Current ex vivo models of glioblastoma (GBM) are not optimized for physiologically relevant study of human tumor invasion. Here, we present a protocol for generation and maintenance of organotypic slice cultures from fresh human GBM tissue. A description of time-lapse microscopy and quantitative cell migration analysis techniques is provided.
Glioblastoma (GBM) continues to carry an extremely poor clinical prognosis despite surgical, chemotherapeutic, and radiation therapy. Progressive tumor invasion into surrounding brain parenchyma represents an enduring therapeutic challenge. To develop anti-migration therapies for GBM, model systems that provide a physiologically relevant background for controlled experimentation are essential. Here, we present a protocol for generating slice cultures from human GBM tissue obtained during surgical resection. These cultures allow for ex vivo experimentation without passaging through animal xenografts or single cell cultures. Further, we describe the use of time-lapse laser scanning confocal microscopy in conjunction with cell tracking to quantitatively study the migratory behavior of tumor cells and associated response to therapeutics. Slices are reproducibly generated within 90 min of surgical tissue acquisition. Retrovirally-mediated fluorescent cell labeling, confocal imaging, and tumor cell migration analyses are subsequently completed within two weeks of culture. We have successfully used these slice cultures to uncover genetic factors associated with increased migratory behavior in human GBM. Further, we have validated the model's ability to detect patient-specific variation in response to anti-migration therapies. Moving forward, human GBM slice cultures are an attractive platform for rapid ex vivo assessment of tumor sensitivity to therapeutic agents, in order to advance personalized neuro-oncologic therapy.
The laboratory study of glioblastoma (GBM), is hindered by a lack of models that faithfully recapitulate the requisite pathologic characteristics of the human disease, namely tumor cell migration and invasion. Comparative studies of 2D and 3D in vitro invasion assays as well as 3D rodent slice culture models have uncovered mechanistically disparate cellular migration programs in these two contexts, potentially limiting the translatability of findings from 2D systems to the human disease1,2,3. The organotypic tumor slice culture and imaging paradigm described here allows for the study of tumor cell migration within slices of ex vivo human tumor tissue obtained from surgical resection. Thus, slice cultures of surgically resected tumor tissue in conjunction with time-lapse confocal microscopy provide a platform to study tumor cell migration in the native microenvironment without tissue dissolution or culture passaging.
There is extensive literature employing rodent brain slice culture models of GBM generated from human tumor xenografts, retroviral-induced tumors, and cellular overlays to study tumor invasion1,2,3,4,5. Recently, several groups have described the generation of organotypic slice cultures directly from human GBM tissue6,7,8,9,10. However, there is marked variation among published protocols with regards to slicing technique and culture media. Further, the use of organotypic slice cultures has focused on static experimental endpoints that have included changes in cell signaling, proliferation, and death. The protocol described herein expands upon prior slice culture paradigms by incorporating time-resolved observation of dynamic tumor cell behaviors through time-lapse laser scanning confocal microscopy. Recent discovery of inter11 and intratumoral12,13 genetic variation in human GBM underlines the importance of linking this heterogeneity with tumor cell behaviors and its implications on tumor response to therapy. Here, we report a streamlined and reproducible protocol for use of direct slice cultures from a human cancer tissue to visualize tumor cell migration in near real-time.
Before collection of patient tissue samples is initiated, informed consent must be obtained from each patient under an approved Institutional Review Board (IRB) protocol. The authors of this protocol received consent for the work described under approved IRB protocols at the University of Colorado Hospital and Inova Fairfax Hospital. Data collected from these slice cultures were not used to direct patient care decisions.
1. Pre-slicing Preparation
2. Day of Surgery: Tissue Acquisition
3. Slice Culture Preparation
NOTE: This protocol requires the use of fresh unfixed human tissue. All samples are presumed to be infectious, and should be handled according to universal blood borne pathogen protocols. Appropriate personal protective equipment should be donned at all times. Forceps and scalpels should be exposed to 15 min of UV light prior to use. During use, intermittently spray the tools with 70% ethanol (EtOH), allowing time for the liquid to evaporate before use. The slicing process is performed in a semi-sterile fashion utilizing a horizontal laminar flow hood with filtered air.
4. Slice Culture Maintenance
5. Tumor Cell Labeling Via Green Fluorescence Protein Expressing Retrovirus
NOTE: Time-lapse microscopy for analysis of tumor cell migration requires stable, long-term fluorescent labeling of cells within the slice culture. Use of retrovirus is suggested because it selectively infects dividing cells, thereby enriching fluorescent labeling within the tumor cell population as opposed to microglia or other cell types present within the slice. Standardization of infection suggests that a viral titer of 104 CFUs/µL results in sufficient green fluorescent protein expression for the tracking and analysis of cell migration. Increased viral titer, use of non-selective virus (i.e. adenovirus, lentivirus), or other means of labeling all cells may preclude identification of clear cell boundaries during migration, thus complicating analysis. Use of alternative fluorescent markers can be utilized and optimized as needed.
