Introduction
Glioblastoma (GBM) is the most common and deadly form of central nervous system cancer [
1,
2]. In the USA, it is estimated that more than 13,000 patients are diagnosed with GBM annually. Unlike many other cancers, GBM rarely metastasizes to a secondary organ, but instead diffusely infiltrates throughout the brain. The current standard of care for treating GBM consists of maximal surgical resection, radiotherapy, and concomitant and adjuvant chemotherapy with temozolomide [
1,
3,
4]. Despite this aggressive treatment strategy, GBM tumors commonly recur with a median survival of less than 18 months, and fewer than 5% of patients surviving to 5 years [
5‐
11]. A central focus for improving GBM therapy is developing new tools to understand pathophysiological processes driving GBM invasion of the brain. Improved therapy will likely require both an improved understanding of which cells within the heterogeneous cell population of GBM tumors invade surrounding tissues, and the extent to which cell-cell crosstalk within heterogeneous cell cohorts contribute to GBM invasion and mortality [
12‐
18].
Microglia (MG) are resident immune cells of the CNS [
19,
20]. In healthy individuals, microglia constantly survey their surroundings and maintain tissue homeostasis by removing apoptotic cells and promoting neuro-network generation [
20‐
22]. Studies of the GBM tumor microenvironment have demonstrated that infiltrative and resident immune cells, such as microglia, may comprise up to a third of the solid tumor mass [
20‐
22]. Morphologically, quiescent microglia typically exhibit a ramified (branching and elongated) shape. Upon stimulation in response to inflammation, disease, or tumor growth, microglia cell processes become hypertrophic and, in some cases, retract causing the cell to take on an ameboid appearance. While the number and phenotype of immune cells have been associated with patient prognosis [
19,
23‐
27], detailed analysis of crosstalk between GBM cells and microglia are difficult to evaluate in vivo. Thus, there is a need for an experimental platform to rigorously investigate interactions between GBM cells and microglia, as well as to identify factors associated with microglia-GBM crosstalk that may alter GBM cell invasion and therapeutic response. Cancer tissue engineering platforms that integrate biomaterial mimics of the tumor microenvironment with primary cells and biomolecules are increasingly used to investigate pathophysiological processes difficult to examine in vivo [
28,
29].
We previously developed a gelatin-based hydrogel model platform to investigate pathophysiological processes underlying GBM cell invasion and therapeutic response. Notably, we observed that biophysical (hyaluronan content and molecular weight) and metabolic (hypoxia) transitions in the GBM tumor microenvironment both significantly alter GBM invasion [
30‐
32]. More recently, we adapted this system to profile cytokine-based crosstalk between cells within the GBM tumor microenvironment, identifying secreted factors generated by an artificial perivascular niche that can accelerate GBM cell invasion [
33]. The objective of the present study was to adapt this established hydrogel platform and cytokine analysis protocols to examine the effects of microglia within the GBM tumor microenvironment on GBM gene expression and invasiveness using patient-derived GBM specimens that maintain patient-specific morphologic and molecular phenotypes [
34,
35].
Discussion
Cellular crosstalk within the tumor microenvironment provides a powerful avenue of interaction that may significantly shape disease progression. Tools to interrogate cellular crosstalk offer an opportunity to identify novel therapeutic compounds to improve treatment of glioblastoma. This work demonstrates the use of a tissue engineering platform to investigate the role of crosstalk between patient-derived GBM specimens and microglia on shifts in the phenotypic, proteomic, and transcriptomic signatures of GBM. GBM-MG crosstalk is bidirectional and induces microglia activation along with shifts in GBM cell activity consistent with the go-or-grow phenomenon [
61]. This work extends technical capabilities beyond more traditional Transwell membrane or mixed culture methods for examining cell-cell crosstalk. Importantly, this effort provides a platform for analysis of individual cell populations, each maintained within discrete multi-dimensional models of the tumor microenvironment, while maintaining the ability to examine the nature and role of cell-cell crosstalk during mixed culture on cell activity.
