-scFv significantly reduced the volume of implanted tumors, when compared with that of the EGFRvIIIscFv-injected controls, and the inhibitory effect of IP10-scFv was significantly greater than that of CTL alone on day 28 post-inoculation. Strikingly, treatment with both IP10-scFv and CTL prevented the growth of implanted tumors in mice up to 28 days after inoculation, and the inhibitory effect of both IP10-scFv and CTL on the growth of implanted tumors was significantly higher than that of treatment with IP10-scFv or CTL alone on day 21 and 28 post-inoculation.

Fig. 2.

Inhibition of tumor growth. a The kinetics of tumor growth. Data are expressed as the mean ± SD of the tumor volumes of each group of mice (n = 10 per group per time point). b The survival curves of tumor-bearing mice were estimated using the Kaplan–Meier method. Data are expressed as mean percentage of each group of mice (n = 10 per group) survived throughout the period. *p < 0.05, **p < 0.01 versus the EGFRvIIIscFv group; ^# p < 0.05, ^## p < 0.01 versus the CTL group; ^{^} p < 0.05, ^{^^} p < 0.01 versus the IP10-scFv group

Similarly, we monitored the survival of tumor-bearing mice and found that, while all the mice in the EGFRvIIIscFv and CTL alone-treated mice died between day 32 and 47 after inoculation, less than 50 % of mice that had been treated with IP10-scFv or with both IP10-scFv and CTL died on day 57 post-inoculation (Fig. 2b). There were 20 % or 40 % of mice that survived throughout the observation period (up to 100 days post-inoculation) in the IP10-scFv-treated and IP10-scFv-/CTL-treated groups of mice, respectively. The effect of treatment with IP10-scFv and CTL on prolonging the survival of tumor-bearing mice was significantly greater than that of treatment with CTL (p < 0.01) or with IP10-scFv alone (p < 0.05). Therefore, treatment with both IP10-scFv and CTL synergistically inhibited the growth of implanted glioma in vivo and prolonged the survival of tumor-bearing mice.

Treatment with IP10-scFv increases the numbers of effector CD8⁺ infiltrates in the brains of mice

We isolated BILs from different groups of mice on day 14 and 28 post-inoculation. The numbers of BILs in the CTL-treated mice were greater than that in the EGFRvIIIscFv group of mice, but were significantly less than that in the IP10-scFv-treated mice (Fig. 3a). The numbers of BILs in the mice that had been treated with both IP10-scFv/CTL were significantly greater than that in the IP10-scFv or CTL-treated mice.

Fig. 3.

Characterization of BILs in mice. a The numbers of BILs. Data are expressed as the mean ± SD of the numbers of BILs from individual mice in different groups at 28 days post-inoculation (n = 10 per group). b Flow cytometry analysis. The isolated BILs (10⁶) from a single mouse (the IP10-scFV/CTL group), pooled two mice (the IP10-scFv group), pooled four mice (the CTL group), or five–six mice (the EGFRvIIIscFv group) were stained with FITC-anti-CD8 and APC-anti-CXCR3, followed by flow cytometry analysis of at least 10,000 events. c Quantitative analysis. Data shown are representative charts or expressed as the mean ± SD of the percentages of CXCR3⁺CD8⁺ T cells of each group from three separate experiments. *p < 0.05, **p < 0.01 versus the EGFRvIIIscFv group; ^# p < 0.05, ^## p < 0.01 versus the CTL group; ^{^} p < 0.05, ^{^^} p < 0.01 versus the IP10-scFv group

In comparison with that in the controls, the frequency of CXCR3⁻CD8⁺ and CXCR3⁺CD8⁺ T cells increased by two–threefold in the CTL-treated mice, suggesting that injection i.v with CTL promoted the infiltration of CD8⁺ T cells into the brain of mice (Fig. 3b, c). The frequency of CXCR3⁻CD8⁺ and CXCR3⁺CD8⁺ T cells in the IP10-scFv-treated mice was significantly higher than that in the controls and the CTL-treated mice (p < 0.01 for both), indicating that IP10-scFv had powerful chemotactic activity. The frequency of CXCR3⁻CD8⁺ (p < 0.01) and CXCR3⁺CD8⁺ (p < 0.05) in the IP10-scFv-/CTL-treated mice was higher than that in the IP10-scFv-treated mice. There was no significant difference in the frequency of CXCR3⁻CD8⁺ and CXCR3⁺CD8⁺ T cells in the brains of glioma-bearing mice between 14 and 28 days post-inoculation (p > 0.05). These data clearly indicated that treatment with IP10-scFv promoted the migration of CD8⁺ T cells and that treatment with both IP10-scFv and CTL synergistically increased the frequency of CD8⁺ T cell infiltrates in the brain of glioma-bearing mice.

