Abstract
Numerous animal models have been developed to provide a deeper insight to tumor progression in the search for new therapeutic leverage. The closer the tumor model represents the real tumor disease, the better. The ideal model provides monitoring, tumor cell detection and quantification, and the physiological events involved in tumor progression and tumor dissemination, simultaneously. Sensitive techniques have been developed which involve fluorescent protein-based methods, developed in order to quantify the tumor cells in a whole organ, and in parallel, to visualize the cells. These genetically encoded fluorescent proteins may also be used to develop biological sensors to monitor the physiological reaction of tumor cells within whole organs in living animals. Here, we aim to review past and present work and to show the perspectives of animal models involving fluorescent protein-transfected tumor cells.
Abbreviations: FP: Fluorescent proteins, GFP: Green fluorescent protein.
In general, a model for a human disease remains by definition only an imperfect representation of the real disease. A multitude of in vitro and in vivo models have been developed. These models have made a tremendous contribution to the understanding of tumor biology and have lead to new and novel approaches in both cancer treatment and prevention. Effective modeling of tumor started with the invention of cell culture systems. Murine sarcoma-180 cells were the first tumor cells to be cultured and passaged. This is why cell culture media components and supplements, such as fetal bovine serum, had to be optimized (1).
In early cancer research, cell lines were mostly used for growth inhibition assays, where the cells were cultured in the presence of drugs. Nowadays, cell cultures can be used to study a variety of aspects of tumor growth, tumor angiogenesis, tumor invasion and immune cell interactions. Despite their convenience, conventional cell culture assays which use monolayers of cells fail to model important features of solid tumors by forcing the cells to adopt to an unnatural environment (2). Since most tumors are diffusion-limited with slow and variable blood flow, monolayer cultures cannot mimic this important feature of the tumor tissues (3). To overcome this inconvenience, several three-dimensional tissue culture systems have been developed that allow cells to grow in a more realistic manner (2, 4, 5). Tumor vasculature as one component of tumor angiogenesis is one of the stromal components which can be exploited as a drug target. On these grounds, assays for endothelial cell migration, differentiation assays and organ culture assays have been developed (6, 7). Comparison of findings from different experiments remains a difficult topic due to differences in the source, viability and passage number of the cells used in these assays. Another elegant model to study tumor cell angiogenesis is the chick chorioallantoic membrane assay. In this assay, a window is carefully cut in an egg shell and tumor cells can then be implanted (8). Thus the development of tumor vasculature may be studied in the context of the immature immune system of a 7 to 8-days-old chicken embryo (9).
Currently, no available tumor cell culture system is able to closely model human disease. In vivo models represent an alternative. Findings which are derived from in vivo models are mostly more relevant for the treatment of human tumors.
The first in vivo models were developed in the mid-1960s: the murine leukemia models P388 and L1210 were selected as the initial systems in which potential agents could be tested for antitumoral activity before further development (10). Syngenic tumor models, which use implanted tumor cells that are derived from the strain in which the tumor originated (11, 12), offer certain scientific and practical valuable advantages: they are reproducible, grow in immunocompetent hosts, cover a wide range of tumor types and are cost-effective (13). The main disadvantages of these models are that the tumour cells have to be implanted. Thus tumor growth may progress differently. Here, certain genetically engineered models have their advantages. In these models, the murine genome has been modified to change the function of genes which are involved in tumor development. Therefore the animals develop tumors spontaneously. The models are convenient if tumor development is to be followed from early time points on (14). Breeding and maintaining a colony large enough to generate sufficient numbers of mice of the same age and gender can be very difficult and costly (13).
The development of human tumor xenograft models also generated powerful tools for oncological research (15). In these models, human tumor cells are transplanted into immunodeficient animals. Historically, Rygaard and Povlsen realized the first successful transplantation of a human tumor in mice in 1969 (16). Since then, the model has been improved significantly. One main concern regarding both syngenic and xenograft models, is the difficulty of visualizing and quantifying the molecular events involved in tumor progression and dissemination. Labelling of tumor cells with fluorescent proteins has made a great contribution to overcome this problem.