While microglia display a quiescent phenotype in single culture, hallmarks of microglial activation were observed via
both morphological changes and increased CD68 expression as a result of GBM-MG co-culture. These results are consistent with hallmarks of microglial activation seen in cases of disease and histopathological analysis of GBM tumors [
24,
62]. The nature of this tissue engineering platform allows significant post-culture analysis of GBM cell activity at functional (invasion, proliferation), transcriptomic (RNAseq), and secretomic levels. We recently reported the use of gelatin hydrogels to profile to role of localized hypoxia on activation of cells associated with the neurovascular unit [
58]. Indeed, while a single hydrogel formulation was used for all microglial culture in this study, significant opportunities exist to use multi-dimensional hydrogels culture to refine our understanding of the role of the matrix microenvironment on microglia activation itself.
While advanced sequencing techniques such as RNA sequencing offer the opportunity to define the transcriptomic signature of cells to aid treatment planning and outcome prediction [
40,
43,
44,
63], the design of the hydrogel culture system reported here enabled analysis of shifts in the transcriptome of individual cell populations as a result of heterotypic cell (GBM, MG) crosstalk. GO analyses revealed GBM-MG
co-culture upregulated genes in a patient-derived GBM specimen associated with cell cycle, RNA/DNA division and metabolic activity. However, genes involved in cell adhesion/migration showed significant downregulation as a result of GBM-MG
co-culture. These findings indicate tradeoffs in GBM proliferation versus invasion due to MG crosstalk consistent with the go-or-grow dichotomy of GBM cells [
32,
61,
64,
65]. Significant decreases were observed in expression of genes associated with
NLR,
TNF,
NF-κB,
MAPK, and
TLR pathways in GBM specimens in response to MG
co-culture.
NLR and
TLR signaling pathways are involved in pathophysiological responses to inflammation and tumor progression [
66‐
68]. Of these, the
NF-κB signaling pathway is known to be sensitive to
TNF signaling [
69‐
71] which plays a major role in immune activation [
72,
73], breast cancer invasion [
74], and driving
TLR and
MAPK signaling involved in cell migration and tumor invasion. These pathways contribute to heightened immune responsiveness and are involved in angiogenesis and cell migration [
66,
67,
69‐
71], suggesting GBM-MG interactions may inhibit GBM invasiveness.
KEGG analysis also showed strong upregulation in
TH and
STAT3 signaling, indicating that secreted factors from microglia may promote GBM proliferation, reduce apoptosis, and enhance chemotherapeutic resistance [
75‐
77]. Recently, our group showed STAT3 is strongly activated in GBM, and inhibiting STAT3 can reduce GBM cell proliferation [
52,
78]. More, GBM-MG
co-culture upregulated
FOXO signaling, which has been linked to therapeutic resistance due to its contribution to DNA repair as well as mediation of oxidative stress. iRegulon analysis showed GBM-MG crosstalk increased enrichment for
SUZ12, previously shown to be increased in high grade astrocytoma and involved in pathways that regulate glioma proliferation and metastasis [
79,
80]. iRegulon analysis also showed GBM-MG crosstalk increased
REST and
RCOR1, known to regulate the oncogenic properties of GBM stem cells [
81] that associate with therapeutic resistance and recurrence [
82,
83]. The
IRF family has been shown to be significant tumor suppressors, inhibit tumor proliferation and loss of
IRF genes may contribute to tumor metastasis and invasion [
84,
85]. Together, analysis of transcriptomic data support the functional responses of increased proliferation but decreased invasion for GBM cells as a result of GBM-MG interactions [
32,
65,
86‐
88]. Inclusion of RNAseq analyses to examine the role of microglia-GBM crosstalk provides a valuable dataset to motivate ongoing efforts. Indeed, while this study highlighted the importance of examining GBM-microglia interactions via RNAseq methods, ongoing efforts are using this approach to consider the role of microglia signaling on GBM subtractions such as glioblastoma stem cells (GSCs) as well as examining the behavior of GSCs within a larger cohort of GBM cells.
The hydrogel platform was subsequently used to experimentally interrogate the influence of GBM-MG crosstalk on GBM proliferative and invasive phenotypes in patient-derived GBM12 cells. GBM12 cells exhibited significantly increased proliferation and significantly inhibited invasion in response to MG co-culture. Strikingly, MG-induced inhibition of GBM12 invasion was observed for multiple combinations of patient-derived GBM specimens and microglia: EGFROE GBM12 cells co-cultured with HMC3 microglia and EGFRvIII GBM39 cells co-cultured with primary mouse neonatal microglia.
Analysis of the combined GBM-MG secretome revealed multiple targets driving the observed shifts in functional and transcriptomic activity. CCL2 and CCL3 are associated with monocyte and macrophage recruitment [
89‐
91] and may act as chemoattractant [
89]. Of these, further study of the role of CCL2 in GBM invasion may be particularly warranted, as expression levels were not only significantly increased in GBM-MG
co-culture (vs.
GBM-single or
MG-single cultures) but also compared to the
Mix control, consistent with synergistic activation of CCL2 secretion due to GBM-MG crosstalk. IGFBP-3, known to regulate cell proliferation, was also increased in GBM-MG crosstalk, though its role in cancer progression remains to be fully understood [
92‐
94]. DPP4 (plasma membrane protein that contributes to immune and metabolic regulation [
95,
96]) and HB-EGF (cell metabolic activity and tumor suppression in other cancers [
95‐
97]) were also strongly upregulated in GBM-MG
co-culture, as was ANG, well-known for its role in angiogenesis and cell proliferation [
98,
99], and Serpin F1, known as for its role in suppression of tumor growth and prostate cancer metastasis [
100‐
103]. Previous study by Shinozaki et al. [
104] also indicated that cytokines produced by microglia could potentially drive astrocytes towards a neuroprotective phenotype upon brain injuries. While results here provide critical data regarding highly expressed factors within the combined MG-GBM secretome, ongoing opportunities exist to consider a wider range of secreted factors, consider the role of alternative signaling pathways such as extracellular vesicles in crosstalk, and to use machine learning algorithms to identify critical subsets of factors most highly associated with GBM cell response. Notably, we recently reported an iterative partial least squares regression machine learning methods to identify [
105] secretome signals generated by niche-associated cells that enhance quiescence of hematopoietic stem cells in hydrogel culture. So, while further efforts are needed to more fully investigate the potential mediators of GBM invasion that arise from GBM-MG crosstalk, we present a robust platform to pursue such investigations here.
The immune system and immune cells and their relationship with cancer have been a hot topic in recent years. Tumor-associated macrophages/microglia, or GBM-associated macrophages/microglia here, has drawn a large amount or research efforts [
19‐
21]. While some studies showed that the infiltrated microglia facilitates the tumor growth and targeting those infiltrated immunes cells could be a promising therapeutic approach [
19‐
21,
106], the exact role of them remains controversial [
107]. In the study, the combination of increased proliferation but decreased invasion aligns with the go-or-grow hypothesis [
64], but more importantly demonstrates that crosstalk between MG and GBM cells in the tumor microenvironment may have powerful effects on GBM activities tied directly to tumor progression and patient survival.
Finally, we note the value and limits of a tissue engineering approach described herein. Glioblastoma tumors contain a heterogeneous mix of cells, including a subpopulation of tumor initiating cells (GBM stem cells, GSCs) [
108‐
110], critical for invasion, recurrence, and mortality [
109,
111‐
117]. Tumors contain a mix of fibrillar matrix (e.g
., collagens, laminins) and hyaluronic acid (HA), complex perivascular niches, regions of hypoxia [
118], and multiple immune-associated cells including microglia and macrophages. Here, we report a multi-dimensional hydrogel platform to examine pathophysiological processes linked to GBM progression and mortality using patient-derived GBM specimens in response to microglia. We have also recently described hydrogel-based platforms to investigate the role of angiocrine signaling from engineered perivascular cultures on GBM cell invasion and resistance to the frontline chemotherapy temozolomide [
57]. We also reported adaptations to the hydrogel environment via localized formation of hypoxic zone to support culture of a broader diversity of cells from the neurovascular unit [
58]. While we acknowledge understanding the role of the coordinated activity of tumor-associated macrophages and microglia on the activity of GBM cell cohorts (or on specific cells from the GBM microenvironment such as GSCs) are essential, this manuscript provides a conceptual framework for pursuing such studies while also providing critical information regarding the role of reciprocal GBM-microglia signaling on GBM invasion.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.