Fundamental Principles of Förster Resonance Energy Transfer (FRET) Microscopy with Fluorescent Proteins

In living cells, dynamic interactions between proteins are thought to play a key role in regulating many signal transduction pathways, as well as contributing to a wide spectrum of other critical processes. In the past, classical biochemistry approaches to elucidating the mechanism of such interactions were commonplace, but weak or transient interactions that might occur within the natural cellular milieu are usually transparent to these techniques. For example, co-localization of suspected protein partners using immunofluorescence microscopy in fixed cells has been a popular method for examining interactions in situ, and numerous literature reports have been presented based on this technique. However, because the resolution of a fluorescence microscope is several hundred times less than the size of a typical protein, co-localization often leads to questionable results. An excellent analogy is that fluorescence microscopy yields information equivalent to the knowledge that two students are present in a large lecture hall. It doesn't offer the resolution necessary to determine whether the students are in the same classroom or, better yet, if they are sitting in adjacent desks.

Typical fluorescence microscopy techniques rely upon the absorption by a fluorophore of light at one wavelength (excitation), followed by the subsequent emission of secondary fluorescence at a longer wavelength. The excitation and emission wavelengths are often separated from each other by tens to hundreds of nanometers. Labeling of cellular components, such as the nuclei, mitochondria, cytoskeleton, the Golgi apparatus, and membranes, with specific fluorophores enables their localization within fixed and living preparations. By simultaneously labeling several sub-cellular structures with individual fluorophores having separated excitation and emission spectra, specialized fluorescence filter combinations can be employed to examine the proximity of labeled molecules within a single cell or tissue section. Using this technique, molecules that are closer together than the optical resolution limit appear to be coincident (and are said to co-localize). This apparent spatial proximity implies that a molecular association is possible. In most cases, however, the normal diffraction-limited fluorescence microscope resolution is insufficient to determine whether an interaction between biomolecules actually takes place.

Co-localization measurements are suggestive at best and misleading at worst, especially considering that many signaling pathways use the same cellular structure, as for example, clathrin-coated pits that are utilized for internalization of many receptor complexes. The knowledge that two molecules or proteins are in fact adjacent, and not just residing in the same neighborhood, provides a significantly more reliable determination of their potential for interactions. The time-honored technique of electron microscopy has ample resolution to meet the needs of high-precision localization, but simply lacks the precise labeling methodology necessary to produce reliable results. Furthermore, many co-localization techniques are generally applied for use within fixed cells, which precludes the highly desirable dynamic measurements attainable by assays in living cells. Fluorescence imaging with multi-color fluorescent proteins readily permits experimentation with live cells, which are necessary for assays of transient interaction, but the approach suffers from having a relatively poor spatial resolution limited to approximately 200 nanometers.

Limitations in determination of the spatial proximity of protein molecules can be overcome by applying Förster (or Fluorescence) Resonance Energy Transfer (FRET) microscopy techniques. FRET occurs between two appropriately positioned fluorophores only when the distance separating them is 8 to 10 nanometers or less. Thus, FRET is well-suited to the investigation of protein interactions that occur between two molecules positioned within several nanometers of each other. Over the past ten years, FRET approaches have gained popularity due to the rise in applications requiring genetically targeting of specific proteins and peptides using fusions to green fluorescent protein (GFP) and its mutated derivatives. FRET between two spectrally distinct fluorescent proteins (known as FP-FRET) has been widely applied for two distinctly separate experimental techniques, as discussed below. Presented in Figure 1 is a Jablonski energy diagram illustrating the coupled excited state transitions involved between the donor emission and acceptor absorbance in FRET. Absorption and emission transitions are represented by straight vertical arrows (blue, green and red), while vibrational relaxation is indicated by wavy yellow arrows. The coupled transitions are drawn with dashed lines that suggest their correct placement in the Jablonski diagram should they have arisen from photon-mediated electronic transitions. In the presence of a suitable acceptor, the donor fluorophore can transfer excited state energy directly to the acceptor without emitting a photon (illustrated by a violet arrow in Figure 1). The resulting fluorescence sensitized emission has characteristics similar to the emission spectrum of the acceptor.

