Protein biosensors based on the principle of fluorescence resonance energy transfer for monitoring cellular dynamics
Genetically-coded, fluorescence resonance energy transfer (FRET) biosensors are widely used to study molecular events from single cells to whole organisms. They are unique among biosensors because of their spontaneous fluorescence and targeting specificity to both organelles and tissues. In this review, we discuss the theoretical basis of FRET with a focus on key parameters responsible for designing FRET biosensors that have the highest sensitivity. Next, we discuss recent applications that are grouped into four common biosensor design patterns—intermolecular FRET, intramolecular FRET, FRET from substrate cleavage and FRET using multiple colour fluorescent proteins. Lastly, we discuss recent progress in creating fluorescent proteins suitable for FRET purposes. Together these advances in the development of FRET biosensors are beginning to unravel the interconnected and intricate signalling processes as they are occurring in living cells and organisms.
KeywordsFluorescence resonance energy transfer (FRET) Genetically coded biosensor Green fluorescent protein (GFP) Intermolecular FRET Intramolecular FRET Protein conformational changes Protein–substrate interaction Substrate cleavage Transgenic organisms
Since the discovery of fluorescent proteins (FPs) that are suitable for fluorescence resonance energy transfer (FRET) (Shaner et al. 2005), protein biosensors have rapidly become important tools for studying live cell molecular events. FRET was first described by Theodor Förster nearly sixty years ago as a non-radiative transfer of energy from a photo-excited donor to an acceptor fluorescent molecule located in close proximity (<100 Å) (Valeur 2002). The distance and orientation between the donor and acceptor governs the efficiency of the energy transfer. This efficiency can be determined by a fluorescence emission spectrum. In FRET biosensors, a biological event induces a conformational change in the biosensor, which in turn causes a detectable change in FRET efficiency as measured by the change of profile in the emission spectrum.
These FRET biosensors have many advantages over methods based on conjugating synthetic dyes: first, their fluorescence is acquired spontaneously; second, they can be constructed by simple genetic manipulations; third, they can be delivered into cells by transient transfection and subsequent expression; lastly, they can be targeted to organelles and tissues allowing imaging from single cells to whole organisms. A potential drawback of FRET biosensors is that FPs are relatively bulky (∼50 Å) compared to synthetic dyes (∼5 Å) and therefore may hinder protein activity. However, the activities of many proteins are not affected by fusion to FPs. A second potential drawback is that FPs gain fluorescence through the rate-limiting step of fluorophore maturation, which in some cases can be as long as 12 h, hindering the study of molecular events occurring during that period. To address this problem, many fast maturing FPs have been developed (Bevis and Glick 2002; Nagai et al. 2002; Pedelacq et al. 2006). A third drawback is that FPs are susceptible to photobleaching as is the case with any organic fluorescent dyes, which restricts their use in long term monitoring of cellular events. This problem can be solved by two-photon microscopy, which significantly lowers the donor FP photobleaching rate. This review will discuss recent applications of FRET biosensors.
The principle of FRET
Applications of FRET biosensors using FPs
List of protein biosensors
Biosensor FRET mechanism
Protein binding interaction
Multimerization of IL-17RA
IL-17RA with itself
Kramer et al. (2006)
GPCR subunit association
Gα with Gβγ
Azpiazu and Gautam (2004)
Transcriptional factor Erg and Jun interaction
Erg with Jun
Camuzeaux et al. (2005)
Protein conformational change
Sensing membrane potential
Potassium channel voltage sensing domain
Sakai et al. (2001)
Activation and signalling of rac and cdc42
Intra M Or Intra S
Cdc42 or rac with GTPase binding domains
Caspase proteolytic substrate
CFP YFP Cerulean Venus
Calpain proteolytic substrate
Stockholm et al. (2005)
Factor Xa proteolytic substrate
Mitra et al. (1996)
MLCK and MLCP
RMLC (regulatory myosin light chain)
Yamada et al. (2005)
Kinetics and potencies of 12 known PKC ligands
Braun et al. (2005)
Detection of PKC activities
Truncated pleckstrin containing PH and DEP domains
Schleifenbaum et al. (2004)
Phosphorylation by insulin receptor
Phosphorylation recognition domain and its binding substrate
Activities of EGFR, Src and Ab1
SH2 with phosphorylation substrates for EGFR, Src and Ab1
Ting et al. (2001)
Activation of Src
SH2 with phosphorylation substrates for Src
Wang et al. (2005)
Glucose binding protein
Periplasmic binding proteins
Fehr et al. (2002)
Glutamate/aspartate binding protein ybeJ
Okumoto et al. (2005)
PKA with cAMP-dependent binding substrate
GKI and PDE
Nikolaev et al. (2006)
Estrogen receptor ligand
Estrogen receptor ligand binding domain
De et al. (2005)
Ca2+ in ER
