2D and 3D in vitro assays to quantify the invasive behavior of glioblastoma stem cells in response to SDF-1α
Abstract
Invasion is a hallmark of cancer and therefore in vitro invasion assays are important tools in cancer research. We aimed to describe in vitro 2D transwell assays and 3D spheroid assays to quantitatively determine the invasive behavior of glioblastoma stem cells in response to the chemoattractant SDF-1α. Matrigel was used as a matrix in both assays. We demonstrated quantitatively that SDF-1α increased invasive behavior of glioblastoma stem cells in both assays. We conclude that the 2D transwell invasion assay is easy to perform, fast and less complex whereas the more time-consuming 3D spheroid invasion assay is physiologically closer to the in vivo situation.
METHOD SUMMARY
We describe 2D transwell invasion assays and 3D spheroid invasion assays for the investigation of effects of the chemoattractant SDF-1α on human glioblastoma stem cells in vitro in a quantitative manner using image analysis. In both in vitro invasion assays, Matrigel was used as the matrix that glioblastoma stem cells invade. The 2D assay is easy to perform, fast and less complex whereas the 3D assay models the in vivo situation more closely.
Invasive behavior is a hallmark of cancer and a major reason why the disease is hard to eradicate, causing poor patient prognosis [1–3]. Invasive cancer cells migrate through the extracellular matrix (ECM), enabled by the secretion of proteases to degrade the ECM and alterations in the cytoskeleton [4–6].
Multiple in vitro assays and models are available to study cellular migration and invasion, such as the scratch assay, 3D bioscaffolds and microfluidic co-cultures [3,7–11]. Frequently used invasion assays are the transwell invasion assay (2D invasion assay) [3,8,12,13] and the spheroid invasion assay (3D invasion assay) [5,12,14–16]. The latter two types of invasion assays are discussed in this article. Both types of in vitro invasion assays are relatively easy to perform, fast, reproducible and cheap. Although the 3D invasion assay is more complex than the 2D invasion assay, it is physiologically more relevant because it represents the in vivo situation in solid tumors more accurately. In 3D invasion assays, cancer cells not only invade through a matrix, but are also affected by other cancer cells in their direct surroundings. Cancer cells need to change and modify their surrounding environment in order to invade, which is also the case in the in vitro invasion assays [1,17]. In vitro invasion assays are generally preferred over in vivo invasion experiments using fluorescently-labeled cells and intravital imaging in animal models because of ethical concerns and the high costs of animal models. Furthermore, the in vitro invasion assays are more controllable, more flexible and easier to modify [1,18,19].
In this article, we describe the in vitro 2D transwell invasion assay and the 3D spheroid invasion assay in detail with the use of a brain tumor patient-derived glioblastoma stem cell (GSC) line. GSCs have been used instead of differentiated glioblastoma cells, because they have a higher invasive potential than differentiated glioblastoma cells [20,21]. The presence of cell surface protein L1CAM and the intracellular PI3K/AKT and Notch signaling pathways are involved in the highly elevated invasive potential of GSCs compared with differentiated glioblastoma cells [21,22]. Therefore, GSCs are a suitable cell type to demonstrate cellular invasion. GSCs as well as differentiated glioblastoma cells express CXCR4, a specific receptor for the chemoattractant SDF-1α, which we found to be highly expressed in hypoxic peri-arteriolar GSC niches in glioblastoma patient samples where GSCs are maintained in the tumors and protected from chemotherapy and radiotherapy because of their quiescence [23–25]. Interactions between SDF-1α and CXCR4 have been reported to be involved in cellular invasive behavior in glioblastoma [26–29] and the SDF-1α–CXCR4 axis enables invasion of CXCR4-positive glioblastoma cells into SDF-1α-rich niches and transformation into GSCs [23–25].
In the 2D and 3D invasion assays, Matrigel is used as a matrix. In this study we aimed to quantitatively determine the invasive cellular behavior of CXCR4-expressing GSCs in the presence and absence of SDF-1α. In addition, we aimed to assess the advantages and disadvantages of both types of in vitro invasion assays.
