The ability of metastatic clones to colonize distant sites depends on their proliferation capacity and/or their ability to survive in the host microenvironment without significant proliferation. Here, we present an animal model that allows quantitative visualization of both types of liver colonization by metastatic clones.
Patients with a limited number of hepatic metastases and slow rates of progression can be successfully treated with local treatment approaches1,2. However, little is known about the heterogeneity of liver metastases, and animal models capable of evaluating the development of individual metastatic colonies are needed. Here, we present an advanced model of hepatic metastases that provides the ability to quantitatively visualize the development of individual tumor clones in the liver and estimate their growth kinetics and colonization efficiency. We generated a panel of monoclonal derivatives of HCT116 human colorectal cancer cells stably labeled with luciferase and tdTomato and possessing different growth properties. With a splenic injection followed by a splenectomy, the majority of these clones are able to generate hepatic metastases, but with different frequencies of colonization and varying growth rates. Using the In Vivo Imaging System (IVIS), it is possible to visualize and quantify metastasis development with in vivo luminescent and ex vivo fluorescent imaging. In addition, Diffuse Luminescent Imaging Tomography (DLIT) provides a 3D distribution of liver metastases in vivo. Ex vivo fluorescent imaging of harvested livers provides quantitative measurements of individual hepatic metastatic colonies, allowing for the evaluation of the frequency of liver colonization and the growth kinetics of metastases. Since the model is similar to clinically observed liver metastases, it can serve as a modality for detecting genes associated with liver metastasis and for testing potential ablative or adjuvant treatments for liver metastatic disease.
Patients with liver metastases from primary colorectal cancers (CRC) are characterized by a poor prognosis. The 5-year survival rate for primary nonmetastatic CRC (stages I – III) is estimated as 75 – 88%3,4, while patients with liver metastases (stage IV) have a 5-year survival rate of only 8 – 12%5,6. However, metastatic patients represent a heterogeneous group, presenting with different numbers of metastases and different recurrence times. Clinical observations indicate that the number of metastases (which may be proportional to colonizing ability or frequency of colonization) and the size of any single metastasis (proportional to the local growth rate) are independent prognostic factors1,7. In other words, the success of metastatic clones colonizing the liver depends on two major properties: their ability to grow and their ability to disseminate and survive in the liver microenvironment.
The design of successful clinical models with the capability of capturing and quantifying the properties of metastatic clones can drastically improve our understanding of liver metastasis biology and provide an effective tool for the design of potential therapeutic approaches. Models of experimental liver metastasis have been previously reported8,9, but neither of them provided the ability to quantitatively capture and describe properties of individual metastatic clones both in vivo and ex vivo.
Here, we present a new, advanced model of liver metastasis that includes the generation of tumor clones with different liver colonization efficiencies and growth properties. We employed a combination of dual-labeling of cancer cells with luciferase and tdTomato fluorescent protein with the generation of monoclonal cell lines that have intrinsic differences in metastatic capacity. In this experimental model, the data indicate that the development of liver metastases can be described in terms of colonization frequency and doubling time (Td), which is consistent with clinical observations. The quantitative nature of this model makes it easily adoptable for drug discovery and diagnostic purposes.
All animal procedures were approved by the Institutional Animal Care and Use Committee at the University of Chicago (Protocol # 72213-09) and performed under sterile conditions.
1. Preparations
2. Generation of Luciferase/tdTomato-labeled Monoclonal Cell Lines 10,11
3. Calibration of Fluorescent Signal Intensity as a Function of Cell Number
4. Animal Model of Liver Metastases
5. In Vivo Bioluminescent Imaging
6. Ex Vivo Fluorescent Imaging
The goal of this experiment was to establish a consistent and easily reproducible animal model with the potential for the serial quantification of the in vivo metastatic tumor burden and for the estimation of the colonizing frequency and growth kinetics of developing liver metastases. Figures 2-6, with legends, are provided from our previous publication under a Creative Commons CC-BY license10.
