The constituents of the oncogene signal transduction pathway are promising targets for anticancer drugs. Despite the wealth of available knowledge regarding their molecular properties, the spatiotemporal regulation of the signaling molecules remains elusive. Biosensors based on the principle of FRET have been developed to visualize the activities of the signaling molecules in living cells. However, difficulties in the development of sensitive FRET biosensors have prevented their widespread use in cancer research. The lack of cell lines constitutively expressing a FRET biosensor has also limited their use. In this review, we will introduce the principle of FRET-based biosensors, describe an optimized backbone of the FRET biosensors, techniques to express FRET biosensors stably in the cells, and discuss the future perspectives of FRET biosensors in cancer research. (Cancer Sci 2012; 103: 614–619)
Cancer is a genetic disease caused by alterations in genes that regulate various aspects of cellular functions. For the last quarter of a century, molecular biology and biochemistry have been the two major fields contributing to an understanding of the genes and gene products involved in the development and progression of cancer cells. Consequently, a substantial portion of the existing knowledge was obtained from homogenized cells. For a more comprehensive understanding of the cancer cells, we will need to obtain spatiotemporal information of genes and gene products in living cells and tissues, thus, live cell imaging is the next promising technique in cancer research.
The application of GFP was an important breakthrough for the understanding of the dynamics of gene products in living cells. Green fluorescent protein and other fluorescent proteins have been used to investigate the dynamic localizations of molecules of interest in living cells and animals. Furthermore, based on the principle of FRET, GFP-based genetically encoded biosensors have been developed for the monitoring of activities of signaling molecules. For example, with FRET biosensors it has been shown that Ras is preferentially activated at the free edge of the cells and that Ras-binding to Raf induces dimerization of Raf.
Previously, we summarized FRET biosensors and the findings obtained using these biosensors in relation to cancer research. Here, we report an optimized backbone of FRET biosensors that facilitates the development of highly sensitive FRET biosensors, the techniques used to establish cell lines stably expressing FRET biosensors, and discuss how these advances contribute to cancer research. The detailed experimental procedures can be found in our previous review article[6, 7] and are also available on our website (http://www.lif.kyoto-u.ac.jp/labs/fret/e-phogemon/index.htm), so we will limit ourselves to describing the recent progress in FRET biosensor technologies and their application to cancer research.
Monitoring the Activities of Signaling Molecules Using FRET-Based Biosensors
Förster (or fluorescence) resonance energy transfer is a process by which a donor fluorophore in an excited state non-radiatively transfers its energy to a neighboring acceptor fluorophore, thereby causing the acceptor to emit fluorescence at its characteristic wavelength. The FRET efficiency depends on several factors: (i) a proper spectral overlap between the donor and the acceptor; (ii) the distance between the two fluorophores; and (iii) the relative orientation of donor and acceptor (Fig. 1A–C). The last two factors are particularly important for the development of FRET biosensors with high sensitivity.[8, 9]
The FRET biosensors generally comprise an acceptor fluorophore, a donor fluorophore, a ligand domain, a sensor domain, and linkers that connect each domain (Fig. 1D). The sensor domain is designated as such because it senses the signal and changes its conformation. Phosphorylation, GTP loading, and phospholipid binding are representative signals that induce the conformational change of the sensor domain. The ligand domain binds to the conformation-changed sensor domain, thereby bringing the acceptor and donor in close proximity to evoke FRET. The fluorophores could be either organic chemicals or fluorescent proteins; however, here we restrict the description only to the fluorescent proteins for the sake of brevity.
The FRET biosensors are classified into intermolecular (or bimolecular) and intramolecular (or unimolecular) types (Fig. 1D). Intermolecular FRET biosensors consists of two molecules, one comprised of an acceptor fluorophore and the ligand (or sensor) domain and the other comprised of a donor fluorophore and the sensor (or ligand) domain. The intermolecular biosensors are particularly useful for detecting protein–protein interaction within the cells. However, for the quantification of FRET, careful correction of bleed-through of donor fluorescence into the FRET channel and the cross-excitation of acceptor fluorophores is essential, making it difficult to use intermolecular FRET biosensors for routine applications.
