Patient-derived xenografts of glioblastoma multiforme can be miniaturized into living microtumors using 3D human biogel culture system. This in vivo-like 3D tumor assay is suitable for drug response testing and molecular profiling, including kinomic analysis.
The use of patient-derived xenografts for modeling cancers has provided important insight into cancer biology and drug responsiveness. However, they are time consuming, expensive, and labor intensive. To overcome these obstacles, many research groups have turned to spheroid cultures of cancer cells. While useful, tumor spheroids or aggregates do not replicate cell-matrix interactions as found in vivo. As such, three-dimensional (3D) culture approaches utilizing an extracellular matrix scaffold provide a more realistic model system for investigation. Starting from subcutaneous or intracranial xenografts, tumor tissue is dissociated into a single cell suspension akin to cancer stem cell neurospheres. These cells are then embedded into a human-derived extracellular matrix, 3D human biogel, to generate a large number of microtumors. Interestingly, microtumors can be cultured for about a month with high viability and can be used for drug response testing using standard cytotoxicity assays such as 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) and live cell imaging using Calcein-AM. Moreover, they can be analyzed via immunohistochemistry or harvested for molecular profiling, such as array-based high-throughput kinomic profiling, which is detailed here as well. 3D microtumors, thus, represent a versatile high-throughput model system that can more closely replicate in vivo tumor biology than traditional approaches.
The most common primary intracranial malignant brain tumors are grade III astrocytomas and grade IV glioblastoma multiforme (glioblastoma or GBM). These tumors offer poor prognoses with median one-year survival between 12 – 15 months with current therapies for GBM in the US1-3. Multimodality therapies include surgery, radiation, and chemotherapy including temozolomide (TMZ) and kinase-targeted agents. Kinase signaling is frequently dysregulated in GBM, including subsets of tumors with amplification or activating mutations in the Epidermal Growth Factor Receptor (EGFR), increases in Platelet Derived Growth Factor Receptor (PDGFR) signaling, increased Phosphatidyl-Inositol-3 Kinase (PI3K) and tumor supporting angiogenic signaling through Vascular Endothelial Growth Factor Receptor (VEGFR) as well as other kinase driven pathways4-6. Current in vitro and in vivo models frequently lose these representative alterations7. Additionally, genetic profiling has not offered the anticipated benefits that may reflect the fact that genetic and epigenetic changes do not always predict changes at the level of protein activity, where most kinase targeting agents act directly, and where therapies with other mechanisms of action may act indirectly.
The traditional immortalized cell line that can be passaged ad infinitum has long been the standard for drug testing due to their ease of maintenance and reproducibility. However, this model suffers from a high nutrient (and artificial) growth environment that selects for fast growing cells that differ greatly from the original tumor. As such, there has been considerable interest in developing more realistic model systems that reflect a more complex tumor biological system as is present in the patient. Tumor xenografts developed directly from a primary tumor grown in mice ("xenoline," patient-derived xenograft or PDX) provide a more reflective model system, particularly in the setting of cancer therapeutics, as they are felt to more reliably predict clinical success.8 Despite the more reflective biology, these models are expensive and are difficult to establish and maintain. Moreover, they are not amenable to high-throughput studies. The need to better develop biologic models that more accurately reflect molecular alterations in the primary tumors, and to profile and test these models using direct measures of kinase activity, not surrogate genetic markers, is clear.
It is well recognized that unlike two-dimensional (2D) monolayer cultures, 3D or multicellular assay models can provide more physiologically relevant endpoints9-11. Common 3D culture approaches involve matrix-coated microcarriers and cell spheroid formation. Tumor spheroids can be generated via cellular aggregation using spinner flask, pHEMA plate and hanging drop techniques. Limitations for these approaches include: inability for some cells to form stable spheroids, variability in growth and challenges with mixed cell types. Alternatively, many synthetic (hydrogel, polymer) and animal-derived Engelbreth-Holm-Swarm (EHS) matrix from mouse sarcomas, bovine collagen) matrices have been developed for 3D culture studies12-14. Mouse EHS matrix is extensively used but known to promote cell growth and differentiation in vitro and in vivo15.
