Abstract
Histone deacetylase inhibitors (HDACi) have been described as multifunctional anticancer agents. The failure of conventional therapy for glioblastoma (GBM) renders this tumor an attractive target for immunotherapy. Innate immune cells, such as natural killer (NK) cells, play a crucial role in antitumor immune responses. Here, we describe how the HDACi trichostatin A (TSA) promotes apoptosis of tumor cells, as well as augments anti-GBM innate immune responses. In vitro treatment of GBM cells with TSA results in an up-regulation of the natural killer group-2 member-D (NKG2D) ligands major histocompatibility complex class I-related chain (MIC)-A and UL16 binding protein (ULBP)-2 at both mRNA and protein levels, rendering them susceptible to NK cell-mediated lysis. In vivo, TSA delays tumor growth of GBM xenografts. Both the in vitro and in vivo antitumor effect of TSA was significantly reduced by blocking NK cell activity. Our data suggest that HDACi, especially in combination with other clinical immunotherapeutical approaches, may be considered in a combined therapeutic approach for GBM.
Glioblastoma (GBM) is the most frequent malignant brain tumor with an incidence of 3.5 cases per 100,000 people per year. GBMs are among the most lethal neoplasms, conferring a median survival of only 12 to 15 months, even when the patients receive standard therapy including surgical resection and radio-chemotherapy (1). Excessive proliferation, resistance to cell death, diffuse infiltration into the parenchyma, massive neoangiogenesis and suppression of antitumor immune surveillance contribute to the aggressive phenotype of GBM. Currently, only the chemotherapeutic drug temozolomide, in combination with tumor irradiation, provides a rather low increase in survival (1), although the methylation status of the O(6)-methylguanine-DNA methyltransferase (MGMT) promoter is predictive for the survival of patients with GBM (2). The lack of effective therapy regimens for malignant GBM is a result of, among others, the molecular mechanisms of epigenetically-mediated reduction of gene transcription, either by methylation of CpG islands in promoter regions, or by histone de-acetylation-mediated chromatin condensation. Gene silencing leads to altered protein expression, especially of proteins regulating cell homeostasis (3), and, in consequence, leads to tumor progression and enhanced malignancy, even in glioma (4, 5).
Acetylation and de-acetylation of core histones and other proteins comprise of non-genomic mechanisms regulating gene expression. Histone acetylation opens chromatin and thereby promotes transcription, whereas its de-acetylation leads to a condensed chromatin structure and results in the opposite effect. The removal of acetyl moieties from histones and non-histone proteins is catalyzed by histone deacetylase (HDAC), and the acetylation by histone acetyl transferases (HAT). Normal cell differentiation and homeostasis in metabolic activity requires well-regulated transcription and balanced activity of HAT and HDAC. Unbalanced expression or function of HAT or HDAC disturbs cell homeostasis and can lead to the initiation of tumors. Accordingly, deletions or inactivating mutations of HAT or elevated activity of HDAC have been associated with tumor progression (3, 6), making pharmacological HDAC inhibition an interesting strategy for cancer therapy. Small-molecule inhibitors of HDAC were indeed found to alter cellular signaling networks relevant to tumorigenesis, and several HDAC inhibitors are currently tested in clinical trials against different types of cancers (7, 8).