Materials & methods
Cell culture
NCH421k GSCs [30,31] were a generous gift from Christel Herold-Mende (Heidelberg University, Heidelberg, Germany) and were cultured as non-adherent 3D spheroids in Neurobasal™ medium (Gibco, Life Technologies, CA, USA) containing 1% penicillin/streptomycin (Sigma, MO, USA), 1% L-glutamine (Sigma), 2% B27 (Gibco), 0.08% bFGF (Gibco), 0.01% EGF (Gibco) and 0.01% heparin (Sigma) at 37°C in a 5% CO2 incubator.
2D transwell invasion assay
The method of the transwell invasion assay is described step by step, to be performed chronologically to reproduce the experiment. NCH421k GSCs growing as 3D spheroids were mechanically resuspended to obtain single cells. Transwell invasion assays as shown in Figure 1 were performed using inserts with 8.0-μm pores (Corning Life Sciences, NY, USA). NCH421k GSCs (80,000 cells/insert) were used that express CXCR4 as described elsewhere [12,24]. Cells were mixed in 0.5 mg/ml Matrigel (Corning) and plated in inserts in a total volume of 50 μl, and the inserts were then placed in 24-well plates (Corning). After incubation for 30 min at 37°C in a 5% CO2 incubator, 50 μl complete Neurobasal medium was added to the inserts to obtain a total volume of 100 μl. For the experimental condition, 600 μl complete Neurobasal medium containing 10 ng/ml SDF-1α (catalog #: 300-28A; Peprotech, NJ, USA) was added to the bottom wells. As a control, complete Neurobasal medium without SDF-1α was added to the bottom wells. After 72 h, invaded cells that accumulated on the bottom surface of the insert membranes were fixed with 4% paraformaldehyde (Merck, Darmstadt, Germany) for 20 min at room temperature, washed with 1× phosphate-buffered saline (PBS; Gibco) for 10 min and the cell nuclei were stained with the DNA dye 4′,6-diamidino-2-phenylindole (DAPI; Sigma) for 5 min, followed by a washing step with 1× PBS for 10 min. After removing the cells and Matrigel out of the inserts with cotton swabs, the membranes were cut out from the inserts with a scalpel and mounted on microscopic slides using Prolong Gold mounting medium (Life Technologies, CA, USA) for quantification of invaded DAPI-positive cells (Figure 1). Images of the membranes were captured using a Nikon Eclipse Ti-E fluorescence inverted microscope and the NIS-Elements AR 4.13.04 software (Nikon, Tokyo, Japan). The total number of cells in ten individual fields per condition was counted using image analysis and the ImageJ software [32,33]. Three experiments were performed, each in triplicate.
Image analysis was performed using ImageJ software and the following tools: → Analyze → Cell counter. Cells were counted by clicking on all DAPI-positive nuclei (Figure 1).
Ki67 expression to determine cell proliferation in transwell invasion assays
In order to determine whether invasion of GSCs is affected by cell proliferation during the transwell invasion assays, transmembranes were stained using the proliferation biomarker Ki67. After fixation of the cells with 4% paraformaldehyde, cell membranes were permeabilized using PBS containing 1% bovine serum albumin and 0.1% Triton-X for 30 min at room temperature. Afterward, transmembranes were incubated with anti-Ki67 antibodies conjugated with fluorescein isothiocyanate (Miltenyi Biotec, CA, USA; catalog number: 130-117-691) in a dilution of 1:50 in PBS at room temperature for 1 h. After one washing step with PBS, transmembranes were stained with DAPI for 5 min, followed by a washing step with 1× PBS for 10 min. Transmembranes were cut out of the inserts and mounted on microscopic slides using Prolong Gold mounting medium for quantification of invaded DAPI-positive and Ki67-positive cells. Images of the transmembranes were taken using a Nikon Eclipse Ti-E inverted microscope and the NIS-Elements AR 4.13.04 software. The total number of cells in ten individual fields per transmembrane was counted using image analysis and Image J software [32,33]. Three experiments were performed, each in triplicate.