In order to replicate 3D tumor biology, a human biomatrix system was developed by Dr. Raj Singh et al.16. The natural, growth factor-free human biogel allows 3D culture scaffolds (beads, discs), which support long-term cultivation of multiple cell types. A series of 3D human biogel culture designs are established for studying tumor growth, adhesion, angiogenesis and invasion properties. Advantages and properties of human biogel as compared to common mouse EHS gels are summarized in Table 1 and Table 2.
Source: | Human Amnions (Pooled tissue) Pathogen-free, IRB-exempt/approved |
ECM nature: | Non-denatured Biogel (GLP-production) |
Key Components: |
Col-I (38%), Laminin (22%), Col-IV (20%), Col-III (7%), Entactin & HSPG (< 3%) |
GF-free: | Undetectable EGF, FGF, TGF, VEGF, PDGF (Non-angiogenic, Non-toxic) |
Table 1: Properties of Human Biogel as Compared to Common EHS Gels.
Human Biogel | EHS gels |
Natural human matrix | Reconstituted mouse matrix |
Controlled cell growth & differentiation | Can promote cell growth & differentiation |
Physiologic gene expression | Variable gene expression |
3D tissue-like culture model | Plate-based culture model |
Table 2: Advantages of Human Biogel as Compared to Common EHS Gels.
NOTE: All xenograft therapy evaluations were done using an orthotopic tumor model for glioblastoma on a protocol approved by the Institutional Animal Care and Use Committee.
1. Isolation of Patient-derived GBM Xenograft Cells
2. Alternate Tissue Disaggregation Protocol
3. Microtumor Generation
4. Morphological and Phenotypic Analysis of Microtumors
5. Kinomic Profiling of Microtumors
We have shown that 3D biogel culture system supports long-term growth and function of multiple cell types. In this collaborative project, patient-derived GBM xenolines (PDX) are used for producing hundreds of microtumors. Dissociated cells (3 x 105) or neurospheres (40 – 50) were embedded in biogel beads (2 mm) and after quick gelation they are cultured in a NB-media filled custom bioreactor. Cellular viability (Calcein-AM), growth profile (MTT), and kinomic activity array-based analysis were determined. PDX microtumors maintained multicellular organization and high viability (> 80%) for > 3 weeks. We believe this in vivo-like biology is due to maintenance of 3D cell-matrix architecture (not possible with 2D or spheroid culture). It is also observed that 3D tumors are difficult to produce with fragile gelatinous protein scaffold, possibly due to lack of collagen I14. A working scheme for microtumor production using PDX cells and 3D biogel system is summarized in Figure 1 and the dissociation process is depicted in Figure 2.
Through live cell imaging via Calcein-AM, we determined cell growth and viability as well as effects of the drugs on the GBM cells derived from the PDX tumor JX10, as indicated by a decrease in fluorescence, signaling cell death. Tumor growth was measured at day 0, 7 and 14 for microtumors displaying continuous growth as shown in Figure 3A. This microtumor, JX10, is known to be resistant to TMZ. When exposed to 1 – 10 µM TMZ, minimal growth suppression was noted (Figure 3B). However, testing PDX tumor JX12, which is known to be sensitive to TMZ, demonstrated sensitivity by MTT assay at 14 days that was confirmed by Calcein-AM imaging (Figure 4).
In order to explore potential mechanisms of drug resistance, we measured kinase signaling in the TMZ resistant microtumors (JX10). Kinomic profiles for 144 tyrosine and 144 serine/threonine phosphopeptide targets were captured for DMSO and 10 µM TMZ treated microtumors. Kinomic phosphorylation intensity for all peptide targets, per cycle, and per multiple exposure times was captured (Data not shown). A representative heatmap displaying peptides that had significantly altered intensities with 10 µM TMZ treatment (p < 0.05, unpaired students T-test) is displayed in Figure 5A. Kinomic profiles that were altered in TMZ treated JX10 microtumors were analyzed using the upstream kinase prediction software that identified SYK, LCK, and CTK kinases as increased in TMZ treated samples relative to DMSO (Figure 5B). Serine/threonine kinome upstream kinase analysis was also performed (Figure 5C).