Intramolecular FRET biosensors combine all components into a single molecule. In contrast to the intermolecular type, the use of intramolecular FRET biosensors is straightforward. The expression of the FRET biosensor and the ratio-imaging of donor and acceptor fluorescence are sufficient to obtain the FRET image. However, the development of intramolecular FRET biosensors is laborious work, mostly because the pair of sensor and ligand domains for the monitoring of protein activities and the order of the four components, that is, the acceptor, donor, sensor, and ligand domains, can be optimized only by trial and error. Nevertheless, 15 years after the first report of an intramolecular FRET biosensor, many research groups have developed a number of intramolecular FRET biosensors that monitor ion concentrations, sugars, phospholipids, protein kinase activities, small GTPase activities, and so on (http://www.lif.kyoto-u.ac.jp/labs/fret/e-phogemon/unifret.htm). As examples, we describe some representative FRET biosensors that monitor the activity of oncogene products (Fig. 2). Raichu-Ras, the FRET biosensor for Ras, shows high Ras activity at lamellipodia induced by epidermal growth factor (EGF) stimulation (Fig. 2A). Prin-c-Raf, the biosensor for c-Raf, shows the recruitment of c-Raf to the plasma membrane and concomitant activation of c-Raf on EGF stimulation. Note that FRET efficiency was inversely correlated with Raf activity in this intramolecular FRET biosensor (Fig. 2B). Picchu-CrkII shows that phosphorylation of CrkII diffusely increased in the cytoplasm on EGF stimulation (Fig. 2C). Recently, a modified version of Picchu-CrkII was used to detect Abl tyrosine kinase activity in CML patients, paving the way to the first clinical use of the genetically encoded FRET biosensors.
Development of an Optimized Backbone for FRET Biosensors
As already described, the developers of FRET biosensors had to expend much effort to design and optimize various parts of the biosensors. To accelerate the development of FRET biosensors, we recently set an optimized backbone for the genetically encoded intramolecular FRET biosensor. First, we determined the optimal donor and acceptor fluorescent proteins. Several research groups have optimized fluorescent proteins for FRET biosensors.[14-16] Because most recent biosensors use a cyan-emitting mutant (CFP) as the donor and a yellow-emitting mutant of GFP (YFP) as the acceptor, we concentrated on the comparison among CFP-like and YFP-like fluorescent proteins such as CyPet, Ypet, Venus, Turquoise, and teal fluorescent protein (TFP). Among the tested fluorescence proteins, enhanced CFP and Turquoise served as the best donor fluorophores with the use of YPet as the acceptor fluorophore.
Other than the FRET pair, the FRET efficiency of FRET biosensors is determined by the distance and the relative orientation of the donor and acceptor fluorophores (Fig. 3). In the distance-dependent mode (Fig. 3A), FRET will increase when the sensor domain binds to the ligand domain. In the orientation-dependent mode, FRET is high before the sensor domain binding to the ligand domain (Fig. 3B). Fluorescent proteins derived from GFP tend to dimerize each other and such dimerization is known to enhance FRET efficiency. Thus, in contrast to the distance-dependent mode, on the binding of the sensor domain to the ligand domain, the FRET efficiency will decrease due to the rotation of the FRET pair (Fig. 3B). Many FRET biosensors show the distance-dependent mode and a smaller number of the FRET biosensors show the orientation-dependent mode.[20, 21] Unfortunately, in the absence of the 3-D structures of the FRET biosensors, it is difficult to predict which of these two modes will be dominant in each biosensor design. We overcame this problem by the use of a very long linker, named the EV linker, that connects the ligand domain to the sensor domain. The EV linker consists of 116 amino acids and thus abolishes the dimerization of the FRET pair, rendering the FRET efficiency mostly dependent on distance. With this system, the gains of FRET biosensors of ERK, PKA, Ras, and Rac1 were markedly increased and new biosensors for S6K and RSK were made without many optimization steps. This EV linker system will accelerate the development of FRET biosensors for kinases and small GTPases, which will help to decipher the role of signaling molecules in living cells.
Techniques to Establish Cell Lines Stably Expressing FRET Biosensors
An unspoken flaw of FRET biosensors is the difficulty of establishing cell lines that stably express FRET biosensors comprised of CFP and YFP. There has been a study using a cell line stably expressing such a biosensor, but most studies with FRET biosensors have used transient transfection for the expression. Transfection/electroporation of linearized plasmid DNAs and retroviral induction are the two major methods for the establishment of cell lines stably expressing exogenous proteins. In our experiments, transfection of a linearized expression plasmid of a FRET biosensor produced cells expressing a functional FRET biosensor for a short period; however, during repeated cloning steps, cells almost always lost the expression of CFP, YFP, or both (Fig. 4A). With retrovirus-mediated gene transfer, most infected cells expressed either CFP or YFP (Fig. 4B). Retroviruses carry two copies of the RNA genome. The RNA genome is often fragmented in the virus particle; therefore, the reverse transcriptase uses the two copies of RNA genome to replicate the full genome without mistake. This means that the reverse transcriptase on one copy of the genome RNA can jump on to the other copy of the genome RNA easily. Thus, we speculated that, due to the high sequence homology between CFP and YFP, the reverse transcriptase may jump from the CFP-coding region to the YFP-coding region or vice versa. Under this assumption, we changed the donor fluorophore from CFP to TFP, a fluorescent protein derived from coral. As we anticipated, FRET biosensors carrying TFP and YFP as a FRET pair were readily expressed by retrovirus-mediated gene transfer, without any recombination (Fig. 4C).
