Recent reports have described reduced immunological responsiveness and stimulatory capacity among monocytes/microglia that infiltrate malignant human gliomas. Herein, we demonstrate that culture of ex vivo human monocytes or primary human microglia with tumor cells isolated from glioblastoma multiforme (GBM) specimens renders them tolerogenic, capable of suppressing the function of ex vivo monocytes in the absence of tumor cells or their soluble factors. We demonstrate that the tolerance induced in monocytes/microglia by GBM tumor cells is not associated with interference with the signaling cascade associated with TLR- or CD40-induced monocyte activation. Rather, these tumor cells appear to up-regulate pathways that antagonize positive signaling pathways, including but not limited to STAT3 and STAT5. Finally, we demonstrate that the tolerogenic properties of GBM tumor cells amplify properties inherent to nontransformed astrocytes. Future studies that identify all of the molecular mechanisms by which astrocytes and malignant gliomas suppress monocyte/microglial function will have dual therapeutic benefits: suppressing these pathways may benefit patients with astrocytic tumors, while enhancing them may benefit patients with autoimmune processes within the CNS, such as multiple sclerosis.
The historical notion of the brain as an immune privileged organ (1) has experienced a change in paradigm over the last several decades (2) in light of evidence for immune responses against self and foreign Ags within the CNS (3). Microglia are the resident mononuclear phagocytes of the CNS and, along with infiltrating dendritic cells (DCs),3 play a predominant role in Ag processing and presentation in the CNS. Resting microglia become activated in response to a wide range of injuries/stimuli (4) acquiring phagocytic ability (5). Ontogenetically, they are related to cells of the mononuclear phagocyte lineage (6). During fetal development, monocyte-like cells infiltrate the CNS and become the parenchymal microglia, participating in Ag presentation at the parenchymal level. In contrast, there is a population of perivascular microglia, constantly repopulated in adult life by monocytes from the periphery, which participates in Ag presentation at the blood-brain barrier level (7).
Glioblastoma multiforme (GBM) is the World Health Organization grade IV astrocytic tumor, the most common and aggressive primary brain malignancy (8), with an annual incidence of 3 per 100,000 individuals (9). It usually presents de novo (primary GBM), without evidence of a precursor lesion. Less commonly, GBM arises in association with progressive genetic alterations after the diagnosis of a lower grade astrocytoma (secondary GBM) (10). Despite aggressive treatment, the median survival of these patients is only 1 year (11).
We and others have observed that microglia can compromise up to one third of a GBM tumor (12). We hypothesized that an impairment in microglial function may account, at least in part, for the lack of productive immunity against GBMs, despite the fact that numerous immune cells infiltrate these tumors (13, 14). Indeed, several studies have reported that GBM-associated monocytes/microglia are hyporesponsive, with a reduced stimulatory capacity (15, 16). However, the mechanism by which this state is induced has not been defined. We demonstrate herein that GBM tumor cells inhibit the ability of monocytes/microglia to produce the proinflammatory cytokine TNF-α in response to diverse stimuli, whereas their presence induces IL-10 secretion, a phenomenon that correlates with up-regulation of STAT3. Using multiparameter flow cytometric cell sorting of ex vivo tumor specimens, we have confirmed the up-regulation of both IL-10 and STAT3 in microglia sorted from GBM tumor specimens but not from meningiomas, tumors that lack significant numbers of infiltrating astrocytes. In addition, we observe that HLA class II and CD80 costimulatory molecule expression in monocytes are suppressed in the presence of GBM tumor cells. Finally, monocytes cultured with GBM tumor cells are rendered tolerogenic, not simply unable to activate CD4+ T cells, rather, able to inhibit subsequent activation of freshly isolated ex vivo monocytes. Interestingly, most of these features are shared, although to a lesser degree, by nontransformed human astrocytes (NHA), suggesting that human astrocytes create a regulatory network that prevents monocyte/microglial activation, an ability that is exacerbated in highly malignant astrocytic human tumors.