Figure 1: Working Scheme for Microtumor Production. GBM-PDX tumors are dissociated from the murine host, and single cells are embedded in biogel matrix to form microtumors. Analyses are then performed on the microtumors. Please click here to view a larger version of this figure.
Figure 2: Dissociation of PDX Tumor Cells. Tumors are removed from host mice and minced prior to mixing with enzyme solution (A-G). Dissociation is shown after 40 min (H). Tumor cell dissociator is shown (I). Trituration in media and filtration (J-N) is shown with an end result of 0.3 ml pellet (O) containing 29.5 x 106 viable cells (P) that form spheroids in NBM in 24 hr (Q). Scale bar is 20 µm. Please click here to view a larger version of this figure.
Figure 3: Growth of Microtumors. JX10 microtumor growth measured by optical density (O.D.) of MTT and representative images over 14 days, both basally (A), and in response to 1-10 µM TMZ treatment (B). Images are shown at 4X magnification (scale bar = 500 µm). Standard deviation indicated. Please click here to view a larger version of this figure.
Figure 4: Growth of Microtumors. JX12 microtumor growth measured by O.D. of MTT and representative images over 14 days, both basally (Day 1), and in response to 0, 1, or 10 µM TMZ treatment (scale bar = 500 µm). Standard deviation indicated. Please click here to view a larger version of this figure.
Figure 5: Kinomic Profiles of TMZ Treated Microtumors. Heatmap of exposure-integrated values (log2 transformed slopes (x 100) of 10, 20, 50, 100, and 200 ms median signal minus background exposure values) are displayed for significantly TMZ-altered altered peptides (p < 0.05). Red indicates increased, and blue indicates decreased relative to DMSO mean per peptide. TMZ altered profiles were compared using the upstream kinase prediction software and altered kinases. Normalized Kinase Statistic (NKS) score > 1.0 and specificity score > 1.3 (red) are considered highly altered. Please click here to view a larger version of this figure.
Critical steps within the protocol predominantly relate to microtumor generation, as well as drug dosing and maintenance. Because the microtumor beads are fragile and easily torn, extreme care is needed in both developmental stages of an assay and maintenance. If an error occurs during either of these processes, experimental interpretation can be compromised, causing extension or unnecessary repetition of the experiments or even exclusion of data.
Modifications and troubleshooting, especially of the microtumor development process, included the design and production of a custom hydrophobic tool for the use of making the microtumor beads. This tool allows for faster and more accurate production of microtumors. Additionally, small modifications to the maintenance of the microtumors helped to speed up the process of changing medium and dosing the cells. Several of these modifications included, but were not limited to, using a multichannel electronic pipette to remove and replace medium from the 96-well plates and pre-mixing fresh drug dosing solutions and arranging the 2x solutions in a corresponding 2 ml 96-well plate.
Major optimizations for culture conditions in 3D modeling includes determining appropriate dosages for drug compounds, as there is a poor relation between 2D modeling and 3D modeling. Therefore, serial dilutions to establish a dose titration curve is often necessary for moving a 2D model system to the 3D microtumor model.
Potential issues to consider with microtumor generation from xenoline tumor cells harvested from a mouse include bacterial or fungal contamination, inconsistent availability of primary xenolines, as unique growth properties in mice can produce differences in harvest time and variance in cell yields from separate dissociations. With each cell line, the number of cells received, and similarly, how many microtumors can be produced will vary. As such, there is a need to prioritize which assays are required and which ones are "optional" should the microtumor yield be insufficient. Potential limitations with the kinomic profiling of microtumors include the difficulty in correcting for inert proteinaceous material loading, as samples are loaded in a manner to correct for BCA determined protein levels, that may not distinguish between biogel based proteins (ECM proteins) and those of the cellular kinase component that is intended to be measured. Measuring of gross microtumor mass via Calcein-AM, or using a housekeeping protein as a correction factor may mitigate this, although issues with housekeeping protein levels in tumors being altered may confound this. For short time-course signaling experiments, assuming equally sized (kinase containing live cell number) and paired treated and untreated samples, this is less of an issue.