When the donor fluorophore was changed from CFP to TFP, however, the gain of FRET biosensors was almost always decreased. We recently found that piggyBac transposon-mediated gene transfer can be used to efficiently establish cell lines stably expressing FRET biosensors comprised of both CFP and YFP. These cell lines are very useful for the assessment of the effect of anticancer drugs on the signaling molecules (Fig. 4D). We adduce a HeLa cell line stably expressing a FRET biosensor for ERK as an example (Fig. 5). ERK activity was chosen as the readout of the canonical oncogene signal transduction cascade comprised of EGFR, Ras, and Raf. There are several merits for using this cell line for the screening of anticancer drugs. First, the cell-based assay guarantees fairly good drug-delivery into the cells for the hit-compounds. Second, multiple potential drug targets in the oncogene signal transduction pathway can be screened simultaneously. Third, the time-course of the effect of drugs can also be acquired in a single experiment. Fourth, the number of cells may be scaled down to as few as 100 cells. Finally, up to three different FRET biosensors can be used in a single experiment. This is because FRET biosensors localized to the cytoplasm, nucleus, and plasma membrane can be distinguished by an image-processing program.
Monitoring Signaling Activities in 3-D Culture and Living Tissues
Next we adduce Rho-family GTPases as examples to indicate the merit of FRET biosensors. Rho-family GTPases play a central role in the regulation of invasion by cytoskeletal re-organization. The coordinated activation and/or antagonistic action of Rho-family GTPases determine the invasion morphologies of cancer cells. Three major members of Rho-family GTPases, RhoA, Rac1, and Cdc42, are associated with the three characteristic subcellular architectures, stress fiber, lamellipodia, and filopodia. In agreement with these proposed roles, FRET biosensors have visualized high RhoA activity at the tail and front, and high Rac1 and Cdc42 activities at the front of migrating cells (Fig. 6A). Thus, FRET biosensors of Rho-family GTPases have been proven to be powerful tools for revealing subcellular activity maps.
How, then, are Rho-family GTPases regulated at the level of tissues? Two-photon intravital microscopy has revealed that cancer cells show diverse invasion morphologies in tissues.[28, 29] It has been proposed that the balance between RhoA and Rac1 may determine the mode of cancer cell invasion. Using the newly developed cell lines stably expressing FRET biosensors for Rho-family GTPases, we have shown the spatial activity maps of Rho-family GTPases with glioma cells invading into brain parenchyma. We found that Rac1 activity was high in the cells invading into the brain parenchyma at the front of glioma cells (Fig. 6B). Two scenarios could be sketched to explain this observation. Glioma cells may be activated at the front of the invasion, probably by certain growth factors. Alternatively, glioma cells showing higher Rac1 activity may guide the other cells showing lower Rac1 activity. To determine which scenario applies, glioma cells stably expressing FRET biosensors for Rac1, Cdc42, or RhoA were grown in 3-D matrigel, wherein the concentration of growth factors should be homogenous (Fig. 6C). We found that cells with high Rac1 activity guided the other glioma cells in 3-D Matrigel. Similar data were obtained for Cdc42. RhoA did not show such a difference and served as a good control.
Future Perspectives of FRET Biosensors in Cancer Research
Now that an optimized backbone for FRET biosensors has been established, a number of FRET biosensors for kinases and small GTPases should be available in the near future. It would be very exciting to see how signaling pathways other than the EGFR-Ras-ERK pathway are spatiotemporally controlled in cancer cells and tissues. We have already reported on the S6K biosensor, which visualizes the activity of the mTORC1 pathway. Biosensors for the Cdk family or others will also shed new light on the spatial control of signaling cascades related to oncogenesis. Use of cell lines stably expressing FRET biosensors could potentially replace SDS-PAGE and immunoblotting analysis. The activity changes of kinases, small GTPases, and/or phosphoinositides in cancer cells will be visualized during the initial transformation, invasion, and metastasis of cancer cells. The screening and validation of anticancer drugs can be accelerated with the cell lines described here and being developed. Moreover, the transposon-mediated gene transfer technique should also be applicable for the generation of transgenic mice. Such mice will provide ideal animals to examine the pharmacodynamics of anticancer drugs in living animals. Thus, stable expression of FRET biosensors will accelerate current trends in cancer research, that is, from cells on a plastic dish to 3-D and/or live tissues, and from biochemistry to live imaging.
This work was supported by a Grant-in-Aid for Scientific Research on Innovative Areas “Fluorescence Live Imaging” (No. 22113002), by a Research Program of Innovative Cell Biology “Cell Innovation” and by an Innovative Techno-Hub for Integrated Medical Bio-imaging Project of the Special Coordination Funds for Promoting Science and Technology. N.K. was supported by research fellowships from the Japan Society for the Promotion of Science for Young Scientists.