One of the major obstacles to the widespread implementation of FRET investigations in living cells has been the lack of suitable methods for labeling specific intracellular proteins with the appropriate fluorophores. The recent development of fluorescent proteins possessing a wide array of spectral profiles and the increasing sophistication of protein chimeras (fusions as well as biosensors) has resulted in a number of potential fluorescent protein pairs that are useful in FRET experiments. Application of fluorescent proteins to FRET involves either integrating a selected pair into a biosensor (a single genetically-encoded construct) or conducting intermolecular measurements between two separate proteins, each fused to a different fluorescent protein. The latter approach has been employed to image a variety of protein interactions, including oligomerization of receptors and elucidating the functions of transcription factors. However, conducting FRET assays on independently expressed protein chimeras is far more difficult due to the variable stoichiometry that inevitably occurs when separate fluorescent entities are expressed in living cells. Regardless of the difficulty, experiments of this nature can yield informative results when appropriate controls are installed and the investigation is conducted with exacting precision.

Fluorescent Protein Biosensors

Fluorescent protein biosensors have found widespread utility in reporting on a diverse array of intracellular processes. By creatively fusing pairs of fluorescent proteins to biopolymers that perform critical functions involved in various aspects of physiological signaling, research scientists have developed a host of new molecular probes that are useful for optical live-cell imaging of important processes such as calcium wave induction, cyclic nucleotide messenger effects, pH changes, membrane potential fluctuations, phosphorylation, and intracellular protease action. An alternative, but quite useful, strategy to biosensor construction involves modifications to the fluorescent protein backbone structure itself, either to split the peptide into individual units that are combined in vivo to produce fluorescence (a technique termed Bi-Molecular Fluorescence Complementation; BiFC) or to join the natural amino and carboxy termini together and create an insertion site within the molecule for a sensor peptide.

The first fluorescent protein biosensor was a calcium indicator named cameleon, constructed by sandwiching the protein calmodulin and the calcium calmodulin-binding domain of myosin light chain kinase (M13 domain) between enhanced blue and green fluorescent proteins (EBFP and EGFP). In the presence of increasing levels of intracellular calcium, the M13 domain binds the calmodulin peptide to produce an increase in FRET between the fluorescent proteins. Unfortunately, this sensor was hampered by a very low dynamic range (a 1.6-fold increase in fluorescence) and was difficult to visualize due to lack of brightness and poor photostability of EBFP. Improved versions using the same template incorporated the cyan and yellow variants ECFP and EYFP to yield higher signal levels, and even better results were obtained when YFP derivatives (termed camgaroos) were generated by inserting the calcium-sensitive peptides at the beginning of the seventh beta-strand in the fluorescent protein backbone. Sensor peptides situated at this unusual position are quite well tolerated with regards to maintaining high levels of fluorescence. Yet another strategy takes advantage of the unique barrel structure common in fluorescent proteins to reconfigure the ends of the protein by linking the natural N and C termini and creating a new start codon in one of several locations within the central region of the structure (usually in the loops). Termed circularly permuted fluorescent proteins, these structurally modified derivatives can be fused to calmodulin and M13 to produce excellent calcium biosensors.

Calcium biosensors were quickly followed by genetic indicators for pH, phosphorylation, and protease activity. Two general approaches can be used to adapt fluorescent proteins as sensors of pH. The first relies on the fluorescence sensitivity of EGFP (pKa = 5.9) and EYFP (pKa = 6.5) to acidic environments coupled to the relative insensitivity of other proteins, such as ECFP (pKa = 4.7) or DsRed (pKa = 4.5). Fusions of EGFP or EYFP with a less sensitive fluorescent protein create a ratiometric probe that can be used to measure the acidity of intracellular compartments. The second approach relies on protonation changes of native (wild-type) GFP that result in a shift in the bimodal spectral profiles of the native protein. A class of probes named pHluorins, derived from wtGFP, exhibits a shift in the excitation peak from 470 to 410 nanometers as the pH decreases. Dual-emission pH sensors have also been developed, which have peaks in the green and blue spectral regions. Although unable to report kinase activity in real time, phosphorylation biosensors consist of a peptide containing a phosphorylation motif from a specific kinase and a binding domain for a phosphopeptide sandwiched between two FRET-capable fluorescent proteins. When the biosensor is phosphorylated by the kinase, the phosphopeptide binding domain binds to the phosphorylated sequence, thus invoking or destroying FRET. This simple strategy has proven to generate robust and highly specific biosensors. As with many other biosensors, the major drawback is reduced dynamic range.

