Pharmacological targeting of the tumor–immune symbiosis in glioblastoma

Targeting the GBM-immune symbiosis

GBM is the most common and fastest growing primary brain tumor in adults [1,2]. Despite the aggressive standard-of-care (SOC) treatment that includes maximal surgical resection followed by radiation and/or chemotherapy with temozolomide (TMZ), the median survival for GBM is only approximately 14–16 months [2]. The low therapeutic efficiency of the SOC treatment relates to the challenges that complete resection of GBM tumors is impossible and the blood–brain barrier (BBB) (see Glossary) can hinder the systemic therapy [3-5]. Genetic profiling of patient tumors has led to identification of several core signaling pathways in GBM cells, thus motivating clinical trials for testing potential targeted therapies. However, all these efforts have failed to improve GBM patient outcomes, probably due to GBM cell genetic instability and heterogeneity [6]. Conversely, noncancerous cells in the GBM tumor microenvironment (TME) are genetically stable. Increasing evidence demonstrates that the TME is critical for supporting GBM progression, and strategies targeting the GBM TME have emerged as a promising therapeutic approach [3,4]. In recent years, studies using advanced technologies, such as single-cell RNA sequencing (scRNA-seq), whole-exome sequencing, and mass cytometry (CyTOF) followed by functional studies, have revealed a dynamic and diverse immune landscape in the GBM TME with respect to different tumor stages and genetic backgrounds [7-9]. These findings highlight a context-dependent tumor–immune symbiotic interaction, which is critical for promoting tumor progression and therapy resistance (e.g., resistance to the SOC treatment and immunotherapies) in GBM [6].

Immunotherapies, including immune checkpoint inhibitor (ICI) therapies, have been shown to improve patient outcomes in multiple cancer types [10,11]. Unfortunately, such ICI therapies only produce modest clinical benefits in GBM, probably due to lack of intratumoral T cell infiltration [2,7,12]. Apart from that, infiltrating myeloid cells, such as glioma-associated macrophages and microglia (GAMs), and myeloid-derived suppressor cells (MDSCs), induce a robust immunosuppressive TME that inhibits the activity and proliferation of cytotoxic T cells, resulting in an even worse immunotherapy response [2,7,13,14]. T cell-based immunotherapy can reshape the composition and status of myeloid cells in the GBM TME. For example, immunotherapy (e.g., the combination of ICI therapy and immunovirotherapy) in GBM mouse models results in a significant increase of immunostimulatory macrophages [15]. Together, these findings support a symbiotic interaction between myeloid cells and T cells and imply that this symbiosis may affect the effectiveness of immunotherapy in GBM.

Knowledge of the crosstalk among GBM cells, myeloid cells, and T cells has motivated great efforts to target these symbiotic interactions, with pharmacological tools as the primary focus for translational studies. Here, we review recent advances in pharmacological targeting of the GBM-immune symbiosis and discuss the role and application of such pharmacological tools for improving the effectiveness of immunotherapies in GBM.

Pharmacological targeting of the GBM–GAM crosstalk

GAMs are the most abundant cell population in the GBM TME (accounting for up to 50% of total live cells) and composed of bone marrow-derived macrophages (hereafter referred as macrophages) and brain-resident microglia. GAMs contribute to tumor progression through various mechanisms, including secretion of distinct cytokines, ligands, and other factors (Box 1) [13]. Given the profound role of GAMs in GBM, targeting the GBM–GAM symbiosis appears to be a promising therapeutic strategy [16]. Based on the types of targets and molecular mechanisms underlying this crosstalk, we discuss pharmacological approaches to: (i) target receptors on GAMs and GBM cells; (ii) target GAM and GBM cell-secreted chemokines and factors; and (iii) trap extracellular signaling in the TME (Figure 1 and Table 1).

Pharmacological targeting of the GBM–MDSC crosstalk

MDSCs have emerged as an important type of myeloid cells contributing to GBM immunosuppression [47]. Depending on their phenotypic and morphological features, MDSCs can be subdivided into polymorphonuclear (PMN) and monocytic (M) MDSCs, which may play different roles in cancer progression and drug treatment response [48,49]. Among them, M-MDSCs are enriched in the tumor tissues of male GBM patients, whereas PMN-MDSCs are widely distributed in the circulating system of female GBM patients [50]. The sexual dimorphism spurs researchers to develop gender-specific MDSC-targeting therapeutic strategies in GBM. Preclinical studies demonstrated that the function of M-MDSCs and PMN-MDSCs in male and female GBM patients could be targeted by antiproliferative agents (e.g., fludarabine) and IL-1β blockade (e.g., anti-IL-1β antibodies), respectively [50] (Figure 2). However, further studies are still needed to elucidate molecular mechanisms underlying the sex-specific manner of MDSCs in GBM.

