Epithelial to mesenchymal transition (EMT) allows cancers to become invasive. To investigate EMT, a neural stem cell (NSC)-based in vitro model devoid of serum and enzymes is described. This standardized system allows quantitative and qualitative assessment of cell migration, gene and protein expression. The model is suited for drug discovery.
Epithelial to mesenchymal transition (EMT) describes the process of epithelium transdifferentiating into mesenchyme. EMT is a fundamental process during embryonic development that also commonly occurs in glioblastoma, the most frequent malignant brain tumor. EMT has also been observed in multiple carcinomas outside the brain including breast cancer, lung cancer, colon cancer, gastric cancer. EMT is centrally linked to malignancy by promoting migration, invasion and metastasis formation. The mechanisms of EMT induction are not fully understood. Here we describe an in vitro system for standardized isolation of cortical neural stem cells (NSCs) and subsequent EMT-induction. This system provides the flexibility to use either single cells or explant culture. In this system, rat or mouse embryonic forebrain NSCs are cultured in a defined medium, devoid of serum and enzymes. The NSCs expressed Olig2 and Sox10, two transcription factors observed in oligodendrocyte precursor cells (OPCs). Using this system, interactions between FGF-, BMP- and TGFβ-signaling involving Zeb1, Zeb2, and Twist2 were observed where TGFβ-activation significantly enhanced cell migration, suggesting a synergistic BMP-/TGFβ-interaction. The results point to a network of FGF-, BMP- and TGFβ-signaling to be involved in EMT induction and maintenance. This model system is relevant to investigate EMT in vitro. It is cost-efficient and shows high reproducibility. It also allows for the comparison of different compounds with respect to their migration responses (quantitative distance measurement), and high-throughput screening of compounds to inhibit or enhance EMT (qualitative measurement). The model is therefore well suited to test drug libraries for substances affecting EMT.
During several stages of embryonic development, epithelial cells lose their strong adherence to each other (e.g., tight junctions) and acquire a migratory phenotype in a process called epithelial to mesenchymal transition (EMT)1. EMT is required for the formation of additional cell types, such as the mesenchymal neural crest cells, a population that segregates from the neuroepithelium2. EMT is not only essential during embryonic stages but also required at later stages of adult life to maintain physiological processes in the adult organism, such as wound healing3and central nervous system (CNS) regeneration in demyelinating lesions4.
Epithelial tumors are known to reactivate EMT as an initiation step for migration, invasion and metastasis, ultimately leading to cancer progression1,3. EMT is indeed centrally linked to strong migration1,3. The cellular steps of conditioning, initiating, undergoing and maintaining EMT are not fully understood and need further investigation.
Here, a standardized in vitro EMT model system based on NSCs, with defined growth factors and media (no serum and no enzyme usage) is presented. This model system is of relevance for scientists working on EMT. The Snail, Zeb and Twist protein families have been shown to be critical for EMT both in development and disease1. The Snail, Zeb and Twist families are also involved in the presented system. The system is based on a specific region of the forebrain that normally does not undergo EMT providing a particular advantage for the study of initial events during EMT induction.
The model system could potentially be applied to study EMT in epithelia outside the CNS, since key EMT inducers, such as the Snail, Zeb and Twist proteins, are also found during EMT in tissue systems outside the CNS. This model system allows the standardized isolation of NSCs from the developing cortex to study stem cell features in general and EMT in particular. Using this system, we isolated NSCs, induced EMT and studied the subsequent migration under the effect of FGF2 and BMP4. We observed that FGF- and BMP-signaling interacts with TGFβ-signaling to promote cell migration, thus validating the model system.
All animal procedures followed the 'Guide for the Care and Use of Laboratory Animals' (NIH publication, 8th edition, 2011) and were approved by the Animal Welfare Committee of Basel (Swiss Guidelines for the Care and Use of Animals). By these guidelines the animal protocol is considered of "lowest animal severity grade".
1. Preparation of Expansion Medium
Note: Work in aseptic conditions as standard for tissue culture.