HDAC inhibitors reduce the invasive ability of GBM cells via reduction of expression of the matrix metalloproteinases (MMP)-2 and -9 (9), enhance apoptosis through jun-N-terminal kinase activation and block telomerase activity (10), induce cell type-specific differentiation (11), sensitize p53-mutant GBM cells towards ionizing radiation (12) and promote growth arrest via induction of tetraspanin/CD81 expression (13). For other tumor entities, HDAC inhibitors have also been described as inducing expression of immunomodulatory surface proteins, rendering tumor cells more susceptible to immune-mediated killing (14-18). A variety of processes mediating tumor immunosuppression have been described: On one hand, major histocompatibility complex class-I molecules (MHC-I), which are necessary for the presentation of (tumor) antigens, are often down-regulated on tumor cells. This could serve as a mechanism of evading the host's immune response (19). The expression of MHC-I and other proteins of the antigen-processing machinery, such as transporters of antigen presentation (TAP)-1, TAP-2, or proteins associated with the proteasome pathway like low molecular mass polypeptide (LMP)-2 and Tapasin, have been described as being regulated by histone(de)acetylation (18, 20). In addition, interaction of the natural killer group-2, member-D (NKG2D), an activating receptor expressed on natural killer (NK) cells, with its ligands on tumor cells, has been described as a very important mechanism of antitumor surveillance. It has been demonstrated that GBM cells often down-regulate expression of NKG2D ligands, this way escaping NK cell recognition and lysis (21). The human NKG2D ligand family includes the stress-inducible surface glycoproteins major histocompatibility complex class I-related chain (MIC)A and MICB, as well as the UL16 binding proteins (ULBP), a multigene family with at least six members of which ULBP1 and ULBP2 are the most common in GBM (22, 23). Like proteins of the antigen-processing machinery, expression of NKG2D ligands is regulated by histone acetylation (14, 15, 17, 18, 24, 25). In glioma cells, surface expression of NKG2D ligands is negatively-regulated by the cytokine transforming growth factor beta (TGF-β), which potently blocks immune cell activation and proliferation (26). Down-regulation of either MICA/B or ULBPs on the surface of GBM cells masks them against NK cell-mediated lysis, whereas re-expression of at least MICA renders them susceptible to cell lysis executed by NK cells (27, 28). In the current proof-of-principle study, we investigated the effects of trichostatin A (TSA), an HDACi, on the immunesurveillance of GBM by the innate immune system, especially by NK cells.
Materials and Methods
Cell lines and reagents. The human malignant glioma cells lines LNT-229 and LN-308 were kindly provided by N. de Tribolet (Lausanne, Switzerland). These cells have been previously characterized in detail by Studer and Ishii (29, 30). LNT-229-Luc cells were generated by stable transfection of LNT-229 cells with pGL4-CMV-Luc expressing firefly luciferase. The cell lines were maintained in Dulbecco's modified Eagle's medium (DMEM; Lonza Group AG, Basel, Switzerland) containing 10% fetal calf serum (FCS), glutamine (2 mM), penicillin (100 IU/ml) and streptomycin (100 μg/ml). Tumor necrosis factor-related apoptosis-inducing ligand (APO2L/TRAIL) was purchased from PeproTech GmbH (Hamburg, Germany). CD95 ligand (CD95L)-containing supernatant was harvested from CD95L-transfected N2A neuroblastoma cells (31). The caspase-3/7 substrate Ac-Asp-Glu-Val-Asp-aminomethylcoumarin (DEVD-amc) was purchased from Bachem (Weil am Rhein, Germany). TSA was from InvivoGen (Toulouse, France). All other reagents were from Sigma (Munich, Germany).
Viability and cell growth assays. For analysis of cytotoxic effects of TSA on cell growth, 104 cells were seeded as triplets in microtiter plates and allowed to attach overnight. For analysis of cell growth over longer periods, 103 cells were seeded as triplets. The cells were treated at different concentration of TSA or for different time periods as indicated. Cell density was determined by crystal violet staining. For the analysis of synergy of TSA and APO2L/TRAIL, 104 cells were seed as triplicates in microtiter plates and allowed to attach overnight. After 24 h of TSA treatment, APO2L/TRAIL was added. The cultures were incubated for a further period of 24 h. Cell density was assessed by crystal violet staining and synergism was calculated by the method of Webb (35). For crystal violet staining, medium was removed and the cells were incubated in crystal violet solution (0.5%, 20% methanol) for 5 min followed by several washes with tap water. Bound crystal violet was dissolved in 50% ethanol/0.1 M sodium citrate and optical density readings were obtained using an ELISA reader (Thermo Electron Multiskan EX, Karlsruhe, Germany) at 560 nm. To analyze cell viability, the cells were stained with trypan blue. Viable cells were defined as trypan blue-negative.