Materials and Methods
In compliance with appropriate institutional review board protocols, PBMCs were isolated from leukopaks or freshly isolated blood from healthy subjects by density gradient centrifugation. Monocytes and naive (CD45RO-depleted) CD4+ T cells were isolated from PBMCs by negative selection using immunomagnetic beads (Miltenyi Biotec). Unless specified otherwise, cells were resuspended in serum-free X-Vivo 15 medium (Cambrex/Lonza). Ex vivo tumor specimens were obtained in compliance with appropriate institutional review board protocols present at the appropriate institutions and processed to a single-cell suspension as we have previously described (17). Briefly, tissue samples were obtained from primary cytoreductive operations washed with cold PBS, minced, and enzymatically digested with 400 g/ml DNase I and 1 ng/ml collagenase IV (Worthington Biochemical). The resulting single-cell suspension was filtered and pelleted by centrifugation. Aliquots were stored at −80°C until assaying. Purified monocyte/microglia populations were obtained by FACS based on expression of both CD11b and CD11c as we have previously described (18). Primary GBM tumor cell lines RCA and 212 (uniformly glial fibrillary acidic protein positive) were obtained by culture and expansion of ex vivo tumor specimens in cell culture medium; cells were used after 10–15 passages. An extensively passaged GBM cell line, T98G, was purchased from American Type Culture Collection. Adult NHA were purchased from two commercial sources (Cambrex/Lonza; All Cells) and cultured in Astrocytes Basal Media supplemented with astrocyte growth medium SingleQuots (Lonza); they were not used beyond passage 3. More than 80% of NHA were glial fibrillary acidic protein positive. Primary human microglia were purchased from two commercial sources (Clonexpress and ScienCell) and cultured in microglia cell culture medium. These primary microglia lacked astrocytic and neuronal cell markers and expressed CD11b, CD68, and CD14.
Ex vivo monocytes (2 × 105/well) or primary human microglia (1 × 105/well) were cultured alone or with NHA or GBM tumor cells (ratios ranging from 2:1 to 40:1 monocytes:GBM/NHA) in triplicate in round-bottom 96-well plates (Corning). Stimuli used were LPS (TLR4 ligand) at 1 μg/ml and flagellin (TLR5 ligand) and sspoly(U) (TLR8 ligand) at 10 μg/ml (all available from InvivoGen). CD40L stimulation of monocytes involved addition of soluble CD40L (R&D Systems) at 5 μg/ml, followed 4 h later by addition of anti-polyhistidine cross-linking Ab (R&D Systems) at 20 μg/ml. All cultures were performed in X-VIVO 15 medium. After 48 h at 37°C, assay supernatants were collected for cytokine analysis. Paired capture and detection mAbs against TNF-α and IL-10 (BD Pharmingen) were used to detect cytokines by ELISA as we have previously described (19). Note that in all monocyte:GBM coculture assays, monocytes and GBM samples were unmatched.
Ex vivo monocytes (2 × 105/well, lower chamber) and GBM tumor cells (1 × 105/well, upper chamber) were cultured with 0.2-μm Anapore membrane tissue culture inserts (Nalge/Nunc) in triplicate in round-bottom 96-well plates (Corning) in the presence of LPS (1 μg/ml). After 48 h at 37°C, assay supernatants were collected for cytokine analysis.
mRNA expression analysis
Ex vivo monocytes (2 × 106/ml/tube) were cultured in the presence of LPS (1 μg/ml) alone or with either NHA or GBM tumor cells (1 × 106/ml/tube) at 37°C in 5 ml of polypropylene round-bottom tubes (Falcon; BD Biosciences). After 4 h, cells were stained with CD11b-PE mAb (BD Pharmingen) and a FACSAria (BD Biosciences) was used to isolate populations of monocytes (CD11b+) at >95% purity. NHA and GBM tumor cells were easily discriminated as negative for CD11b as well as by their larger size and greater granularity (forward scatter and side scatter) compared with CD11b+ monocytes. As controls, ex vivo monocytes, NHA, and GBM tumor cells were cultured alone for the same time points. Total RNA from sorted populations and from single culture conditions was obtained using RNeasy Mini columns (Qiagen) following the manufacturer’s instructions. Samples were analyzed by quantitative RT-PCR (TaqMan) for IL-10 and STAT3 transcripts (primer and probes were from Applied Biosystems). RNA levels reported are relative to β2-microglobulin, normalized using the equation 1/2ΔCt.