Three-dimensional spheroid invasion assay
The method of the spheroid invasion assay is described step by step, to be performed chronologically to reproduce the experiment. NCH421k GSCs were seeded in complete Neurobasal medium containing 4% methylcellulose in U-bottomed 96-well plates with 3000 cells/well (BD Biosciences, CA, USA). The 4% methylcellulose was prepared as follows: 6 g of pure methylcellulose powder (Sigma, catalog # m-0512) was autoclaved and dissolved in 250 ml pre-heated complete Neurobasal medium (60°C) for 20 min using a magnetic stirrer. Then 250 ml complete Neurobasal medium (room temperature) was added, to a final volume of 500 ml, and this solution was mixed for 2 h at 4°C. The final stock solution was aliquoted and cleared by centrifugation (5000×g, 2 h, room temperature). Only the clear highly viscous supernatant – approximately 90–95% of the stock solution – was used for the spheroid invasion assay. The cells in U-bottomed 96-well plates were centrifuged at 850 × g and 31°C for 90 min and incubated at 37°C in a 5% CO2 incubator for 72 h to form a single spheroid of similar size in each well. Spheroids were transferred into 24-well plates (Corning) and embedded in 6.0 mg/ml Matrigel (Figure 2). After 30 min of incubation at 37°C in a 5% CO2 incubator, spheroids were covered with complete Neurobasal medium containing SDF-1α (10 ng/ml). As a control, complete Neurobasal medium without SDF-1α was used to cover the spheroids. Microscopical images were captured after 72 h using a Nikon Eclipse Ti inverted fluorescence microscope and NIS-Elements AR 4.13.04 software. The number of GSCs that invaded from the spheroids into the Matrigel was determined using image analysis and the NIS-Elements software (Figure 2) [32,33]. GSC spheroids with a minimum diameter of 40 μm were considered as spheroids [11]. Three experiments were performed, each in triplicate.
Statistical analyses
Data was processed in Excel 2013 (Microsoft, WA, USA) and GraphPad Prism 6 (GraphPad, CA, USA) for statistical analyses using one-way analysis of variance; p-values lower than 0.05 were considered to indicate significant differences.
Results & discussion
CXCR4-positive GSCs are attracted to SDF-1α in 2D transwell invasion assays
Transwell invasion assays were performed using CXCR4-positive NCH421k GSCs in Matrigel-containing medium in inserts in the presence or absence of the chemoattractant SDF-1α in the bottom wells. Imaging of DAPI-positive GSCs on the bottom surface of insert membranes was performed at 72 h (Figure 3). Our data demonstrate a significant (sevenfold) increase in the average number of invading GSCs in the presence of SDF-1α (Figure 3B & C) as compared with the control condition, where empty medium was added to the bottom wells (Figure 3A & C). Only cells that had accumulated on the bottom surface of insert membranes were included as no cells were found in the bottom wells.
CXCR4-positive GSCs invade into Matrigel in the presence of SDF-1α in 3D spheroid invasion assays
In the 3D invasion assay, spheroids of CXCR4-positive NCH421k GSCs were embedded in Matrigel and covered with medium with or without SDF-1α (Figure 4). At 72 h, imaging was performed and the number of GSCs that invaded out of the spheroid was quantified. Our data demonstrate that the presence of SDF-1α resulted in a 1.8-fold higher number of GSCs invading from the spheroid into the Matrigel-containing medium (Figure 4B & C), as compared with the control condition (Figure 4A & C).
Proliferation of GSCs is not affected during invasion assays
To determine whether or not proliferation of GSCs is affected during invasion assays, antibodies against the proliferation biomarker Ki67 were used to detect proliferating cells on the membrane of inserts in 2D transwell invasion assays (Figure 5A). Our data demonstrate that proliferation of NCH421K GSCs was not affected; the percentage of Ki67-positive GSCs did not differ between control conditions and the wells where SDF-1α was added (Figure 5A). In the 3D spheroid invasion assays, we assessed that cells that invaded out of spheroids were always single cells and not clusters of cells (Figure 5B), confirming that GSC proliferation did not take place during invasion.
Taken together, our data indicate that SDF-1α attracts CXCR4-positive GSCs in 2D transwell invasion assays and induces invasion of CXCR4-positive GSCs into the Matrigel in 3D spheroid invasion assays. Proliferation is not affected during invasion of GSCs in either type of in vitro invasion assay.
In this article, we describe two quantitative assays for the evaluation of GSC invasion in vitro – the 2D transwell invasion assay and a 3D spheroid invasion assay – and assess the advantages and disadvantages of each.
