Glioblastomas (CNS5 WHO Grade 4 astrocytomas, IDH-wildtype, henceforth called GBM) are a devastating disease, with extremely poor clinical outcomes evidenced by the median survival time of 15 months post-diagnosis [1]. New therapies are desperately needed; however, many recent approaches have failed to improve outcomes [2,3,4,5], including immunotherapeutic approaches. Insufficient understanding of the tumor microenvironment is a critical challenge impairing the development of effective therapies. Resident (microglia) and infiltrating macrophages are a substantial component of the tumor environment and the tumor itself, constituting 30–50% of the cellular content [6,7], and are primary mediators of inflammatory responses [8]. Macrophages are a key cellular component of a successful anti-tumor response in which they identify and phagocytose cancerous tissues [9,10]. In GBM and other cancers, this capability is impaired or blocked entirely by tumor-induced suppression of anti-tumor immune responses [11,12,13]. In an increasing number of instances, this is recognized as being induced through extracellular vesicle (EV) intercellular communication [12]. EVs are lipid-enclosed vesicles released from cells that contain proteins, nucleic acids, and other biological mediators, allowing for intercellular communication in both health and disease [14,15]. GBM/macrophage EV crosstalk is a critical component of the tumor microenvironment [11,12]. With improved understanding, these interactions could be harnessed—or discouraged—to engage endogenous anti-tumor responses and develop novel targeted therapies. 

Macrophages are a key component of the innate and adaptive immune response. Circulating monocytes exit the bloodstream into numerous tissues and mature into resident macrophage populations throughout the body [16]. Microglia, the resident macrophage population in the central nervous system, are unique in that this population is established during embryonic development [17]. Functionally, however, monocyte-derived macrophages and microglia retain extensive similarities. In innate immunity, macrophages contain numerous receptors and sensors to identify the presence of infection, cellular damage, or cancer [16]. These macrophages can then induce numerous signaling cascades, inducing localized inflammation (production of reactive oxygen species, lymphocyte and Natural Killer (NK) cell activation, recruitment of neutrophils from the bloodstream, etc.). Similarly, these macrophages can induce systemic responses, such as a fever [16]. Once the activating stimulus is resolved, macrophages also produce factors contributing to cellular repair and the resolution of inflammation [18]. In adaptive immunity, macrophages are an important antigen-presenting cell which identify microbial antigens, or in the case of cancer, antigens from tumors that are either not expressed by healthy cells or antigens that are expressed much more highly in tumor cells [16]. Through this antigen expression, macrophages are able to activate adaptive immune responses, notably T-cells, to specifically target that antigen throughout the body [19]. Unfortunately, in tumors, the cancer often adapts to evade and manipulate the immune response. For example, tumors can begin to produce proteases, clipping off the ligand necessary for the macrophage-activated NK cells to attack the tumor cells [20]. Further, the tumor can signal directly to nearby macrophages to manipulate their activities, as discussed in detail in this review.

Macrophages are an extremely malleable cell type capable of adopting a wide spectrum of activation states [21,22]. These activation states control many aspects of macrophage physiology, including morphology, phagocytic activity, cytokine production, inflammatory profiles, and many other components of the immune response [23,24]. While macrophage activation is complex and multifaceted, for practical purposes this is often simplified to a pro-inflammatory “M1” subset and a second anti-inflammatory “M2” subset. The M1/M2 nomenclature is overly simplistic and does not cleanly apply across most disease states [25]; however, for general scientific discourse, it remains the best option available so long as its limitations are understood. In GBM, it is generally understood that pro-inflammatory “M1”-polarized macrophages promote a robust immune response and anti-tumor effects, whereas anti-inflammatory “M2”-polarized macrophages promote pro-tumor effects and dampen the broader immune response [13,26]. M0 is sometimes used in reference to naïve unstimulated macrophages [27]. Intracellularly, these macrophage activation states are regulated by several canonical and non-canonical signaling pathways. For M1 macrophages, this is typically via the NF-κβ/STAT1 [28,29] signaling axes, often modeled in vitro using lipopolysaccharide and interferon gamma stimulation of TLR-4 and the interferon-gamma receptor [30]. Conversely, M2 macrophages are more typically associated with the STAT3/STAT6 signaling axes [28], modeled with Interleukin 4 and/or 13 stimulation [31]. In the context of GBM, however, countless potential environmental stimuli likely induce complex activation states not fully in line with the M1/M2 nomenclature [32,33]. Further, given that macrophages are highly impressionable by extremely localized environmental factors, there are likely diverse activation states across the GBM macrophage populations [21]. Continued advances in single-cell RNA sequencing technologies will allow for an improved understanding of how the macrophage activation spectrum impacts GBM pathophysiology [34].