6. Time-Lapse Single Photon Laser Scanning Confocal Imaging of Tumor Cell Migration
NOTE: After successful transduction and health of the culture is confirmed, cells may be imaged under control conditions, followed by an equal period of imaging under treatment conditions. Using this protocol, cells were successfully imaged and tracked for 12 hours in each condition. However, shorter or longer periods of imaging and environmental manipulation may also be informative.
7. Image Post-Processing and Tumor Cell Tracking
NOTE: Many confocal imaging systems are equipped with proprietary image-processing software. The processing steps discussed below comprise a general protocol, which can be performed across software platforms. Specific instructions will be given for the open-source platforms, NIH ImageJ and MTrackJ15.
Our group has successfully generated slice cultures from over 50 patients undergoing initial GBM resection. This slice generation, culture, retroviral-labeling, imaging, and migration analysis protocol has been streamlined into a reproducible workflow (Figure 1). Critically, these organotypic GBM slices demonstrate concordance with originating tumor tissue throughout culture, including maintenance of pathologic hallmarks and microglia up to 15 days in culture (Figure 2). In addition, we have utilized this system to perform functional assays of tumor response to microenvironmental changes. As a metric of physiologic integrity, we examined how GBM slice cultures responded to hypoxia (1% O2) by measuring the production of vascular endothelial growth factor (VEGF), a process that occurs abundantly within the GBM microenvironment in vivo17,18. We demonstrated that by placing the slice cultures in hypoxic conditions, the slices mounted a rapid physiologic response, inducing VEGF release into the media (Figure 3).
To evaluate qualitative and quantitative aspects of tumor cell migration we utilized the time-lapse images to generate detailed migration maps. These maps demarcate all GBM cells tracked within the confines of tumor microregions (1 mm2), providing a static visualization of the dynamic migratory behavior of the tumor population. Quantitative measures of migration speed and directionality (cell displacement / total distance traveled) were calculated for each cell, allowing for investigation of changes in migration parameters across tumor regions, tumor samples, and in response to treatment (Figure 4).
The cell labeling and imaging protocol described here also provides sufficient spatial and temporal resolution to evaluate changes in cell morphology during migration through the native tumor microenvironment. We observed the presence of morphologically distinct motile tumor cells and microglial intermingled within the slice culture (Figure 5A). Tumor cell movement was characterized by a "search and burst" process, which involved repeated protrusion and retraction of filopodia from a static cell, followed by a short period of efficient movement. Imaging slice regions approximately every 10 min also provided adequate temporal resolution to record time-lapse images of tumor cells undergoing cell division. These dividing cells paused from migration, completed mitosis, and the daughter cells re-initiated migration without delay, all within a 3-h timeframe (Figure 5D). In contrast, microglia migrate at a higher and more consistent speed, with lower directionality than adjacent tumor cells, demonstrating their relatively inefficient migration (Figure 5C, 5E-G). Such observations may be important for gaining insight into the biology underlying patient- or cell-specific responses to treatment.
Finally, we used this protocol to demonstrated patient-to-patient variability in cell migration parameters at the population level, including a correlation of epidermal growth factor receptor (EGFR) genomic amplification with augmented migratory potential of tumor cells16. In addition, time-lapse microscopy of tumor slices before and after treatment with the anti-invasive drug, gefitinib, demonstrated a significant reduction in migration, which was specific to EGFR amplified tumor slices16.
Figure 1: Human GBM Organotypic Slice Culture Infection, Imaging, and Cell Migration Analysis Workflow. Tumor tissue is localized to a specific region via intraoperative navigation equipment. One week post slicing, the ZsGreen expressing retrovirus is added to the slice cultures to label the mitotically active tumor cells. 3 days post-infection, slices are prepared for confocal imaging. The 3D imaging data is post-processed into 2D images for cell migration path tracking, generation of tumor cell migration maps, and calculation of tumor cell migration parameters. Portions of this figure were originally published in Parker et al, 201316 and reproduced with permission of Oxford University Press. Please click here to view a larger version of this figure.