Osibow et al. (2006)
CFP YFP BFP GFP
Miyawaki et al. (1997)
Specific RNA sequence
HIV-1 Rev protein
Endoh et al. (2005)
FRET using intermolecular interactions
Binding interactions between proteins are essential for signal transduction and catalytic activation. For instance, many protein pathways are activated by the association of membrane receptors, which activate enzymes that propagates the signals. To study these binding interactions, FRET biosensors are created by fusing the donor and acceptor FP separately to the interacting proteins of interest. When the intermolecular interaction of these separate fusion proteins occur, the donor and acceptor are consequently brought closer together to create an intermolecular FRET signal corresponding to the location and time of the interaction (see Fig. 2A). Biosensors of this class have proven to be as invaluable tools to study the association mechanisms of membrane receptors as well as the cytoplasmic proteins. Recently, an inflammatory cytokine receptor subunit, IL-17RA, was shown to multimerize and preassemble in the plasma membrane of HEK293 cells. When HEK293 cells were coexpressed with IL-17RA fused to CFP and YFP, the plasma membrane of transfected HEK293 cells displayed a strong FRET signal indicating the association of CFP and YFP tagged IL-17RA (Kramer et al. 2006). Using a similar approach, a biosensor was constructed to study the association of G-protein coupled receptors (GPCRs) with G-proteins. Gα was tagged with CFP and Gβγ with YFP (Azpiazu and Gautam 2004). Before activation by GPCRs, Gα-CFP and Gβγ-YFP subunits bind together, producing a high FRET signal. After activation, Gα-CFP and Gβγ-YFP separate, resulting in the loss of FRET. Studies using this biosensor suggested the mechanism behind the specificity of G-protein signalling pathways was the stochastic collision between GPCRs and G-proteins. In another example, the direct interaction of transcription factors Erg and Jun was demonstrated by fusing them to YFP and CFP respectively (Camuzeaux et al. 2005).
In addition to studying protein associations, intermolecular FRET has also been applied to probe the effects of small molecules in signal transduction pathways. Here, the biosensor mechanism usually relies on the small-molecule-dependent binding of two separate domains that bring a FRET pair into close proximity and consequently, increasing the FRET signal. This mechanism has been used to construct biosensors for cAMP where YFP is fused to protein kinase A (PKA) and CFP to its cAMP-dependent binding substrate (Lissandron et al. 2005; Zaccolo et al. 2005). Using this biosensor, it was discovered that cAMP is generated by cells in discrete functional compartments inside cardiac myocytes (Zaccolo et al. 2002, 2005; Zaccolo and Pozzan 2002). We have seen from the examples described above that intermolecular FRET biosensors provide direct evidence of protein binding events, which is ideal for studying protein associations in vivo. The available intermolecular biosensors to this date cover only a small fraction of the total protein binding interactions, hence, we expect to see a broader the range of protein binding interactions being detected with new biosensors in the near future. Currently, one of the limitations of intermolecular FRET biosensors is that using one FP pair can detect only one pair of protein association. The solution to the problem is to use multiple FRET pair as will be discussed below.
FRET using intramolecular interactions
The changes of protein conformation are usually resulted from biochemical stimulation, including either environmental changes such as the concentration of small molecules and membrane voltage or modifications performed by other proteins such as kinases and phosphatases. By monitoring these protein conformation changes with FRET, we can indirectly detect their inducing molecular events (see Fig. 2B). Indeed, sensory proteins with relatively large conformation changes such as IP3-receptor, cGMP-dependent protein kinase I, estrogen receptor ligand binding domain, HIV-1 Rev protein, glucose-binding protein, periplasmic binding protein and glutamine-binding protein ybeJ exhibit sufficient conformational changes to be used to detect concentration changes of IP3 (Remus et al. 2006; Tanimura et al. 2004), cGMP (Nikolaev et al. 2006), estrogen receptor ligand (De et al. 2005), specific RNA sequence (Endoh et al. 2005), glucose (Fehr et al. 2003, 2004; Ye and Schultz 2003), maltose (Fehr et al. 2002) and glutamate (Okumoto et al. 2005), respectively. These small molecules are important in cellular signalling and metabolic pathways and hence, these biosensors will help to study their dynamics in the cell. For example, a glucose sensor capable of sensing glucose concentration within the physiological range was employed to study the dynamics of glucose metabolism in COS-7 cells (Fehr et al. 2003). The study showed the rapid glucose consumption in COS-7 cells by probing glucose concentration in the cytosol while changing the extracellular glucose concentration or inhibiting glucose transport by cytochalasin B.