Fluorescence-activated cell sorting (FACS)-based analysis. Surface protein expression was analyzed by flow cytometry using a phycoerythrin (PE)-conjugated anti-mouse antibody against CD49b (clone DX5; eBioscience, Frankfurt, Germany) for the detection of NKG2D on mouse splenocytes. For the detection of NKG2D ligands on human GBM cells, cells were incubated with hybridoma supernatants against MICA (clone AMO1), MICB (clone BMO1), ULBP1 (clone AUMO1), or ULBP2 (clone BUMO1) [all kindly provided by Alexander Steinle, Institute for Molecular Medicine, Frankfurt am Main, Germany; (22)], and the following secondary fluorescent-labeled antibodies were used: Cy3-conjugated mouse IgG1 (Dianova GmbH, Hamburg, Germany) and PE- or allophycocyanin (APC)-conjugated mouse IgG1 or IgG2. Isotype controls were mouse IgG1 or IgG2 (ImmunoTools GmbH, Friesoythe, Germany). In brief, the cells were detached using Accutase (PAA, Pasching, Austria) and subjected to FACS analysis as described elsewhere (26), using a Cyan-ADP cytometer (DAKO GmbH, Hamburg, Germany). To measure changes in surface protein expression, the specific fluorescence index (SFI) was calculated (fluorescence of specific antibody/fluorescence of isotype control). Changes are shown as the SFI of treated cells/SFI of control cells.
Protein detection and quantification. The preparation of cellular lysates, immunoblot procedure and detection of specific protein content has been previously described (32). For detection of histone acetylation, an antibody specific for acetyl-histone H3 was used (clone K14; Merck Millipore, Schwalbach, Germany). Equal protein loading was controlled by the detection of the housekeeping gene glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (clone 6C5; Millipore).
PCR analysis. For real time reverse transcription-polymerase chain reaction (RT-PCR), total RNA was prepared using Highpure Roche RNA isolation kit (Roche, Mannheim, Germany), and cDNA was prepared as described elsewhere (33). PCR was performed on a ABI 7500 Real-Time PCR System (Life Technologies, Darmstadt, Germany) using the ABsolute QPCR Sybr low ROX Mix (Thermo Scientific, Blenheim, Surrey, UK). The following primer pairs were used: Actin, forward: TGTTTGAGACCTTCAACACCC; Actin, reverse: AGCACTGTGTTGGCGTACAG; MICA, forward: GTGCCCCAGTCCTCCAGAGCTCAG; MICA, reverse: GTGGCATCCCTGTGGTCACTCGTC; MICB, forward: GGCGTCAGGATGGGGTATCTTTGA; MICB, reverse: GGCAGGAGCAGTCGTGAGTTTGCC; ULBP1, forward: CTGCAGGCCAGGATGTCTTGTGAG; ULBP1, reverse: TGAGGGTGGTGGCCATGGCCTTGG; ULPB2, forward: CTGCAGGCAAGGATGTCTTGTGAG; and ULPB2, reverse: TGAGGGTGGTGGCTGTGGCCCTGA. Relative expression of mRNA content was analysed using the comparative ΔΔCT method as described (http://www3.appliedbiosystems.com).
Caspase activity assay. The assay has been described in detail elsewhere (34). Briefly, the cells were seeded in microtiter plates and allowed to adhere. After treatment with TSA, or with CD95L as a control for caspase induction, the cells were washed with PBS, lysed, and 12.5 μM DEVD-amc was added. Fluorescence was measured after 30 min using a Mithras LB940 fluorimeter (Berthold Technologies, Bad Wildbad, Germany).