Figure 2. Human GBM Organotypic Slices Retain Histological Features Throughout Ex Vivo Culture. (A) T1 contrast enhanced MRI sequences were used to localize and document the region(s) of tissue acquisition (arrow). (B) H&E staining of initial donor tissue (OR) and slices at day 8 of culture from a slice culture generated from tissue obtained from the region highlighted in A. (C) Microenvironmental pathologic and cellular features of GBM, in vivo, are maintained throughout slice culture. (I, II) Immunohistochemistry for CD68, a microglia/macrophage marker, at low (top) and high (bottom) magnification, demonstrates microglial persistence in slices after 15 days of culture. H&E staining on day 4 of slice culture confirms maintenance of pseudopallisading necrosis, a pathologic hallmark of GBM, at both low (iii) and high (iv) magnification. Scale bars = 200 µm. Please click here to view a larger version of this figure.
Figure 3: Human GBM Slice Cultures Secrete VEGF in Response to Hypoxia. (A) The experiment utilized individual culture inserts containing 3 similar sized tumor slices generated from the same tumor region. The slices were maintained in normoxia for 12 h, followed by two 12 h intervals of hypoxia, with new media added before each interval. (B) VEGF secretion into the media measured by ELISA (mean ± standard deviation) from slice cultures generated from two representative tumors, was significantly increased under hypoxia than normoxia (p <0.05). (C) A pooled analysis of slice cultures from 4 different tumors demonstrated increased VEGF secretion after sequential 12 h intervals of hypoxia compared to normoxia (p <0.05). Please click here to view a larger version of this figure.
Figure 4: Tumor Cell Path Track Data Allows Quantitative Determination of Cell Speed and Directionality. (A) Tumor cells with a directionality of 1 represent perfect efficiency along a straight vector, whereas those with lower directionality engage in inefficient meandering paths, as represented in this schematic16. Reprinted with permission from Oxford University Press. (B) Analysis of representative path track data ("low" resolution ~55 min cell tracking intervals) demonstrates cellular variability in speed and directionality. (C) Directionality versus migration speed for each cell track is plotted to visualize migration behavior across the cell population (each dot represents an individual cell). Please click here to view a larger version of this figure.
Figure 5: Microglia within GBM Slice Cultures are Characterized by High Migration Speed and Low Directionality Relative to Tumor Cells. (A) Replicating tumor cells selectively expressing ZsGreen retrovirus and microglia tagged with an Isolectin-IB4-647 conjugate exist intermingled in a representative tumor micro-region from a representative slice culture. (B-C) The paths of individually tracked GBM (B) and microglia (C) in the same tumor micro-region demonstrate discordant migration behaviors. Scale bars = 200 µm. (D) An actively migrating tumor cell pauses, retracts its processes (arrowhead), undergoes cell division, and two daughter cells migrate away in opposing directions (arrows). Scale bars = 50 µm. (E and F) Microglia demonstrate increased migration speed (p <0.0001) and decreased directionality (p <0.0001) when compared to tumor cells in the same region. (G) The distribution of tumor and microglial cells based on speed and directionality demonstrates the unique migratory phenotypes of the two cell populations. Please click here to view a larger version of this figure.
Organotypic slice cultures from human cancer tissue provide an attractive and underutilized platform for pre-clinical translational experimentation. Understanding of population-level behaviors of tumor cells with regards to migration, proliferation, and cell death in the native tumor microenvironment is lacking. Critically, studying tumor response to therapy in a dynamic, time-resolved fashion at the level of cell behavior may shed light on novel mechanisms of treatment resistance. Human tumor slice cultures provide a link between the human disease process and current ex vivo and in vivo modeling techniques19. We recently validated the technique described here as a method to study GBM migration, reporting for the first time measurable inter-tumoral variations in cell migration behavior related to EGFR amplification and signaling16. This study also utilized the slice culture model to test the patient-specific effectiveness of the EGFR inhibitor, gefitinib, as a potential anti-invasive therapy for GBM 16.
Several of the common pitfalls during tissue slicing, retroviral infection, imaging, and image analysis paradigm have been discussed above. However, the retroviral infection protocol warrants further attention. Given patient-to-patient variation, it may prove challenging to titrate the density of virally labeled tumor cells within each slice. If insufficient numbers of cells are labeled by the viral construct, add additional 5-10 µL aliquots of retroviral supernatant to the surface of each slice daily until desired concentration of labeled tumor cells is achieved. In primary slice cultures, the percent of replicating cells is generally lower than in transformed cell lines, thus limiting the subset of cells permissive to retroviral incorporation at any time point. Alternatively, if too many cells are labeled, preventing accurate demarcation of cell migration paths, dilute the viral supernatant with neuronal media to achieve a lower effective titer. Tumor associated microglia incorporated the fluorescent protein expressing retrovirus at a frequency of approximately of 1% of all labeled cells. We were able to visually isolate these cells for separate analysis by the use of a microglia binding lectin for post-imaging data analyses (Figure 5).