In the above examples, target molecules are detected by directly binding to the biosensors. Other intramolecular FRET biosensors are constructed such that instead of binding, the sensory domain is modified by enzymes and is thus able to detect molecular events associated with the activities of the modifying enzyme. For example, a biosensor able to detect phosphorylation of regulatory myosin light chain (RMLC) was constructed utilizing the conformational change of RMLC upon phosphorylation and dephosphorylation by myosin light chain kinase (MLCK) and myosin light chain phosphatase (MLCP), respectively (Yamada et al. 2005). In another example, a Ca2+ biosensor without the parasitic Ca2+-buffering properties was created using the apoK1-er domain, which undergoes a reversible conformational change in a Ca2+-dependent reaction with calreticulin and a protein disulfide isomerase (Osibow et al. 2006). Because of this unique Ca2+-dependent reaction, the biosensor is sensitive to Ca2+ concentration levels in the physiological range of the endoplasmic reticulum (10–700 μM). Proteins may undergo conformational changes without chemically reacting with other molecules. For instance, the voltage-sensing domain of a potassium channel protein changes conformation when triggered with changes in membrane potential. A membrane potential biosensor was cleverly constructed utilizing the rotational conformation change of this protein that causes changes in the orientation factor rather than separation between the FRET pair (Sakai et al. 2001).
In contrast to what was described above, not all proteins undergo significant conformational changes to induce observable FRET differences. In these cases, a sensory domain is fused with its binding substrate and sandwiched between a FRET pair. When the sensory domain is stimulated by the molecular event of interest, it binds to the substrate protein inducing a large overall conformational change that changes the FRET signal (see Fig. 2C). This substrate-binding mechanism has been demonstrated in early studies on Ca2+ biosensors using calmodulin and its binding peptide (Miyawaki et al. 1997). The binding of calmodulin to the binding peptide is reversible depending on the Ca2+ concentration. When Ca2+ concentration rises to 1 μM, calmodulin binds to the peptide to bring CFP and YFP closer, increasing FRET; when Ca2+ concentration drops below 0.1 μM, calmodulin dissociates from the peptide, decreasing FRET. Further improved variants of Ca2+ biosensors have been created to monitor cellular Ca2+ dynamics and signalling (Mank et al. 2006; Miyawaki et al. 1999; Truong et al. 2001). For example, by targeting mitochondria with a yellow cameleon Ca2+ biosensor (Miyawaki et al. 1997), researchers have shown that mitochondria in mouse skeletal muscle cells take up Ca2+ during neuron stimulated contraction and release Ca2+ during relaxation. Furthermore, the Ca2+ dynamics in the mitochondria is several milliseconds behind that of the cytosol (Rudolf et al. 2004). The cameleon biosensors (Miyawaki et al. 1997, 1999) have also been used in transgenic flies (Diegelmann et al. 2002; Fiala and Spall 2003; Fiala et al. 2002; Mank et al. 2006), mice (Hara et al. 2004; Nyqvist et al. 2005; Tsujino et al. 2005), zebrafish (Higashijima et al. 2003) and nematodes (Kerr et al. 2000) to detect cellular activities under external stimulations. For instance, researchers were able to visualize Ca2+ concentration changes in olfactory projection neurons of Drosophila brain when stimulated by odorant, providing a system to model olfaction (Fiala et al. 2002). Biosensors using a similar substrate-binding mechanism have also been created to detect protein activities such as kinases (Braun et al. 2005; Sato et al. 2002; Sato and Umezawa 2004; Schleifenbaum et al. 2004; Ting et al. 2001; Wang et al. 2005) and GTPases (Itoh et al. 2002; Seth et al. 2003).
Presently, the major application of intramolecular biosensors is to detect enzyme activities and the concentration of small molecules as we have seen in the examples discussed above. In the coming years, we expect to see biosensors for most of the important signalling and metabolic molecules. On the technical side, it is easier to characterize the FRET signal from intramolecular biosensors because it is less sensitive to the relative concentration of the biosensor than intermolecular biosensors. Hence, this type of biosensors is preferred in quantitative molecular cell biology studies. However, one of the challenges is to improve the dynamic range of intracellular biosensors in order to achieve higher signal-to-noise ratio. At the moment, intramolecular biosensors having the largest dynamic range is constructed using interacting sensory domains, which could be further optimized by circularly permutated FPs (cpFPs) as will be discussed below.
