Orthotopic intracranial injection of tumor cells has been used in cancer research to study brain tumor biology, progression, evolution, and therapeutic response. Here we present fluorescence molecular tomography of tumor xenografts, which provides real-time intravital imaging and quantification of a tumor mass in preclinical glioblastoma models.
Tumorigenicity is the capability of cancer cells to form a tumor mass. A widely used approach to determine if the cells are tumorigenic is by injecting immunodeficient mice subcutaneously with cancer cells and measuring the tumor mass after it becomes visible and palpable. Orthotopic injections of cancer cells aim to introduce the xenograft in the microenvironment that most closely resembles the tissue of origin of the tumor being studied. Brain cancer research requires intracranial injection of cancer cells to allow the tumor formation and analysis in the unique microenvironment of the brain. The in vivo imaging of intracranial xenografts monitors instantaneously the tumor mass of orthotopically engrafted mice. Here we report the use of fluorescence molecular tomography (FMT) of brain tumor xenografts. The cancer cells are first transduced with near infrared fluorescent proteins and then injected in the brain of immunocompromised mice. The animals are then scanned to obtain quantitative information about the tumor mass over an extended period of time. Cell pre-labeling allows for cost effective, reproducible, and reliable quantification of the tumor burden within each mouse. We eliminated the need for injecting imaging substrates, and thus reduced the stress on the animals. A limitation of this approach is represented by the inability to detect very small masses; however, it has better resolution for larger masses than other techniques. It can be applied to evaluate the efficacy of a drug treatment or genetic alterations of glioma cell lines and patient-derived samples.
Cancer is one of the leading causes of illness-related deaths in humans in the industrialized world. With an extremely high death toll, new treatments are urgently required. Glioblastoma multiforme (GBM) is an extremely lethal type of brain cancer, composed of heterogeneous populations of brain tumor, stromal, and immune cells. According to the Central Brain tumor registry of the USA, the incidence of primary malignant and non-malignant brain tumors is approximately 22 cases per 100,000. Approximately 11,000 new cases are expected to be diagnosed in the USA in 20171.
Preclinical studies investigate the likelihood of a drug, procedure, or treatment to be effective prior to testing in humans. One of the earliest laboratory steps in preclinical studies is identifying potential molecular targets for drug treatment by using cancer cells implanted in a host organism, defined as human xenograft models. Within this context, intracranial brain tumor xenograft models using patient-derived xenografts (PDXs) have been widely used to study brain tumor biology, progression, evolution, and therapeutic response, and more recently for biomarkers development, drug screening, and personalized medicine2,3,4.
One of the most affordable and non-invasive in vivo imaging methods to monitor intracranial xenografts is bioluminescence imaging (BLI)5,6,7,8. However, some BLI limitations include substrate administration and availability, enzyme stability, and light quenching and scattering during imaging acquisition9. Here we report the infrared FMT as an alternative imaging method to monitor preclinical glioblastoma models. In this method, signal acquisition and quantification of intracranially implanted PDXs, expressing a near-infrared fluorescent protein iRFP72010,11 (henceforth termed as FP720) or turboFP635 (henceforth termed as FP635), is performed with a FMT imaging system. Using the FMT technology, the orthotopic tumors can be monitored in vivo before, during, or after treatment, in a non-invasive, substrate-free, and quantitative manner for preclinical observations.
The use of experimental research animals and infectious agents, such as lentivirus to transduce the cancer cells, require prior approval by the institutional animal care program and by the institutional biosafety committee. This protocol follows the animal care guidelines of the University of California San Diego (UCSD).
1. Labeling of Glioblastoma Cells with FP635 or FP720 Construct
2. Intracranial Injection of iRFP-tagged Glioblastoma Cells into Immunodeficient Mice
NOTE: Before starting the surgery, make sure that the surgical room and tools are clean for the procedure. Use immunodeficient athymic nude (Foxn1nu) male or female mice, between 4-5 weeks old and 17-19 g for intracranial injections. Animals should be housed for at least 3 days after arrival and before surgery.
3. FMT Imaging
NOTE: According to the experimental aim, the iRFP-tagged glioblastoma cells can be monitored in vivo before, during, or after treatment. For imaging purpose, anesthetize animals using an isoflurane induction chamber, maintain in an imaging cassette during the scanning, and image in the docking station of the FMT imager.