2. Preparation of Passaging Medium
3. Preparation of Growth Factors
4. Coating of Cell Culture Dishes
5. Standardized Dissection and Preparation of the Cortical Subventricular Zone (SVZ)
6. Preparation of Explant Cultures
7. Preparation of NSC Single Cell Culture
8. Migration Assessment
9. Invasion Assessment
This EMT model system is based on the standardized isolation of NSCs both as single cells or as explants from a specific region of the developing neural tube, the central cortex (Figures 1 and 2). For quantitative assessment, explants were seeded right at the center of a 500 µm grid culture dish (Figure 3). Explants from the central cortex were first exposed to FGF2 for two days, followed by additional two days in different combinations of growth factors (Table 1). We started with the cortex which is larger than other regions of the developing neural tube. We observed that only explants cultured in BMP4 showed a migratory response (Table 1 and Supplementary Figure 2).
Next podoplanin (PDPN) expression was analyzed under defined conditions. PDPN is a transmembrane sialomucin-like protein that has been associated with invasion in multiple cancers11,12 and also in NSCs8. The proportion of PDPN expressing cells is increased in high-grade gliomas13. PDPN is also associated with poorer survival in glioblastoma patients14. Further, PDPN has been shown to be a marker of malignant progression in multiple tumors including breast, lung, colon carcinomas15,16. PDPN is expressed both in invasive as well as migratory NSCs8. Using the above protocol, PDPN expression was compared to control, FGF2 only, in explants exposed to TGFβ1 and BMP4 alone or in combinations. PDPN expression was detected in all BMP4 and TGFβ1/BMP4-treated explants. In contrast, no PDPN was observed in the control explants in FGF2 alone or in explants with TGFβ1 alone (Table 2 and Figure 4). In addition cells with a migratory phenotype were induced in BMP4 and TGFβ1/BMP4-exposed explants, whereas the control explants (only FGF2) and TGFβ1 alone did not show migratory cells (Table 1). The observation of migratory cells provides a low cost, straight-forward qualitative assessment.
Further, we tested the migratory response and EMT induction of several regions of the developing neural tube to BMP4 (Table 2). 400 µm explants were prepared from the central cortex, the posterior cortex (labeled posterior pole), the MGE and the mesencephalon, as indicated in Figure 1. The explants were all exposed to FGF2 for two days, followed by two days in FGF2 with BMP4. Migration was defined by the appearance of flat migratory cells with a leading and trailing edge (Figure 5 and Supplementary Figure 2). A strong induction of migratory cells from the posterior cortex and an intermediate response from the MGE and the mesencephalon was observed (Table 2). 145 of 146 explants of the central cortex showed a migratory response in FGF2/BMP4. Thus, the central cortex demonstrated the most robust induction of migratory cells of all explants tested (Table 2). Omission of BMP4 abolished any migratory response (Table 1 and 2, and data not shown).
For quantitative assessment of growth factor potency, migration distance was measured in explants cultured in multiple growth factors. Explants were cultured as above for two days in FGF2, then for additional two days in FGF2 without factors (control) or single or combined BMP4 and TGFβ1. In the control explants a strong proliferation was observed in response to FGF2, as expected. The explants are derived from the cortical SVZ and contain to a large part NSCs which are known to proliferate in response to FGF217. The control cells reached 689 ± 14, the TGFβ1 cells 582 ± 49 µm (Figure 5.). The BMP4-group showed a mean migration distance of 935 ± 91 µm. In comparison the TGFβ1/BMP4-cells migrated significantly further, to 1,150 ± 23 µm (Figure 5). The results show that BMP4 is inducing a migration in FGF2-exposed NSCs. This migration can be further enhanced by TGFβ1. The combination of FGF2, BMP4 and TGFβ1 is therefore the most effective to induce migration. In summary, the cell culture model system allows both qualitative as well as quantitative migration assessment.