Conventional Tumor Cell Quantification
Sensitive quantification of the tumor load in different animal organs is essential in order to monitor the development of tumor metastasis and to identify the treatment effect of different antitumoral drugs (17). However there are currently only few animal models that allow quantification of tumor load in a whole organ. These methods are based mostly on the staining of tumor in histological sections or fluorescence activated cell sorting (FACS)-analysis of live cell suspensions after tissue dissociation. These are often performed in only few sections of the relevant organ, which may be misleading, since an equal distribution of tumor cells within an organ is unlikely (18). Alternatively, preparation of cell suspensions for FACS-analysis may be performed, which involves many destructive steps, such as enzymatic tissue digestion by perfusion, organ mincing and filtering of the suspension through gauzes, which may lead both to the loss of tumor and healthy cells.
Accurate tumor cell quantification is particularly important in early tumor development since the size of micrometastases and the number of tumor cells arrested in the target organs can be very small. They can therefore evade conventional staining methods. Quantitative estimation of metastasis was typically based on the formation of macroscopic metastatic nodules. However, there are also tumor models in which organs are also diffusely infiltrated by scattered metastatic cells (18). To estimate the tumor load at early tumor stages in the case of diffuse metastasis, molecular techniques have been widely used for the detection of tumor cells. Polymerase chain reaction (PCR) was shown to be superior to conventional techniques in detecting occult tumor cells, allowing the identification of one malignant cell mixed with up to 10 million normal cells (19). A major strategy for the detection of occult tumor cells in the clinical context is PCR amplification of tumor-specific abnormalities present in the DNA or mRNA of these cells, such as tumor antigens. Recently, quantification of colorectal cancer metastasis in lymph nodes has been reported based on detection of tumor markers such as the carcinoembryonic antigen (CEA) by RT-PCR (20). The transfection of tumor cells with fluorescent proteins such as green fluorescent protein (GFP) allows for easy tumor quantification since the amount of GFP-marker protein mRNA will always correlate with the tumor load within the studied organ. Furthermore, it allows for easy visualization of the tumor.
GFP and its Variants
The cloning and heterologous gene transfer of wild-type GFP from the jellyfish Aequorea victoria brought about a revolution in cell biology. GFP cDNA encodes a 238 amino acid polypeptide that requires no other co-factors or substrates to fluoresce (21). As a fully genetically encoded label that can be linked with almost any protein, GFP has become a unique tool that enables direct visualization of cells and subcellular structures. Nowadays, advances in multiphoton fluorescent microscopy and the engineering of GFP into mutants with improved properties and altered colors have provided the basic tools that allow not only the direct visualization of cellular components, but also the quantification of signaling molecules and events with high spatial and temporal resolution in individual cells, both in vitro and in vivo (22). Especially in the Anthozoans (e.g. sea anemones), new fluorescent proteins have been exploited and modified through random site-directed mutagenesis. Recently the two-photon absorption spectra of over 48 fluorescent proteins, from enhanced BFP- (blue) and CFP- (cyan) series, the ‘fruit’ fluorescent proteins (e.g. mBanana, mOrange, tdTomato) to far-red variants, such as mRaspberry, mKate and Katushka2, have been characterized (23). Further modifications of these will lead to brighter mutants with improved properties for multiphoton applications. However, besides overall brightness, there are other factors of concern such as folding dynamics, monomerization, and greater photostability. Due to their long-wavelength emission, the next-generation sensors with emission peaks in the far red spectrum are of special interest for in vivo imaging.