4. Image Analysis
Glioblastoma cells U87EGFRvIII (U87 cells over-expressing the EGF receptor variant III) were cultured according to the step 1.2. Lentivirus was produced and purified according to step 1.1. The viral concentration was determined by p24 ELISA analysis. Cells were transduced with lentivirus carrying infrared fluorescent proteins according to step 1.8. The plasmid encoding FP72010,11 was kindly provided by Dr. V.V. Verkhusha and the FP635 vector was purchased from a commercial vendor. Target cells were FACS sorted to enrich for the top 10% most fluorescent cell population (Figure 1A). Increasing the fluorescent signal improves the sensitivity of the fluorescence tomography system. To validate the intensity of the signals of fluorescently-labeled cells prior to orthotopic implantation, we used the phantom cassette of the FMT imager. Specifically, U87EGFRvIII-TurboFP635 and U87EGFRvIII-iRFP720 cells were dissociated with trypsin, and 1 x 105, 1 x 106, and 1 x 107 cells were resuspended in 100 µL of ice cold PBS, respectively and analyzed. The FMT imager allows for detection in 4 channels: 635 nm, 680 nm, 750 nm, and 780 nm. Far infrared channels 750 nm and 780 nm are not suitable for this analysis because there is no emission at such wavelengths from the near-infrared fluorescent proteins used (data not shown). Therefore, we analyzed the emission signals at 635 nm and 680 nm. Reported in Figure 1, the fluorescent signal image and intensity quantification for both FP635 (Figure 1B, C) and FP720 (Figure 1D, E), which increases proportionally as the cell number increases. Of note, the emission for FP635 is visible in the 635 nm channel but not in the 680 nm channel. In contrast, the emission for FP720 is visible in the 680 nm channel but not in the 635 nm channel. This makes these two candidate fluorescent proteins suitable for studies where two populations of cells can be labeled and monitored simultaneously using two near infrared fluorescent labels.
U87EGFRvIII cells labeled with near infrared fluorescent proteins were then used for orthotopic injection and in vivo tumor growth monitoring, as described in steps 2.1-2.4. We used the same cell line, i.e., U87EGFRvIII, labeled with different fluorescent tags, to exclude tumor growth difference between different cancer cell types. The mice were injected with U87EGFRvIII-TurboFP635 cells, U87EGFRvIII-iRFP720 cells, or an equal combination (1:1) of the two cell lines. A total of 5 x 105 cells in a 5 µL volume of PBS were injected intracranially into 4-5 weeks old immunocompromised athymic nude mice using a stereotactic system and a Hamilton syringe, following steps 2.3-2.4. Anesthesia was accomplished by intraperitoneal injection of ketamine/xylazine. The animals were kept warm throughout the injection using a water thermal pad and checked for any signs of distress. Tumor engraftment and growth was then monitored by the FMT imaging system. During the scanning sessions, the mice were temporarily anesthetized using a 3% isoflurane chamber (steps 3.1-3.12). The mice were then placed in an imaging cassette and imaged inside the laser scanning camera. Each mouse was scanned using the 635 nm and 680 nm channels. The signal quantification was recorded over time and mice were euthanized at the onset of neurological symptoms, according to the Institutional Animal Care and Use Committee (IACUC) guidelines. As indicated in Figure 2A, U87EGFRvIII-TurboFP635 emission signal was evident in the 635 nm channel and minimal background signal was recorded in the 680 nm channel (Figures 2B, C). In contrast, U87EGFRvIII-iRFP720 cells emitted in the 680 nm channel showing little background in the 635 nm channel (Figures 2D–F). Finally, mice injected with a 1:1 mix of U87EGFRvIII-TurboFP635 cells and U87EGFRvIII-iRFP720 cells showed increased signal in both channels for the duration of the experiment (Figures 2G–I). These data show that it is possible to utilize infrared fluorescent proteins for dual in vivo imaging of cancer cell populations. The data also indicate that different fluorescent proteins have a different intensity in their signal emission. Here, FP720 signal intensity was superior to FP635, with increased sensitivity in detecting smaller populations of cancer cells in a non-invasive monitoring setting.