EMT has been linked to transcription factors (TFs), such as Snail1, Snail2, Zeb1, Zeb2, Twist1, Twist2 in various systems including epithelial cancers, such as breast cancer, colon cancer, lung cancer and also in brain tumors1,18. We had previously shown that FGF2 and BMP2 or BMP4 can induce Snail1 and Snail28,9. Using the above system we investigated the expression levels of the other EMT-related TFs. No changes in Twist1 expression could be detected; with Twist1 being at low expression levels during FGF2 exposure (data not shown). In the cell culture model an upregulation of Zeb1 and Zeb2 during FGF2-exposure and of Twist2 during FGF2/BMP4-exposure was observed (Figure 6).
NSCs are known to contribute to the population of oligodendrocyte precursor cells (OPCs)8,19,20. Several lines of evidence indicate that NSCs and in particular OPCs may be the cell of origin for gliomas21,22. To characterize the above model further, we investigated the expression of OPC markers. We observed that FGF2-exposed NSCs demonstrated co-expression of Olig2/Nestin and Sox10/Nestin in more than 90% (Figure 7). This observation demonstrates that the NSCs show OPC-features in our system. Indeed, the OPC-features correlated with upregulation of Zeb'1 and Zeb'2 (Figures 6 and 7). These results suggest that FGF2-expansion infers an OPC identity, as judged by Olig2 and Sox10, while at the same time initiates first Zeb-based steps of EMT.
Figure 1: Standardized Embryo and Forebrain Dissection for NSC Isolation and EMT Induction. (A) Left forelimb of rat E14.5 embryo after caesarean section. Every fore limb digit is marked with a black arrow tip. Note the beginning digit formation typical for E14.5. (B) Rat E14.5 embryo after hull removal. Note the hash (#) at the dorsal diencephalon which is ideal to start removal of the skin/skull anlage. (C) The skin/skull is already mostly removed. The border between removed and unremoved skin is marked with two triangular arrowheads. The skin can be peeled anteriorly from the anterior arrow and posteriorly from the posterior arrow. (D) The skin/skull is now completely removed. Note the notch of the midbrain-hindbrain boundary (arrow) and the cut just caudally to it, marked with an arrowhead. (E) The telencephalon-diencephalon-mesencephalon (TEL-DI-MES) block is removed from the rest of the embryo. (F) Superior view of TEL-DI-MES block. Cranial is left. (G) Oblique view from above. (H) Inferior view of TEL-DI-MES block. (I) The mesencephalon and part of the diencephalon are separated from the telencephalon with the anterior part of the diencephalon. Top: Anterior view of both telencephalic vesicles. Bottom: Right lateral view of mesencephalon, top faces cranially. (J) Top: both telencephalic vesicles without diencephalon. Inferior view to illustrate complete removal of diencephalon. Asterisk depicts MGE. Bottom: anterior part of diencephalon. (K) Top: Both hemispheres are separated in the interhemispheric fissure. Left hemisphere on left side. Asterisk depicts MGE. Plus sign depicts LGE. (L) Magnification of left hemisphere with median view. Right faces cranially, left caudally, top is dorsal, bottom ventral, respectively. Asterisk (*) is located at MGE. Plus sign (+) is located at LGE. ch, cortical hem; FL, forelimb; di, diencephalon; mes, mesencephalon; MHB, midbrain-hindbrain boundary; HL, hind limb; LGE, lateral ganglionic eminence; MGE, medial ganglionic eminence; ob, olfactory bulb; pc, posterior cortex; rho, rhombencephalon; tel, telencephalon. The grid on every image shows 2 mm thick lines with four intersection with thin lines every 400 µm. Scale bar on all images: 2 mm. Please click here to view a larger version of this figure.