The Use of GFP in Tumor Models
GFP was the first fluorescent protein to be used for visualization of Chinese hamster ovary cells that were implanted orthotopically in nude mice (24). Chishima et al. stably transfected cancer cells with GFP and transplanted them into several mouse models. As a result, visualization of single metastatic cells in the tissues was possible for the first time without involvement of any post mortem histological technique (Figure 1). The recent explosion in the diversity of fluorescent proteins and a better understanding of their functionality under physiological conditions, allowed development of a variety of in vivo application modalities. Nowadays, fluorescent proteins can be used to visualize any type of cancer process, including primary tumor growth, tumor cell motility and invasion, metastatic seeding and colonization, angiogenesis, and the interaction between the tumor and its microenvironment (25). Transfected cell lines have been used to study tumor‘host interaction, tumor immunology and tumor angiogenesis. Transfection of tumor cells with GFP allows in vivo observation of their fate in living organisms by video microscopy (17, 26). Here, mouse models with syngenic EGFP-transfected colon cancer cells, imitated the portal metastatic route and allowed intravital observation by intravital microscopy. Circulating tumor cells were easily monitored. Thus it was possible to monitor the tumor cell arrest within the liver vasculature which preceeded the tumor cell invasion (17, 27). These models permit real time, as well as post mortem observation of the pathophysiological processes involved in metastatic cancer growth and serve as a tool for monitoring the specific key steps of the metastatic cascade at the single-cell level. Therefore, these models can be used to clarify cell biological interactions and signalling cascades which are involved in the multistep process of tumor dissemination (17, 28). Importantly, not only the fluorescent attributes but also the number of DNA copies can be used as a marker to determine the effectiveness of novel therapeutic approaches (29, 30).
GFP-based Biological Sensors
Beside ‘passive’ fluorescent reporters for the measurement of protein expression, localization, and tracking within live cells, recent advances in the development of ‘active’ markers allow the monitoring of more complex cellular processes such as second messenger dynamics (e.g. Ca2+, cAMP), receptor or enzyme activation and protein-protein interaction. For example all known second messengers are small molecules without complex structural features. They mediate a wide spectrum of cellular responses such as the regulation of proliferation, metabolism, cell migration, and/or cell death. Notably, impaired calcium signaling has recently been implicated in tumor cell migration and development of a metastatic cell phenotype (31). However, different signals might arise within one cell at the same time, rendering it necessary for the cell to achieve signal specificity. Clearly, as in the case of intracellular calcium, this specificity can be attributed to amplitude, pattern (e.g. oscillations), and subcellular compartmentalization of signals. Due to the de-differentiation of cells under two-dimensional culture conditions, the latter can only be imperfectly reconstituted in vitro in a reliable manner, making it necessary to study such processes in the context of an intact organism. The use of fluorescent protein-based sensors can now deliver such insights into many second messenger-dependent processes in vitro and in vivo (21, 32, 33). Most probes for the measurement of such dynamics show an environment-dependent change in their spectral characteristics. The most commonly used approach to monitor e.g. calcium, cAMP signaling and protein-protein interactions exploit Förster resonance energy transfer (FRET), usually between cyan and yellow fluorescent proteins, to transform conformational changes in the sensor molecule into measurable changes in their fluorescence spectra (e.g. Cameleon-based calcium sensors, RIα-EPAC-based sensors for cAMP). These sensors can be targeted to subcellular compartments and their ratiometric nature ensures that changes in probe quantity and movement artifacts are inherently corrected. Different types of FRET imaging can be applied, but not all of them can be easily used for quantitative measurements. As such, FRET-fluorescent lifetime imaging microscopy (FLIM) might be the most quantitative, but it is still fairly slow and, at least to our knowledge, has not yet been described in living organisms. Although it is not as quantitative as FLIM, we used dynamic ratiometric FRET-FLIM imaging, which relies on increased FRET acceptor (YFP) emission at the expense of donor (CFP) emission upon donor excitation, when FRET efficiency is rising. A good example are sensors based on EGFP and tHcred1 which we have recently developed for a functional imaging of caspase-3 activity. These sensors allowed monitoring of caspase-3 activation in cells by FRET-FLIM imaging microscopy or fluorescence correlation spectroscopy (34). The construct is based on EGFP and tHcred1 and a 14-amino acid linker containing the caspase-3 cleavage site for readout of apoptotic activity in cancer cells by FRET-FLIM (Figure 2). FRET between EGFP and tHcred1 in intact sensor, results in reduction of fluorescence lifetime of the donor‘EGFP. This reduction can be correctly and reproducibly detected by FLIM microscopy. Activation of caspase-3 in the process of apoptosis disrupts the covalent linkage between EGFP and tHcred1 and effectively eliminates FRET, thus conveying that a cell is apoptotic. EGFP phase lifetimes of 1,6-2.0 ns means cells are considered as being healthy, and longer than 2.0 ns as being apoptotic (Figure 3). Sensor (tHcred1-DEVD-EGFP)-transfected colorectal tumor cells such as SW-480 cells, can be used for reproducible single-cell-based apoptosis detection and quantification (34).