The tumor microenvironment including aspects of nutrient delivery, cell-cell interactions, and extracellular matrix all play a role in the pathogenesis of GBM20. Direct human GBM slice cultures eliminate the need for passaging within small animal models or disseminated cell culture, while providing a close recapitulation of the human tumor microenvironment. Further, slice cultures provide uniform access to nutrients across samples, while maintaining cell-cell and cell-ECM interactions. By reducing variations in cellular access to nutrients which are known to occur within tumors, we propose observed differences in the cultures shed light on intrinsic differences among tumor cell behaviors (i.e. migration) at the population level. However, interpreting data collected across slice cultures generated from human tumors is complicated by inherent inter- and intra-tumoral heterogeneity. Critically, further study is needed to characterize the potential genetic and epigenetic shifts that may occur during ex vivo maintenance of human tumor slice cultures.
The use of human tumor slice cultures in parallel with Phase I/II clinical trials is a promising strategy to correlate slice parameters with patient clinical outcomes. Validation of these potential predictive/prognostic parameters is necessary before slice cultures can be used to personalize oncologic therapy. Our work, as well as that of others, demonstrates feasibility of biomarker validation21, as well as rapid ex vivo testing of therapeutic agents in slice cultures from GBM9,16. Similar human slice culture techniques using lung22, colon22, head and neck23, breast24, and prostate cancer25 tissues suggest this approach is generalizable across human cancers.
The authors have nothing to disclose.
We would like to thank Dr. Lee Niswander and Dr. Rada Massarwa for their technical expertise and contributions to the slice culture confocal imaging protocol described here. Further thanks to Dr. Kalen Dionne who provided expertise regarding optimizing brain tumor tissue slicing and culture parameters.
DMEM High Glucose | Invitrogen (Gibco) | 11960-044 | |
Neurobasal-A Medium, minus phenol red | Invitrogen (Gibco) | 12349-015 | |
B-27 Supplement (50X), serum free | Invitrogen (Gibco) | 17504-044 | |
Penicillin-Streptomycin (10,000 U/mL) | Invitrogen (Gibco) | 15140-122 | |
GlutaMAX Supplement | Invitrogen (Gibco) | 35050-061 | |
L-Glutamine (200 mM) | Invitrogen (Gibco) | 25030-081 | |
HEPES (1 M) | Invitrogen (Gibco) | 15630-080 | |
Nystatin Suspension | Sigma-Aldrich | N1638-20ML | 10,000 unit/mL in DPBS, aseptically processed, BioReagent, suitable for cell culture |
UltraPure Low Melting Point Agarose | Invitrogen (Gibco) | 16520-050 | Melts at 65.5 C, Remains fluid at 37 C, and sets rapidly below 25 C. |
Isolectin GS-IB4 from Griffonia simplicifolia, Alexa Fluor 647 Conjugate | Thermo Fisher (Molecular Probes) | I32450 | Used in media to label Microglia/Macrophages |
pRetroX-IRES-ZsGreen1 Vector | Clonetech | 632520 | |
Retro-X Concentrator | Clonetech | 31455 | Binding resin for non-ultracentrifugation concentration of viral supernatants |
pVSG-G Vector | Clonetech | 631530 | part of the Retro-X Universal Retroviral Expression System |
GP2-293 Viral packaging cells | Clonetech | 631530 | part of the Retro-X Universal Retroviral Expression System |
Cyanoacrylate Glue (Super Glue) | Sigma-Aldrich | Z105899 | Medium-viscosity |
Equipment | |||
Peel-A-Way Embedding Mold (Square – S22) | Polysciences, Inc. | 18646A-1 | Molds for tumor sample embedding |
Stainless Steel Micro Spatulas | Fisher Scientific | S50823 | Bend instrument 45 degrees at the neck of the spoon blade |
Curved Fisherbrand Dissecting Fine-Pointed Forceps | Fisher Scientific | 08-875 | |
Single Edge Razor Blade (American Safety Razors) | Fisher Scientific | 17-989-001 | Blade edge is 0.009" thick. Crimped blunt-edge cover is removed before loading onto vibratome. |
Leica VT1000 S Vibratome | Leica Biosystems | VT1000 S | |
Hydrophilic PTFE cell culture insert | EMD Millipore | PICM0RG50 | 30 mm, hydrophilic PTFE, 0.4 µm pore size |
35 mm Glass Bottom Dishes | MatTek | P35G-1.5-20-C Sleeve | 20mm glass diameter. Coverslip glass thickness 1.5 |
LSM 510 Confocal Micoscope | Zeiss | LSM 510 | 10x Air Objective (c-Apochromat NA 0.45) |
PECON Stagetop Incubator | PeCON Germany | (Discontinued) | Incubator PM 2000 RBT is a comprable product designed for use with Zeiss Microscopes. |