Figure 2: Standardized Cortical SVZ dissection for NSC Isolation and EMT Induction. (A) Median view of left hemisphere. Through the foramen of Monro the MGE and LGE are visible, marked by asterisk and plus sign. (B) The olfactory bulb is removed just cranially of the MGE-LGE. (C) The posterior pole of the cortex is removed just behind the CGE. (D) Top: the cortical hem is separated from the cortex. Left: posterior pole. Middle: complex containing the central portion of the cortex and the CGE-LGE-MGE block. MGE and LGE face to the right. Right: olfactory bulb. (E) The CGE-LGE-MGE-cortex block is flipped horizontally. Now the outer surface of the telencephalon is facing the microscope. MGE and LGE now face to the left. (F) The reddish meninges are removed from the bright translucent cortex. The CGE-MGE-LGE-ctx block is now flipped back horizontally for the next step. (G, H) The same procedure is repeated with the right hemisphere. The central cortex is separated away from the CGE-MGE-LGE block. Note that there is an intermediate cortex zone that is left at the CGE-MGE-LGE block, marked with two arrows. This intermediate cortex thickness corresponds to half of the diameter of the LGE. Dotted line representing the border between LGE-CGE and the cortex. (I) The central cortex is separated into explants of less than 400 µm diameter. Scale bar: 2 mm; grid on every image shows 2 mm thick lines with four intersections (thin lines) every 400 µm. Asterisk (*) is located at MGE. Plus sign (+) is located at LGE. Abbreviations: ant-ctx, anterior cortex; cen-ctx, central cortex; cge, caudal ganglionic eminence; ch, cortical hem; cp, choroid plexus; ctx, cortex; LGE, lateral ganglionic eminence; men, meninges; MGE, medial ganglionic eminence; ob, olfactory bulb; pc, posterior cortex. Please click here to view a larger version of this figure.
Figure 3: Explant Seeding onto 35 mm Grid Cell Culture Dishes. Place 1 ml of expansion medium at the center of the 500 µm grid dish. The drop is contained by the grid rim (small arrows). Place explants right at the center of the 1 ml drop. If the explants are spread outside the grid, swirl the dish in a slow circular motion and the explants will move to the center by centripetal force. Note: the explants are only barely visible by eye (black arrow heads). Scale bar: 35 mm.
Figure 4: PDPN is Induced in the Presence of BMP4. Explants were cultured according to the protocol section 8 (2 days in FGF2 only, then followed by FGF2 with factors as indicated). Control (A) (FGF2 alone) and TGFβ1 (C) explants did not contain PDPN-positive cells (green). Nuclei are stained with DAPI (blue). BMP4 (B) and TGFβ1/BMP4 (D) explants showed a high proportion of PDPN-positive cells. The explant center is at the left, the periphery at the right. Note: The PDPN-positive cells were mostly found at the periphery and not in the center of the explant. The TGFβ1/BMP4 explants contained flatter, more elaborated PDPN-positive cells. (E) The percentage of PDPN-positive cells in explants at different conditions is represented. Control and TGFβ1 both are significantly different from the conditions with BMP4. Means are shown ± SEM. Control vs. TGFβ1 is not significant (ns). Control and TGFβ1 vs. BMP4: p < 0.0001 (***). TGFβ1 vs. TGFβ1/BMP4: p < 0.0001 (***). BMP4 vs. TGFβ1+BMP4 is not significant: p = 0.0981 (ns). Control vs. TGFβ1+BMP4: p < 0.0001 (***). Scale bar for all images, as illustrated in (A): 50 µm. Please click here to view a larger version of this figure.
Figure 5: Quantitative Migration Assessment. BMP4 and TGFβ1 show Additive Effect on Cell Migration. Explants were cultured according to protocol section 8. (A) Control explant. (B) Explant in BMP4 alone. (C) Explant in TGFβ1 alone. (D) Explant in the combination of TGFβ1 and BMP4. (E) Migration distance in µm. The 500 µm grid was used as reference. The control and the TGFβ1 explants are showing a strong proliferation with a larger explant diameter than in the other conditions. Means are shown ± SEM. Note that the BMP4 and TGFβ1/BMP4 explants partly disintegrate the explant core and cells emigrate away from it centrifugally. Control vs. TGFβ1 is not significant (ns). TGFβ1 vs. BMP4: p = 0.0353 (*). Control vs. BMP4: p = 0.0351 (*). BMP4 vs. TGFβ1+BMP4: p = 0.0372 (*). Control vs. TGFβ1+BMP4: p < 0.0001 (***). Scale bar: 500 µm. Center of explant is marked by a plus sign (+). Outer edge of migrating cells is marked with triangular arrow. Please click here to view a larger version of this figure.
Figure 6: Upregulation of EMT-related Transcription Factors by qRT-PCR. EMT is linked to the key transcription factors of the Zeb- and Twist-family. (A, B) Zeb'1 and Zeb'2 were upregulated during the first phase of FGF2-exposure. (C) Twist'2 was upregulated during the second phase of FGF2/BMP4-exposure. Relative expression levels based on qRT-PCR are shown (n = 2). The mRNA levels were normalized to GAPDH. Means are shown ± SEM. At day 0 of the FGF2 period mRNA was harvested, further mRNA was harvested after four days in FGF2, then after one day in FGF2/BMP4. Zeb'1: Control vs. FGF2, p = 0.0002. Zeb'2: Control vs. FGF2, p = 0.0206. Twist'2: Control vs. FGF2+BMP4, p = 0.003.
Figure 7: NSCs show OPC Characteristics in the Model System. NSCs were isolated from the central cortex and cultured as single cells in the presence of FGF2 (according to Section 5). After 8 d in culture (passage after 4 d according to Section 7.9) the cells formed small NSC colonies and were stained for Olig2 (red, in B, C), Sox10 (red, in E, F) and Nestin (green, A, C, D, F). A high proportion of cells co-expressed Olig2/Nestin (A-C) and Sox10/Nestin (D-F). (G) Co-expression of Olig2/Nestin and Sox10/Nestin is shown in percentage of all cells. Means are shown ± SEM. Scale bar for all images as illustrated in (A) and (D): 20 µm. Please click here to view a larger version of this figure.
Condition | Total explants (central cortex) | Explants with migratory response | % | Explants with Podoplanin expression | % |
Control | 33 | 0 | 0 | 0 | 0 |
TGFb1 | 21 | 0 | 0 | 0 | 0 |
BMP4 | 34 | 34 | 100 | 34 | 100 |
TGFb1+BMP4 | 22 | 22 | 100 | 22 | 100 |
Table 1. BMP4 and the Combination of TGFβ1 with BMP4 Induce a Robust Migratory Response. Explants from the central cortex were maintained in expansion medium and FGF2 for a total of four days. After the first two days, the explants either continued to receive FGF2 alone (control) or FGF2 and in addition TGFβ1, BMP4 or TGFβ1/BMP4 as combination. Control- and TGFβ1- explants did neither show a migratory response nor PDPN expression. BMP4 and TGFβ1/BMP4 induced both migratory cells as well as PDPN expression in all explants.
Region | Total explants | Explants with migratory response in FGF2/BMP4 | % |
Central cortex | 146 | 145 | 99.3 |
Posterior cortex | 52 | 48 | 92.0 |
MGE | 56 | 29 | 51.7 |
Mesencephalon | 53 | 28 | 52.8 |
Table 2. The central Cortex shows the most Robust Migration in Response to FGF2/BMP4. Explants were prepared from different regions of the developing rat E14.5 neural tube: the central cortex, the posterior cortex, the medial ganglionic eminence (MGE) and the mesencephalon. All explants were cultured identically in FGF2/BMP4 (Section 8). Explants treated without BMP4 did not show any migratory response (data not shown). All explants were screened for appearance of any migratory cells per explant. All regions were able to respond to FGF2 with BMP4 with migration. The central cortex, however, showed the most robust response. Note that BMP4 was required to induce any migratory cells. Explants cultured in FGF2 alone showed proliferation but no migration; and no flat migratory cells were observed in FGF2 alone (Table 1 and Supplementary Figure 2).
Supplementary Figure 1. 400 µm Grid File. The grid file may be printed and used as reference during the above dissection steps and as shown in Figures 1 and 2. The large intersections represent 2 mm with 400 µm subdivisions. Please click here to download a PDF version of this figure.
Supplementary Figure 2. BMP4-exposed Explants Show Flat Migratory Cells. Magnification of cells shown in Figure 5. (A) Control- (FGF2 alone) and (C) TGFβ1-explants demonstrated a strong cell division with small rounded cells. (B) BMP4- and (D) TGFβ1/BMP4-explants showed consistently flat elongated cells. Scale bar for all images, as illustrated in (A): 100 µm. Please click here to view a larger version of this figure.
Supplementary Figure 3. Summary of in vitro Model System for EMT investigation. E14.5 rat central cortex contains the NSC-containing SVZ. The central cortex is either used as explant pieces or as single cells. Explant pieces are more convenient for quantitative migration analysis (Section 8). Single cells are suited for qualitative analysis, such as gene or protein analysis (Section 9.13).
In this study a standardized system for EMT analysis utilizing NSCs is described (summarized in Supplementary Figure 3). The standardization ensures reproducibility (Table 1 and 2). The NSCs are derived from the developing cortex, a tissue that normally does not undergo EMT. This is of advantage for the analysis of early steps in EMT. Initial steps in EMT cannot be adequately studied in tumor cells that have accumulated genetic changes and may have already adopted EMT features. Moreover, primary tumor samples are not ideal to understand EMT since most malignant tumors are heterogeneous containing both invasive and non-invasive cells. The presented protocol provides EMT researchers with a novel model to study early steps of EMT induction. Here we show that the key EMT inducers of the Zeb and Twist family are sequentially activated: during the first phase Zeb1 and Zeb2 are upregulated, during the second phase Twist2 (Figure 6). Previously, we had shown that Snail1 is already upregulated several hours after BMP4-exposure8. We had also shown that upon BMP4-exposure the neuroepithelial stem cells from the central cortex differentiate into mesenchymal cells, positive for smooth muscle actin (SMA) and calponin9. The results above complete the previous study in demonstrating that all of the three known key EMT regulators are involved in the system. In addition, we observed that TGFβ1 can significantly enhance the migratory effect of FGF2/BMP4 alone. These results point to a network between FGF-, BMP- and TGFβ-signaling during EMT induction.
The most critical steps for successful isolation of NSCs for EMT and migration analysis are: correctly identifying the embryonic age, identifying anatomical landmarks of the developing embryo, preparation of fresh culture media and (at least overnight) coated plates, use of fine-tipped instruments and proper dissection of the central portion of the developing cortex to establish cortical explants.
The presented techniques may be enhanced by some modifications. As suggested above, multiple explants are plated on a single dish with a grid. For screening of multiple compounds, however, it may be more convenient to place single explants in every well of a 96-well plate. Further, the model system can be refined for large scale drug discovery. As described above, PDPN is a valuable marker for newly acquired migratory features since PDPN is found in invasive NSCs8. A PDPN reporter can be transfected in normal NSCs before EMT induction. As a consequence PDPN expressing cells can serve as an internal control for cell transformation since PDPN is not expressed in normal NSCs (Figure 4 and Table 1). Several additional NSC markers are available, such as Nestin, that are lost after EMT induction8. As a consequence, both the initial state, "non-migratory/non-invasive/Nestin+", and the final state, "migratory/invasive/PDPN+", can be monitored, which enables automated drug discovery. In large cell culture plates with multiple wells (e.g., 96-well plates and larger) initial cells may be seeded and treated with established drugs or compounds without known function. Thus, multiwell plates can be automatically scanned for expression of PDPN or other migration/invasion markers. If an inhibitor is the main target of the screen, disappearance of PDPN expression can help to identify putatively interesting novel inhibitory compounds.
During testing of a novel substance the explants may not show migration and/or the explants may round up and detach from the dish. In this situation, it is best to change only half of the media every day (instead of full media change every other day). This reduces stress by surface tension when replacing with fresh media. The explants may not attach after the initial plating. In this situation, fibronectin has to be in contact with the dish surface for at least 36 h. Fibronectin used for coating should be freshly prepared without exposure to a plastic tube for more than 10 min since fibronectin may stick to the plastic. Further, the explants may settle outside of the grid. In this situation, it is advisable to swirl the dish in a slow circular motion. This allows the explants to reposition to the center by centripetal force (cf. Figure 3).
There is strong evidence that gliomas are derived from NSCs and/or OPCs derived from NSCs23,24. As a consequence other groups have established glioma growth models on the basis of NSCs: Sampetrean et al. isolated NSCs from Ink4/ARF-deficient mice and forced overexpression of H-Ras25. High-grade malignant tumors resulted that showed proliferation and invasion after transplantation25. McNeill et al. also isolated NSCs from genetically modified mice (floxed Rb1, Nf1, Kras, Pten alone and in combinations)26. The genes were inactivated by Cre-virus infection and the NSCs were transplanted26. Both these model systems have the advantage that invasion can be monitored in vivo and potential new anti-invasion drugs can be tested. Their main disadvantage is that the onset of invasion is not well defined and it is unclear whether the cells show glioma-typical EMT invasion (EMT genes) or only local migration. Further only one drug can be tested per animal and the analysis requires several weeks. To identify a novel anti-invasion drug, pharmaceutical companies need (a) to screen for several thousand compounds or more at once, (b) to replicate the tests and (c) to try different drug concentrations. To prepare several thousand transplanted animals, treat them with different drugs and finally test for effects by histochemical or bioluminescence analysis is very resource-consuming. Here a novel NSC-based model system for EMT-related invasion is described that allows the cost-efficient screening of thousands of compounds.
Although this model enables high throughput screening of compounds, it is a primary screening platform only. Once compounds of interest are identified in this model, they will require additional validation. In vivo testing is necessary for safety and efficacy parameters for any compound identified and transplantation models as described above become important25,26. The model is also limited by the fact that it is based on rodent cells. Although the model can be used to investigate rat or mouse EMT, it is unclear if the same mechanisms are in place in human cells or in the human patient. Further experiments are needed to investigate if NSCs are not only invasive in vitro, but also after transplantation. Several lines of evidence support the role of NSCs in glioma formation. Glioma progression has been linked to the following genes: Tenascin C, Hey1, SPARC, Snail1 and Snail2, FGFR+, BMPR1a, EGFR, PDGFRβ, Sox2, Podoplanin, Gli3 and p75NGFR8. All of these genes are also expressed during the transformation of normal non-invasive NSCs to invasive mesenchymal cells8. At the time being, it is unclear whether NSCs can be transformed into tumors by external growth factors only.
Tumor-initiating stem cells (TICs) from different tumors have been isolated and are used as models to understand tumor progression27. TICs are, however, not well suited for EMT analysis, since EMT did already occur in TICs27. To understand early and also late steps of EMT induction, a primary non-neoplastic cell population is needed. Ideally this population was not meant to undergo EMT in the first place. It had been previously shown that embryonic NSCs were suitable candidates for this purpose9,28. Namely, migration and invasion could be induced in NSCs by FGF2 and BMP49,28. FGF2/BMP4-induced migration was related to single EMT-related genes, however, complete evidence was lacking since the key EMT families Zeb and Twist had not been investigated8. The results above show that a specific combination of four factors, FGF2, BMP4, TGFβ1 and insulin cause a very strong and complete EMT induction, not observed before. The present study demonstrates that the key EMT genes of the Zeb- and Twist-families are also upregulated. This study therefore provides the first evidence that there is not selective upregulation of single genes, but the results show that all key EMT families are active in the proposed cell culture system.
EMT has also been observed in epithelial cancers outside of the brain, such as lung, breast, colon and gastric cancer29. Several models to study EMT are in use for these tumors30-32, which come, however, with significant limitations: (a) usage of transformed tumor cell lines carrying multiple genetic and epigenetic changes33; to identify inhibitors of EMT for early cancer stages, late-stage cancer cells are inadequate; (b) serum-dependent cultures which contain various undefined growth factors; this renders identification of critical factors very difficult. Further, experiment reproducibility is impaired since serum quality varies from batch to batch; (c) serum contains inhibitors and enzymatic activities possibly inactivating potentially useful exogenous compounds; (d) need of enzymes for cell passaging that degrade cell surface molecules; (e) to identify EMT agonists to promote physiological EMT, normal cells are needed to test induction. Tumor cell models are inadequate to find substances that promote regeneration in normal stem cells.