Flow cytometric analysis
Ex vivo monocytes (2 × 106/ml/tube) were cultured with or without NHA or GBM tumor cells (1 × 106/ml/tube) at 37°C in 5-ml polypropylene round-bottom tubes (BD Biosciences) in the presence of LPS (1 μg/ml). After 24 h, cells were stained with PE mouse CD11b-allophycocyanin, FITC mouse anti-HLA-DR, PE mouse anti-CD80, and FITC mouse anti-CD33 mAbs (BD Pharmingen). Data were collected on a FACSCalibur flow cytometer (BD Biosciences) and analyzed with FlowJo (Tree Star).
Monocyte: T cells or monocyte coculture assays
Ex vivo monocytes (2 × 106/ml/tube) were cultured with or without GBM tumor cells or NHA (1 × 106/ml/tube) in the presence of LPS in 5-ml polypropylene round-bottom tubes (BD Biosciences) at 37°C for 72 h, after which they were reisolated by positive selection using immunomagnetic beads conjugated with mouse anti-human CD14 (Miltenyi Biotec). Normalized numbers (2.5 × 104/well) of treated (LPS, GBM/LPS, or NHA/LPS) monocytes were incubated with either ex vivo CD4+ T cells (5 × 104/well) in triplicate in round-bottom 96-well plates (Corning) in the presence of 1 μg/ml soluble anti-CD3 (OKT3) mAb or with freshly isolated ex vivo monocytes (1 × 105/well) in triplicate in round-bottom 96-well plates in the presence of 1 μg/ml LPS. After 48 h at 37°C, assay supernatants were collected for cytokine analysis, and T cell cocultures were pulsed with 1 μCi/well [3H]thymidine for the last 24 h. Cells were harvested and [3H]thymidine uptake was determined.
Unstimulated monocytes, LPS-stimulated monocytes, or cocultures of LPS-stimulated monocytes and tumor cells were incubated at 37°C for 0.5, 1, or 4 h. Analysis of phosphorylation was performed as previously described (20). Fixation of cells and preservation of phosphorylation status was achieved by adding prewarmed 1× BD Biosciences Cytofix and incubating in a 37°C water bath. Cells were permeabilized by incubating cells in BD Biosciences Perm Buffer III on ice for 30 min, at which point they were washed twice with 2% FBS/PBS and stained using BD Biosciences Staining Buffer. To identify monocytes, cells were stained using allophycocyanin mouse anti-CD11b; for phospho-site-specific analysis, the following Abs were used: Alexa Fluor 488 mouse anti-NF-κB p65 (pS529), PE mouse anti-STAT3 (pY705), Alexa Fluor mouse anti-STAT5 (pY694), and PE mouse anti-ERK1/2 (pT202/pY204; all from BD Biosciences). Data were collected on a FACSCalibur flow cytometer (BD Biosciences) and analyzed with FlowJo (Tree Star).
SD for experimental values was calculated using Prism 4.0 software (GraphPad).
GBM tumor cells potently modulate the ability of human monocytes/microglia to respond to a variety of stimuli
Previous studies have described altered cytokine secretion among ex vivo glioma-associated microglia by intracellular staining (15, 16) or immunohistochemistry (21, 22); all of these studies involved ex vivo microglia that were analyzed for cytokine expression without activation. We hypothesized that GBM tumor cells were directly responsible for the hyporesponsiveness of glioma-associated microglia and their altered cytokine secretion profile. To directly evaluate the functional effect of ex vivo GBM tumor cells on monocyte/microglial activation, we cocultured allogeneic ex vivo peripheral monocytes and primary GBM tumor cells in the presence of different monocyte-activating stimuli. These included pathogen (TLR4, TLR5, and TLR8)- and T cell-derived (CD40) stimuli. In the absence of tumor cells, all of the stimuli elicited secretion of large amounts of the proinflammatory cytokine TNF-α by the monocytes. However, the presence of the GBM tumor cells almost completely inhibited TNF-α secretion (Fig. 1,A). Concomitantly, the presence of the GBM tumor cells induced secretion of the anti-inflammatory cytokine IL-10 (Fig. 1 B). Monocytes without activating stimuli and GBM tumor cells cultured alone (with or without stimuli) did not produce any cytokines (data not shown).