Purification and activation of immune effector cells. Peripheral blood mononuclear cells (PBMCs) were isolated from EDTA-anti-coagulated peripheral venous blood of healthy donors by density gradient centrifugation (Biocoll, Biochrom KG, Berlin, Germany). PBMCs were depleted of monocytes by adherence to plastic. The non-adherent fraction was co-cultured with irradiated RPMI 8866 feeder cells (ATCC, Manassas, VA, USA) and 25 U/ml of recombinant human Interleukin (IL)-2 (ImmunoTools) to obtain lymphokine-activated (LAK) cells. LAK cells were used as effectors in the lysis assays on day 11 of their culture. Where indicated, CD56+ cells were purified from bulk LAK cultures on day 11 using magnetic cell separation (MACS) NK isolation kit II (Miltenyi Biotec, Bergisch Gladbach, Germany) according to the manufacturer's protocol. Flow cytometric analysis to determine the purity of NK cells showed that >90% of the cells were CD56+CD3−. These cells were used as effectors in the lysis experiments on the same day. For the assessment of NK cell depletion in mice, NK cells were isolated from spleens of mice according to www.protocol-online.org. Erythrocytes were lysed for 5 min using ACK buffer (Invitrogen, Darmstadt, Germany). Splenocytes were counted and directly subjected to FACS-based detection of NK cells.
Cellular cytotoxicity assay. NK cell cytotoxicity against human GBM cells was analyzed using a non-radioactive assay measuring luciferase activity of LNT-229-Luc cells. Target LNT-229-Luc cells were incubated with TSA (2 μM) or control medium for 24 h, washed, and LAK or CD56+ NK cells were added to the target cells and incubated for 3 h at 37°C. For blocking experiments, 10 μg/ml neutralizing monoclonal antibody to NKG2D (clone 149810) or a control IgG1 antibody (both from R&D Systems, Wiesbaden, Germany) were added to the NK cells 30 min prior to co-culture. Viable LNT-229-Luc cells were determined by measuring luciferase activity using a Mithras LB940 Luminometer (Berthold Technologies). Maximum lysis was determined by sodium dodecyl sulphate treatment. The experimental lysis was corrected by substraction of the spontaneous lysis of target cells at the corresponding TSA concentration. The percentage of lysis was calculated as follows: 100× (corrected experimental lysis-spontaneous lysis)/(maximum lysis-spontaneous lysis). Experiments were carried out in triplicates.
Mouse experiments. Athymic CD1-deficient NMRI nude mice were purchased from Janvier (St. Berthevin, France); NMRI mice harbor functional NK cells, but do not develop functional T-cells. Female mice of 6-12 weeks of age were used in all experiments. The animal experiments were licensed by the regional board of Tübingen and were performed according to the German law, Guide for the Care and Use of Laboratory Animals. LNT-229 cells were treated with TSA (2 μM, 24 h) or left untreated. The cells were trypsinized, counted and viability was assessed by trypanblue staining. For subcutaneous (s.c.) tumor implantation groups of seven mice were injected s.c. with one million viable cells into the right flank. Mice were examined regularly for tumor growth using a metric caliper and sacrificed when tumors reached 500 mm2. To avoid artifacts due to cytotoxicity or proliferation inhibition induced by TSA, growth and viability of the cells used for inoculation was analyzed in parallel. For NK cell depletion, mice were injected intraperitoneally (i.p.) with 70 μl of an antibody to asialo-GM1 or with control IgG (both Wako Chemicals, Neuss, Germany) according to the manufacturer's instructions two days prior to tumor cell implantation and twice (day 2 and 5) after. Depletion of NK cells in mice was evaluated by flow cytometric analysis. In brief, lymphocytes were isolated from spleens of animals. The population of NK cells was analyzed using a PE-coupled anti-mouse CD49b antibody (clone DX5) or control IgM isotype antibody (both from eBioscience). Since NMRI nude mice do not develop T-cells, no counterstaining using a T-cell marker was used.