Perhaps the most widely used biosensor design to screen new or improved FRET pairs involves a protease cleavage assay (see Figure 2). The simple motif consists of two fluorescent proteins linked together by a short peptide that contains a consensus protease cleavage site. In general, the sensor exhibits very strong energy transfer that is completely abolished upon cleavage of the linker sequence. Because the technique usually features high dynamic range levels, it can be used to screen new cyan and green FRET donors with yellow, orange, and red acceptors. The largest family of protease biosensors incorporates a cleavage site sensitive to one of the caspase family of proteases, which enables the sensor to be examined during induction of apoptosis. Over the past several years, a large number of novel biosensors using both sensitized fluorescent proteins and FRET pairs have been reported. Despite the continued limitations in dynamic range of FRET sensors using ECFP and EYFP derivatives, this strategy has been widely adopted, probably due to the simplicity of ratiometric measurements and ease of probe construction. New strategies will no doubt emerge using more advanced fluorescent protein combinations that serve to increase the dynamic range and other properties of this highly useful class of probes.

Basic Principles of FRET

The fundamental mechanism of FRET involves a donor fluorophore in an excited electronic state, which may transfer its excitation energy to a nearby acceptor fluorophore (or chromophore) in a non-radiative fashion through long-range dipole-dipole interactions. The theory supporting energy transfer is based on the concept of treating an excited fluorophore as an oscillating dipole that can undergo an energy exchange with a second dipole having a similar resonance frequency. In this regard, resonance energy transfer is analogous to the behavior of coupled oscillators, such as a pair of tuning forks vibrating at the same frequency or a radio antenna. In contrast, radiative energy transfer requires emission and re-absorption of a photon and depends on the physical dimensions and optical properties of the specimen, as well as the geometry of the container and the wavefront pathways. Unlike radiative mechanisms, resonance energy transfer can yield a significant amount of structural information concerning the donor-acceptor pair.

Resonance energy transfer is not sensitive to the surrounding solvent shell of a fluorophore, and thus, produces molecular information unique to that revealed by solvent-dependent events, such as fluorescence quenching, excited-state reactions, solvent relaxation, or anisotropic measurements. The major solvent impact on fluorophores involved in resonance energy transfer is the effect on spectral properties of the donor and acceptor. Non-radiative energy transfer occurs over much longer distances than short-range solvent effects and the dielectric nature of constituents (solvent and host macromolecule) positioned between the involved fluorophores has very little influence on the efficacy of resonance energy transfer, which depends primarily on the distance between the donor and acceptor fluorophore.

The phenomenon of FRET is not mediated by photon emission, and furthermore, does not even require the acceptor chromophore to be fluorescent. In most applications, however, both donor and acceptor are fluorescent, and the occurrence of energy transfer manifests itself through quenching of donor fluorescence and a reduction of the fluorescence lifetime, accompanied also by an increase in acceptor fluorescence emission. The theory of resonance energy transfer was originally developed by Theodor Förster and, in honor of his contribution, has recently been named after him. The Förster theory shows that FRET efficiency (E) varies as the inverse sixth power of the distance between the two molecules (denoted by r):

EFRET = 1/[1 + (r/R0)6]

where R(0) is the characteristic distance where the FRET efficiency is 50 percent, which can be calculated for any pair of fluorescent molecules (this variable is also termed the Förster radius and is discussed below in greater detail). The FRET efficiency of a theoretical fluorophore pair (enhanced cyan and yellow fluorescent proteins) is graphically demonstrated in Figure 3(a). Because of the inverse sixth power dependence on the distance between the two molecules (r), the curve has a very sharp decline. For distances less than R(0), the FRET efficiency is close to maximal, whereas for distances greater than R(0), the efficiency rapidly approaches zero. The useful range for measuring FRET is indicated by the red shaded region in Figure 3(a) with limits of 0.5 and 1.5 x R(0). FRET can be effectively used as a molecular ruler for those distances close to R(0), and indeed FRET has been adapted for such purposes in structural biology by using precision spectroscopic approaches. For most applications in cell biology, however, the signal-to-noise ratios available limit FRET experiments to a more binary readout. In effect, a measurement will often be only able to distinguish between high-FRET and low-FRET, or simply between the presence and absence of FRET.