Potency and complexity of immunomodulation by GBM tumor cells and normal human astrocytes
We next determined whether the immunoinhibitory property of the GBM tumor cells was tumor-specific or whether it was an inherent property of human astrocytes. We thus stimulated monocytes with LPS in the presence or absence of NHA. To assess the potency of this modulatory effect, we titrated the number of GBM tumor cells or NHA in the coculture, observing a 63% reduction of TNF-α production at a 40:1 monocyte:tumor cell ratio (Fig. 1, C and D). Although less potently, NHA behaved similarly to GBM tumor cells in terms of their ability to suppress monocyte activation (44% reduction at a 40:1 monocyte:NHA ratio). The ability of GBM tumor cells to suppress TNF-α secretion was consistently more pronounced than their ability to induce IL-10 secretion (Fig. 1,D); NHA, while able to suppress TNF-α secretion in coculture with stimulated monocytes, induced very little IL-10 secretion. Given that extensive culturing of GBM tumor cell lines in vitro may alter their immunological properties (17), we compared the suppressive activity of two primary GBM cell lines established from ex vivo specimens as well as an extensively passaged, commercially available GBM cell line. All GBM cell lines suppressed monocyte TNF-α secretion in response to LPS stimulation to comparable extents at a variety of ratios (Fig. 1 E). Assays performed with a primary, pilocytic (low-grade) glioma cell line similarly suppressed monocyte responsiveness to LPS stimulation (data not shown).
We similarly stimulated monocytes with LPS in the presence or absence of GBM tumor cells or NHA for 24 h, after which we stained with mAbs to CD11b (monocyte gate) as well as to HLA-DR and CD80. CD80 induction was suppressed in the presence of both GBM tumor cells and NHA (inhibition of mean fluorescence intensity (MFI) of 53 ± 4% and 62 ± 5% for coculture with GBMs and NHA, respectively; p < 0.05), suggesting an intrinsic capacity of human astrocytes to impair CD80 up-regulation in monocytes upon activation. Similarly, HLA-DR expression was down-modulated in the presence of GBM tumor cells and to a lesser extent NHA (inhibition of MFI of 46 ± 7% and 30 ± 16%, respectively; p < 0.05).
To determine whether soluble and/or surface molecules were responsible for the suppressive effects of the tumor cells on monocyte activation, we performed Transwell experiments. Results of these studies suggested that the inhibition of TNF-α is both soluble and cell-contact dependent, whereas the induction of IL-10 appears to be entirely cell-contact dependent (Fig. 2, A and B). Neutralization of IL-10 in coculture experiments did not prevent GBM tumor cell inhibition of TNF-α secretion from stimulated monocytes, demonstrating that induction of IL-10 was not the mechanism responsible for inhibition of TNF-α secretion. This conclusion is supported by the observation that NHA cells could readily inhibit TNF-α secretion, but induced very little up-regulation of IL-10 (Fig. 1, C and D). Collectively, these data suggest that GBM tumor cells exploit an ability of human astrocytes to inhibit a variety of mechanisms that promote a productive immune response within the CNS.
During fetal development, monocyte-like cells infiltrate the CNS and become parenchymal microglia. Thus, monocytes are frequently used to model the response of human microglia. Nevertheless, to formally demonstrate that GBM tumor cells could modulate the activity of human microglia, we performed coculture experiments with primary human microglial cells. Consistent with results found with peripheral monocytes, we observed significant TNF-α suppression and a tendency to up-regulate IL-10 secretion in primary human microglia in the presence of GBM tumor cells (Fig. 3).