Statistical analysis. Figures show representative data from three to five independent experiments. Quantitative data were assessed using the t-test. To calculate synergy, the fractional product method was used (35). Box-whisker-plot analysis to evaluate differences in tumor size was carried out using JMP software (SAS, Cary, NC, USA) and Student's t-test.
Results
TSA acts synergistically with APO2L/TRAIL in inducing cell death of GBM cells. We first analyzed whether TSA can induce histone acetylation in GBM cell lines, and consequently open the chromatin structure leading to gene transcription. Greatest histone acetylation was observed by treatment of the cells for 24 h with 2 μM TSA (Figure 1A). Under these conditions, only marginal cytotoxic/cytostatic effects were observed, whereas prolonged treatment (>36 h) revealed both toxic and growth-inhibitory effects (Figure 1B). Even with higher doses of TSA (data not shown) or longer exposure times, approximately 40-60% of the cells survived (Figure 1B). In this context, p53-negative LN-308 GBM cells were more resistant to TSA-induced cell death than p53 wild-type LNT-229 cells. To assess whether the observed decrease in cell density was due to apoptosis induction, we analyzed caspase-3 and -7 activity. Caspase activity was moderately induced after 24 h of TSA treatment (Figure 1C). We chose to use 2 μM of TSA for our further studies, since this concentration was sufficient to induce histone acetylation with marginal cell death and low caspase activation at 24 h post treatment.
Trichostatin A (TSA) induces caspase activity and acts synergistically with tumor necrosis factor-related apoptosis inducing ligand (APO2L/TRAIL). A: LNT-229 cells were treated with 2 μM TSA or left untreated (Co) for 24 h. Histone acetylation was assessed by immunoblot. Ac-H3: acetylated histone H3. GAPDH: Glyceraldehyde 3-phosphate dehydrogenase. B: LNT-229 and LN-308 cells were treated with 2 μM TSA for increasing periods of time. Cell density was assessed by crystal violet staining (n=5-6, SEM; *p<0.05; **p<0.01; t-test). C: LNT-229 cells were treated with TSA (2 μM, 24 h) and caspase activity was assessed using DEVD-amc. Treatment of cells with CD95L (300 U/ml) plus cycloheximide (CHX, 10 μg/ml) for 6 h served as a control (n=2, SEM; *p<0.05). D: After 24 h of TSA treatment, APO2L/TRAIL (100 or 1000 ng/ml) was added to cultures incubated for a further 24 h. Cell density was assessed by crystal violet staining and synergism was calculated by the method of Webb (35) (n=3, one representative experiment is shown). Numbers indicate the fraction of surviving cells (%).
Since HDAC inhibitors have been described as inducing cell death in various cancer cell lines and as acting synergistically with the death ligand APO2L/TRAIL in B-cell chronic lymphocytic leukemia cells (36), we were interested in examining whether this also holds true for GBM cells. To this end, we pre-treated LNT-229 and LN-308 cells known to be resistant to APO2L/TRAIL with either TSA or vehicle for 24 h and then APO2L/TRAIL was added to the cultures. As shown in Figure 1D, treatment of the cells with TSA and APO2L/TRAIL in combination worked synergistically and was more potent than either single treatment. However, even with co-treatment, approximately 15% of the LNT-229 and 34% of the LN-308 cells survived. These findings demonstrate that TSA sensitizes, at least partially, APO2L/TRAIL-resistant GBM cells to death ligand-induced apoptosis.
TSA treatment induces up-regulation of NKG2D ligands and immunogenicity in GBM cells. NK cells have been described as key players in eliminating and preventing tumor growth of GBM (37). Recognition of tumor cells by NK cells is regulated by the expression of a variety of activating and inhibitory factors. The NKG2D/NKG2D ligand system is well-recognized for its importance in immune recognition of cancer cells by NK cells (26, 27, 38). Since TSA is a HDACi and should increase transcription, we first analyzed mRNA expression of different NKG2D ligands in LNT-229 cells. We observed a robust up-regulation of MICA and ULBP2, whereas TSA treatment did not affect mRNA expression of MICB and ULBP1 (Figure 2A). To examine whether these changes in mRNA also correlate with surface protein levels, the expression of all the aforementioned proteins was assessed by flow cytometry. As shown in Figure 2B and C, the general basal expression of MICA and MICB, as well as of ULBP1 and ULBP2, was low in control LNT-229 cells. In accordance with the PCR data, TSA increased the surface expression of MICA, and, to a lower extent, that of ULBP2, whereas no up-regulation of MICB and ULBP1 was detected. The most prominent up-regulation of surface protein expression post-24-h TSA treatment was that of MICA, even at a low TSA concentration (1 μM).
