Profound tumor-specific Th2 bias in patients with malignant glioma

All peptides were synthesized using standard FMOC chemistry to 95% purity (New England Peptide Company). All peptides were at least 15 amino acids in length. The following peptides were used for stimulation of PBMCs: MAGE-A3_112-127 (KVDELAHFLLRKYRAK); MAGE-A3_121-136 (LRKYRAKELVTKAEML); MAGE-A3_143-160 (WQYFFPVIFSKASSSLQL); IL13Rα2_341-355 (LLRFWLPFGFILILV); IL-13Rα2_351-365 (ILILVIFVTGLLLRK); MBP_85-99 (ENPVVHFFKNIVTPR). HLA class II alleles predicted to bind the peptides were determined using ProPred HLA class II binding algorithm [15], summarized in Table 1. All peptides were predicted to be very promiscuous, binding to multiple (up to 9 in several cases) alleles. The validity of the predictions is supported by experimental data for the MBP_85-99 peptide, which has been shown to bind to both the DRB1*04 and *15 alleles as predicted [16].

Twenty-five mL of blood from patients or healthy subjects was obtained under an IRB-approved protocol. All samples of blood from patients were taken at the time of surgery. Ages of the patients ranged from 48 to 76 among patients with primary GBMs (median age 55), from 41 to 69 among patients with recurrent GBMs (median age 52), and from 40 to 73 among patients with meningiomas (median age 62). A greater number of men had primary or recurrent GBMs (6 of 8 patients and 5 of 6 patients, respectively), while more women than men had meningiomas (5 of 8 patients). Primary and recurrent GBMs were located in temporal, parietal, and frontal lobes with comparable frequencies. All tumor-bearing patients received similar doses of steroids and anti-epileptic medications at the time of tumor debulking surgery prior to obtaining peripheral blood for these studies. T cell responses in patients with meningiomas controlled for influences of steroids on antigen responsiveness and cytokine balance. Tumor tissue was independently confirmed in all cases by formal pathological analysis. PBMCs were purified from heparinized blood by density gradient centrifugation using Ficoll-Hypaque (GE Healthcare Biosciences), and cells were then washed with PBS and viable cells quantified by trypan-blue staining.

Freshly isolated PBMCs were plated at 2 × 10⁵ cells/well in 200 μl of serum-free X-VIVO15 (X15) media (Lonza) in 96-well round-bottom cell culture plates. Candidate peptides, in addition to a negative control peptide derived from MBP were added at a concentration of 10 μg/mL and anti-CD3 mAb was added at a concentration of 1 μg/mL. Six T cell cultures were established for each condition in each subject, and 100 IU/ml of IL-2 was added on the following day. Plates were incubated at 37°C and 5% CO₂ for 14 days, with media changed as needed, and the supernatant was harvested to evaluate T cell responses (cytokines) induced by each condition using ELISA. In a limited number (n = 3) of patients, we assessed cytokine production after both 7 and 14 days. Tumor-specific responses were apparent at day 7, and the frequency of positive responses did not change significantly at day 14, but the cytokine values did increase significantly (data not shown). An IFN- ELISPOT assay was performed as previously described [17].

To detect T helper cell responses directed against the candidate peptides, IFN-γ and IL-5 were measured by ELISA using commercially available kits supplied by BD bioscience. IFN-γ was used as a prototypic Th1 cytokine and IL-5 was chosen as a prototypic cytokine released by Th2 cells because unlike IL-4 there would be no potential consumption by antigen-specific T cells in our culture conditions [18]. The Th2-associated transcription factor GATA-3 directly binds and regulates both the IL-4 and IL-5 gene promoters [19] and a positive correlation has been reported among GATA-3, IL-4, and IL-5 gene expression during human T cell differentiation [20], providing further support for analysis of IL-5 as a representative Th2 cytokine. Initial experiments also examined the secretion of IL-10 in response to peptide stimulation, which was not detected. Flat-bottom microtiter plates (Immulon) were coated with primary antibody (IFN-γ or IL-5) diluted 1:1000 in NaHCO3 and incubated overnight at 4°C. Coating solution was then removed, plates blocked with PBS + 10%FBS at 25°C for 2 hours, rinsed 3 times with diluted wash buffer (dH20, Tween 20, PBS 20X), and standards were then added in duplicate at 0, 62.5, 125, 250, 500, 1000, 2000, and 4000 pg/mL (diluted in X15 media). Supernatants (50 μl/well) from T cell assays were then added to wells. Plates were incubated for 2 hours at 25°C and subsequently rinsed 3 times. Wells were then coated with a secondary biotinylated antibody diluted 1:1000 in PBS + 1%FBS and incubated for 1 hour at 25°C. Plates were again rinsed 3 times and avidin-peroxidase diluted 1:1000 in PBS+10%FBS was added and incubated for another 1hour. After rinsing 6 times, TMB (tetramethylbenzidine) (BD biosciences) was added to wells, which were allowed to develop. The reaction was stopped by adding 50 μL of sulfuric acid and absorbance was measured at 455 nm by an ELISA plate reader (BIO-RADR). A standard curve was generated by plotting absorbance against each reference standard, and sample concentrations were extrapolated from this curve. Appropriate statistical tests and analyses based on our data were determined using Prism 5.0 (GraphPad software).

Six primary T cell cultures were established from each patient against all stimuli. Anti-CD3 mAb was used to stimulate and expand T cells to confirm T cell viability and to examine global, nonspecific T cell cytokine responses among the different cohorts. Relative to healthy subjects, anti-CD3 mAb-induced IFN-γ levels in patients with GBMs (primary and recurrent) and meningiomas were modestly lower (Figure 1a). More strikingly, anti-CD3 mAb stimulation uniquely induced secretion of high amounts of IL-5 from patients with recurrent GBMs (P < 0.0001). A recent clinical trial examined the IFN-γ/IL-5 ratio after polyclonal stimulation of PBMCs in patients with metastatic melanoma treated with immunomodulators given to restore the Th1/Th2 balance [21], and we performed a similar analysis of our data (Figure 1b). The IFN-γ/IL-5 ratios in both primary GBM patients (geometric mean 3.7) and recurrent GBMs (geometric mean 0.9) were significantly lower than those in healthy subjects (geometric mean 16.0) and meningioma patients (geometric mean 10.0) (p<0.001). This antigen-nonspecific bias towards a Th2 response in patients with primary and recurrent GBMs is consistent with past reports [22‐25]. There was no significant difference in the global IFN-γ/IL-5 ratio between healthy subjects and meningioma patients, indicating that neither treatment with steroids or antiepileptic medications nor the simple presence of a CNS tumor were responsible for the deviation in global T cell responses.

Both glial cells and melanocytes derive from neural ectoderm [26] and several studies have demonstrated that melanoma-associated tumor antigens are also expressed by gliomas, including MAGE-A3 [27‐29]. Like MAGE-A3, IL-13Rα2 is a cancer testes antigen that is over-expressed in gliomas [30, 31]. In trying to identify novel glioma-associated HLA class II-restricted T helper cell epitopes, we hypothesized that epitopes identified in patients with melanoma may similarly be expressed by patients with gliomas. Two such epitopes with homology to MAGE-A3 were identified [32‐34], which we modified to incorporate adjacent HLA class I-restricted CTL epitopes [35‐38] as well as several amino acid substitutions. The amino acid substitutions altered the hydrophobicity of the peptides but not their charge (Ala to Asp, Leu to Arg) and potentially their secondary structure (Pro to Leu). Similarly, two overlapping 15mer IL-13Rα2 epitopes were identified, one of which was modified to incorporate a CTL epitope [39]. The five epitopes used in this study in relation to previously described epitopes are depicted in Table 2.

Measurement of antigen-specific T cell responses in the peripheral blood in humans differs depending on whether responses are high affinity interactions with foreign (viral) epitopes or lower affinity interactions involving recognition of self-antigens. We have previously demonstrated that high frequencies of T cells directed against the self-antigen MBP peptide 85–99 in the peripheral blood of patients with multiple sclerosis (MS) fail to proliferate when stimulated with antigen but readily secrete high levels of cytokine [40]. Given that T cell responses directed against MAGE and IL-13Rα2 antigens also involve T cells with low affinity to these self-antigens, we quantified antigen-specific responses based on cytokine secretion, as recently described in a phase I study of patients with MS [41]. We quantified cytokine production by ELISA, defining a positive T cell response for each patient as the amounts of IFN-γ or IL-5 that were > 50 pg/mL and two standard deviations above the mean cytokine levels secreted after stimulation of cells from that patient with negative control MBP peptide. The mean cut-off for a positive cytokine response based on cytokine induced by stimulation with control MBP peptide was 895 pg/ml (range: 13–1298) and 314 pg/ml (range: 72–852) for IFN-γ and IL-5 among healthy subjects, and was 123 pg/ml (range: 0–286) and 312 pg/ml (range: 59–1347) for IFN-γ and IL-5 among GBM patients. Use of a traditional IFN-γ ELISPOT assay, in which quantification of spots can at times be ambiguous, confirmed that memory T cell responses could be detected with these peptides in patients with primary GBMs, as peptide specific cytokine production could be detected with 48 hours of culture (Figure 2).

T cells responding to all five peptides examined among healthy subjects exhibited a predominant Th1 response (high IFN-γ and low IL-5 secretion) (Figure 3). In marked contrast, the majority of peptide-specific T cell responses among both primary and recurrent GBM patients were Th2 polarized (low IFN-γ and high IL-5 secretion). Frequencies of response to the individual peptides were most prevalent among healthy subjects and patients with primary GBMs, both in terms of the number of subjects responding to a given peptide and the number of positive lines. Responses in one or a few subjects did not dominate among any of the cohorts examined; at least half of the subjects (in some cases all) in each cohort responded to all of the epitopes tested (Table 3). Patients with meningiomas generally had less frequent responses, though strong responses could be detected against both MAGE-A3 and IL-13Rα peptides. Mean IFN-γ/IL-5 ratios were significantly lower (p<0.05) for patients with primary GMBs (geometric means for MAGE-A3_112-127, MAGE-A3_121-136, MAGE-A3_143-160, IL13Rα2_341-355, and IL-13Rα2_351-365 were 0.2, 0.1, 0.3, 0.3, 0.3, respectively) relative to healthy subjects (geometric means were 4.9, 8.1, 4.3, 1.6, 2.0, respectively) in response to all of the epitopes (Figure 4). The Th2 bias was even more profound among patients with recurrent GBMs (geometric means for MAGE-A3_112-127, MAGE-A3_121-136, MAGE-A3_143-160, IL13Rα2_341-355, and IL-13Ra2_351-365 were 0.04, 0.06, 0.4, 0.02, 0.05), and was significantly lower than that of patients with primary GBMs for the MAGE-A3_143-160 and IL-13Rα2_351-365 epitopes (P<0.05).