Treatment with IP10-scFv and CTL synergistically induces “Th1-like” splenic IFN-γ responses in mice

Antigen-specific IFN-γ-producing Th1 and Th1-like CD8⁺ T cell responses are crucial for anti-tumor immunity. We next examined the impact of different treatments on the frequency of splenic IFN-γ-secreting T cells ex vivo by ELISPOT assays. As shown in Fig. 4, the frequency of splenic IFN-γ-secreting T cells in the IP10-scFv-treated mice was significantly higher than that in the EGFRvIIIscFv-treated mice (p < 0.01), but was significantly less than that in the CTL-treated mice (p < 0.05). Furthermore, the frequency of splenic IFN-γ-secreting T cells in the IP10-scFv-/CTL-treated mice was significantly higher than that in other groups of mice (p < 0.05).

Fig. 4.

ELISPOT analysis of splenic glioma-specific IFN-γ-secreting T cells in mice. Splenic mononuclear cells were isolated from different groups of mice and challenged with glioma cells in vitro for 48 h. The glioma-specific IFN-γ-secreting T cells were determined by ELISPOT assays. Data are expressed as mean ± SD of the SFC numbers of individual groups of mice from three separate experiments (n = 6 per group). The background SFC in glioma and T cell alone wells was <3 per well. *p < 0.05, **p < 0.01 versus the CTL group; ^# p < 0.05, ^## p < 0.01 versus the EGFRvIIIscFv group; ^{^} p < 0.05, ^{^^} p < 0.01 versus the IP10-scFv group

Next, we examined T cell immunity in the target brain tissues by testing the cytotoxicity of the isolated BILs against glioma GL261-EGFRvIII cells in vitro. There was a low level of cytotoxicity in the mice that had been treated with IP10-scFv alone, which was similar to that in the control mice, regardless of the ratios of effectors to targets (Fig. 5a). In contrast, the percentages of GL261-EGFRvIII cells lyzed by the BILs from the CTL-treated mice increased significantly with the increased ratios of effectors to targets (p < 0.05–0.01), and the frequency of GL261-EGFRvIII cells lyzed by the BILs from the IP10-scFv-/CTL-treated mice was significantly higher (p < 0.05). Similarly, the in vitro generated CTL had potent cytotoxicity against glioma cells, but failed to kill other tumor cells and non-tumor cells (Fig. 5b). Hence, treatment with both IP10-scFv and CTL promoted strong Th1-like T cell immunity, which had potent cytotoxicity against glioma cells.

Fig. 5.

The glioma-specific cytotoxicity. a The cytotoxicity of BIL cells. The BILs were isolated from the IP10-scFv and CTL, IP10-scFv, CTL, or EGFRvIIIscFv-treated mice on day 14 post-inoculation. b the cytotoxicity of CTL generated in vitro. Data are expressed as the mean ± SD of the percentages of glioma-specific cytotoxicity in different groups of mice from three separate experiments. *p < 0.05, **p < 0.01 versus the CTL and 3LL group; ^# p < 0.05, ^## p < 0.01 versus the EGFRvIIIscFv and Hepa1-6 group; ^{^} p < 0.05, ^{^^} p < 0.01 versus the IP10-scFv and NIH3T3 group

Treatment with IP10-scFv and CTL synergistically induces glioma cell apoptosis in vivo

Finally, we examined whether the induced T cell responses could trigger glioma cell apoptosis in vivo. As shown in Fig. 6, while the frequency of apoptotic glioma cells in the CTL-transfused mice was significantly higher than that in the EGFRvIIIscFv group (p < 0.05), but was significantly lower than that in the IP10-scFv-treated mice (p < 0.05). Furthermore, the frequency of apoptotic glioma cells in the IP10-scFv-/CTL-treated mice was significantly higher than that in other groups. These data clearly indicated that treatment with both IP10-scFv and CTL synergistically increased the frequency of apoptotic glioma cells, which contributed to inhibiting the growth of implanted glioma in mice.