The effects of gene silencing of cancer-related genes on tumorigenicity in preclinical models can also be study with this protocol. For this aim, GBM-spheres (GSC23 and GSC11) were cultured (step 1.2) and transduced with FP720-lentivirus (step 1.8). Cells expressing FP720 were FACS sorted as described previously in steps 1.11-1.13. FP720-labeled cells were co-transduced with lentivirus-encoding shRNA targeting DAXX (shDaxx,) or shRNA control (shLuc) at MOI 516. Cells were cultured for 5 days, dissociated into a single cell suspension, and 1 x 106 cells were resuspended in 3 µL of PBS for intracranial injection. The efficacy of the shRNA targeting DAXX was confirmed by reverse transcription quantitative polymerase chain reaction (RT-qPCR) in GSC23 (Figure 3A) and GSC11 (Figure 3C). GBM-iRFP720/shRNA cells were stereotactically injected into the striatum of immunodeficient mice, according to steps 2.3-2.4. Animals were observed for neurological signs and the relative fluorescence signal of the xenografts were analyzed by the FMT imaging system, quantified using analysis software, according to steps 3.1-4.6. Representative images of FMT show a decrease fluorescence signal in mice engrafted with shDAXX/GBM-spheres in comparison with animals implanted with shControl (Figure 3B and Figure 2D) for both GBM-spheres models, confirming our previous finding16 that DAXX inhibition represses tumor growth in preclinical glioblastoma models.
Figure 1: Standard curves using cells for phantom and FACS analysis. (A) Representative FACS histograms of FP720- and FP635-U87EGFRvIII expressing cells (left and right panel, respectively). Glioma cells were infected with lentivirus encoding infrared fluorescent proteins and FASC sorted to enrich for the highest expression of the FP720 or FP635 proteins. Representative images obtained from the imager and analyzer software of U87EGFRvIII-TurboFP635 (B) and U87EGFRvIII-iRFP720 (D) cell suspensions using the phantom cassette. Upper panels represent the signal in the 635 nm channel and bottom panels in the 680 nm channel. Increase in the cell number corresponded to an increase in signal visualization. Signal quantification of U87EGFRvIII-TurboFP635 (C) and U87EGFRvIII-iRFP720 (E) cells in 635 nm and 680 nm channels. Relative fluorescence units are indicated in the color bar. Error bars represent S.E.M. from 3 different scans per cell suspension. Please click here to view a larger version of this figure.
Figure 2: Brain tumor xenograft model using U87EGFRvIII cells scanned in different channels. Representative images of fluorescent signal in 635 nm and 680 nm channels from animals intracranially implanted with U87EGFRvIII-turboFP635 (A), U87EGFRvIII-iRFP720 (D), and a mix combination (1:1) of U87EGFRvIII-iRFP720:U87EGFRvIII-turboFP635 cells (G). Images were acquired at day 5 after first scanning. (B, E, H) Signal intensity monitoring for each mouse over a period of 5 days. Each mouse was scanned in the 635 nm channel (blue bars) and in the 635 nm channel (red bars). The lightest shade of the bar color represents day 1 of scanning and the darkest shade represents day 5. (C, F, I) Relative fluorescence quantification showing a direct comparison of signal intensity from the 635 nm channel (blue) and 680 nm channel (red) in the entire cohort of mice. Relative fluorescence units are indicated in the color bar. Data show the mean values with standard deviation from 3 different animals. Please click here to view a larger version of this figure.
Figure 3: Gene silencing in preclinical glioblastoma models using FMT. Gene expression analysis of DAXX by RT-qPCR in GSC23 (A) and GSC11 (C) glioma patient-derived cells transduced with lentivirus-encoded shRNAs targeting DAXX (shDaxx) or shRNA Control (shLuc) and FP720. Representative images of fluorescent signal from the 680 nm channel of orthotopically engrafted mice implanted with GSC23-iRFP720 (B) and GSC11-iRFP720 (D) patient-derived cells. Strong fluorescence signal is observed in animals implanted with shRNA Control (shLuc) (top part of the panel) in comparison with mice implanted with shRNA-DAXX (shDaxx) (bottom part of the panel), indicating that DAXX inhibition suppresses tumor growth in preclinical glioma models determined by fluorescence molecular tomography. Relative fluorescence units are indicated in the color bar. Please click here to view a larger version of this figure.