Critically, in the context of GBM, macrophage activation status determines whether the macrophage attacks the cancerous tissue or promotes continued growth and broader immune suppression [17,25]. M1 macrophages express high levels of antigen-presenting MHC complexes, through which they can display tumor-derived antigens and induce robust adaptive immune responses against the tumor [35,36]. M1 macrophages also express elevated levels of the costimulatory molecule CD87 (B7-2) allowing for improved activation of naïve T-cells [31]. Similarly, M1 macrophages produce elevated levels of cytotoxic factors such as reactive oxygen species, nitric oxide, and pro-inflammatory cytokines to directly target nearby tumor cells. Conversely, M2 macrophages are generally anti-inflammatory and immunosuppressive [26]. With reduced MHC and costimulatory molecule expression, M2 macrophages have a reduced ability to activate the adaptive immune response [35]. Further, under certain conditions, M2 macrophages can directly promote tumor growth in response to tumor signals, as discussed in this review. GBM macrophages are a heterogenous cell population derived from both tissue-resident microglia and infiltrating myeloid-derived macrophages. While these populations share many similarities and are often referred to collectively as macrophages, some distinctions can occur [25,37]. When examined separately experimentally, specific terminology will be utilized for clarity.

EVs are small lipid-enclosed vesicles released from cells into the extracellular space [38,39]. EVs contain a broad mixture of biological materials, including proteins, lipids, nucleic acids, and other metabolites, whose contents can vary depending on the cellular source and EV subtype [40]. Even the lipid composition of the EV membrane can influence biological activity [41]. Historically, EVs were further distinguished into several EV subtypes based on size and cellular source or biogenesis. Exosomes are a subtype of EVs derived from the endosomal/multivesicular body system and range from 30–150 nm in diameter. Microvesicles are a subtype of EVs formed from the plasma membrane through outward budding and range from 100 nm to 1 µm in diameter. Apoptotic bodies form from dying cells. Notably, apoptotic bodies differ in that they contain additional cellular material, such as intact organelles and chromatin [42]. In recent years, the EV research field has updated its guidelines on EV terminology [43]. When specified in the source literature, the original terminology will be used; however, where sufficient method detail is provided, reaserchers will also include the current terminology. Notably, these recent guidelines describe EVs in terms of their physical characteristics, molecular composition, and cellular origin via biogenesis [43]. Here, this will support a shift in terminology from the previous term “exosome” towards the modern terminology, “small EVs” (sEVs) for EVs under 200 nm diameter. Reaserchers note that the biogenesis of “exosomes” versus “microvesicles” is seldom known when collecting EVs from biofluids or conditioned cell/tissue culture media, and thus the term “EVs” refers to both such populations [44].

While EVs were originally thought to be a form of cellular waste, it has since been appreciated that EVs conduct an intricate form of cellular communication [40]. They allow for cellular crosstalk both locally and across great distances [45]. These communications occur both in normal physiological conditions and in disease [15]. GBM tumor cells show an increased production of EVs relative to healthy tissue [46,47]. Critically, in GBM, EVs are thought to be a primary means through which the tumor manipulates its surroundings to invoke a more conducive environment for further growth and anti-cancer drug resistance [44,48,49]. 

EVs are a critical mechanism utilized by GBM tissue to evade the immune system [59]. Macrophages are the predominant immune cells in the tumor vicinity; however, they are ineffective at targeting tumors [60]. Worse, some macrophages begin to support the tumor, allowing for increased and faster growth [61]. GBM tissue supports this maladaptive macrophage response through numerous EV-dependent mechanisms as detailed below.