Fig. 6.

The glioma cell apoptosis. Glioma cell suspension was prepared from different groups of mice at 21 days post-inoculation and stained with APC-Annexin V and PI. The percentages of apoptotic glioma cells were determined by flow cytometry. Data are expressed as the mean ± SD of the percentages of apoptotic glioma cells in different groups of mice from three separate experiments. *p < 0.05, **p < 0.01 versus the EGFRvIIIscFv group; ^# p < 0.05, ^## p < 0.01 versus the CTL group; ^{^} p < 0.05, ^{^^} p < 0.01 versus the IP10-scFv group

Discussion

The success of immunotherapeutic strategies against cancer depends on the generation of effective tumor antigen-specific T cells that can not only infiltrate the tumor tissue, but also effectively attack tumor cells. Clinically, the numbers of circulating tumor antigen-specific CD8⁺ CTL in individuals with cancer are not correlated with the numbers of CTL infiltrated into cancer tissues and tumor regression [24]. Therefore, therapeutic approaches to recruit CTL into the tumor may improve the efficacy of immunotherapies for patients with tumors [25, 26].

IP-10 is a potent chemotactic factor of CTL [27], and a single chain of antibody has been an attractive carrier of bioactive molecules into the tumor tissues due to its antigen-specific and small size. In this study, we tested whether local treatment with IP-10 could effectively recruit glioma-specific CTL into the tumor and inhibit the growth of glioma in mice. First, we generated a recombinant IP10-scFv that specifically recognized glioma-specific EGFRvIII antigen commonly expressed on the membrane surface of glioma cells. The recombinant IP10-scFv from E. coli may possibly contaminate with LPS. During the generation of recombinant proteins, we purified these recombinant proteins using the TOYOBO MagExtractor fusion purification Kit (Toyobo, Tokyo, Japan). As we stated, the purity of IP10-scFv used in this study was >93 % by SDS-PAGE and silver staining, and the EGFRscFv had similar purity (Lanjing Bisynthesis, China). More importantly, we found that treatment with the control EGFRscFv had no beneficial effect on inhibiting the growth of implanted tumor, similar to that of treatment with PBS (data not shown). These data suggest that the possibly contaminating LPS may be insufficient in interfering with the growth of implanted tumor or resulting in any significant effect on anti-tumor immunity we observed. Following the purification, we found that the generated IP10-scFv had similar affinity to that of scFv and retained the immunoreactivity of scFv and the chemotactic activity of IP-10. Given that dendritic cells have potent antigen-presenting activity and can induce T cell activation, we isolated dendritic cells from naïve mice and loaded them with glioma cell lysates, followed by stimulating CD8⁺ T cells to generate CTL in vitro. We found that the generated CTL had selective cytotoxicity against glioma in vitro. Given that IP-10 is a conservative protein among various mammals [28], the generated IP10-scFv may be used for testing its therapeutic effect in human glioma xenograft in humanized nude mice. Together, these unique reagents provide a basis for testing a combination of IP10-scFV and CTL in vivo.

We studied the therapeutic effect of IP10-scFv and/or glioma-specific CTL in a mouse model of implanted glioma and found that treatment with either IP10-scFv or CTL inhibited the growth of implanted glioma and prolonged the survival of tumor-bearing mice. Treatment with both IP10-scFv and CTL synergistically inhibited the growth of tumors and prolonged the survival of mice. Indeed, treatment with either IP10-scFv or CTL increased the numbers of inflammatory infiltrates and the frequency of CXCR3⁺CD8⁺ T cells in the brain of mice, and treatment with both synergistically enhanced inflammatory infiltration and elevated the percentages of effector T cells in the brain of mice. Furthermore, treatment with both IP10-scFv and CTL synergistically enhanced splenic glioma-specific IFN-γ responses and CD8⁺ T cell-mediated cytotoxicity against glioma cells in vitro. In addition, treatment with both IP10-scFv and CTL triggered a higher frequency of glioma cell apoptosis, similar to that in another type of tumor [29]. These novel data indicated that the injected IP10-scFv in the tumor recruited effector T cells, which attacked the tumor cells by inducing glioma cell apoptosis, leading to the inhibition of tumor growth and the prolonged survival of tumor-bearing mice. Our data extend previous findings that vaccination with IP-10 combined with glioma antigen-loaded DCs induces potent CTL and inhibits the growth of glioma in mice [28, 30]. We are interested in further investigating whether vaccination with tumor antigen-loaded DCs together with treatment with IP10-scFv can induce potent anti-tumor immunity in vivo. Therefore, our findings may provide a proof of principle that therapeutic strategy of both local chemokines and systemic administration of tumor-specific CTL is valuable for inhibiting the growth of glioma.