IL-10 induction in cocultures of monocytes and GBM tumor cells is associated with STAT3 up-regulation
To elucidate which of the two populations of cells was responsible for the production of IL-10 in our system, we cocultured ex vivo human monocytes with GBM tumor cells or NHA, after which we reisolated the two populations by FACS, based on the expression of CD11b by the monocytes and on the size and granularity (forward scatter and side scatter) of the tumor cells. Total RNA was extracted from sorted populations for expression analysis by quantitative PCR. We observed induction of IL-10 in both monocytes and GBM tumor cells after coculture (Fig. 4,A). Since STAT3 is associated with IL-10 induction in tumor-associated monocytes isolated from several peripheral tumors (23), we also assessed RNA levels of STAT3. Culturing monocytes with GBM tumor cells consistently induced STAT3, which correlated with up-regulation of IL-10 (Fig. 4 A). Of note, we also consistently observed that coculture of monocytes with GBM tumor cells induced STAT3 in tumors cells, indicating that bidirectional signaling was occurring between monocytes and tumor cells.
Microglia directly isolated from high-grade glioma specimens express high levels of IL-10 and STAT3 RNA
To determine how well our in vitro model of GBM-mediated suppression of monocytes/microglia reflected processes in vivo, we used FACS to isolate populations of infiltrating monocytes/microglia from ex vivo human CNS tumor specimens (18). We sorted CD11b+CD11c+ monocytes/microglia from malignant human gliomas and meningiomas. We hypothesized that monocytes/microglia present in malignant gliomas would be affected by the tumor cells, evidenced by induction of STAT3 and IL-10, whereas monocytes present in meningiomas, not located in the parenchyma and lacking significant numbers of astrocytes, would have comparably little STAT3 and IL-10 induction. Consistent with our hypothesis, we readily observed RNA levels of both STAT3 and IL-10 in high-grade gliomas, with little evidence of either transcript in meningiomas (Fig. 4 B).
GBM tumor cells do not abrogate LPS-induced signaling pathways in monocytes
We reasoned that GBM tumor cells suppress TNF-α secretion by either blocking proximal LPS-induced signaling pathways necessary for TNF-α secretion or by inducing additional molecular pathways that antagonize the transcription of TNF-α. To distinguish between these possibilities, we used a flow cytometric-based assay with which we measured the activation status of molecules documented to be involved in LPS-induced TNF-α transcription in monocytes and microglia (24). We first confirmed using this technique that whereas LPS stimulation of monocytes induced TNF-α secretion measured by intracytoplasmic staining, no TNF-α staining was detected when LPS was used to stimulate monocytes in the presence of GBM tumor cells (Fig. 5,A). Nevertheless, LPS stimulation resulted in stronger and more rapid activation (phosphorylation) of both ERK1/2 and NF-κB when in the presence of GBM tumor cells. Among two comparable, independent experiments, the following mean fold induction of pNF-κB, pERK, pSTAT3, and pSTAT5 after LPS stimulation for 30 min in the absence vs presence of GBM tumor cells were observed: pNF-κB (2.01 vs 2.45), pERK (1.43 vs 2.89), pSTAT3 (1.03 vs 1.75), and pSTAT5 (1.10 vs 2.12). Thus, there was no evidence that the presence of GBM tumor cells inhibited proximal signaling events associated with TNF-α secretion. The presence of the tumor cells also led to stronger and more rapid activation of STAT3 and particularly STAT5 (Fig. 5 B), both of which may contribute to the ability of the tumor cells to suppress TNF-α secretion. That coculture of LPS-stimulated monocytes with GBM tumor cells suppressed TNF-α secretion when measured using intracytoplasmic staining further indicated that expression of TNF receptors on GBM tumor cells, acting as a sink for secreted TNF-α, did not explain tumor cell suppression of TNF-α secretion.
GBM- and NHA-treated monocytes suppress T cell and monocyte activation
We next investigated how GBM- and NHA-treated monocytes affect T cell activation. To address this question, we cocultured monocytes either alone or in the presence of GBM tumor cells or NHA (all three conditions in the presence of LPS). After 72 h of culture, we reisolated the monocytes and cocultured them with ex vivo CD4+ T cells obtained from an allogeneic donor in the presence of soluble anti-CD3 for a period of 72 h. We found that GBM- and NHA-treated monocytes were unable to activate CD4+ T cells, as evidenced by dramatically reduced proliferative responses following anti-CD3 stimulation, compared with the response of CD4+ T cells stimulated with anti-CD3 in the presence of monocytes initially treated with LPS (Fig. 6,A). The effector function of T cells was also compromised, as IFN-γ secretion was suppressed when T cells were cultured with GBM- and NHA-treated monocytes (Fig. 6 B). There was no evidence for induction of immunomodulatory cytokines such as IL-10 or TGF-β (data not shown).