Using this biosensor for caspase-3 activity and apoptosis, we transferred this approach to a mouse model of peritoneal and liver metastasis (35). For this, the tHcred1-DEVD-EGFP construct was stably transfected into C26 murine colon-carcinoma cells. These cells were then implanted by intra-portal and intraperitoneal injection into syngenic Balb/c mice to generate models for liver and peritoneal metastasis.
Capase-3 sensor also allowed quantification of the apoptosis rate in fresh tumor tissue samples derived from peritoneal and liver metastases. Besides these in vivo measurements, apoptosis can also be measured in an ex vivo setting (Figure 4). For this purpose, sensor-transfected C26 cells were isolated from omental tissue of the tumor-bearing mice and were studied further under cell culture conditions. The cells were isolated both from 5FU-treated and non-treated mice. Further 5FU treatment of the cells derived from 5FU-treated mice have caused significantly less apoptosis, than treating intact sensor containing cells and cells isolated from non-treated mice. We have explained this phenomenon with increasing resistance of the cells towards longer treatment with 5FU in mice. Since tHcred1-DEVD-EGFP transfection of C26 cells allowed an accurate quantification of the apoptosis rate in response to chemotherapy, the model provided a platform for chemoresistance studies. Furthermore, it was possible to measure the tumor load in the whole organ through real-time PCR quantification of EGFP DNA copies. Directly after removing organs, DNA was extracted from the omentum and the liver using a modified DNA purification protocol. Tumor load was determined by real-time PCR for EGFP with the genomic DNA (Figure 5).
Tumor load determination in the liver and omental tissues demonstrated the effect of chemotherapy. Under ongoing 5FU treatment, the tumor load in omentum decreased at 72 h compared to the control and the previous treatment time point. Interestingly, after 240 h of treatment, the tumor load again significantly increased. This is likely due to increasing chemoresistance to 5FU, which correlated with the FLIM data.
In summary, the novel fluorescent protein sensor allowed quantification of apoptosis and of the tumor load in omental and liver metastases of colorectal cancer in the mouse. Using this model, we are currently investigating different therapeutic approaches to overcome resistance development and to study the ways of overcoming the resistance.
Future Perspectives
Advances in microscopy and optical imaging and the generation of new fluorescent biosensors have revolutionized optical imaging. Live cell imaging has been used to study a variety of cells in different organisms. Optical imaging of specific molecular targets and pathways in living cells has recently become possible through continued developments in microscopic imaging technology, and more importantly, due to the availability of genetically encoded fluorescent biosensors (36-38). Rational genetic engineering of GFP molecules will in future lead to further evolution in terms of brightness and biocompatibility (39). These fluorescent reporters have already further promoted live-cell imaging of biochemical processes in a variety of cells from different organisms (37, 40). Interactions between two or more proteins are presently used to study protein reactions in living cells by FRET (41, 42) within living cells. The development of new and brighter mutants in the blue (43) and in the red region of the light spectrum (44) will enable the development of new biosensors with a significantly enhanced FRET efficiency. Quantitative co-expression of two or more proteins has recently been achieved with little cell-to-cell variability. This finding enables reliable co-expression of donor and acceptor-tagged proteins for FRET studies, which is of particular importance for the development of novel biomolecular sensors that can be expressed from a single plasmid (43). Advances in microscopy such as in vivo two-photon FLIM microscopy will permit monitoring of the parallel function of two or more biosensors within a living tumor animal model.
- Received May 16, 2012.
- Revision received June 3, 2012.
- Accepted June 5, 2012.
- Copyright © 2012 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved