We report a technique permitting live imaging of microtubule dynamics in glioblastoma (GBM) cells invading a vertebrate brain tissue. Coupling the orthotopic injection of fluorescently tagged GBM cells into a transparent zebrafish brain with high-resolution intravital imaging allows the measurement of cytoskeleton dynamics during in situ cancer invasion.
With a dismal median survival time in real populations-between 6 to 15 months-glioblastoma (GBM) is the most devastating malignant brain tumor. Treatment failure is mainly due to the invasiveness of GBM cells, which speaks for the need for a better understanding of GBM motile properties. To investigate the molecular mechanism supporting GBM invasion, new physiological models enabling in-depth characterization of protein dynamics during invasion are required. These observations would pave the way to the discovery of novel targets to block tumor infiltration and improve patient outcomes. This paper reports how an orthotopic xenograft of GBM cells in the zebrafish brain permits subcellular intravital live imaging. Focusing on microtubules (MTs), we describe a procedure for MT labeling in GBM cells, microinjecting GBM cells in the transparent brain of 3 days post fertilization (dpf) zebrafish larvae, intravital imaging of MTs in the disseminating xenografts, altering MT dynamics to assess their role during GBM invasion, and analyzing the acquired data.
Cell motility is a stereotyped process requiring polarity axis establishment and force-generating cytoskeletal rearrangements. Actin polymerization and its association with myosin are recognized as the main contributors to protrusive and contractile forces required for cell movement1. Microtubules are considered to be the main actors in cell polarization and directional persistence during migration2. In recent years, MTs have also been shown to create and stabilize protrusions to support mechanocompressive forces during cell invasion in 3D3. More recently, MTs have been directly involved in mechanotransduction at focal adhesions and mechanosensitive migration4. The dynamic instability that characterizes MT-plus end dynamics is made of repeated phases of polymerization (growth) and depolymerization (shrinkage), which are controlled by a plethora of microtubule-binding proteins and intracellular signaling cascades, such as those governed by RHO-GTPases5,6,7. The role of the MT network in cell migration and invasion has made the investigation of MT dynamics a key element to better understand the mechanisms of immune cell homing, wound healing, and cancer invasion.
The ability of cancer cells to escape the primary tumor core, spread in the tissues, and generate secondary tumors is a critical step in preventing global success in the war against cancer declared 50 years ago8,9. One of the biggest hurdles has been understanding how cancer cells actively invade the tissue. Key invasion mechanisms rely on the same principles as those governing non-tumorous cell migration10. However, cancer cell migration specificities have emerged11, triggering the need for better characterization of this type of migration. Specifically, because the tumor microenvironment appears as a key player in cancer progression12, observing and analyzing cancer cell invasion in a relevant physiological context is essential to unravel the mechanisms of cancer cell dissemination.
MTs are central to cancer progression, to sustain both proliferation and invasion. Precise analysis of MT dynamics in situ can help identify MT-altering agents (MTA) in both processes. MT dynamics vary drastically upon a change in environment. In vitro, treatment with MT-destabilizing agents such as nocodazole prevents cell protrusion formation when cells are embedded in gels in 3D, whereas it has little effect on 2D cell migration13,14. Although technically challenging, advances in intravital imaging permit in vivo analysis of MT dynamics during cancer cell invasion. For instance, the observation of MTs in subcutaneously xenografted fibrosarcoma cells in mice revealed that tumor-associated macrophages affect MT dynamics in tumor cells15. However, these mouse models involve extensive surgical procedures and remain unsatisfying for less accessible cancers, such as the highly invasive brain tumor, GBM.
Despite a dismal 15 month average survival time16, little is known about GBM's mode of dissemination within the brain parenchyma or the key molecular elements sustaining GBM cell invasion in the brain tissue. Improvement in the mouse orthotopic xenograft (PDX) model and the establishment of cranial windows offered new prospects for GBM cell invasion studies17,18. However, due to suboptimal imaging quality, this model has mostly permitted longitudinal imaging of superficial xenografts and has not been successfully used to study subcellular imaging of cytoskeleton proteins so far. Furthermore, in the wake of the "3Rs" injunction to reduce the use of rodents and replace them with lower vertebrates, alternative models have been established.
Taking advantage of the primitive immunity observed in zebrafish (Danio rerio) larvae, orthotopic injection of GBM cells in the fish brain was developed19,20,21. Injection in the vicinity of the ventricles in the developing midbrain recapitulates most of human GBM pathophysiology21, and the same preferred pattern of GBM invasion as in humans-vessel co-option-is observed22. Thanks to the transparency of the fish larvae, this model allows the visualization of GBM cells invading the brain from the peri-ventricular areas where most GBMs are thought to arise23.