MDSCs respond to GBM cell- and/or GAM-secreted chemokines and cytokines, such as C-C motif chemokine ligand 2 (CCL2), macrophage migration inhibitory factor (MIF), C-X-C motif ligand 1/2 (CXCL1/2), and G-CSF, in the TME [48,51-53]. A growing body of evidence demonstrates that targeting the cytokine-receptor interaction during the GBM–MDSC symbiosis appears to be a primary therapeutic approach (Figure 2). For example, inhibition of this symbiosis by blockade of the MIF-CD74 axis using the CD74 inhibitor Ibudilast in GBM-bearing mice decreases M-MDSC recruitment and GBM cell proliferation, and increases CD8⁺ T cell infiltration [48]. In addition, GBM cell-derived CCL20 and osteoprotegerin (OPG) upregulate CCL2 production in GAMs, which, in turn, increases the infiltration of M-MDSCs through the C-C motif chemokine receptor 2 (CCR2) and CCR4. Pharmacological inhibition of CCR2 and CCR4 (using CCX872 and C021, respectively) significantly extends the survival of GBM-bearing mice by decreasing MDSC infiltration [51,54]. Similarly, inhibition of MDSC infiltration using anti-CXCL1 and anti-CXCL2 neutralizing antibodies exhibits a significant antitumor effect in several different GBM mouse models [52]. Worth noting, in addition to MDSCs, the CCL2-CCR2 axis and CXCL1/2 may also affect the biology of GAMs and other immune cells [55,56]. Therefore, identification of specific MDSC-related chemokines and their potential clinical translation could accelerate personalized drug development and avoid unexpected side effects.

These recruited MDSCs need to be further activated to gain immunosuppressive function in the GBM TME. Multiple cytokines have been reported to activate MDSCs, including M-CSF, GM-CSF, IL-6, IL-10, TGF-β, B7-H1, and INFγ [47,57]. Additionally, exosomal miRNAs have been reported to be essential for MDSC differentiation and activation [58-61]. Functional studies have revealed that GBM-derived extracellular vesicles containing miR-1246 induce MDSC differentiation from donor monocytes under hypoxic conditions [59]. Inhibition of miR-1246 transcription and exosomal packaging using 2-methoxyestradiol impairs GBM tumor growth and MDSC infiltration [59].

Pharmacological targeting of GBM–T cell crosstalk

A growing body of studies using CyTOF and scRNA-seq have revealed a unique landscape of T cell populations and T cell receptors in brain tumors [7,62-64]. Compared with IDH-mut glioma, IDH-WT tumors express higher T cell-specific genes and cytotoxicity signatures [63], suggesting a potential GBM-T cell crosstalk. Here, we discuss the recent progress of pharmacological tools targeting T cell exhaustion and T cell tolerance in GBM (Table 1).

In the GBM TME, exhausted T cells exhibit reduced effector function and increased expression of immune checkpoints (e.g., PD1, CTLA4, TIM3, TIGIT, LAG3, BTLA, 2B4, CD39, and CD160) [14,65,66]. Pharmacological approaches to target these immune checkpoints have been developed to treat GBM patients. For instance, a randomized, multi-institution clinical trial demonstrated that neoadjuvant anti-PD1 therapy significantly improves the overall survival and progression-free survival of recurrent GBM patients [67]. This result is consistent with another single-arm Phase II clinical trial (NCT02550249) in which researchers observed a significant antitumor effect of neoadjuvant anti-PD1 therapy on newly diagnosed or relapsed GBM [68]. However, a recent clinical trial (NCT02017717) with anti-PD1 therapy failed to increase patient overall survival in recurrent GBM patients [69]. We speculate that these controversial results may relate to tumor genetic status and altered core signaling pathways in GBM cells, and their associated GBM-T cell symbiosis. Indeed, genomic profiling in GBM patient tumors has revealed that PTEN mutations are enriched in anti-PD1 therapy nonresponders, whereas MAPK pathway alterations (e.g., PTPN11 and BRAF) are enriched in responders [28]. A recent study further demonstrated that phospho-ERK1/2 expression in GBM cells is predictive of overall survival following adjuvant anti-PD1 therapy in recurrent GBM patients [70]. Together, these findings suggest that context-dependent GBM–T cell crosstalk is critical for designing effective ICI therapy in GBM. Additional clinical trials are underway for testing immunotherapies, including anti-LAG3 combined with anti-PD1 (NCT02658981), anti-TIGIT combined with anti-PD1 (NCT04656535), and anti-CD39 (NCT04306900), in GBM patients.