As previously discussed, R(0) can be readily calculated for any pair of fluorescent molecules. The value of R(0) in an aqueous (or buffered) solution is determined by a fairly simple equation with the well-established input parameters:

R0 = [2.8 x 1017 × Κ2 × QD × ΕA × J(λ)]1/6 nanometers

where Κ(2) or kappa squared represents the orientation factor between the two fluorophore dipoles (see Figure 3(b) for a summary of angles used to calculate the orientation factor), Q(D) is the donor quantum yield, Ε(A) is the maximal acceptor extinction coefficient in reciprocal moles per centimeter, and J(λ) is the spectral overlap integral (see Figure 4) between the normalized donor fluorescence, F(D)(λ), and the acceptor excitation spectra, E(A)(λ), according to the equation:

J(λ) = FD(λ) × ΕA(λ) × λ4

Although the mathematics may appear complicated, most of the parameters are constants that are easily found in the literature. The two most important terms that generally require further explanation are Κ(2) and J(λ), the overlap integral. The orientation angle variable (Κ(2)) simply indicates that the FRET coupling depends on the angle between the two fluorophores in much the same manner as the position of a radio antenna can affect its reception. If the donor and acceptor are aligned parallel to each other, the FRET efficiency will be higher than if they are oriented perpendicular. This degree of alignment defines Κ(2). Although Κ(2) can vary between zero and 4, it is usually assumed to be 2/3, which is the average value integrated over all possible angles. For almost any realistic situation Κ(2) is close to 2/3, and there is usually nothing that an investigator can do to adjust this value (although some have attached fluorescent proteins rigidly to their target proteins of interest, which could lead to dramatic effects). The overlap integral, J(λ), is the region of overlap between the two spectra, as illustrated in Figure 4. The other parameters that can affect FRET are the quantum yield of the donor and the extinction coefficient of the acceptor. Thus, in order to maximize the FRET signal, the researcher must choose the highest quantum yield donor, the highest absorbing acceptor, and fluorophores having significant overlap in their spectral profiles. This theory has been repeatedly verified by experiment, and there are no other mechanisms to maximize FRET for non-aligned fluorescent probes.

It should be noted that each of the parameters discussed above affects the Förster radius calculation only by the sixth power. Thus, doubling of the donor quantum yield results in only a 12.5 percent change in R(0). Because almost all fluorophores used in FRET imaging experiments have high quantum yields (greater than 0.5) and extinction coefficients (greater than 50,000), the range of possible Förster radius values is limited to between 4 and 6 nanometers, and most FRET pairs have an average value of R(0) ~ 5 nanometers. Given that FRET efficiency is strongly dependent on the distance separating the FRET pair as well as the relative orientation of the fluorophores, FRET can be used to detect changes in protein-protein interactions that arise from changes in the affinity between the two proteins or changes in the conformation of their binding. It is worth repeating that, for most FRET imaging applications in cell biology, experiments generally differentiate only between two states (FRET and no FRET) and additional information is necessary to aide in the molecular interpretation of the observed FRET changes.

Factors Affecting FRET Measurements

In practice, a wide spectrum of issues can complicate and/or compromise FRET measurements, ultimately leading to ambiguous or meaningless results. One of the principal issues is that the donor and acceptor fluorophores might exhibit significantly different brightness levels when imaged together. Although in theory this discrepancy should not be a problem, however in practice because most instruments can measure only a limited dynamic range, dual fluorophore imaging may result in one channel that is saturated (for the brighter fluorophore) while the other channel is dominated by systematic noise (for the dimmer fluorophore). Thus, whenever possible it is best to use a donor and acceptor that are of comparable brightness.

Another factor that can limit the detection of FRET is a donor-to-acceptor stoichiometry that lies outside the range between 10:1 and 1:10. This factor can be a serious limitation in FRET measurements of protein-protein interactions in which one partner might be in excess concentration. The primary problem is the measurement of a small level of FRET against a background of fluorescent labels that are not undergoing FRET. Due to the fact that there is really nothing that can be done to improve this situation, a host of possible protein-protein interaction experiments falling into this category are simply unsuitable for examination by FRET techniques. For the fluorescent protein biosensors described above, which are constructed with only a single donor and acceptor, the stoichiometry is fixed and guaranteed to be 1:1; thus, this issue never arises and the signal level remains constant, regardless of the biosensor concentration.