We also evaluated whether GBM- and NHA-treated monocytes could affect activation of allogeneic, freshly isolated, ex vivo monocytes. To do so, we cocultured monocytes in the absence or presence of GBM tumor cells or NHA in the presence of LPS. After 72 h, we reisolated the monocytes and cultured them with ex vivo monocytes isolated by negative selection and stimulated them with CD40L for 48 h. First, we found that monocytes previously cocultured with GBM tumor cells or NHA in the presence of LPS do not reacquire the ability to respond to restimulation (even if this “secondary” stimulus is derived from a different signal than the “primary” stimulus) (data not shown). This suggested the induction of a long-lasting regulatory phenotype. Indeed, we found that monocytes that were previously cocultured with GBM tumor cells or NHA suppress the ability of ex vivo monocytes to respond to stimulation, as evidenced by reduced TNF-α production in coculture conditions after CD40L stimulation, compared with monocytes cocultured with monocytes initially treated with LPS alone (Fig. 6 C).
Using ex vivo and primary human cells, we have investigated how human malignant gliomas, as well as normal human astrocytes, modulate monocytic/microglial activation. The role of microglia in initiating CNS inflammation has been extensively documented (25); furthermore, experimental paralysis of microglia can inhibit the development and maintenance of inflammatory CNS lesions (26). The observation that microglia can comprise a substantial part of a GBM tumor mass (12) led us to hypothesize that some degree of functional impairment may explain the weak host immunity directed against GBMs. The majority of studies that investigate Ag presentation in the context of peripheral human tumors focus on DCs; however, due to the predominance of microglial cells relative to DCs within GBMs (D. E. Anderson, unpublished observations), we decided to look specifically at monocytic/microglial activation.
To test our hypothesis, we first investigated the effect of GBM tumor cells on ex vivo human monocyte activation. We found that the presence of GBM tumor cells down-regulated the production of the proinflammatory cytokine TNF-α by monocytes after LPS stimulation. A concomitant production of the anti-inflammatory cytokine IL-10 was induced. These phenomena were reproduced with both T cell (CD40L, a member of the human TNF superfamily)- and pathogen-derived (TLR5 and TLR8 ligands) stimuli, demonstrating that GBM tumor cells affect monocyte activation regardless of the nature of the stimulus. We further demonstrated the remarkable potency of the tumor-mediated suppression: a reduction of TNF-α secretion greater than 50% is observed at ratios as low as 1 GBM tumor cell to 40 monocytes. Comparable results were obtained when we cultured GBM tumor cells with primary human microglia. These data provide a mechanism to explain recent reports that proinflammatory cytokine production and tumor cytotoxic activity are reduced among microglia isolated from ex vivo GBM tumor specimens (15, 16). The ability of GBM tumor cells to induce IL-10 in monocytes/microglia in our in vitro model is consistent with in situ studies that have localized IL-10 to monocytes/microglia present within GBM tumor microenvironments (21, 22). Nevertheless, GBM-induced IL-10 in stimulated monocytes/microglia is not responsible for the suppression of TNF-α secretion.
Besides TNF-α production, we were interested in the modulation of other parameters of monocytic activation. Given the nature of immune infiltrates in the microenvironment of GBMs, which includes monocytes, microglia, and CD4+ and CD8+ T cells, we investigated HLA class II and CD80 surface expression on monocytes: both HLA class II and CD80 are not only activation markers, but also influence the ability of APCs to promote T cell activation and elicit an effective cellular immune response. The up-regulation of CD80, a hallmark feature of LPS-stimulated monocytes, did not occur on monocytes cocultured with GBM tumor cells or NHA. These monocytes also exhibited down-modulated HLA-DR expression. Our in vitro observations are consistent with a study performed in rats (27) that demonstrated a down-regulation of surface expression of CD80 and MHC class II on microglia present in glioma-bearing rodents and cell surface analysis of ex vivo microglia isolated from human GBMs that showed reduced expression of costimulatory molecules, including CD80 (15).