T cell tolerance represents the programmed induction of unresponsiveness due to misexpressed self-antigens in GBM [14], a process that is regulated by the expansion of Treg cells [71]. Mechanistically, GSCs express and secrete distinct factors (e.g., TGF-β and CCL2) to control Treg cell infiltration and expansion in GBM [72]. Several pharmacological tools have been developed to target Treg cells in GBM. First, glucocorticoid-induced TNFR-related receptor (GITR) has been shown to be critical for Treg cell differentiation into CD4 effector T cells [71]. Targeting GITR using an agonistic antibody (anti-GITR) improves the survival of GBM-bearing mice via converting Treg cells to Th1-like CD4 T cells [71]. Moreover, the anti-GITR therapy synergizes with anti-PD1 therapy in GBM-bearing mice, and the synergy is further amplified when combined with the SOC treatment [71]. The other approach is to block indoleamine 2,3-dioxygenase 1 (IDO1), given previous studies have shown that GBM cell IDO1 promotes Treg cell expansion [73]. Although IDO1 inhibitor BGB-5777 alone is not enough to inhibit GBM tumor growth, treatment with this inhibitor synergizes with anti-PD1 therapy in GBM mouse models [73].

Pharmacological targeting of tumor–immune symbiosis to improve the effectiveness of immunotherapy

Given the critical role of the tumor–immune symbiosis in regulating innate and adaptive immunity in GBM, blockade of this symbiosis may affect the effectiveness of immunotherapies. This concept is also supported by the emerging evidence showing that high infiltration of myeloid cells correlates with increased immunotherapy resistance in GBM patients [70,74,75]. This section summarizes recent findings highlighting pharmacological targeting of myeloid cells (e.g., GAMs and MDSCs) to improve immunotherapy efficiency in GBM (Table 1).

GAMs are a heterogeneous population of cells exhibiting a potent immunosuppressive function in GBM. A growing body of evidence demonstrates that blockade of GAM immunosuppressive function through different strategies may overcome immunotherapy resistance. First, distinct subsets of GAMs may play different roles in affecting ICI therapy efficiency [41,76,77]. Unbiased CyTOF and scRNA-seq studies in GBM tumors have revealed a unique population of CD73^high macrophages that persist following anti-PD1 therapy. Depletion of CD73 in GBM-bearing mice exhibits a robust synergistic antitumor effect with anti-PD1 and anti-CTLA4 [74]. Although anti-CD73 antibody has been developed [78], further studies are needed to validate the antitumor effect of the combination therapy with anti-CD73 and ICIs in GBM. The second approach is to suppress cytokine/ligand-receptor interactions during the GBM–GAM symbiosis. GBM cell-derived IL-6 is essential and necessary for PD-L1 expression in tumor-associated myeloid cells, including macrophages. Inhibition of IL-6 with neutralizing antibodies synergizes with anti-PD1 therapy in GL261 tumor-bearing mice [79]. However, the antitumor effect of anti-IL-6 antibodies is context dependent. For example, EC-derived IL-6 can induce macrophage immunosuppression in a genetic GBM mouse model, but anti-IL-6 therapy is insufficient to activate antitumor immune response and does not sensitize tumors to ICIs [41]. The failure of synergy between EC IL-6 inhibition and ICIs may relate to the dual effect of IL-6 in GBM. In addition to the protumor effect, IL-6 inhibits tumor growth by stimulating CD40 expression [41], suggesting a therapeutic potential of dual targeting IL-6 and CD40. Indeed, combination of anti-IL-6 therapy and CD40 stimulation induces a robust antitumor immunity and synergizes with ICIs (e.g., anti-PD1 and anti-CTLA4) in GBM mouse models [41]. In a similar way, blockade of the PROS1-AXL axis-mediated GAM-GSC symbiosis using AXL inhibitor BGB324 synergizes with anti-PD1 therapy in GBM-bearing mice [17]. The third approach is to disrupt the GBM–GAM crosstalk via blockade of the CD47-signal regulatory protein alpha (SIRPα) pathway. CD47 is a ‘don’t eat me’ signaling that helps GBM cells to evade GAM-mediated phagocytosis [80]. In vivo pharmacological studies have demonstrated that anti-CD47 blockades significantly extends survival of GBM-bearing mice by regulating GAM-mediated innate immune response [81,82], and that this antitumor effect is further amplified upon TMZ treatment [83]. In addition to regulating the innate immune response, anti-CD47 therapy also actives adaptive immunity. Combining anti-CD47 and TMZ treatment significantly sensitizes GBM tumor to anti-PD1 therapy [83]. The final appealing strategy is to reprogram GAMs from an immunosuppressive to an immunostimulatory phenotype. For example, inhibition of macrophage immunosuppressive polarization by combined rapamycin and hydroxychloroquine treatment not only reduces the expression of the CD47-SIRPα signaling axis in GBM cells and GAMs but also synergizes with anti-PD1 therapy in GBM-bearing mice [84]. Consistently, blockade of GAMs immunosuppressive polarization using additional several pharmacological approaches (e.g., MAGL-specific inhibitor JZL184, Robo1Fc, CSF-1R inhibitor AFS98) shows robust synergy with ICIs (e.g., anti-PD1, anti-CTLA4, or anti-4-1BB) in different GBM mouse models [30,43,75].