With this research, the goal is to provide a more cost-effective and disease-representative preclinical model for accurate drug therapy testing in comparison to orthotopic patient-derived xenografts, the current gold standard model for GBM testing. In recent years, several groups have attempted to model the in vivo GBM environment for drug testing purposes. Some groups have simply tried to grow GBM tumor cells in non-differentiating conditions to preserve GBM stem cells or at least brain tumor initiating cells (BTIC)24,25. These tumorsphere or neurosphere cultures can be used for drug testing and are likely superior to traditional models. However, since they still lack the tumor-stroma interplay that is preserved in our microtumor model, the tumorsphere approach is still limited. Other groups have attempted to apply engineering principles to improve GBM models26-28. These approaches have included flow chambers, variable ECM proteins and tumor cell/normal cell mixtures. While promising, almost all of these reports have used immortalized cells lines with the caveats mentioned previously. As such, we believe that the PDX-based microtumor model described here is more clinically relevant.
In the future, the microtumor model could be used as either a true tumor avatar or a 'proband' system. With the proband system, the PDX developed from a particular patient does not directly inform the clinician regarding that specific patient's therapy, as with the tumor avatar system. Instead, a patient's tumor is "matched" to a pre-existing, well-characterized (both molecularly and phenotypically) PDX29. Indeed, this model could serve as a 'go-to' avatar or reside in a proband library as a comparative profile. With traditional avatars, growing a patient's tumor cells in mice in parallel to the patient's treatment is not only time consuming, as it may take several months for the primary passage, but also costly, as testing several drug therapies in mice generate an overwhelming price tag. To combat this, the primary patient tumor can be implanted directly into the biogel matrix. With the microtumors, results could be generated quickly and at a lower cost.
As a proband model, the microtumor kinome data profiles and drug response could be added into a proband 'library,' along with profiles and data from previous 3D in vitro models, existing PDXs, animal studies, other patients, etc. In theory, the patient's tumor cells could then be 'omically' (genomically, kinomically, transcriptomically, etc.) matched to a similar existing microtumor profile to determine accurate treatment for the best possible outcome.
Summary: Here we identify that these microtumor models can be used to measure both phenotypic growth and can be queried at both the basal kinase activity level and in response to drug treatment that may be useful as a translational tool for further preclinical research. Developing accurate, and molecularly valid testable models for human tumors is imperative for effective drug development.
The authors have nothing to disclose.
Supported by NIH R21 grant (PI: C. Willey, CA185712-01), Brain Tumor SPORE award (PD: G.Y. Gillespie, P20CA 151129-03) and SBIR contract (PI: R. Singh, N43CO-2013-00026).