Down-regulation of NKG2D, an important activating receptor involved in NK cell-mediated lysis, has been shown to compromise NK cell-mediated immune surveillance in patients with GBM (28), whereas up-regulation of the NKG2D ligand MICA on GBM cells stimulates tumor immunity (27). Due to this fact and since it has been shown in other tumor entities that HDACi can stimulate NK cell-mediated antitumor responses (14, 15, 18, 24, 25), we analyzed whether the observed TSA-mediated up-regulation of MICA and ULBP2 on GBM cells could promote their recognition and lysis by LAK cells. Indeed, TSA significantly rendered GBM cells more susceptible to LAK cell-mediated cell lysis (Figure 3A). The observed enhancement in cellular cytotoxicity was not a result of TSA toxicity or reduced luciferase activity, since at the TSA concentration and duration of treatment we used, no reduction in cell density (Figure 1B), reduction of luciferase activity, or enhanced cell lysis was detectable when LNT-229 cells were treated with TSA in the absence of immune effector cells (Figure 3A). Furthermore, cytotoxicity was significantly reduced when isolated CD56+ NK cells were pre-treated with NKG2D-neutralizing antibody prior to co-cultivation with GBM cells (Figure 3B). In summary, these results clearly demonstrate that TSA, besides its direct cytotoxic effect on GBM cells after long-term exposure (Figure 1B), and its synergistic effect on cell death, induced by APO2L/TRAIL (Figure 1D), also promotes GBM cell lysis by NK cells via up-regulation of NKG2D ligands on the cell surface.
TSA treatment of GBM cells reduces tumor growth in nude mice. To further substantiate our findings, we studied the stimulatory capacity of TSA on NK cell effector function in a sub-cutaneous xenograft GBM model in NMRI nude mice. For this, we implanted either mock- or TSA-treated LNT-229 cells sub-cutaneously and measured tumor growth over time. In order to exclude that putative tumor growth reduction was induced by TSA-mediated cell death, we measured cell viability prior to implantation and implanted the same numbers of viable cells in the mock- and TSA-treated groups. In parallel, growth and survival of cells used for implantation was measured over a period of ten days. We found no significant differences between mock- and TSA-treated cells (Figure 4A, inset). Although no differences in proliferation between mock- and TSA-treated LNT-229 cells were observed under these conditions, growth of TSA-treated LNT-229 tumors was significantly delayed. Mock-treated LNT-229 tumors became primarily visible 14 days post-tumor cell implantation and had reached a size of 480±120 mm2 at the end of the experiment. In contrast, mice harboring TSA-treated cells developed smaller tumors (180±100 mm2), and tumors became visible not earlier than day 20 after implantation (Figure 4A, and data not shown). Intra-peritoneal injection of the antibody to asialo-GM1-prior to tumor cell implantation led to a more than 10-fold reduction of NK cells in the mice (Figure 4B) and abrogated the tumor growth-inhibitory effect of TSA. At the end of the experiment, tumors developed from TSA-treated cells in mice which had NK cells (Figure. 4C) reached a similar size to those derived from TSA-treated cells in untreated mice (Figure 4A), whereas in NK cell-depleted mice, tumors developed from TSA-treated cells (Figure 4C) reached approximately the same size as those derived from untreated LNT-229 cells (Figure 4A) indicating that NK cell activity is essential for the observed TSA-mediated reduction in tumor size.