Because MTs are essential for GBM cell invasion in vitro24,25, a better characterization of MT dynamics and the identification of key regulators during cell invasion is needed. However, to date, the data generated with the zebrafish orthotopic model has not included subcellular analysis of MT dynamics during the invasion process. This paper provides a protocol to study MT dynamics in vivo and determine its role during brain cancer invasion. Following stable microtubule labeling, GBM cells are microinjected at 3 dpf in zebrafish larvae's brains and imaged in real time at high spatio-temporal resolution during their progression in the brain tissue. Live imaging of fluorescent MTs allows the qualitative and quantitative analysis of MT plus-end dynamics. Furthermore, this model makes it possible to assess the effect of MTAs on MT dynamics and on the invasive properties of GBM cells in real time. This relatively non-invasive protocol combined with a large number of larvae handled at a time and the ease of drug application (in the fish water) makes the model an asset for preclinical testing.
Animal experiments were conducted according to European Union guidelines for the handling of laboratory animals. All protocols were approved by the Ethical Committee for Animal Experimentation of Institut Pasteur – CEEA 89 and the French Ministry of Research and Education (permit #01265.03). During injections or live imaging sessions, animals were anaesthetized with Tricaine.At the end of the experimental procedures, they were euthanized by anesthetic overdose. See the Table of Materials for details related to the materials, equipment, and software used in this protocol. The general workflow of the protocol is described in Figure 1.
1. Generation of glioblastoma cells stably expressing α-tubulin-mKate2
NOTE: The following steps are performed in a biosafety cabinet BSL2+.
2. Prepare zebrafish larvae for microinjection
3. Xenotransplantation procedure
4. Intravital imaging of the glioblastoma xenografts
5. Image analysis
To analyze the role played by MTs during in vivo GBM invasion, we describe here the major steps to perform stable MT labeling in GBM cells by lentiviral infection, orthotopic xenotransplantation of GBM cells in 3 dpf zebrafish larvae, high-resolution intravital imaging of MT dynamics, MTA treatment and its effects on GBM invasion, and image analysis of MT dynamics and in vivo invasion (Figure 1). MT dynamics are measured either by building kymographs along growing and shrinking MTs (Figure 3C) or by manually tracking individual MT edges over time (Figure 3D). An example of drug treatment administered in the larvae medium and its reversible effect on MT network organization is given in Figure 4. Treatment with a low dose of nocodazole (200 nM) leads to progressive shrinkage of the MT network and disappearance of glioblastoma cell protrusion 4 h later (Figure 4A). Washing out the drug restored the capacity of glioblastoma cells to form protrusions. The cells resumed migrating along the vasculature 12 h after the washout (Figure 4B). These data suggest treatment with 200 nM nocodazole is sufficient to disrupt the MT network and rapidly blocks in vivo glioblastoma cell invasion. A 3 day-long analysis of the same treatment on global glioblastoma cell invasion reveals that nocodazole 200 nM halts long-term glioblastoma cell invasion in vivo, without affecting the general health of the fish, compared to a control (Figure 4C).
Figure 1: Protocol workflow diagram. Abbreviations: dpf = days post fertilization; hpi = hours post injection; MT = microtubule; MTA = microtubule-altering agent; GBM = glioblastoma multiforme. Please click here to view a larger version of this figure.