Our data demonstrate that both in vitro invasion assays are feasible experimental methods to investigate the cellular invasive behavior of cancers with invasive properties, such as glioblastoma. As an example, we demonstrate that CXCR4-positive GSCs are attracted toward SDF-1α in bottom wells in the 2D transwell invasion assays (Figure 3) and in Matrigel-containing medium around spheroids in the 3D spheroid invasion assay (Figure 4). GSCs invaded out of spheroids as single cells (Figures 4 & 5B), as reported in our previous study [12]. We also demonstrate that cell proliferation is not affected in either type of in vitro invasion assay (Figure 5). This is in line with studies that reported the ‘go or grow’ hypothesis, which states that cancer cells either invade or proliferate, because cellular invasion and proliferation do not occur simultaneously [34,35].
The major advantages of the 2D transwell and 3D spheroid invasion assays for the in vitro evaluation of GSC invasion are that they are technically easy to perform, relatively cheap and reproducible [3]. They can also be applied to other types of cancer (stem) cells and are now also used for drug testing, to determine whether drugs can inhibit the invasive behavior of cancer (stem) cells and their spheroid formation ability [36–41]. Chemotaxis, such as attraction between CXCR4 and SDF-1α, can be studied in an accurate manner using both assays [12]. In both assays, the presence of SDF-1α clearly results in increased invasive behavior of CXCR4-positive GSCs (Figures 3 & 4). Another advantage of 2D transwell and 3D spheroid invasion assays is that qualitative data are obtained by imaging (differences in numbers of invading GSCs can be clearly observed microscopically) and quantitative data are obtained using image analysis, which is relatively easy to perform and accurate at the same time. Alternatively, a quantitative microscopic analysis has been described for the 3D spheroid assay in which the distance of invading cells from the spheroid or invasive area is determined [15,42].
Both in vitro invasion assays can be performed in more complicated experimental settings by using multiple cell types – for example, CXCR4-positive GSCs in the inserts and SDF-1α-producing cells in the bottom wells, such as mesenchymal stem cells (MSCs) that are known to infiltrate glioblastoma tumors and are major SDF-1α producers [24]. Compared with experiments with exogenous SDF-1α present in the bottom wells, such experiments are both more biologically relevant (because SDF-1α is produced by MSCs in GSC niches in vivo) and cheaper (fewer reagents are needed because MSCs produce SDF-1α). However, it should be noted that MSCs produce a plethora of chemokines, not only SDF-1α [12,43]. Thus, to determine whether the obtained effect was caused by MSC-secreted SDF-1α, control experiments need to be performed using selective CXCR4 inhibitors like plerixafor [44]. In the 3D spheroid invasion assays, spheroid co-cultures of different fluorescently-labeled cell types can also be used to study cancer cell invasion into the 3D matrix [14]. A disadvantage of these two invasion methodologies is that the tumor microenvironment is not accurately reproduced [3,8]. More advanced and expensive in vitro 3D invasion methods are now available, such as 3D microfluidic co-culture systems that allow the construction of a 3D tumor microenvironment, real-time cell tracking, evaluation of interactions between different cell types, mimicking of blood flow and studies of phenomena such as cell proliferation and invasion [3,8,11,45–47]. The most true-to-nature systems are the 3D tumor organoids, in which the tumor complexity (i.e., tumor structures including blood vessels, stroma, ECM) is intact and tumor heterogeneity is present. Organoids can be used for drug testing as well for the investigation of phenomena such as cellular invasive behavior [48–55]. In future research, we aim to introduce 3D tumor organoids as a model to study glioblastoma invasion and interactions between glioblastoma cells/GSCs with other cell types in the tumor microenvironment, but both 2D transwell invasion assay and 3D spheroid invasion assay will continue to be used as well because of their own advantages. The 2D transwell invasion assay is a faster method and is less complex than the 3D spheroid invasion assay. However, it is more expensive than the 3D invasion assay because of the costs of inserts. Therefore, we suggest that the 2D transwell assay is applied first to determine whether the hypothesized chemoattractant-receptor interactions are involved in the process of cellular invasion. Afterward, cellular invasion can be studied using the more time-consuming 3D spheroid invasion assays that are more complex and closer to the in vivo situation. A limitation of the present study is that both in vitro invasion assays have been performed with the use of a single GSC line only. The application of this protocol to additional patient-derived GSC lines in the future may enable the optimization of this methodology for individual cell lines.