Generation of Double-labeled Tumor Cell Monoclones
First, a panel of luminescent and fluorescently labeled monoclonal cell lines was generated with different metastatic abilities. Following successful transfection of luminescent (Luc2) and fluorescent (tdTomato) proteins in parental HCT116 colorectal cancer cells, 16 double-labeled monoclones were generated by diluting the parental HCT116-L2T cells to 1 cell/200 µL in 96 well plates (Figure 1)10. In vitro fluorescence and luminescence of HCT116-L2T was then quantified using an increasing amount of cells to estimate the number of tumor cells from in vivo and ex vivo liver tumor imaging (Figure 2)10.
Metastases Development by Clones with Differential Liver Colonization
Next, to generate liver metastases, the 16 previously generated individual HCT116-L2T clones were intrasplenically injected into mice; a splenectomy was then performed. The development of liver metastases was monitored by weekly in vivo bioluminescent imaging, followed by ex vivo fluorescent imaging after sacrificing the mice. Out of the 16 monoclones generated, we selected clones designated as P1, P2, O1, and O2, due to their different observed colonization and growth capabilities in the liver. Clones O1 and O2 had a smaller tumor burden, representing "oligometastatic" clones, and potentially recapitulating a phenotype of limited metastatic disease. Clones P1 and P2 had a larger tumor burden, representing widespread dissemination, which is also referred to as polymetastatic disease. The bioluminescence was higher in the P1 and P2 mice (6.7×106 ± 4.7×106 and 3.6×106 ± 2.5×106 photons/s/cm2/steradian) as compared with the O1 and O2 mice (2.0×105 ± 1.3×105 and 1.7×105 ± 9.1×104 photons/s/cm2/steradian) 3 weeks after the spleen injection (Figure 3A, B)10. The quantification of ex vivo fluorescence of the livers at 4 weeks after the spleen injection demonstrated that P1 and P2 livers had higher fluorescent intensities compared with O1 and O2 livers (Figure 3C). The tumor distributions in the ex vivo fluorescent images were consistent with macroscopic findings (Figure 3C, D)10.
Colonization and Growth Kinetics of Individual Clones
The numbers and sizes of tumors visualized macroscopically were counted and measured in each individual liver. In addition, ex vivo fluorescence imaging was done on each individual tumor colony for each monoclone (Figure 4)10. The total number of tumors in P1 was higher, as compared to P2, O1, and O2, while the size of tumors was larger in P2 than in P1, O1, and O2 (p <0.001; Figure 5A, B)10. However, the total tumor volume in the liver in P1 and P2 was larger as compared to O1 and O2 (Figure 5C)10. The Fc, which indicates the ability of tumor cells to colonize the liver, is calculated as the number of tumors in the liver divided by the number of cells injected. Using this method, there is an assumption that the number of tumors in the liver corresponds to the number of cells that have an ability to colonize the liver. Thus, P1 had an enhanced ability to colonize the liver when compared to the other clones (Fc = 8.2×10−5 ± 1.1×10−5, 6.0×10−6 ± 2.3×10−6 in P1 and O1, respectively, p <0.001; Figure 5D)10. Though P2 had a similar ability to colonize the liver as clones O1 and O2, the size of the tumors generated was larger than the other clones. The total number of cells per tumor colony, the Td, and the number of cell divisions can be calculated from this data by using the calibration previously generated from the correlation between fluorescent intensity and the number of cells (Figures 2 and 5E – G)10. P1, which generated many small tumors in the liver, as seen by macroscopic visual inspection, had a larger Fc, though it had a Td similar to O1 and O2 (p <0.05). P2, which generated a limited number of macroscopic large tumors, had a similar Fc but a shorter Td compared to O1 and O2 (p <0.05). Taken together, these results indicate that the Fc and Td are two key parameters contributing to metastatic potential.
The 3D distribution of metastatic tumors in the liver was demonstrated by DLIT (Figure 6)10. Using this imaging technique, it is possible to monitor spatial-temporal dynamic behavior of the metastases by the detection and measurement of the bioluminescence of individual hepatic metastatic colonies.