Figure 2: Microinjecting glioblastoma cells into 3 dpf zebrafish larvae brains. (A) Photograph of the equipment used for xenotransplantation: 1, oil microinjector; 2, mechanical micromanipulator; 3, universal capillary holder; 4, glass capillary. The microinjection plate is under a stereomicroscope. (B) Photograph showing anesthetized 3 dfp zebrafish larvae aligned in a trench and ready to be microinjected. The tip of a microcapillary loaded with a red dye is visible on the right of the photograph. Details of the patterned trenches built in the agarose plate are seen in the inset on the bottom right corner. Scale bar = 3 mm. (C) Scheme of a representative microinjection plate. Ready-to-be-injected larvae are placed laterally (center of the plate). Note that cells are concentrated at the tip of the microcapillary (black arrow) before proceeding to the injection. Injected larvae are shown on the right of the plate, placed ventrally. (D) Scheme of a transversal slice in the microinjection plate (along the black dotted line in C) showing the trenches where the larvae are placed during injection. (E) Photograph showing the tip of the capillary ready to penetrate the optic tectum (dotted line). The ventricles are delineated by the white lines. Scale bar = 150 µm. (F) Scheme of a 3 dpf fli1a:gfp larva expressing gfp in the endothelial cells, indicating the OT region where the cells are injected, just above the middle cerebral vein. (G) Fluorescence image of a 3 dpf gfap:gfp larva (gfp expressed in neural stem cells) placed laterally after microinjection of U87-mkate2 cells (white circle and white arrow). High autofluorescence in red is caused by the iridophores in the eyes. Scale bar = 100 µm. (H) Confocal fluorescence image of a successfully injected fli1a:gfp larva at 16 hpi. Scale bar: 50 µm. (I–K) Confocal fluorescence images of unsuccessfully injected fli1a:gfp larvae at 96 hpi (I) and 4 hpi (J,K). GBM cells have been injected in ventricles (I,J) or in multiple foci in the brain (K). Scale bars = 50 µm. Abbreviations: dpf = days post fertilization; OT = optic tectum; MCeV = middle cerebral vein; gfp = green fluorescent protein; gfap = glial fibrillary acidic protein; hpi = hours post injection. Please click here to view a larger version of this figure.
Figure 3: Visualizing in vivo microtubule dynamics in glioblastoma cells. (A) Representative fluorescence image of xenografted U87 cells expressing tubulin-α1-mkate2 in the OT of a fli1a:GFP zebrafish larva at 20 hpi. (B) Maximum intensity projected fluorescence image of the MT network in a single xenografted U87 cell. (C) Kymograph along the red dotted line in B, showing the growing, the pause, and the catastrophe phases of MT dynamic instability. (D) Time-lapse sequence of the boxed region in B highlighting the tracking of three MT + ends. Scale bars = 10 µm. Abbreviations: OT = optic tectum; GFP = green fluorescent protein; hpi = hours post injection; MT = microtubule; G = growing phase; P = pause phase; C = catastrophe phase. Please click here to view a larger version of this figure.
Figure 4: Visualizing the effects of the microtubule altering agent on glioblastoma invasion in vivo. (A) Time-lapse sequence of U87 cells expressing tubulin-α1a-mkate2 in zebrafish larvae treated with nocadazole (200 nM). Arrows point to the extremity of the protrusion in two different cells invading the brain along a blood vessel. Note the retraction of the protrusion upon treatment with nocodazole. Scale bar = 10 µm. (B) Time-lapse sequence representing the effect of nocodazole washout on U87 cell invasion. At 500 min after the washout, the cell marked with the white asterisk elongates an MT-based protrusion (white arrow), which permits its resumption of the invasion along a blood vessel. Scale bar = 20 µm. (C) 3D representations of a xenografted larvae brain treated with DMSO or nocodazole (200 nM) for 72 h. U87 cells signal has been segmented (in red) and integrated to the fli1a-GFP fluorescence signal (in white). Note the decreased dissemination of U87 cells treated with nocodazole. Scale bar = 30 µm. Abbreviations: GFP = green fluorescent protein; hpi = hours post injection; MT = microtubule. Please click here to view a larger version of this figure.
Figure 5: Image analysis of in vivo glioblastoma invasion. (A) Fluorescent images of xenografted U87 cells expressing cytosolic mKate2 in fli1a-GFP zebrafish larvae, 4 hpi. White asterisks underline typical autofluorescence from eye iridophores. Scale bar = 40 µm. (B) Fluorescent image of fli1a-GFP larvae coupled with the segmented surface corresponding to the U87 cells signal in A. The centroid of the tumor mass appears green. Scale bar = 40 µm. (C) Fluorescent image representing the red channel signal in A and the newly defined "distance to centroid" channel (in blue). Scale bar = 30 µm. (D) Red channel fluorescence image superimposed with colored spots, whose color represents the distance of the cell to the centroid (in green), violet being the closest to the centroid and white being the furthest apart.Scale bar = 20 µm. (E) An example of sequential analysis of global GBM invasion. 3D distances are determined at 4 hpi and 72 hpi, and the invasion index (II) is calculated according to the formula in the inbox. Scale bar = 20 µm. Abbreviations: GFP = green fluorescent protein; hpi = hours post injection. Please click here to view a larger version of this figure.
Imaging tumor xenografts at single-cell resolution is likely to become an indispensable tool to improve our understanding of GBM biology. Live imaging in mouse PDX models has led to valuable discoveries on how GBM collectively invades the brain tissue18. However, to date, the spatiotemporal resolution is not high enough to reveal the dynamics of proteins controlling GBM invasion. We reasoned that by coupling the orthotopic engrafting of GBM cells in transparent zebrafish larvae with high-resolution intravital imaging, cytoskeletal proteins such as MTs could be analyzed in sufficient detail to analyze their dynamics during in situ GBM invasion.