FRET detecting proteolytic cleavage
Proteases belong to a major class of enzymes essential to many cellular processes such as the initiation and propagation of the apoptosis pathway. Because proteases usually cleave their substrate irreversibly, the detection of protease activity inevitably requires the cleavage of the biosensor (see Fig. 2D). A frequent target for this class of FRET biosensors are caspases due to their importance in the apoptosis pathway and high peptidal substrate specificity (Chiang and Truong 2005; Jones et al. 2000; Nagai and Miyawaki 2004; Onuki et al. 2002; Xu et al. 1998). The caspase biosensors is constructed by fusing a FRET pair to the N- and C-terminal of their peptide substrates, which is usually four residues with the sequence specific to the caspase of interest. When caspase cleaves the substrate in the biosensor, the FRET signal dramatically decreases as the FP pair is separated. By monitoring the FRET signal, the dynamics of caspase-8 activation in single cell was studied showing that caspase-8 activation occurs earlier than caspase-3 during apoptosis, which suggested along with other evidences that caspase-3 activation is dependent on caspase-8 (Luo et al. 2003). Recently, by using both caspase-3 and Ca2+ biosensors in a cell co-culture, the two molecular events were imaged simultaneously showing that the activation of caspase-3 during apoptosis is accompanied by a cytosolic Ca2+ concentration rise and fall (Chiang and Truong 2005). Since the discovery of caspases in the early 90s, caspase biosensors have played critical roles in the discovery of the biochemical properties of caspases and their biological roles inside cells. Beyond caspases, biosensors for other proteases, such as calpain (Stockholm et al. 2005) and Factor Xa protease (Mitra et al. 1996), were also made based on the same principle to image in vivo protease activities.
FRET using multiple FPs
As was mentioned in the previous sections, FRET experiments using one pair of FPs provide information on a single molecular event, however in many cases signalling and metabolic pathways are composed of multiple simultaneous molecular events. To image two molecular events, FRET biosensors using multiple FPs have been used. As each FRET pair occupies a large portion of the visible spectrum, it is difficult to introduce a second distinct FRET pair. To solve the problem, a triplet of CFP-YFP-mRFP was used to measure three distinct FRET signals—CFP to YFP, CFP to mRFP, YFP to mRFP (see Fig. 2E) (Galperin et al. 2004; He et al. 2005). As a proof-of-concept experiment to simultaneously monitor the three FRET events, the three-domain complex of Rab5 and EEA.1sh was detected and the interaction between EGFR with Grb2 and Cbl was confirmed (Galperin et al. 2004). In a similar experiment, the trimerization of TRAF2 was confirmed in living cells (He et al. 2005). Currently, the only triplet FPs for multiple FRET is CFP-YFP-RFP. Since the CFP-YFP FRET is very high and the YFP-RFP FRET is barely detectable, only a handful of studies were conducted using multiple FRET to this date. We expect multiple FRET to become more popular as more RFP mutants are created with better properties as FRET acceptors.
Fluorescent proteins for FRET
Spectral properties of several FRET pairs
Ex (Steinmeyer et al. 2005)
Em (Steinmeyer et al. 2005)
Extinction coefficient (EC)
Quantum yield (QY)
Relative brightness (ECxQY)
Rizzo et al. (2004)
Nagai et al. (2002)
Griesbeck et al. (2001)
Nguyen and Daugherty (2005)
Nguyen and Daugherty (2005)
Karasawa et al. (2004)
Karasawa et al. (2004)
Zapata-Hommer and Griesbeck (2003)
Shaner et al. (2004)
Yang et al. (2005)
Conclusion and perspectives
FRET biosensors are valuable tools in monitoring molecular events in living cells. Here, we have reviewed the concepts and applications of FRET biosensors. They are widely used for their unique advantages in cellular imaging, including spontaneous fluorescence, simple genetic manipulations, ease of delivery inside cells, and flexibility in targeting to organelles and tissues. One of the challenges in FRET biosensor design is the availability of good FRET pairs. Consequently, many recent studies on FPs are focused on creating variants that are especially designed for FRET applications (Karasawa et al. 2004; Nguyen and Daugherty 2005; Rizzo et al. 2004). Another challenge in biosensor design is ensuring that the sensory domain traverses conformations with distance and orientation factors favourable for the FRET signals. Looking into the future, we anticipate FRET biosensors will allow us to simultaneously image the many coordinated and complex molecular events of signalling and metabolic pathways within transgenic animals.
This work was supported by grants from the Canadian Foundation of Innovation (CFI) and the National Science and Engineering Research Council (NSERC).
- Valeur B (2002). Molecular fluorescence: principles and applications. Wiley-VCH, WeinheimGoogle Scholar