Migration is also required for normal processes, such as wound healing and regeneration of the injured brain18. After traumatic brain injury and stroke, NSCs are necessary to migrate to the lesion site to participate in the regeneration4,34. The current system may also be used to identify substances that promote migration and invasion. As shown above, migration is induced upon start with BMP4-treatment. If the cells are not exposed to BMPs but instead to novel compounds, these may substitute for the BMP action which will help to identify novel BMP agonists. With a reporter system that indicates PDPN expression, large scale testing for novel substances becomes feasible; if a novel compound can substitute for BMPs, PDPN expressing cells can be automatically identified.
The proposed system is also useful to investigate cell-signaling interactions. Our results show that the BMP effect on migration could be enhanced by TGFβ1 activation. The results uncover an additive interaction between BMP- and TGFβ-signaling. However, additional signaling pathways, such as Notch-, Wnt- or EGFR-/MAPK- and other signaling cascades may be involved. Therefore, further investigations on BMP/TGFβ-interactions and cross-talk to other pathways are needed. In addition, the responsive central cortex can be isolated from genetically modified mice, which can help to elucidate underlying mechanisms driving EMT. In summary, the EMT model system described here may be helpful in the fields of stem cell biology and regeneration, as well as in cancer research. It may be used for screening of drug libraries for substances inhibiting or enhancing migration and invasion.
The authors have nothing to disclose.
The study was supported by the University of Basel Science Foundation and the Swiss National Science Foundation by a grant to MHS and AG (SNF IZLIZ3_157230). We thank: Dr. Tania Rinaldi Burkat for generously providing infrastructure; all members of the Bettler group for discussions and comments. We thank Gerhard Dorne (Leica Microsystems, Switzerland) for professional and competent installation of the Full HD MC170 video camera (Leica Microsystems, Switzerland).
BMP4, rhBMP4 | RnD Systems | 314-BP-01M | |
Bovine pancreas insulin | Sigma | I1882 | |
Boyden chamber, CytoSelect cell invasion assay | Cell Biolabs | CBA-110 | 24 well plate system |
Cell culture dish with grid | Ibidi 500 mm dish, 35 mm | 80156 | |
CellMask Orange | Life Technologies | C10045 | Plasma membrane dye, use at 1:1000 . |
DAPI | LifeTechnologies | D1306 | Stock at 5mg/ml. Use at 1:10000. Cancerogenic. Appropriate protection (gloves, coat, goggles) required. |
DMEM/F12 1:1 medium bottle | Gibco Invitrogen | 21331-020 | |
FGF2, rhFGF2 | RnD Systems | 233-FB-01M | |
Fibronectine, bovine | Sigma | F4759 | |
Glutamax supplement | Gibco Invitrogen | 35050-061 | |
Graphics software with pixel measurement feature | Fiji | fiji.sc/Fiji | version 2.0.0-rc-30/1.49s |
HBSS media | Sigma | H9394 | |
Human apo-Transferrin | Sigma | T1147 | Possible lung irritant. Avoid inhalation. Use appropriate protection. |
L-glutamine | Gibco Invitrogen | 25030-024 | |
Nestin, Mouse anti Nestin antibody | Genetex | GTX26142 | Use at 1:100, 4% PFA fixation, Triton X100 at 0.1% |
Olig2, Rabbit anti Olig2 antibody | Provided by Hirohide Takebayash | Personal stock | Use at 1:2000, 4% PFA fixation, Triton X100 at 0.1% |
Penicillin/Streptomycin/Fungizone | Gibco Invitrogen | 15240-062 | |
Podoplanin, Mouse anti Podoplanin antibody | Acris | DM3614P | Use at 1:250, 4% PFA fixation, avoid Triton X100 |
Poly-L-ornithine | Sigma | P3655 | |
Putrescine | Sigma | P5780 | Skin and eye irritant. Appropriate protection required. |
Sodium selenite | Sigma | S5261 | |
Sox10, Rabbit anti Sox10 antibody | Millipore Chemicon | AB5774 | Use at 1:200, 4% PFA fixation, Triton X100 at 0.1% |
TGFb1, rhTGFb1 | RnD Systems | 240-B-010 | |
Uncoated Petri dishes | Falcon Corning | 351029 |