The critical steps of the methodology lie in the preparation of the GBM cells and the microinjection procedure. Unhealthy and inadequately dissociated cells will stick together or to the capillary borders and block the flow of injection. In addition, cells need to be sufficiently concentrated in the capillary to minimize the injected volume and implant them in bulk. Injecting a higher volume of more diluted cells will result in multiple, sometimes intermixed tumor foci, whose invasive indexes become difficult to measure. In our hands, manual handling of the oil-based microinjector allows better control of the injection flow than a pressurized air-based electronic microinjector that has been used previously in a similar model19. This is critical to prevent excess flow pressure inside the brain, thereby avoiding subsequent tissue damage and ventricular aggregation of the injected cells.
Some limitations of this model include the necessity to perform the experiment at suboptimal temperatures for both species. Zebrafish are usually raised at 28 °C, whereas human cells are cultured at 37 °C. Above 32 °C, zebrafish embryo development is altered and these changes can be lethal35. However, similar to what is done in adult zebrafish xenograft models36, sequentially acclimating the zebrafish larvae to a temperature of 32 °C increases the survival of transplanted animals compared to the rapid change in temperature post transplantation from 28 °C to 32 °C. However, increasing the temperature further leads to increased animal deaths in accordance with the sensitivity of zebrafish embryos to temperatures above 32 °C35.
Interpreting the in vivo MT dynamics data has to be done carefully as MT dynamics change when the temperature drops below 37 °C37. Parallel in vitro measurement of MT dynamics at 37 °C and 32 °C in the same GBM cells with the same MTA treatment will help validate the differences seen between various GBM cells or between in vivo treatments. It should help confirm that the differences are not caused by variation in temperature sensitivity but by different regulation pathways (for GBM comparison analysis) or by MTA treatment (for MTA effect analysis). This will be of interest should the MT dynamics heterogeneities be linked to different invasion abilities.
Another limitation is the short time window during which invasion can be monitored (72 to 96 h), preventing the measurement of invasion plasticity driven by potential changes in MT dynamics38. After 96 h, we noticed a sharp decrease in GBM cell invasion. At 6 days post injection, the number of GBM cells declined rapidly, presumably due to a host immune response caused by the accumulation of neutrophils and macrophages in the tumor microenvironment39. Delivering MTAs to the whole brain is likely to affect nearby neuronal host cells, which depend on MTs for their activity and whose alteration might subsequently affect GBM invasion40. This approach needs to be complemented with shRNA or optogenetics assays restricting MT alteration to the GBM cells, but remains a good platform to screen for new anti-invasive compounds.
The orthotopic injection of MT-labeled GBM cells in the zebrafish brain is of particular interest to decipher the role of MTs during cancer cell invasion, as very few animal models permit in situ subcellular imaging of cancer cell migration in their tissue of origin15. To date, studies of MT functions during GBM migration rely mostly on in vitro and ex vivo assays and lack in vivo validation24,41,42,43. Coupled with gene-of-interest knock down or an unbiased gene-based screening approach, the assay presented here will help reveal new MT regulators that are important for GBM invasion in vivo.
GBMs are highly heterogeneous tumors whose invasive properties differ greatly between specimens44. Understanding the molecular mechanisms underlying their different mode of invasion will help define ad hoc therapeutic treatments to block GBM dissemination. Systematic measurement of the invasive index, mode of invasion, and cytoskeletal properties such as MT dynamics in various GBM samples will reveal new correlations between frequent genomic mutational profiles and cell invasion patterns relying on specific cytoskeletal properties. Revealing how these mutations affect the change in MT dynamics not only would add to our knowledge on MT regulation during cell migration45,46 but could also lead to long-awaited patient-specific, anti-invasive therapeutics.
The relative ease of microinjecting in zebrafish combined with the high number of larvae available and the ease of injection of drugs make this procedure suitable for personalized medicine47,48. Moreover, in contrast to intravital imaging of GBM xenografts in mice, that only occurs in the upper 500 µm part of the cortex49,50, using zebrafish allows for the visualization of GBM infiltration in the whole CNS. The model presented here meets the criteria to become an invaluable tool for rapid analysis of glioblastoma's invasive capacities and its response to treatments.
The authors have nothing to disclose.