Besides the SDF-1α–CXCR4 axis, various other signaling axes are involved in the invasive behavior of GSCs, such as the intracellular PI3K/AKT and Notch signaling pathways [21,22]. In addition, proteases such as matrix metalloproteinases (e.g., MMP-2, MMP-9 and MMP-13), membrane-type matrix metalloproteinases and cathepsins, which degrade the ECM, are also involved in increased cancer cell invasion [5,6,56–58]. Pro-invasion soluble factors secreted from non-cancerous cells such as endothelial cells and immune cells in the tumor microenvironment [25,56,59–61] have also been reported to be involved in the invasive behavior of GSCs.
In conclusion, both 2D transwell invasion assays and 3D spheroid invasion assays are accurate and reproducible in vitro methodologies to study GSC invasion and are also applicable to other types of cancer. We conclude that the 2D transwell invasion assay is faster and less complex and is therefore more suitable for the initial experimental determination of specific chemoattractant receptor-mediated cellular invasion, which can then be studied using the more time consuming and complex 3D spheroid invasion assay that is physiologically closer to the in vivo situation than the 2D transwell invasion assay.
Future perspective
The described in vitro invasion assays are excellent experimental tools for translational research to increase our understanding of glioblastoma biology and to facilitate the development of novel therapies that inhibit invasion of GSCs. The methodological principles are not exclusive for glioblastoma and can also be applied to other types of cancer (stem) cells. In addition, novel therapies designed on the basis of these methodologies can work synergistically with current therapeutic approaches such as chemotherapy, radiotherapy and/or immunotherapy.
Author contributions
Study conception: V Hira, B Breznik, R Molenaar. Experiments: V Hira, B Breznik. Analysis: V Hira, B Breznik. Manuscript drafting: V Hira, B Breznik, R Molenaar. Manuscript editing: all authors. Supervision: C Van Noorden, T Lah, R Molenaar. Funding: T Lah, R Molenaar. Final approval of the manuscript: all authors.
Financial & competing interests disclosure
This study was financially supported by the Dutch Cancer Society (KWF; UVA 2014-6839 and UVA 2016-10460). V Hira was supported by the IVY Interreg Fellowship 2018, R Molenaar was supported by the Fondation pour la Recherche Nuovo-Soldati 2019, T Lah was supported by the Slovenian Research Agency (Program P10245) and the European Program of Cross-Border Cooperation for Slovenia-Italy Interreg TRANS-GLIOMA (Program 2017) and B Breznik was supported by postdoctoral grant of Slovenian Research Agency (Project Z3-1870). The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.
No writing assistance was utilized in the production of this manuscript.
Ethical conduct of research
The study was approved by the National Medical Ethics Committee of the Republic of Slovenia (approval no. 0120-190/2018/4).
Open access
This work is licensed under the Attribution-NonCommercial-NoDerivatives 4.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-nd/4.0/
References
- 1. . 2D or 3D? How cell motility measurements are conserved across dimensions in vitro and translate in vivo. Bioeng. Transl. Med. 5(1), e10148 (2020).
- 2. . Angiogenesis in cancer and other diseases. Nature 407(6801), 249–257 (2000).
- 3. . An experimenter's guide to glioblastoma invasion pathways. Trends Mol. Med. 24(9), 763–780 (2018).
- 4. . Possible future issues in the treatment of glioblastomas: special emphasis on cell migration and the resistance of migrating glioblastoma cells to apoptosis. J. Clin. Oncol. 23(10), 2411–2422 (2005).
- 5. The regulation of cysteine cathepsins and cystatins in human gliomas. Int. J. Cancer 131(8), 1779–1789 (2012).
- 6. . Antiprotease therapy in cancer: hot or not? Expert Opin. Biol. Ther. 6(3), 257–279 (2006).
- 7. . Microregional extracellular matrix heterogeneity in brain modulates glioma cell invasion. Int. J. Biochem. Cell Biol. 36(6), 1046–1069 (2004).
- 8. . Toward 3D biomimetic models to understand the behavior of glioblastoma multiforme cells. Tissue Eng. Part B Rev. 20(4), 314–327 (2014).
- 9. . New dimensions in cell migration. Nat. Rev. Mol. Cell Biol. 13(11), 743–747 (2012).