We reasoned that the inhibitory effect of GBM tumor cells may be intrinsic to human astrocytes. Indeed, in the context of Alzheimer’s disease, human astrocytes can attenuate TNF-α production by macrophages upon amyloid β stimulation (28). We confirmed that human astrocytes have the intrinsic capacity to abrogate TNF-α secretion after LPS stimulation. However, transformed astrocytes (GBM tumor cells) have a more potent inhibitory effect on monocyte/microglial activation than do primary astrocytes. Whether GBM tumor cells up-regulate normal inhibitory molecules/pathways utilized by nontransformed astrocytes, acquire new inhibitory mechanisms, or become more immunosuppressive for both reasons remain unclear. In this regard, numerous pathways and mechanisms have been shown to play important roles in GBM tumor immune evasion (29). In our system, blocking experiments targeting IL-10, COX2, TGF-β, PGE2, IL-6, and CD200 did not reverse the immunomodulatory phenotype induced in GBM-treated monocytes (data not shown).
Interestingly, we found that induction of IL-10 was associated with up-regulation of STAT3 both in monocytes and in GBM tumor cells after coculture, consistent with the immunosuppressive role of STAT3 activation in many solid tumors other than human malignant gliomas (30). No consistent up-regulation of STAT3 was observed when monocytes and NHA were cocultured (data not shown). Our in vitro model appears to reflect what occurs in situ, as analysis of monocytes/microglia isolated by FACS from ex vivo GBM tumor specimens confirmed transcripts for both IL-10 and STAT3 that were not present in monocytes sorted from meningiomas which lack astrocytes. It is known that LPS induces expression and phosphorylation of STAT3; following its binding to the promoter of the human IL-10 gene, STAT3 regulates IL-10 expression (31). Moreover, it has recently been shown that IL-10 induces IL-10 secretion in human monocytes in an autocrine fashion via STAT3 activation (32). STAT3 is also involved in oncogenic pathways; its role in malignant gliomas survival has recently been addressed by demonstrating that inhibition of the pathway resulted in apoptosis of human glioma cells (33). Additionally, STAT3 is an important activator of many genes that are crucial for immunosuppression; STAT3-deficient animals have enhanced inflammatory activity, developing chronic enterocolitis (34) and enhanced antitumor activity (23). Accordingly, pharmacologic inhibition of STAT3 activity may be a promising therapeutic strategy for the treatment of patients with malignant gliomas due to its potent immune adjuvant responses (35). Nevertheless, we also observed potent induction of STAT5, and recent reports have identified additional molecules capable of repressing NF-κB-mediated transcriptional activity that is necessary for TNF-α production (36, 37). Whether there is a single, dominant molecule, or instead a complex network of inhibitory pathways, responsible for the GBM-mediated suppression of monocyte/microglial function remains an important, unanswered question. Indeed, we believe a comprehensive, array-based approach will be necessary to fully appreciate the extent of pathways/mechanisms by which GBMs suppress monocyte/microglial activation.
We were intrigued by the potent ability of GBM tumor cells to suppress monocyte/microglia activation at frequencies as low as 1 tumor cell per 40 monocytes. We demonstrate that monocytes cultured with either GBM tumor cells or NHA in the presence of LPS were unable to stimulate CD4+ T cell proliferation or cytokine secretion, as has previously been shown (15, 35).
In exploring this phenomenon further, we were able to demonstrate that coculturing monocytes with human astrocytes or GBM tumor cells confers a tolerogenic state, as they prevent subsequent activation of freshly isolated monocytes, therefore creating a self-perpetuating loop of immunosuppression. Thus, monocytes rendered tolerogenic by GBM tumor cells suppress the activation of both T cell and APC populations. Deciphering the molecular mechanism(s) by which GBM tumors cells suppress monocyte/microglia activation could be used in concert with current GBM-specific tumor vaccines to improve efficacy.
We thank Francisco Quintana for helpful comments on this manuscript and Deneen Kozoriz for assistance with cell sorting.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by an American Cancer Society Research Scholar Grant (to D.E.A.).
Abbreviations used in this paper: DC, dendritic cell; GBM, glioblastoma multiforme; NHA, nontransformed human astrocyte; MFI, mean fluorescence intensity.