Similar to inhibition of the GBM–GAM crosstalk, targeting the GBM–MDSC symbiosis could also enhance the effectiveness of ICI therapies (Table 1). This conclusion is supported by the recent findings showing that treatment with CCR2 inhibitor CCX872 or IDO1 inhibitor BGB-5777 not only impairs MDSC infiltration but also shows robust synergy with anti-PD1 therapy in GBM-bearing mice [54,73]. In addition to ICI, immunostimulatory gene (e.g., TK/Flt3L) therapy is also affected by MDSCs. A very recent study demonstrates that due to the high infiltration of PMN-MDSCs in the TME, IDH1 WT gliomas do not respond to the TK/Flt3L therapy [53]. However, reprogramming immunosuppressive PMN-MDSCs into nonsuppressive granulocytes using recombinant G-CSF significantly enhances TK/Flt3L therapeutic efficacy in GBM-bearing mice [53].

Concluding remarks and future perspectives

With its vital role in regulating tumor progression and the effectiveness of immunotherapies, the tumor–immune symbiosis embodies critical therapeutic targets for GBM. This review has outlined recent pharmacological approaches to target the tumor–immune crosstalk (e.g., GBM–GAM, GBM–MDSC, and GBM–T cell crosstalk) in GBM, which not only directly inhibit tumor progression but also turn the TME from ‘cold’ to ‘hot’, thus improving the effectiveness of immunotherapies. Also, a range of pharmacological tools have been developed to target the GBM-immune symbiosis (Figures 1 and 2, and Table 1), demonstrating tremendous clinical translation potential.

GBM has a unique immunosuppressive TME with infiltration of various types of immune cells (e.g., macrophages, microglia, MDSCs, neutrophils, Treg cells, and T cells), and each type of these immune cells exhibit phenotypic heterogeneity and multifaceted functions [7,8,62,85,86]. Despite the success of developing many pharmacological approaches in GBM mouse models (Table 1), many challenges remain regarding how to translate these preclinical findings into the clinic, and how to develop novel, effective and specific pharmacological tools targeting the GBM-immune symbiosis (see Outstanding questions). Apart from few examples, this concept has not been translated into the clinic for GBM treatment. One reason would be the choice of GBM mouse models, which may not completely recapitulate the immune landscape and genomic heterogeneity of GBM patients. Studies using humanized mice with genetically engineered GBM system via CRISPR/Cas9 could better evaluate the drug effect on blocking the GBM-immune symbiosis [23,87,88]. Additionally, organoids and tumor-on-a-chip systems may provide more comprehensive platforms for rapid drug screening [21,89]. The second reason would be the limitation of effective therapeutic targets. Further studies using both bottom-up and top-down strategies will help to develop new and effective therapeutic tools aiming to block the GBM–immune crosstalk (Figure 3). The classical treatment design starts with the investigations focusing on the cellular and molecular mechanisms underlying the GBM-immune symbiosis. The alternative research strategy may begin with drug screening by determining which pathways/genes are affected by drug candidates [38,90]. Machine-learning approach could help to zero in on pathways/genes that are essential for the G BM-immune symbiosis. The last challenge could be the major barriers (e.g., BBB, blood–cerebrospinal fluid barrier, and brain-resident lymphatic barrier) that can limit drug delivery into the GBM TME. To overcome this challenge, both invasive and noninvasive approaches have been developed to improve drug delivery into the brain by hijacking the cellular and molecular barriers of the BBB [5]. In addition, other approaches (e.g., therapeutic strategies with therapeutic vaccines, adoptive cell therapy, and oncolytic viruses) have also been tested for targeting the GBM TME, although they are not classical catalog of pharmacological drugs [15,91,92]. However, further studies are still needed to optimize appropriate pharmacological treatments combining the GBM-immune symbiosis-targeted therapy and the strategy of enhancing the BBB-penetrating ability. Although a growing body of preclinical data has largely accelerated drug discovery, clinical trials are still needed to validate the clinical benefit of these drug candidates targeting the GBM-immune symbiosis. We anticipate that success in translating current known pharmacological tools into the clinic, and developing novel therapeutic strategies targeting the GBM-immune symbiosis will ultimately improve GBM patient outcomes.