Discussion
In this study, we investigated the potential use of HDACi as a putative strategy for GBM treatment. We demonstrate that TSA is effective as an HDAC blocking agent, inducing protein acetylation and cell death in a fraction of GBM cells and acting synergistically with APO2L/TRAIL. We have shown that even co-treatment of GBM cells with TSA and APO2L/TRAIL does not result in complete eradication of all cells, leaving a cell population resistant to TSA- and/or to APO2L/TRAIL-induced cell death. Interestingly, p53-deficient LN-308 GBM cells were more resistant to long-term TSA treatment than were p53 wild-type LNT-229 cells. Wang et al. recently described that TSA, via induction of DNA damage, transactivates the p53/p21 pathway and, in consequence, induces cell-cycle arrest or apoptosis if damaged DNA cannot be repaired (39). TSA-mediated apoptosis induction might therefore work better in GBM with p53 wild-type status than in those with mutated p53, the latter being more resistant to HDACi-induced cytotoxicity.
Trichostatin A (TSA)-mediated alterations on mRNA and surface protein expression of natural-killer group-2, member-D (NKG2D) ligands. A: TSA-treated LNT-229 cells (2 μM TSA, 24 h) were analyzed for MICA, MICB, ULBP1 and ULBP2 mRNA expressions by real time quantitative-polymerase chain reaction (n=4, SEM). Cells treated with interferon gamma (IFN-γ; 500 U/ml, 48 h) were used as a positive control. B: Cells were cultured for 24 h with either 1 μM or 2 μM TSA, or with medium alone (Co). Surface expression of NKG2D ligands was analyzed by flow cytometry. C: TSA-mediated up-regulation of surface expression of NKG2D ligand proteins assessed by flow cytometry was determined by calculating the specific fluorescence index (SFI) values (TSA: 2 μM, 24 h; n=2, SEM; *p<0.05).
Many features contribute to the lethal nature of GBM. GBM cells are resistant to apoptosis and, more importantly, they can directly induce immune cell death or inactivation, mainly by secreting the immunosuppressive cytokine TGF-β (40). Furthermore, GBM cells can evade immune cell recognition by down-regulation of surface proteins necessary for cytotoxic T-cell-mediated (e.g. MHC I) or NK cell-mediated (e.g. NKG2D ligands) immune response (26, 41). This suggests that the levels of immune-activating proteins expressed on GBM cells are often too low to trigger antitumor immunity. It has been well-established that NK cell activity is strictly regulated by a fine balance of activating and inhibitory signals transmitted by cell surface receptors (21). Our study suggests that HDAC inhibitors, besides their direct cytotoxic effect, also enhance the immunorecognition of GBM cells by inducing the expression of ligands necessary for their interaction with NK cells. TSA-induced expression of the NKG2D ligands MICA and ULBP2 resulted in enhanced NK cell-dependent GBM cell lysis (Figure 3A). Whether up-regulation of NKG2D ligand expression in TSA-treated GBM cells is the sole result of an increase in mRNA of NKG2D ligands, or whether this effect is also due to HDACi-mediated reduction of secretion of MMPs (9), proteins known to cleave MICA and release it as a soluble form (42, 43), needs further investigation. The immunostimulatory effect of TSA was shown to be dependent on the presence of NK cells, since pre-incubation of effector cells with a neutralizing antibody targeting NKG2D significantly reduced GBM cell lysis. Interestingly, lysis of GBM cells by NK cells was not completely abrogated by the addition of the blocking NKG2D antibody. This may be explained by a sub-optimal concentration of blocking antibody used in our experiments. Alternatively, a different receptor/ligand system like that of the DNAX accessory molecule (DNAM)-1 on NK cells with the DNAM ligands CD112 and CD115 (44) on GBM cells, could also play a role in NK cell-mediated GBM cell lysis.
