Human glioblastoma stem-like cells are more sensitive to allogeneic NK and T cell-mediated killing compared with serum-cultured glioblastoma cells

Abstract

Glioblastoma multiforme (GBM) is the most dramatic primary brain cancer with a very poor prognosis because of inevitable disease recurrence. The median overall survival is less than 1 year after diagnosis. Cancer stem cells have recently been disclosed in GBM. GBM stem-like cells (GSCs) exhibit resistance to radio/chemotherapeutic treatments and are therefore considered to play an important role in disease recurrence. GSCs are thus appealing targets for new treatments for GBM patients. In this study, we show that GBM cells with stem cell characteristics are resistant to lysis mediated by resting natural killer (NK) cells because of the expression of MHC class I molecules. However, GSCs are killed by lectin-activated NK cells. Furthermore, in experiments using the therapeutic antibody CetuximAb, we show that GSCs are sensitive to antibody-mediated cytotoxicity. We confirm the sensitivity of GSC to cytotoxicity carried out by IL2-activated NK cells and tumor-specific T cells. More importantly, we show that GSCs are more sensitive to NK and T cell-mediated lysis relatively to their corresponding serum-cultured GBM cells obtained from the same initial tumor specimen. Altogether, these results demonstrate the sensitivity of GSC to immune cell cytotoxicity and, therefore, strongly suggest that GSCs are suitable target cells for immunotherapy of GBM patients.

Figure 6

Glioblastoma multiforme (GBM) neurosphere (NS) cells are preferentially killed by lectin‐ and IL2‐activated natural killer (NK) cells. PKH26‐labeled GBM#2 NS or adherent (Adh) cells alone or mixed at 1:1 ratio with unlabeled GBM#2 NS or Adh cells were used as target cells in a 4‐h cytotoxicity assay. PHA‐ and IL2‐activated NK effector cells [for lectin‐dependent cytotoxicity (LDC) and LAK lysis, respectively] were used at 10:1 and 20:1 ratio, respectively, with PKH26+ target cells alone; and at 10:1:1 and 20:1:1 ratio, respectively, with mixed unlabeled and PKH26+ target cells. The percentages of dead cells of mixed unstained and stained target cell populations were assessed by flow cytometry using 7‐amino‐actinomycin D (7AAD) staining. Activated NK effector cells were previously labeled with carboxyfluorescein succinimidyl ester and then were excluded from the analysis. A and B. Targets: GBM#2 NS and Adh cells. A. Representative flow cytometry profiles obtained with NS and Adh cells using phytohemagglutinin (PHA)‐activated NK cells are shown. B. Results are presented as percentages of specific lysis of the tested target cells as described in Materials and Methods. GBM#2 cell line is shown (see Supporting Information Figure S8 for the other cell lines).

Figure 1

Characterization of glioblastoma multiforme (GBM) primary cell lines. Cells from GBM specimen were expanded in a serum‐free medium [neurosphere (NS) condition] or a fetal calf serum‐containing medium [adherent (Adh) condition]. A. Typical cell morphology of neurospheres for NS cell lines and adherent for Adh cell lines. A representative primary cell line GBM#2 is shown (see Supporting Information Figure S1A for the others cell lines). B. Cumulated number of neurospheres and cells obtained with NS cell lines during the different passages. C. Expression of stem cell and progenitor markers. Cells from NS and Adh GBM lines were dissociated and stained with isotype controls (open histograms) or specific labeled antibodies against CD133, nestin, SOX2, SSEA1, A2B5 and NG2 molecules (closed histograms), and then analyzed by flow cytometry. A representative primary cell line GBM#2 is shown (see Supporting Information Figure S1B for the others cell lines). D. The mean of specific fluorescence intensity of the protein expression obtained in each NS and Adh cell line (n = 8 and 5, respectively) was determined in at least three different experiments as described in Materials and Methods. The thick gray bar indicates the mean of specific fluorescence intensity of protein expression obtained in all cell lines. Cell lines were considered as positive for the expression of proteins of interest when the specific fluorescence intensity was more than 2 (dashed line). ***P < 0.01; *P < 0.05. ns = not statistically different.

