Mesenchymal Stromal-Like Cells in the Glioma Microenvironment: What Are These Cells?

Simple Summary We review here what is currently known about the role of mesenchymal stromal-like cells, which complicate our understanding of the glioma microenvironment. We provide an overview of the major studies on these cells, highlighting their role in tumor progression and prognosis. Researchers and clinicians should consider these cells to be an integral component of the glioma microenvironment, and of considerable potential value for future prognostic and therapeutic perspectives. Abstract The glioma microenvironment is a critical regulator of tumor progression. It contains different cellular components such as blood vessels, immune cells, and neuroglial cells. It also contains non-cellular components, such as the extracellular matrix, extracellular vesicles, and cytokines, and has certain physicochemical properties, such as low pH, hypoxia, elevated interstitial pressure, and impaired perfusion. This review focuses on a particular type of cells recently identified in the glioma microenvironment: glioma-associated stromal cells (GASCs). This is just one of a number of names given to these mesenchymal stromal-like cells, which have phenotypic and functional properties similar to those of mesenchymal stem cells and cancer-associated fibroblasts. Their close proximity to blood vessels may provide a permissive environment, facilitating angiogenesis, invasion, and tumor growth. Additional studies are required to characterize these cells further and to analyze their role in tumor resistance and recurrence.


General Characteristics of GASCs
GASCs have been isolated from primary cultures of cells from mouse orthotopic transplantation tumor models [7,8], human low-grade gliomas (LGGs), human high-grade gliomas

Origin of GASCs
The origin of GASCs has yet to be determined. Hossain et al. [3] suggested that GASCs may differentiate from glioma stem cells (GSCs). This hypothesis is consistent with reports showing that GSCs can differentiate into stromal cells, including endothelial cells in particular [27]. However, most GASCs are genetically different from GSCs. They are typically diploid and do not harbor the genetic alterations commonly seen in GSCs, such as losses of chromosome 10 or gains of chromosome 7, suggesting that GASCs are probably recruited to the tumor from a source other than GSCs [3,22]. GASCs may be derived from an epithelial-to-mesenchymal like transition of reactive astrocytes which acquire stem cell properties [20]. These reactive astrocytes may originate from local astrocytes or from the migration and differentiation of neural progenitor cells found in the subventricular zone [28]. Alternatively, as suggested for CAFs, GASCs may arise from the transdifferentiation of pericytes and vascular smooth muscle cells, and from endothelial cells through the process of endothelial-to-mesenchymal transition [29]. GASCs can also originate from MSCs. This latter hypothesis is supported by many studies showing the migration of MSCs isolated from different sources such as the bone marrow towards established gliomas and their integration into the tumor vasculature [30][31][32]. The mechanism underlying the tropism of MSCs for gliomas has not been fully elucidated, but numerous chemotactic factors have been implicated in this process, including VEGF, SDF-1/CXCL12, and IL-8 [33][34][35]. Cells similar to MSCs in terms of in vitro growth, surface markers, and trilineage mesenchymal differentiation have also been isolated from normal brains [36][37][38]. Conflicting results have been reported, but several studies have shown that MSCs promote tumor progression [30,39].

Heterogeneity of GASCs:
-Two different subsets of GASCs, differing in their expression of the CD90 surface marker, were discovered after cell sorting.

Heterogeneity of GASCs: -GASCs derived from
LGGs and HGGs have different proteome profiles.
-Molecules associated with mesenchymal cells (vimentin and transglin), and tumor aggressiveness with potential secretory behavior (e.g., cathepsin B) were among those for which differential gene expression was detected.

Functions of GASCs
The interaction between "naïve" MSCs and tumor cells remains poorly understood [30] but experimental evidence in favor of a tumor growth-enhancing role of GASCs has been obtained. We found that GASCs isolated from the peritumoral region of GBs had tumor-promoting effects on human GB cell lines in vitro and in vivo [22]. The subcutaneous injection of these cells, together with U87MG GB cells, into nude mice resulted in the induction of tumors larger than those induced by the injection of U87MG cells alone or together with control stromal cells obtained from non-GB peripheral brain tissues [22]. Other studies have shown that GASCs isolated from fresh human glioma specimens can promote the proliferation and self-renewal of tumor-initiating GSCs in vitro, and enhance GSC tumorigenicity in intracranial models in vivo [3,16]. These effects may be mediated by both the secretion of soluble growth-promoting factors, such as IL-6, and the exosomal delivery of specific oncogenic miRNAs [2,3,19].
GASCs also have angiogenic properties. A comparison of GASCs from the GB peritumoral zone with control stromal cells derived from non-GB peripheral brain tissues, based on iTRAQ labeling and mass spectrometry, showed that GASCs overproduced several proteins involved in the promotion of tumor angiogenesis or in blood vessel development, including CSPG4/NG2, CRYAB, CNN1, CALD1, and VASP [23]. Furthermore, the secretion of angiogenesis factors, such as SDF-1/CXCL12, and HGF, was upregulated in GASCs. Consistent with the overproduction of these proteins by GASCs, the inoculation of nude mouse striatum with both these cells and U87MG cells promoted angiogenesis, leading to an increase in the number of small vessels [23]. Similarly, Kong et al. [16] observed enhanced microvessel formation in mice receiving injections of GSCs cocultured with GASCs relative to mice receiving injections of GSCs cultured alone. GB-conditioned medium may also induce GASC differentiation into pericytes and enhance the attachment of these cells to tube-like vessels formed by human umbilical vein endothelial cells on Matrigel, stabilizing capillary-like structures in vitro [5]. Two subpopulations of GASCs differing in terms of CD90 expression were recently sorted from fresh glioma tissues (WHO II-IV gliomas) [6,11]. Both in vivo and in vitro experiments have shown that CD90 high GASCs significantly promote glioma cell growth, and that CD90 low GASCs promote angiogenesis via pericyte transition [6].

Heterogeneity of GASCs
Taghipour et al. [14] found that GASCs isolated from newly diagnosed LGGs and HGGs had different proteomic profiles. Proteins associated with mesenchymal cells (vimentin and transglin), and with tumor aggressiveness with potential secretory behavior (e.g., cathepsin B) were among proteins for which differential expression was observed. GASCs isolated from LGGs and HGGs also differed in terms of cancer cell adhesion [18]. The adhesion strength between GASCs and GSCs from HGGs was significantly lower than that observed between GASCs and GSCs from LGGs. In addition to the differences between the GASCs of LGGs and HGGs, GASCs may also differ between gliomas with the same histopathological classification. As indicated above, two subpopulations of GASCs differing in terms of CD90 expression have been sorted from the same glioma tissues, with CD90 low GASCs more abundant than CD90 high GASCs [6,11]. We have shown that by analyzing the transcriptome and methylome of GASCs from GB-free surgical margins and control stromal cells derived from non-GB peripheral brain tissues, that two surgical margin microenvironments can be encountered in GB patients: a surgical margin microenvironment containing GASCs with procarcinogenic properties, and another containing GASCs without such properties [24]. Thus, the genetic background of tumor cells may be a key determinant of the GASC-tumor relationship. Consistent with this hypothesis, we observed that the migration pattern of MSCs and their effect on glioma cell growth depended on the tumor cell line used [40]. Similarly, Breznik et al. [41] demonstrated that MSC/GB crosstalk differed between two established GB cell lines, U87 and U373, with MSCs inhibiting invasion by U87 cells but enhancing that by U373 cells. Differential gene expression between U87 and U373 cells may account for the different responses of these cell lines to MSCs.

GASCs: A Prognostic Marker for Gliomas
The deletion of both chromosome 1p and 19q, O-methylguanine DNA methyltransferase (MGMT) promoter methylation, and mutations of the isocitrate dehydrogenase (IDH) 1 and 2 genes are the principal prognostic factors for gliomas identified to date [42,43]. Several studies have reported that GASCs may be another prognostic factor for gliomas. For example, Bourkoula et al. [19] defined a GASC score for predicting the prognosis of human LGGs. This score was based on the levels of nine surface proteins: three stem cell antigens (CD271, CD133, and ABCG2), three adhesion proteins (CD49a, CD49d, and E-Cadherin), and three mesenchymal markers (CD90, CD73, and CD105). Specifically, GASCs obtained from LGG patients with poor prognosis were characterized by higher levels of stem cell-related markers, integrin downregulation, and a variable modulation of mesenchymal markers. Moreover, in multivariate analysis with various covariates, including IDH mutations, 1p/19q co-deletions, and MGMT promoter methylation, GASC score was the only independent predictor of overall survival (OS) and malignant progression-free survival (PFS) for LGG patients. Ius et al. [26] used next-generation sequencing to compare the gene expression profiles of GASCs isolated from LGGs with a good prognosis, and GASCs isolated from LGGs undergoing rapid anaplastic transformation. They identified an NF-κB signature composed of 14 genes that was able to predict OS in a dataset for 530 newly diagnosed patients with diffuse LGG from The Cancer Genome Atlas (TCGA). Moreover, the levels of the NF-kB-p65 protein in the nucleus, assessed by an immunohistochemical method, were found to be an independent predictor of both OS and malignant PFS in 146 grade II LGG patients. He et al. [25] showed that CXCL14 is overexpressed in GASCs and predicts clinical outcome. They found that CXCL14 expression was negatively correlated with OS in patients with glioma in the TCGA dataset. Yoon et al. (2016) [15] were able to isolate GASCs from 58.5% of patients with GB via primary culture. They found that the GB patients from whom they were able to isolate GASCs had poorer survival than those from whom no GASCs were isolated. Similarly, Shahar et al. [4] analyzed GASCs from surgical HGG specimens, based on coexpression of the MSC markers CD105, CD73, and CD90, and determined the fraction of triple-positive cells in the fresh tumor mass or in cultured tumors. They found that the percentage of GASCs in gliomas was variable between tumors and that patients with high percentages of GASCs in their tumors (fresh or cultured) had a poorer OS than those with a low percentage of these cells. All these data highlight the potential utility of GASCs as a prognostic marker for human LGGs and HGGs.

GASCs and Cellular Models
The development of new models of human glioma providing new insight into tumor biology and improving the prediction of response to treatment in patients is challenging. GSCs are difficult to isolate from LGGs. Ius et al. [26] showed that patient-derived GASCs could potentially be used instead, for the identification of novel LGG prognostic/predictive biomarkers. As described above, they demonstrated the potential of an NF-κB signature extrapolated from their GASC study for predicting LGG prognosis. Methods for constructing self-organizing three-dimensional (3D) coculture systems, known as organoids, have been developed, to mimic in vivo tumors. Hermida et al. [44] described 3D bioprinting methods using bioinks based on a modified alginate for the preparation of tumor models incorporating tumor and stromal cells from GB. They observed that the bioprinting of GSCs, together with both patient-derived GASCs and human microglia, had no adverse effect on the viability of these cell types. The use of such 3D-bioprinted GB cell constructs is promising for the future preclinical drug sensitivity testing and studies of the tumor microenvironment.

GASCs: A Future Treatment Target or Therapeutic Tool?
Given the tumor growth-enhancing role of GASCs, these cells are promising new targets for glioma treatment. However, additional studies are required to identify candidate target molecules. We recently turned this problem around by using the properties of these cells to trap tumor cells, rather than targeting them directly [45]. The concept of tumor cell traps, which emerged from the ecological trap strategy, was developed for brain tumors as a means of attracting the residual cancer cells surrounding the surgical cavity to the cavity, where they are trapped within a biomaterial support that can be targeted by treatment, such as stereotactic radiosurgery [46,47]. As indicated above, we and others have shown that GASCs increase the invasiveness of GB cells [10,12,13,17,24,25]. Based on this knowledge, we used GASC-conditioned medium as a source of chemoattractants to load the trap. We chose a bacterial cellulose (BC) scaffold as the trap matrix, because cellulose-based materials are already used in clinical practice, as exemplified by Surgicel ® , which is widely used in neurosurgery due to its hemostatic effects and high tissue compatibility. BC is also highly flexible, and is therefore easy to introduce into the tumor bed after resection, and its visibility on MRI may facilitate stereotactic radiosurgery. We found that the structure of BC membranes, with their random assembly of nanofibrils, was ideal for the trapping of tumor cells, which, once attached to the surface of the membrane, were unable to move, pass through the membrane or escape, even in the presence of an attractive medium in close proximity [45]. We also demonstrated that BC membranes loaded with GASC-conditioned medium could release and attract tumor cells in vitro [45]. Advanced methods should be developed to transform this trap into a chemotaxis device for the diffusion of chemoattractants over large distances and at high enough concentrations to establish a concentration gradient extending into the surrounding environment, for the trapping of GB cells infiltrating tissues several centimeters away from the resection cavity.

Conclusions
All these data indicate that GASCs are important stromal cells in the microenvironment of gliomas that should not be ignored (Figure 1). Their close proximity to blood vessels may provide a permissive environment, facilitating angiogenesis, invasion and tumor growth. Additional studies on the impact of GASCs on the response to treatments for gliomas, such as radiotherapy, chemotherapy, anti-angiogenic therapy, and laser interstitial therapy, are urgently required.
on the impact of GASCs on the response to treatments for gliomas, such as radiotherapy, chemotherapy, anti-angiogenic therapy, and laser interstitial therapy, are urgently required. LGGs and HGGs, alongside blood vessels, immune cells and neuroglial cells. They may be recruited from local brain sources or from the bone marrow, and are mostly found around blood vessels: arrows indicate S100A4 + cells in a GB region; scale bar = 50 μm. Cultured GASCs have properties similar to those of MSCs, such as adherence to plastic (scale bar = 100 μm), expression of surface antigens characteristic of MSCs (CD73, CD90, CD105), mesenchymal differentiation potential and a lack of tumorigenesis potential. They also have phenotypic and functional properties in common with the CAFs described in the stroma of carcinomas. In particular, GASCs express markers associated with CAFs, including FSP1/S100A4, and have tumor-promoting effects mediated by the secretion of soluble factors and exosomes. These cells are of potential interest for prognostic and therapeutic applications.   LGGs and HGGs, alongside blood vessels, immune cells and neuroglial cells. They may be recruited from local brain sources or from the bone marrow, and are mostly found around blood vessels: arrows indicate S100A4 + cells in a GB region; scale bar = 50 µm. Cultured GASCs have properties similar to those of MSCs, such as adherence to plastic (scale bar = 100 µm), expression of surface antigens characteristic of MSCs (CD73, CD90, CD105), mesenchymal differentiation potential and a lack of tumorigenesis potential. They also have phenotypic and functional properties in common with the CAFs described in the stroma of carcinomas. In particular, GASCs express markers associated with CAFs, including FSP1/S100A4, and have tumor-promoting effects mediated by the secretion of soluble factors and exosomes. These cells are of potential interest for prognostic and therapeutic applications.
Author Contributions: A.C. wrote the manuscript with support from P.M. All authors have read and agreed to the published version of the manuscript.
Funding: There are no funding sources.