We are extremely grateful to Dr. P. Herbomel (Institut Pasteur, France) and his laboratory, especially Valérie Briolat, and Emma Colucci-Guyon for providing us with the zebrafish lines and the plastic mold for microinjection plates, and for their valuable expertise on zebrafish experimental procedures. We gratefully acknowledge the UtechS Photonic BioImaging (C2RT, Institut Pasteur, supported by the French National Research Agency France BioImaging, and ANR-10-INBS-04; Investments for the Future). This work was supported by the Ligue contre le cancer (EL2017.LNCC), the Centre National de la Recherche Scientifique, and Institut Pasteur and by the generous donations of Mrs. Marguerite MICHEL and Mr. Porquet.
Glioblastoma cell culture | |||
Foetal calf serum | Eurobio | CVFSVF00-01 | Reagent |
MEM NEAA | Gibco | 11140-050 | Reagent |
Modified Eagle's medium | Eurobio | CM1MEM18-01 | Reagent |
Penicillin–streptomycin | Gibco | 15140-122 | Reagent |
U-87 MG | ECACC | 89081402-1VL | Cells |
Lenitivirus production | |||
BD FACSAria III | BD bioscience | Instrument | |
BD FACSDiva software v8.0 | BD bioscience | Software | |
HEK-293T | Merck | 12022001 | Cells |
pMD2.G | Addgene | Plasmid #12259 | Reagent |
psPAX2 | Addgene | Plasmid #12260 | Reagent |
Ultracentrifuge Optima XPN-80 | Beckman Coulter | Instrument | |
Cell passaging and staining | |||
dPBS | Gibco | 14190-094 | Chemical |
Hoechst 34580 | Sigma-Aldrich | 63493 | Chemical |
Trypsin-EDTA (0,05%) | Gibco | 25300-054 | Reagent |
Zebrafish husbandry | |||
Fluorescence stereomicroscope LEICA M165FC | LEICA | https://www.leica-microsystems.com/fr/produits/stereomicroscopes-et-macroscopes/informations-detaillees/leica-m165-fc/ | Instrument |
Methylene Blue hydrate | Sigma-Aldrich | M4159 | Chemical |
N-Phenylthiourea (PTU) | Sigma-Aldrich | P7629-25G | Chemical |
Transfer Pipettes fine tips | Samco Scientific | 232 | Equipment |
Transfer Pipettes Large Bulb3mL | Samco Scientific | 225 | Equipment |
Tricaine (Ethyl 3-aminobenzoate methanesulfonate) | Sigma-Aldrich | Cat#: A5040 | Chemical |
Volvic Source Water | DUTSCHER DOMINIQUE SAS | 999556 | Reagent |
Xenotransplantation | |||
24-well plate | TPP | 92024 | Equipment |
Borosilicate glass capillaries (1.0 ODx0.58IDx150L mm) | Harvard Apparatus | (#30-0017 GC100-15 | Equipment |
CellTram oil vario microinjector | Eppendorf | 5176000.025 | Instrument |
Microloading pipet tips (Microloader) 20µL | Eppendorf | 5242956003 | Equipment |
Micromanipulator | NARISHIGE | https://products.narishige-group.com/group1/injection/english.html | Equipment |
Mineral Oil | Sigma | M8410-100ml | Equipment |
Stereomicroscope | Olympus | KL 2500 LCD | Instrument |
Universal capillary holder | Eppendorf | 5176190002 | Equipment |
Vertical Pipette puller | KOPF (Roucaire) | Model 720 | Instrument |
Intravital Imaging | |||
3.5cm glass-bottom videoimaging dish | MatTek Life Sciences, MA, USA | P35G-1,5-14-C | Equipment |
Acquisition software: NIS-Elements-AR version 5.21 | Nikon | Software | |
Heat-Block | Techne | DRI-BLOCK DB-2D | Equipment |
Microscope head Nikon Ti2E | Nikon | Instrument | |
sCMOS camera Prime 95B | Photometrics | Instrument | |
sCMOS camera Orca Flash 4 | Hammatsu | Instrument | |
Ultrapure Low melting point agarose | Invitrogen | 16520-050 | Chemical |
Yokagawa CSU-W1 spinning disk unit | Hammatsu | Instrument | |
Drug Treatment | |||
DMSO | Sigma-Aldrich | D2650-100ML | Chemical |
Nocodazole | Sigma-Aldrich | M1404-2MG | Chemical |
Image Analysis | |||
Imaris 9.5.1 software | Oxford Instruments | Software | |
ImarisFileConverter 9.5.1 | Oxford Instruments | Software |