Figure 2

Expression of MHC molecules and tumor antigens on glioblastoma multiforme (GBM) primary cell lines. A. GBM neurosphere (NS) and adherent (Adh) cells were stained with isotype controls (open histograms) or specific labeled antibodies against HLA‐ABC, HLA‐DR, IL13Rα2, EGFRvIII and EGFR (closed histograms), and then analyzed by flow cytometry. For EGFR expression, NS cell lines were grown for 48 h without (open histograms with thick lines) or with EGF (closed histograms). A representative primary cell line GBM#2 is shown (see Supporting Information Figure S4 for the others cell lines). B. The mean of specific fluorescence intensity of the protein expression obtained in each NS and Adh cell line was determined as described in Figure 1 (n = 8 and 5, respectively). The thick gray bar indicates the mean of specific fluorescence intensity of protein expression obtained in all cell lines. Cell lines were considered as positive for the expression of proteins of interest when the specific fluorescence intensity was more than 2 (dashed line). *P < 0.05. ns = not statistically different. C. Tumor antigen mRNA expression on GBM cell lines was analyzed by transcriptomic microarray experiments. Total mRNA from GBM cell lines was extracted and analyzed for a gene expression profile. Results are expressed as the mRNA expression fluorescence intensity.

Figure 3

Sensitivity of glioblastoma multiforme (GBM) neurosphere (NS) cell lines to natural killer, lectin‐dependent, antibody‐dependent and lymphokine‐activated lysis mediated by natural killer (NK) cells. A. Unstimulated and IL2‐stimulated NK cells were stained with specific labeled antibodies against T and NK cell markers CD3, CD56, NKp46 and NKp44. B. GBM NS cell lines were incubated with human IgG (open histograms), TrastuzumAb and CetuximAb therapeutic antibodies (closed histograms) and stained with labeled antibodies against human IgG and then analyzed by flow cytometry. A representative primary cell line GBM#2 is shown (see Supporting Information Figure S6A for the others cell lines). C. GBM NS cell lines were labeled with 51‐Cr and used as target cells in a 4‐h cell cytotoxicity assay with NK cells or IL2‐activated NK effector cells. Different effector : target (E : T) ratios were used and specific lysis was calculated as indicated in Materials and Methods. Effector cells were added alone (□) or in the presence of anti‐HLA‐ABC blocking antibodies (), the lectin phytohemagglutinin (PHA) (◆), the therapeutic antibodies TrastuzumAb (○) or CetuximAb (). A representative primary cell line GBM#2 is shown (see Supporting Information Figure S6B for the others cell lines). D. Similar results were obtained with NK cells prepared from four different healthy donors. E : T ratio: 10:1 (see Supporting Information Figure S6C for the others cell lines). ***P < 0.01; *P < 0.05. ns = not statistically different (comparison with lysis with NK cells alone); LDC = lectin‐dependent cytotoxicity; ADCC = antibody‐dependent cell cytotoxicity.

Figure 4

Sensitivity of MelanA‐loaded glioblastoma multiforme (GBM) neurosphere (NS) cell lines to MelanA/HLA‐A2‐specific T cells. A. MelanA/HLA‐A2‐specific T cell lines were generated in vitro after co‐culture with autologous dendritic cells loaded with MelanA peptides. MelanA/HLA‐A2‐specific T cell lines obtained were stained with MelanA/HLA‐A2 tetramers and anti‐CD3, ‐CD4, ‐CD8 antibodies, and then analyzed by flow cytometry. B. MelanA/HLA‐A2‐specific T cell lines were used as effectors in a 4‐h cell cytotoxicity assay against 51Cr‐labeled unloaded (□) or MelanA‐loaded () T2 cells (used as positive controls) and GBM NS cell lines in the absence or in the presence of blocking antibodies against HLA‐ABC (). Different effector : target (E : T) ratios were used and specific lysis was calculated as indicated in Figure 3. A representative primary cell line GBM#2 is shown (see Supporting Information Figure S7A for the other cell lines). C. Results are representative of results obtained with three different MelanA/HLA‐A2‐specific T cell lines. E : T ratio: 10:1 (see Supporting Information Figure S7B for the other cell lines). ***P < 0.01; ns, not statistically different (comparison with unloaded target cells). ⁺⁺⁺ P < 0.01; ⁺ P < 0.05 (comparison with MelanA‐loaded target cells).

Figure 5

Comparison of glioblastoma multiforme (GBM) neurosphere (NS) and adherent (Adh) primary cell lines sensitivity to natural killer (NK) and T cell lysis. Unloaded or MelanA‐loaded GBM NS and Adh cell lines obtained from the same initial tumor specimen were used as target cells against NK cells in the presence of the blocking antibody anti‐HLA‐ABC (), the lectin phytohemagglutinin (PHA) (◆); or IL2‐activated NK cells (□); or MelanA‐specific T cell effectors (). Results are representative of results obtained with four different donors for NK cell effectors and three different MelanA/HLA‐A2‐specific T cell lines. E : T ratio: 10:1. ***P < 0.01; *P < 0.05. ns = not statistically different.