In 2015, de Vrij et al. applied GBM-derived EVs (presumably small EVs) onto monocyte-derived macrophages and observed several changes [48]. Notably, the application of GBM-derived EVs resulted in an increased expression of CD163, a marker associated with the M2 pro-tumor phenotype, as measured by flow cytometry. The macrophages were found to produce increased levels of VEGF and IL-6. These cytokines have been previously found to support tumor growth (IL-6) [62] and tumor hyper-vascularization (VEGF) [63]. The authors hypothesized that these effects were caused by small RNA molecules enriched within GBM EVs, as had been demonstrated elsewhere with EVs from different tumor types [64]; however, they did not identify an exact candidate. Work by van der Vos et al. (2016) supported this hypothesis by visualizing the extensive uptake of GBM EVs (presumably small EVs) by primary microglia and noting the increase in GBM EV-cargo microRNAs in the microglia and macrophage. These studies suggest the transfer of microRNAs from EVs to the microglia/macrophage [65].

Zhao et al. (2022) also demonstrated an M2 polarization shift in macrophages treated with GBM EVs (presumably small EVs). They found that GBM EVs are enriched in microRNA-27a-3p and have even higher levels under hypoxic conditions. Further, they demonstrated that microRNA-27a-3p inhibits enhancer of zeste homologue 1 (EZH1) [50]. Inhibition of EZH1 has been previously demonstrated to promote M2 macrophage polarization [66]. Indeed, the authors observed that microRNA-27a-3p-treated macrophages had elevated levels of Arginase-1 (M2 phenotypic marker) and reduced iNOS (M1 phenotypic marker). Myeloid-cell-derived arginase-1 (via arginine depletion) has a noted ability to suppress T-cell responses [67]. A similar study confirmed the transfer of miR-21 from GBM to microglia in an in vivo mouse model and observed substantial microglial reprogramming [51]. This was an important study in that the host mice were miR-21 null, directly implying the microRNA was transferred from GBM EVs to cells of the tumor microenvironment. This is further evidence that GBM EVs may be preferentially taken up by tumor-associated macrophages to promote a supportive (immune-suppressed) tumor environment. These effects are mediated by GBM EV microRNAs such as microRNA-27a-3p and miR-21, although other EV components could certainly contribute towards this effect.

Proteomics analysis has demonstrated a diverse set of proteins enriched in GBM EVs with likely bioactive effects [68]. Indeed, in 2018, Gabrusiewicz et al. also found that GBM stem-cell-derived exosomes (presumably small EVs) applied to monocytes induced a shift in monocyte-derived macrophage polarization state [49]. Specifically, through flow cytometry they found that macrophages exposed to these exosomes had reduced expression of M1 indicators (MHCII and CD80) and increased M2 indicators (CD163 and CD206). The authors found that the exosomes contained the transcription factor STAT3, a transcription factor heavily associated with M2 macrophage polarization [49,69]. This suggested that GBM-derived STAT3 is transferred to the macrophages inducing their phenotypic shift towards a tumor-supportive phenotype. Further, they demonstrated the exosomes used had a propensity to be taken up by macrophages relative to other cell types [49]. Additional work to understand how GBM-derived EVs target macrophage populations could unveil an advantageous therapeutic target.

Additional studies have also investigated the roles of GBM EVs on macrophage physiology. Yang et al. (2019) found that the GMB EV microRNA miR-214-5p was associated with poor clinical prognosis and targeted microglial CXCR5 transcripts, and thus reduced protein expression [52]. CXCR5 has numerous roles in cancer biology, notably through chemoattractant interactions with its ligand CXCL13 [70]. Xu et al. (2021) found that exosomes (presumably small EVs) from hypoxic glioma cells also induce macrophages towards an M2 state through an induction of autophagy pathways. These results were greatly subdued in EVs derived from normoxic glioma cells [53].

Collectively, it is clear that GBM EVs are taken up by macrophage populations and subsequently modulate cellular functions towards a tumor-supportive phenotype. While this is frequently addressed as an M2 macrophage or M2-like, it is likely far more complex, with numerous caveats. In 2015, de Vrij et al., for example, saw a macrophage shift towards the M2 phenotype after GBM EV exposure, yet observed increased IL-6 production. While IL-6 has been documented as a tumor-supportive cytokine in this context, it is typically associated with M1 macrophages [31,71]. This highlights the uniqueness of the GBM microglial/macrophage population to consider when developing macrophage-targeted therapies developed in other contexts.