Tumor-specific CTLs are crucial for the control tumor growth and anti-tumor immunity. A previous study has shown that the presence of a low-to-moderate number of functional CTLs is sufficient in the control of tumor growth [31]. Activated CD8⁺ T cells can secrete perforin, granzymes, and IFN-γ, which can directly kill target tumor cells [32]. Furthermore, IFN-γ can activate other infiltrated inflammatory cells in tumors, limit TAM development, and enhance their cytotoxicity [33]. In addition, IFN-γ can also enhance MHC I molecule expression on tumor cells and increases their susceptibility to CTL-mediated cytotoxicity [34, 35]. We detected a high frequency of splenic glioma-specific IFN-γ-secreting CD8⁺ T cells and strong cytotoxicity against glioma cells in the brains of the mice that had been treated with IP10-scFv and CTL. Conceivably, these IFN-γ-producing CTL may mediate the inhibition of glioma growth in mice. We are interested in further investigating whether other types of functional CD8⁺ and CD4⁺ T cells also participate in anti-tumor immunity in this model.

It is notable that the effect of treatment with IP10-scFv alone on inhibiting the growth of glioma was significantly stronger than that treatment with CTL. These data suggest that the injected IP10-scFv in the tumor is sufficient in recruiting inflammatory cell infiltration into the brain and controlling glioma growth in mice. Indeed, we detected greater numbers of BILs and a significantly higher frequency of apoptotic glioma cells. Furthermore, we detected a higher frequency of CXCR3⁺CD8⁺ T cells in the brains in the IP10-scFv-treated mice than that in the CTL-treated mice. These data suggest that CXCR3⁺CD8⁺ T cells in the brain may be sufficient in triggering glioma cell apoptosis and inhibiting glioma cell growth in mice. Alternatively, the injected IP10-scFv may recruit other inflammatory cells that inhibit the growth of gliomas in mice. These data support the notion that IP-10 is crucial for inducing tumor antigen-specific T cell responses and recruiting them into the tumor site [36]. In addition, although we detected a higher frequency of splenic glioma-specific IFN-γ-secreting T cells and stronger cytotoxic CD8⁺ T cells in the brain, there was only moderate inhibition of glioma growth in the CTL-treated mice. The low inhibitory effect of CTL treatment may stem from poor CTL infiltration into the brain of mice in the absence of high levels of IP-10. The weak inhibitory effect of CTL treatment on the growth of glioma suggests that the frequency of tumor-specific IFN-γ-secreting T cells may not reflect anti-tumor immunity in the brain and may not be a good biomarker for evaluating tumor-specific T cell immunity in the target tissues. Indeed, the numbers of circulating tumor antigen-specific CD8⁺ CTL in individuals with cancer in the clinic are not correlated with the numbers of CTL infiltrated into cancer tissues and tumor regression [24]. Therefore, it is important to search new biomarkers for the evaluation of immunotherapy of tumors.

In summary, our data indicated that treatment with IP10-scFv and CTL synergistically inhibited the growth of gliomas and prolonged the survival of tumor-bearing mice, accompanied by enhancing inflammatory cell infiltration and CTL-mediated cytotoxicity as well as glioma cell apoptosis in the brain and splenic glioma-specific IFN-γ-secreting T cell responses in mice. Our data clearly demonstrate that combination of local chemokine treatment with transfusion of antigen-specific CTL is a promising strategy for inducing potent antigen-specific T cell immunity and for controlling tumor growth. Therefore, our findings may provide new insights into glioma-specific immunity and aid in the design of new immunotherapies for the intervention of glioma.