Collagenase-I | Sigma-Aldrich | CO130 | |
Trypsin EDTA (10X) | Invitrogen | 15400-054 | |
Neurobasal-A | Life Technologies | 10888-022 | |
N-2 Supplement | Life Technologies | 17502-048 | 1x final concentration |
B-27 Supplement w/o Vitamin A | Life Technologies | 12587-010 | 1x final concentration |
Recombinant Human FGF-basic | Life Technologies | PHG0266 | 10 ng/mL final concentration |
Recombinant Human EGF | Life Technologies | PGH0315 | 10 ng/mL final concentration |
L-Glutamine | Corning Cellgro Mediatech | 25-005-CI | 2 mM final concentration |
Fungizone | Omega Scientific | FG-70 | 2.5 ug/mL final concentration |
Penicillin Streptomycin | Omega Scientific | PS-20 | 100 U/mL Penicillin G, 100 ug/mL Streptomycin final concentration |
Gentamicin | Life Technologies | 15750-060 | 50 ng/mL final concentration |
MTT | Life Technologies | M6494 | prepared to 5 mg/mL in PBS and sterile filtered, 1 mg/mL in well |
SDS | Fisher | BP166 | for MTT lysis buffer, prepared to 10% in 0.01M HCL, 5% in well |
HCl | Fisher | A144SI-212 | for MTT lysis buffer, prepared to 0.01M with SDS, 5 mM in well |
Calcein AM | Life Technologies | C1430 | 1 mM in DMSO stock, 2 uM in PBS staining solution, 1 uM in well |
Halt’s Protein Phosphatase Inhibitor cocktail | Pierce ThermoScientific | 78420 | 1:100 ratio in MPER |
Halt's Protein Protease Inhibitor | Pierce ThermoScientific | 87786 | 1:100 ratio in MPER |
Mammalian Protein Extraction Reagent (MPER) | Pierce ThermoScientific | PI78501 | |
Trypan Blue | Pierce ThermoScientific | 15250-061 | |
DMSO | Fisher | BP231 | for dissolution of calcein AM & compounds |
Phosphate-Buffered Saline without Ca/Mg | Lonza | 17-517Q | diluted to 1X with MiliQ ultrapure water and sterile filtered (for cell culture) |
Dulbecco's Phosphate-Buffered Saline with Ca/Mg | Corning Cellgro Mediatech | 20-030-CV | diluted to 1X with MiliQ ultrapure water (for pre-fixation wash) |
10% Neutral Buffered Formalin | Protocol | 032-060 | |
Trypan Blue | Pierce ThermoScientific | 15250-061 | |
High Density Hubiogel | Vivo Biosciences | HDHG-5 | |
Halt's Protein Phosphatase Inhibitor | Pierce | 78420 | |
Halt's Protein Protease Inhibitor | Pierce | 87786 | |
Mammalian Protein Extraction Reagent (MPER) | Thermo Scientific | 78501 | |
Protein Tyrosine Kinase (PTK) Array Profiling chip | PamGene | 86312 | |
PTK kinase buffer | PamGene | 36000 | 300 µl 10X PK buffer stock in 2.7 ml dH20, catalog number for PTK reagent kit |
ATP | PamGene | 36000 | catalog number for PTK reagent kit |
PY20- FITC-conjugated antibody | PamGene | 36000 | catalog number for PTK reagent kit |
PTK Additive | PamGene | 32114 | |
PTK-MM1 tube (10X BSA) | PamGene | 36000 | catalog number for PTK reagent kit |
Serine/Threonine Kinase (STK) Array Profiling chip | PamGene | 87102 | |
STK kinase buffer | PamGene | 32205 | catalog number for STK reagent kit |
STK Primary Antibody Mix (DMAB tube) | PamGene | 32205 | catalog number for STK reagent kit |
FITC-conjugated Secondary Antibody | PamGene | 32203 | |
STK-MM1 tube (100X BSA) | PamGene | 32205 | catalog number for STK reagent kit |
STK Antibody Buffer | PamGene | 32205 | catalog number for STK reagent kit |
Equipment | |||
#11 Blades, sterile | Fisher | 3120030 | |
#3 scalpel handles, sterile | Fisher | 08-913-5 | |
100mm glass Petri dishes | Fisher | 08-748D | |
Semicurved forceps | Fisher | 12-460-318 | |
Trypsinizing flask | Fisher | 10-042-12B | |
Magnetic stirrer | Fisher | 14-490-200 | |
3/4" stir bar | Fisher | 14-512-125 | |
B-D cell strainer | Fisher | #352340 | |
B-D 50ml Centrifuge tube | Fisher | #352098 | |
PamStation 12 | PamGene | ||
BioNavigator 6.0 kinomic analysis software | PamGene | ||
Evolve Kinase Assay Software | PamGene | ||
UpKin App software (upstream kinase prediction) | PamGene | ||
gentleMACS Dissociator | Miltenyi Biotec | 130-093-235 | |
Rotary Cell Culture System (RCCS) | Synthecon | RCCS-D | with 10 mL disposable HARV |