Tumor xenografts have been extensively used in cancer research and a number of well-established imaging techniques has been developed: BLI; magnetic resonance imaging (MRI); positron emission tomography (PET), computed tomography (CT); FMT. Each of these approaches comes with pros and cons, but ultimately complement each other with the type of information provided. One of the most commonly used in vivo imaging technology is BLI5,6,7,8. BLI and FMT both require cell engineering (with luciferase enzymes or infrared fluorescent proteins) that can affect gene expression. BLI is more sensitive for small tumor masses but requires intraperitoneal injection of a substrate (luciferin), which reaches maximum distribution in the body in 10-12 min; but light quenching and scattering during imaging acquisition, and substrate clearance frequently occur9. FMT, instead, does not require any additional substrate administration to the animals and its signal is stable and independent of time or enzyme stability. Additionally, FMT, when used in the same experimental conditions, offers reproducible semi-quantitative data. BLI and FMT both do not offer spatial resolution; however, coupling these methods with CT scan addresses this issue.
A CT scan measures the attenuation of photons when its signal crosses the body of an animal and it is ideal in identifying body structures. However, soft tissue contrast, which might interfere with the detection of a small intracranial tumor, is a limitation for this approach17,18. One additional disadvantage of CT is the use of radiation and contrast media, which can induce molecular changes in the target cells. A CT scan can be combined with PET, which uses radiolabeled tracers like the glucose analogue, fluorodeoxyglucose 18F-FDG; but this can also interfere with some physiological processes in small animals. In contrast, FMT does not need any injected substrate to measure the tumor mass and it is possible to measure the tumor metabolism, inflammation, angiogenesis, apoptosis, and other markers using fluorescently labeled biomarkers19.
Overall one of the most useful and versatile techniques for in vivo imaging is the MRI. Minor limitations of the MRI can be found in the low acquisition time, lower sensitivity than PET, and some instrument-related variations18,20.
Here, we report the FMT method as an alternative method for preclinical glioblastoma models. Pre-labeling of target cells with infrared fluorescent proteins (Figure 1A–E) and scanning post-implantation in immunocompromised animals can be achieved with a fluorescence tomography imaging system. We reported the FMT method as an alternative method for preclinical glioblastoma models. With this method imaging is applied in a substrate-free and non-invasive manner, animals are kept in a low stress environment, and deep-tissue quantification can be performed (Figure 2A–I, Figure 3A, B). This protocol can be applied to study the efficacy of combinatorial treatments in preclinical glioma models21, to identify new therapeutic targets for brain tumors16, and to investigate new druggable metabolic molecules22, epigenetic pathways23, or new chemotherapeutic agents in combination with radiotherapy (unpublished data), and applied to other preclinical studies. We also propose here the use of the FMT as a dual in vivo imaging approach, when two co-existing, but different, tumor cell populations are meant to be analyzed.
Some limitations with FMT must be taken into consideration, such as auto-fluorescence generated from some organs that can interfere with the signal. For example, murine spleen and liver emit auto-fluorescence in the 680 nm channel but not at 635 nm. Therefore, FMT-related experimental procedures involving these organs must be performed using other infrared fluorescent proteins, like FP635. We used nude foxn1-mutated nude mice to avoid interference in the fluorescence signal recording; if a different mouse model must be used, it is recommended to shave the area of interest of the animals. Additionally, this approach is also limited in detecting very small tumor masses. In our experience, the signal-to-noise ratio improves as the tumor mass becomes larger, yet before the onset of any neurological symptom in the mice. In fact, in previous experiments conducted on orthotopically injected PDXs, we were able to detect tumors during a drug regimen for at least 2 weeks in mice without any sign of distress due to tumor burden or drug administration21. Ultimately, the aggressiveness of each cell line and the number of injected cells will determine the length of each experiment. The data show a linear proportional relationship between number of cells and signal (Figure 1B–E), and tumor mass and signal (Figure 2); but a general signal correlation between tumor mass and number of cells cannot be applied across different cell lines and infrared proteins. The signal for each cell line/infrared fluorescent protein combination must be determined empirically. Moreover, the signal obtained from the analysis of the cells using the phantom (Figure 1B, D) does not correspond to the signal from intracranial xenografts as an attenuation in signal occurs due to the skull of the mouse.