Enhanced natural killer (NK) cell-mediated glioblastoma (GBM) cell lysis after trichostatin A (TSA) treatment. A: LNT-229 cells were incubated with 2 μM TSA or control medium for 24 h prior to the cytotoxicity assay, using lymphokine-activated (LAK) cells from healthy donors, as described in Materials and Methods. B: CD56+/CD3− NK cells were purified from bulk LAK cell cultures by negative selection and were pre-incubated with 10 μg of a natural-killer group-2, member-D (NKG2D) blocking antibody or control IgG before co-culture with untreated or TSA-treated LNT-229 target cells. A,B: Cellular toxicity assays were carried out up to six times, one representative experiment is shown. E: Effector cells; T: LNT-229-Luc target cells; bars=SD; *p<0,05, **p<0.01, t-test.
In our mouse GBM model, we substantiated our in vitro data by demonstrating that TSA delays tumor growth of sub-cutaneously growing LNT-229 xenografts in vivo (Figure 4). This effect was independent of TSA-mediated cell death and proliferation inhibition, since a parallel growth assay showed no significant changes in cell density between mock- and TSA-treated GBM cells for more than ten days, indicating that an effective antitumor immune response is responsible for mitigation of tumor growth. Since NMRI nude mice lack functional T-cells, we chose this model to study the stimulatory effect of TSA on effector cells of the innate immune system. Indeed, the antitumor immune effect of TSA was dependent on NK cell activity, and was at least in part dependent on the expression of NKG2D on the effector cells, since inhibition of receptor ligand interaction using a NKG2D blocking antibody significantly reduced NK cell cytotoxicity in vitro (Figure 3B). Furthermore, depletion of NK cells delayed tumor growth in mice harboring TSA-treated GBM cells in vivo (Figure 4).
The capacity of TSA to inhibit HDAC and to stimulate immunity is of great interest, since HDAC inhibitors are currently used as therapeutic agents in patients with cancer, mainly because of their apoptosis-inducing effect in cancer cells. In 2006, the HDACi vorinostat (SAHA/Zolinza), which is less toxic than TSA, was given FDA approval for the treatment of cutaneous T-cell lymphoma (45). In GBM, it has been reported that inhibition of HDAC promotes cell differentiation (46), mediates proliferation (13), induces apoptosis (10) and finally, sensitizes GBM cells to radiation-induced cell death (47, 48). Here we describe a mechanism of HDACi in promoting innate immune responses against glioma. We demonstrate that TSA mediates differential regulation of MICA and ULBP2 expression rendering the surviving GBM cell fraction more susceptible to NK cell-mediated lysis. In this proof-of-principle study we have demonstrated that TSA improves the immunorecognition of glioma cells by inducing MICA and ULBP2 expression. This finding brings an interesting new dimension to the usage of HDAC inhibitors as anticancer drugs, particularly since expression of MICA and ULBP2 is induced in the surviving fraction of GBM cells. Therefore, our results suggest that HDAC inhibition represents a promising strategy to prime GBM cells in vivo for innate immune cell-mediated killing.
Trichostatin A (TSA)-mediated up-regulation of expression of natural-killer group-2, member-D (NKG2D) ligands delays the growth of subcutaneously growing human glioblastoma (GBM) xenografts in nude mice. A: The growth of s.c. LNT-229 tumors was monitored two-times a week (n=7 mice; bars=SD). Inset: Growth profile of control and TSA-treated LNT-229 cells used for tumor cell implantation into nude mice as determined by crystal violet staining. B: Abrogation of CD49b-expressing cells in mice treated with antibody to asialo-GM1 (n=2, one representative experiment is shown). C: Box-whisker-plot of tumor size. The mice were depleted of natural killer (NK) cells using asialo-GM1-antibody or were injected with IgG isotype antibody prior to and after tumor cell implantation, as described in Materials and Methods (n=6 mice; *p=0.014).
- Received February 26, 2013.
- Revision received March 13, 2013.
- Accepted March 14, 2013.
- Copyright© 2013 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