Statistical Analysis
The data generated was analyzed using standard data analysis software, and error bars were demonstrated as the mean ± standard deviation. Pearson's correlation coefficients were used to evaluate associations between parameters. P values were assessed using 2-tailed Student's t-tests, with a p < 0.05 considered to be statistically significant.
Figure 1: Generation of a Panel of Monoclones of Labeled HCT116 Cells. (A) Grow parental HCT116 cells labeled with luciferase and tdTomato. (B) Dilute parental HCT116 cells to 1 cell/200 µL. (C) Plate 200 µL of the dilution in a 96 well plate. (D) Grow monoclones from each well in the 96 well plate. (E) Inject monoclones intrasplenically into mice to generate liver metastases. Please click here to view a larger version of this figure.
Figure 2: Estimation of the Number of Cells by Fluorescent Intensities. (A) The luminescent image of different numbers of cells triplicated in a 96 well plate. (B) The correlation between the number of cells and the luminescence (R = 0.99, p < 0.0001). The scale bar represents luminescent intensity (photons/s/cm2/steradian). (C) The fluorescent image of different numbers of cells triplicated in a 96 well plate. Intensities were acquired as radiant efficiencies with a 535 nm excitation and a 580 nm emission. (D) The correlation between the number of cells and the fluorescence (R = 0.99, p < 0.0001). The scale bar represents the fluorescent intensity in an arbitrary unit of radiant efficiency ((photons/s/cm2/steradian)/(µW/cm2)). Reprinted with permission10. Please click here to view a larger version of this figure.
Figure 3: Bioluminescence and Ex Vivo Fluorescence of the Liver in Selected Clones. (A) Representative bioluminescent images. The scale bar represents luminescent intensity (photons/s/cm2/steradian). (B) Bioluminescence measured weekly from 1 - 3 weeks in P1 and P2, and 1 - 4 weeks in O1 and O2. Data were represented as the mean ±standard deviation. (C) Representative ex vivo fluorescent images of livers. Intensities were acquired as radiant efficiencies with a 535 nm excitation and a 580 nm emission. The scale bar represents fluorescent intensity in an arbitrary unit of radiant efficiency ((photons/s/cm2/steradian)/(µW/cm2)). (D) Macroscopic liver images (arrows indicate small white tumors). P1 and P2: polymetastatic clones, O1 and O2: oligometastatic clones. Reprinted with permission10. Please click here to view a larger version of this figure.
Figure 4: Quantification of Fluorescent Intensities of Individual Liver Colonies. Images on the left are representative macroscopic findings and images on the right are fluorescent intensities in each clone. The fluorescent intensities were acquired as radiant efficiencies with a 535 nm excitation and a 580 nm emission. P1 and P2: polymetastatic clones, O1 and O2: oligometastatic clones. Reprinted with permission10. Please click here to view a larger version of this figure.
Figure 5: Quantitative Estimation of Colonizing Ability and Growth Kinetics. (A) The macroscopic total number of tumors, (B) macroscopic size of individual tumors, (C) calculated total tumor volume in the liver, (D) colonizing fraction of tumor colony forming cells (Fc), (E) total cell number per colony, (F) doubling time (Td), and (G) number of cell divisions of each clone. P1 and P2: polymetastatic clones, O1 and O2: oligometastatic clones. Data were represented as the mean ±standard deviation for all figure panels in which error bars were shown. Reprinted with permission10. Please click here to view a larger version of this figure.
Figure 6: Quantification of Bioluminescence of Individual Tumor Colonies using Diffuse Luminescent Imaging Tomography (DLIT). (A) 3D overhead view. The scale bar represents fluorescent intensity in an arbitrary unit of radiant efficiency. (B) Coronal section. (C) Sagittal section. (D) Transaxial section. (E) Macroscopic image of liver. (F) Ex vivo fluorescent image of the liver. Reprinted with permission10. Please click here to view a larger version of this figure.
The animal model presented in the current report is based on two major approaches. First, in order to ensure the ability to observe metastatic clones with different propensities to colonize and proliferate in the liver, a panel of highly heterogeneous monoclonal cell lines was established, rather than an established unfractionated cancer cell line12,13. The monoclonal approach to metastasis development is justified by recent genomic data14 and was successfully used previously to model the metastatic process10,15,16. Second, double-labeling of the parental cell line by luciferase and tdTomato markers was incorporated in order to provide enhanced analysis of metastatic growth both in vivo and ex vivo. The luminescent and fluorescent labeling allows for a detailed anatomical examination and potential quantification of both the amount of metastatic tumors (considered as the frequency or efficacy of liver colonization) and the size of any individual colony. The latter can easily be converted to an actual number of cells using a simple calibration procedure described in step 2 and Figures 210 and 410. Using this data, it is possible to estimate the Td for each metastatic node in the liver and to evaluate the growth kinetics of liver metastases for each initial monoclone injected (Figure 5)10.
Three different colorectal metastatic phenotypes were observed in this experimental system, designated as P1, P2, and O1/O2. These metastatic phenotypes differed when comparing the data of frequency of colonization and local growth rates for each metastatic clone. The first metastatic phenotype, P1, demonstrated an enhanced ability to colonize the liver but a relatively low local growth rate, corresponding to many small metastatic tumors throughout the liver. The second metastatic phenotype, P2, had a lower frequency of colonization but a significantly higher local growth rate, resulting in a relatively small number of large secondary tumors. Lastly, the third metastatic phenotype, O1 and O2, demonstrated both a low frequency of colonization and slow local growth rates. This third metastatic phenotype may be considered as a representation of clinical oligometastatic liver disease (Figures 310 and 510). In addition, gene expression analysis comparing these clones10 identified significant differences in expression patterns. For example, when comparing P1 with O1 and O2 clones, there were 756 differentially expressed genes, including a subset of genes involved in inflammatory responses and interferon signaling. This study, as well as others, have implicated these genes in playing key roles in tumor growth and survival21, 22. In addition, when comparing P2 to O1 and O2, there were 461 differentially expressed genes, with a subset involving genes critical to cellular growth and proliferation mediated through ERK1/2 signaling.
The diversity of metastatic phenotypes produced by this experimental model is consistent with the observed clinical representations of metastatic heterogeneity. For example, it has been shown that the size of liver metastases (a parameter related to growth potential) and their numbers (a parameter related to colonization ability) are independent risk factors for distant metastasis1. With enhanced growth potential or colonization ability, the properties observed in P1 and P2 are consistent with these data. In addition, recent observations of patients treated with Stereotactic Body Radio-therapy (SBRT) or with surgically resected lung metastases point both to the number of tumors and the rate of progression as key factors determining overall and disease-free outcomes17,18. Thus, colonization and growth abilities seem to be among the most critical factors in determining the development of metastatic clones at distant sites19,20. Experimental models, which are based on these properties of metastatic clones, may provide useful tools for understanding the mechanisms of metastasis development, as well as provide a means for testing new treatment modalities for the cure of liver metastatic disease.
While an orthotopic model capable of producing spontaneous metastases is ideal, there remains a paucity of easily reproducible and consistent spontaneous models. While models utilizing intracecal implantation or injections have been described, they demonstrated variable rates of uptake and metastasis formation23-26. Other methods have utilized an ectopic implantation approach by intrasplenic or intraportal injection to generate hepatic metastases. While the overall technique of injection is the same, the model presented here provides a sensitive approach to understanding the mechanisms of metastasis development, starting from the stage of Circulating Tumor Cells (CTC). Sensitivity and reproducibility of our approach is determined by the double-labeling of the cell lines to allow for both bioluminescent and fluorescent imaging to track the dynamics of the metastatic process. Due to the ability to quantify the metastatic burden from the luminescent and fluorescent data, it is possible to systematically measure and track the growth kinetics of metastases over time. Although alternative approaches, such as ultrasound-based imaging, have been used, these modalities are much more user-dependent, with potentially increased inter-user variability27. In addition, by generating a panel of monoclonal cell lines, we are able to provide reproducible and distinct metastatic phenotypes in order to better model the variability of metastatic disease seen in the clinical setting.
While mastering this experimental technique, specifically the intrasplenic injection for the in vivo mouse model (see step 4), there are a few critical steps that may require modification. Of note, performing the splenic injection is the most important portion of the model, as an inadequate injection of the spleen may result in poor results. Leakage of cells during the injection, or leakage caused by the inappropriate placement of the microclip, may result in a lower number of liver metastases and/or a significant amount of extra-liver tumors or peritoneal carcinomatoses. In addition, it is important to inject the tumor cells slowly, for at least as long as described in the protocol, in order to avoid an increase in the intrasplenic pressure, which may lead to extravasation of the cells to adjacent tissues and the growth of extra-liver tumors around the stump of the spleen dissection.
Lastly, in step 4.3, there is a range of cells given for the splenic injection (1.2 – 2 × 106 cells). The purpose of this range is to account for individual operators of this technique. It may take time to calibrate an optimal number of cells to inject. If too few cells are injected, liver metastases may not develop consistently, and if too many cells are injected, portal venous occlusion may occur, leading to the early demise of the animal. It may be prudent to initially try several different concentrations of tumor cells when starting this model in order to find the optimal number.
One pitfall of the presented model is that it is based on a xenograft approach and exploits immunodeficient nude mice. One alternative method proposed is the generation of a similar panel of monoclonal derivatives of the murine colorectal cell line MC38, double-labeled with luciferase and tdTomato or EGFP proteins. These clones can then be used in syngeneic animal models in order to capture the interactions of developing liver metastases with the host immune system.
The authors have nothing to disclose.
We would like to thank Dr. Geoffrey L. Greene (University of Chicago) for the Luc2-tdTomato plasmid and the HCT116 cell line, Mr. Ani Solanki (Animal Resource Center) for the mice management, and Dr. Lara Leoni for the assistance with the DLIT. Quantifications of fluorescent and luminescent intensities were performed in the Integrated Small Animal Imaging Research Resource at the University of Chicago on an IVIS Spectrum (PerkinElmer, Hopkinton, MA). This work was supported by the Virginia and D.K. Ludwig Fund for Cancer Research, the Lung Cancer Research Foundation (LCRF), the Prostate Cancer Foundation (PCF), and the Cancer Center Support Grant (P30CA014599). The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.
IVIS Spectrum In Vivo Imaging System | Caliper Life Sciences | 124262 | In vivo imaging system |
LivingImage 4.0 Software | Caliper Life Sciences | 128165 | Imaging software |
VAD-MGX Research Anesthetic Machine | Vetamac | VAD-MGX | Inhalation anesthesia machine |
DMEM | Gibco | 11965-118 | Cell culture reagents |
DPBS | Gibco | 14190250 | Cell culture reagents |
Penicillin-Streptomycin, liquid (10,000 units penicillin;10,000 μg streptomycin) | Invitrogen | 15140163 | Cell culture reagents |
HBSS | ThermoFisher | 24020117 | Cell culture reagents |
Buprenex Injection (0.3mg/mL) | Reckitt Benckiser Healthcare Ltd. | 12496-0757-5 | Buprenorphine hydrochloride |
Gemini Cautery System | Braintree Scientific | GEM 5917 | Hand-held cautery for splenectomy |
Micro Clip; Straight; 70 Grams Pressure; 1.5mm Clip Width; 10mm Jaw Length | Roboz Surgical Instrument | RS-5426 | Hemoclip: Hemostasis instruments after spleen injection |
D-luciferin, potassium salt | Goldbio Technology | LUCK-1G | Luciferin potassium salt |
Opti-MEM I Reduced Serum Medium | Gibco | 31985062 | Reduced Serum Medium |
TC20 Automated Cell Counter | BIO-RAD | 1450102 | Automatic cell counter |
JMP10 software | SAS Institute | Data analysis software | |
BD FACSAria II cell sorter | BD Biocsiences | Cell sorter |