The imager and analyzer software allow for thresholding. Although we did not use any threshold here, scanning animals without any implanted cells gives background noise which tends to be eliminated by reconstruction in presence of a true signal. A few factors influence the signal intensity of the xenograft and must be kept in consideration: each infrared fluorescent protein has different emission spectrum and intensity, with FP720 being one of the most fluorescent proteins used experimentally10,11; each cell line allows for an abundant, yet limited, amount of protein production, thus influencing the maximum fluorescence obtainable. Finally, the FMT imager is setup with internal standard curves for specific compounds, therefore, it is important to consider which compound has the closest emission spectrum to the infrared fluorescent protein used to label the cells. Although well tolerated, it is recommended to determine potential cytotoxic effects mediated by the overexpression of the infrared proteins8,9. Moreover, we recommend using cohorts of mice of at least 8 per experimental group, as loss of data and mice, due to surgical procedure or variability in the cancer cell engraftment, might occur; nonetheless, this adds statistical significance to the study. We also recommend to serially scan the animal cohort for more than one day and analyze the signal emission at the same time, since variations in the imaging cassette depth and fluorescent signal reconstruction might interfere with the final signal quantification. To solve this issue, the analyzer software allows for minor changes in laser intensity (from normal to high and very high), although in our experience the final outcome is not dramatically different. One final note regarding cost-effectiveness is that the FMT imaging system can be easily shared as a core instrument among different laboratories, which helps defray its purchase and maintenance cost. In conclusion, the FMT is a valuable resource that makes experimental and preclinical evaluation of in vivo tumor growth easy and reliable.
The authors have nothing to disclose.
We thank Dr. Frederick Lang, MD Anderson Cancer Center for GBM-PDX neurospheres. This work was supported by the Defeat GBM Research Collaborative, a subsidiary of National Brain Tumor Society (Frank Furnari), R01-NS080939 (Frank Furnari), the James S. McDonnell Foundation (Frank Furnari); Jorge Benitez was supported by an award from the American Brain Tumor Association (ABTA); Ciro Zanca was partially supported by an American-Italian Cancer Foundation postdoctoral research fellowship. Frank Furnari receives salary and additional support from the Ludwig Institute for Cancer Research.
DMEM/High Glucose | HyClone/GE | SH30022.1 | |
DMEM/F12 1:1 | Gibco | 11320-082 | |
FBS | HyClone/GE | SH30071.03 | |
Accutase | Innovative cell technologies | AT-104 | |
Trypsin | HyClone/GE | SH30236.01 | |
B27 supplement | Gibco | 17504044 | |
human recombinant EGF | Stemcell Technologies | 2633 | |
human recombinant FGF | Stemcell Technologies | 2634 | |
DPBS | Corning | 21-031-00 | |
FACS tubes | Falcon | 352235 | |
DAPI | ThermoFisher Scientific | 62248 | |
Blasticidin | ThermoFisher Scientific | A1113903 | |
p24 ELISA | Clontech | 632200 | |
Xylazine | Akorn | NDC 59399-110-20 | |
Ketamine | Zoetis | NADA 043-403 | Controlled substance |
Ointment | Dechron | NDC 17033-211-38 | |
Absorbable suture | CpMedical | VQ392 | |
5 ul syringe | Hamilton | 26200-U | Catalog number as sold by Sigma-Aldrich |
Cell Sorter | Sony | SH8007 | |
Mouse stereotaxic frame | Stoelting | 51730 | |
Motorized stereotaxic injector | Stoelting | 53311 | |
Micromotor hand-held drill | Foredom | K1070 | |
Mouse warming pad | Ken Scientific Corporation | TP-22G | |
Fluorescence Tomography System | PerkinElmer | FMT 2500 XL | |
TrueQuant Imaging Software | Perkin Elmer | 7005319 | |
Ultra-centrifuge Optima L-80 XP | Beckman Coulter | 392049 | |
Tissue Culture 100mm Dishes | Olympus Plastics | 25-202 | |
Tissue Culture 150mm Dishes | Olympus Plastics | 25-203 | |
Tissue Culture Flasks T75 | Corning | 430720U | |
50 mL conical tubes | Corning | 430290 | |
15 mL conical tubes | Olympus Plastics | 28-101 | |
Centrifuge Avanti J-20 | Beckman Coulter | J320XP-IM-5 | |
Tube, Polypropylene, Thinwall, 5.0 mL | Beckman Coulter | 326819 | |
Tube, Thinwall, Polypropylene, 38.5 mL, 25 x 89 mm | Beckman Coulter | 326823 | |
Athymic nude mice | Charles River Laboratories | Strain Code 490 (Homozygous) | Prior approval by the Institutional Animal Care Program and by the Institutional Biosafety Committee required. |