The presence of bleed-through (also termed crosstalk and crossover) and cross excitation between spectrally overlapping fluorophores are also important issues that can hamper FRET investigations (see Figure 5). In some cases, the acceptor can be directly excited with light in the wavelength region chosen to excite the donor (Figure 5(a)). Additionally, fluorescence from the donor can leak into the detection channel for the acceptor fluorescence, especially when the emission spectral profiles of the donor and acceptor exhibit significant overlap (Figure 5(b)). Because these two sources of crosstalk arise from the photophysics of organic fluorophores and will most certainly be present for any FRET pair, they must be addressed when FRET is measured. Choosing fluorophores that are well-separated spectrally is an excellent mechanism to reduce crosstalk. However, in most cases the increased spectral separation also reduces the overlap integral, (J(λ)), which in practice usually translates to a reduced ability to detect the FRET signal.

Finally, the level of a FRET signal can be reduced if the two fluorophores are not properly aligned (for instance, having a Κ(2) value of approximately zero) or if they are simply not positioned within the Förster radius (greater than 6 nanometers). As an example, if two labeled proteins interact, but the fluorescent labels are located on opposite sides of the complex, then there might not be a detectable FRET signal, even though the proteins of interest are bound. In general practice, this type of false negative is quite common, especially with fluorescent protein FRET partners. Often, several labeling strategies are required before a sufficient and reliable FRET signal is detected. However, each of the issues described above can be mitigated (or partially so) by an informed choice of the fluorophore pair to be used prior to making vector constructs or conducting synthetic labeling experiments.

Presented in Figure 5 is the overlap in the excitation and emission spectral profiles of ECFP and mVenus, currently one of the most preferred fluorescent proteins pairs for FRET investigations. These two proteins exhibit considerable overlap in both the excitation (Figure 5(a)) and emission (Figure 5(b)) spectra. Direct excitation of the FRET acceptor (mVenus; red curve) can be significant depending on the wavelength used for excitation of the donor (ECFP; cyan curve or mCerulean; blue curve) due to the higher extinction coefficient of the yellow protein as compared to the cyan proteins. This overlap is especially problematic when ECFP is used as the donor and can be partially offset by using CFP variants with high extinction coefficients, such as mCerulean. Note that the excitation curves in Figure 5(a) are drawn to scale in order to reflect the differences in extinction coefficient between the yellow and cyan proteins. Excitation at 458 nanometers produces a much higher level of mVenus excitation crosstalk than does excitation at 405 or 440 nanometers. The broad fluorescence emission spectrum of ECFP (Figure 5(b)) exhibits considerable intensity overlap throughout the region of mVenus emission.

FRET Techniques in Cell Biology Applications

Investigators employing fluorescent protein biosensors, or attempting to match the stoichiometry of fluorescent probes fused to separate interacting targets, should use as many different FRET analysis techniques as feasible to establish the methodology for a given experiment. Such an effort is warranted because each of the fluorescent protein FRET pairs exhibits a distinct pathology that complicates its use, necessitating a clear understanding of the optical microscopy parameters applied to measuring the relatively small signal differences produced in most FRET assays. Once the system and the possible results are well established, then the simplest approaches can be used for ongoing procedures. The list of techniques that have been developed to image FRET is quite extensive. In general, all of the existing strategies for measuring FRET can be applied to fluorescent protein experiments but, on the basis of practical considerations, five general approaches have proven particularly useful:

  • Sensitized Emission - Two-channel imaging using an algorithm that corrects for excitation and emission crosstalk

  • Acceptor Photobleaching - Also known as donor dequenching, this technique measures increased donor emission when the acceptor is photobleached

  • Fluorescence Lifetime Imaging Microscopy (FLIM) - Fluorescent protein (or other fluorophore) donor lifetime measurement changes

  • Spectral Imaging - Exciting at one or two wavelengths and measuring the complete spectral profiles of donor and acceptor

  • Fluorescence Polarization Imaging - Measure polarization parallel and perpendicular to excitation with high signal-to-noise

Each of the FRET approaches listed above has strengths and weaknesses. For example, on one hand, two-channel imaging is the simplest method, but requires the most complicated set of controls. On the other hand, FLIM can yield an unambiguous measurement of FRET efficiency, but instrumentation to measure the nanosecond lifetimes is expensive and not yet widely available to most cell biology laboratories.