- 10. . Bioengineered scaffolds for 3D analysis of glioblastoma proliferation and invasion. Ann. Biomed. Eng. 43(8), 1965–1977 (2015).
- 11. . Multiscale cytometry and regulation of 3D cell cultures on a chip. Nat. Commun. 8(1), 469 (2017).
- 12. Cathepsin K cleavage of SDF-1alpha inhibits its chemotactic activity towards glioblastoma stem-like cells. Biochim. Biophys. Acta 1864(3), 594–603 (2017).
- 13. . PREX1 integrates G protein-coupled receptor and phosphoinositide 3-kinase signaling to promote glioblastoma invasion. Oncotarget 8(5), 8559–8573 (2017).
- 14. . Mesenchymal stem cells differentially affect the invasion of distinct glioblastoma cell lines. Oncotarget 8(15), 25482–25499 (2017).
- 15. INSIDIA: a FIJI macro delivering high-throughput and high-content spheroid invasion analysis. Biotechnol. J. 12(10), (2017).
- 16. Evolution of the brain tumour spheroid model: transcending current model limitations. Acta Neurochir. 145(9), 819–824 (2003).
- 17. . Can anti-migratory drugs be screened in vitro? A review of 2D and 3D assays for the quantitative analysis of cell migration. Med. Res. Rev. 27(2), 149–176 (2007).
- 18. . Intravital imaging of metastasis in adult zebrafish. BMC Cancer 17(1), 660 (2017).
- 19. Intravital imaging reveals how BRAF inhibition generates drug-tolerant microenvironments with high integrin beta1/FAK signaling. Cancer Cell 27(4), 574–588 (2015).
- 20. Glioma infiltration of the corpus callosum: early signs detected by DTI. J. Neurooncol. 112(2), 217–222 (2013).
- 21. Elevated invasive potential of glioblastoma stem cells. Biochem. Biophys. Res. Commun. 406(4), 643–648 (2011).
- 22. . Invasive glioblastoma cells acquire stemness and increased Akt activation. Neoplasia 12(6), 453–463 (2010).
- 23. CD133+ and nestin+ glioma stem-like cells reside around CD31+ arterioles in niches that express SDF-1alpha, CXCR4, osteopontin and Cathepsin K. J. Histochem. Cytochem. 63(7), 481–493 (2015).
- 24. Similarities between stem cell niches in glioblastoma and bone marrow: rays of hope for novel treatment strategies. J. Histochem. Cytochem. 68(1), 33–57 (2020).
- 25. . The hypoxic peri-arteriolar glioma stem cell niche, an integrated concept of five types of niches in human glioblastoma. Biochim. Biophys. Acta Rev. Cancer 1869(2), 346–354 (2018).
- 26. Hypoxia-inducible factor 1 and VEGF upregulate CXCR4 in glioblastoma: implications for angiogenesis and glioma cell invasion. Lab. Invest. 86(12), 1221–1232 (2006).
- 27. . CXCR4 expression mediates glioma cell invasiveness. Oncogene 25(19), 2801–2806 (2006).
- 28. . The role of hypoxia in glioblastoma invasion. Cells 6(4), 45 (2017).
- 29. SDF-1 and CXCR4 are up-regulated by VEGF and contribute to glioma cell invasion. Cancer Lett. 236(1), 39–45 (2006).
- 30. Transmembrane protein CD9 is glioblastoma biomarker, relevant for maintenance of glioblastoma stem cells. Oncotarget 7(1), 593–609 (2016).
- 31. . Paracrine effects of mesenchymal stem cells induce senescence and differentiation of glioblastoma stem-like cells. Cell Transplant. 24(4), 631–644 (2015).
- 32. . NIH image to ImageJ: 25 years of image analysis. Nat. Methods 9(7), 671–675 (2012).
- 33. . Image cytometry: protocols for 2D and 3D quantification in microscopic images. Prog. Histochem. Cytochem. 47(4), 211–333 (2013).
- 34. The ‘go or grow’ potential of gliomas is linked to the neuropeptide processing enzyme carboxypeptidase E and mediated by metabolic stress. Acta Neuropathol. 124(1), 83–97 (2012).
- 35. . ‘Go or grow’: the key to the emergence of invasion in tumour progression? Math. Med. Biol. 29(1), 49–65 (2012).
- 36. . Cancer stem cells in neuroblastoma: expanding the therapeutic frontier. Front. Mol. Neurosci. 12, 131 (2019).
- 37. Sphere-formation assay: three-dimensional in vitro culturing of prostate cancer stem/progenitor sphere-forming cells. Front. Oncol. 8, 347 (2018).
- 38. . Drug repurposing towards targeting cancer stem cells in pediatric brain tumors. Cancer Metastasis Rev. 39(1), 127–148 (2020).
- 39. The Akt/mTOR pathway in cancer stem/progenitor cells is a potential therapeutic target for glioblastoma and neuroblastoma. Oncotarget 9(71), 33549–33561 (2018).
- 40. . Medulloblastoma cancer stem cells: molecular signatures and therapeutic targets. J. Clin. Pathol. 73(5), 243–249 (2020).
- 41. Metformin and Ara-a effectively suppress brain cancer by targeting cancer stem/progenitor cells. Front. Neurosci. 9, 442 (2015).
- 42. . Post-translational regulation of cathepsin B, but not of other cysteine cathepsins, contributes to increased glioblastoma cell invasiveness in vitro. Pathol. Oncol. Res. 15(4), 711–723 (2009).
- 43. . Cytokines play a key role in communication between mesenchymal stem cells and brain cancer cells. Protein Pept. Lett. 22(4), 322–331 (2015).
- 44. . CXCR4 antagonists as stem cell mobilizers and therapy sensitizers for acute myeloid leukemia and glioblastoma? Biology (Basel) 9(2), 31 (2020).
- 45. . Endothelium-induced three-dimensional invasion of heterogeneous glioma initiating cells in a microfluidic coculture platform. Integr. Biol. 9(9), 762–773 (2017).
- 46. . A novel microfluidic cell co-culture platform for the study of the molecular mechanisms of Parkinson's disease and other synucleinopathies. Front. Neurosci. 10, 511 (2016).
- 47. A versatile valve-enabled microfluidic cell co-culture platform and demonstration of its applications to neurobiology and cancer biology. Biomed. Microdevices 13(3), 539–548 (2011).
- 48. A patient-derived glioblastoma organoid model and biobank recapitulates inter- and intra-tumoral heterogeneity. Cell 180(1), 188–204.e22 (2020).
- 49. . Modeling development and disease with organoids. Cell 165(7), 1586–1597 (2016).
- 50. Genetically engineered cerebral organoids model brain tumor formation. Nat. Methods 15(8), 631–639 (2018).
- 51. . Glioblastoma model using human cerebral organoids. Cell Rep. 23(4), 1220–1229 (2018).
- 52. . Spontaneous glioblastoma spheroid infiltration of early-stage cerebral organoids models brain tumor invasion. SLAS Discov. 23(8), 862–868 (2018).
- 53. Modeling patient-derived glioblastoma with cerebral organoids. Cell Rep. 26(12), 3203–3211.e5 (2019).
- 54. A three-dimensional organoid culture system derived from human glioblastomas recapitulates the hypoxic gradients and cancer stem cell heterogeneity of tumors found in vivo. Cancer Res. 76(8), 2465–2477 (2016).
- 55. . Applications of human brain organoids to clinical problems. Dev. Dyn. 248(1), 53–64 (2019).
- 56. . Molecular and microenvironmental determinants of glioma stem-like cell survival and invasion. Front. Oncol. 7, 120 (2017).
- 57. Cancer stem-like cells of glioblastoma characteristically express MMP-13 and display highly invasive activity. Int. J. Oncol. 37(5), 1121–1131 (2010).
- 58. . The role of gelatinases in colorectal cancer progression and metastasis. Biochim. Biophys. Acta 1705(2), 69–89 (2004).
- 59. . Tumor cell invasion in glioblastoma. Int. J. Mol. Sci. 21(6), 1932 (2020).
- 60. . Tumor microenvironment tenascin-C promotes glioblastoma invasion and negatively regulates tumor proliferation. Neuro Oncol. 18(4), 507–517 (2016).
- 61. . CXCL12 modulation of CXCR4 and CXCR7 activity in human glioblastoma stem-like cells and regulation of the tumor microenvironment. Front. Cell. Neurosci. 8, 144 (2014).