Sensitized Emission

Also commonly referred to as two-color ratio imaging with controls, sensitized emission is perhaps the simplest method of imaging FRET. The donor fluorophore is excited by a specific wavelength (in a widefield or confocal microscope), and the signal is collected by using emission filters chosen for the donor fluorescence and the acceptor fluorescence. In the (unrealistic) absence of crosstalk between the excitation and fluorescence of the two fluorophores, then sensitized emission would be a perfect method. However, crosstalk between fluorescent proteins is a significant problem and extensive control experiments are usually required to establish the presence or absence of FRET. Thus, it is difficult to obtain quantitatively accurate FRET data with this approach. Sensitized emission is relatively simple to configure on a widefield fluorescence microscope, available in many laboratories, but the necessary control experiments require considerable image processing to subtract crosstalk components, which significantly increases the noise level and uncertainty in the measurements.

A variety of corrective approaches have been developed for sensitized emission FRET imaging. The basic concept involves the use of different filter combinations with multiple samples that contain: only the donor, only the acceptor, and the putative FRET sample with both the donor and acceptor. The emission values from these samples permit the investigator to determine the amount of expected crosstalk in both excitation and emission channels and to subtract it from the FRET measurement. In theory this approach works nicely, but the requirement for image processing unfortunately increases the noise level in all of the images. Thus, if the FRET signal is minute, then it may be difficult to measure FRET using this approach.

Despite the difficulties mentioned above, sensitized emission measurements can be useful for rapid dynamic experiments in which FRET signals are large due to the ability to acquire both images simultaneously. Sensitized emission is a particularly attractive technique when examining fluorescent protein biosensors where the FRET dynamic range is large and the stoichiometry of the donor and acceptor is fixed in a 1:1 ratio. A good example is the protease biosensor illustrated in Figure 2. This chimera has been engineered to have a high FRET efficiency that drops essentially to zero when the peptide linker is enzymatically cleaved. The result is a large and readily measurable FRET change that demonstrates a specific protease activity at a given time and region within the living cell.

Acceptor Photobleaching

Although limited to only a single measurement, acceptor photobleaching (or donor dequenching) is also a simple technique that often yields excellent results. The underlying concept takes advantage of the fact that donor fluorescence is quenched during FRET because some of the donor fluorescence energy is channeled to the acceptor. Photobleaching the acceptor fluorophore irreversibly eliminates the quenching effect and increases the level of donor fluorescence. If FRET is occurring between the fluorophores, the donor fluorescence must increase when the acceptor is removed. In general, it is important to ensure that acceptor photobleaching does not degrade the donor fluorescence, and that the acceptor is photobleached to approximately 10 percent of its initial value. Both of these constraints are easily met with a laser scanning confocal microscope, but can also be accomplished with widefield or spinning disk microscopes equipped with a specialized illumination system.

The acceptor photobleaching technique has the advantage of being very straightforward, quantitative, and performed using only a single sample. The FRET efficiency can be calculated by subtracting the donor intensity in the presence of the acceptor from its intensity after photobleaching the acceptor, and then normalizing this value to the donor intensity after bleaching. The primary disadvantage is that acceptor photobleaching is destructive and can be used only once per cell, limiting its application to those experiments not involved with dynamic measurements. Furthermore, photobleaching is a relatively slow process that often requires several minutes or longer. Nevertheless, it is almost always worthwhile to perform an acceptor photobleaching measurement at the end of an experiment, regardless of whichever methods are being used to assay FRET.

Presented in Figure 6 are examples of sensitized emission and acceptor photobleaching FRET assays using live cell imaging. Figure 6(a) illustrates a human cervical carcinoma epithelial cell (HeLa line) expressing a cameleon biosensor comprised of mCerulean and mVenus fused together with an intervening calcium-sensitive peptide containing calmodulin and the M13 domain (described above). Prior to the addition of a calcium-inducing agent (ionomycin), excitation of the cell with 440-nanometer illumination produces cyan fluorescence indicating a lack of FRET between the cyan and yellow fluorescent proteins (Figure 6(a)). Upon addition of ionomycin, time lapse two-color ratio imaging (sensitized emission) records a calcium wave traversing the cytoplasm as the biosensor responds with an increase in the level of FRET between the fluorescent proteins (Figures 6(b) and 6(c); FRET is pseudocolored yellow-red). The African green monkey kidney cell (COS-7 line) in Figure 6(d)-(f) was labeled with the synthetic cyanine dyes, Cy3 (Figure 6(d); green) and Cy5 (Figure 6(e); red), conjugated to cholera toxin B-subunit and targeting the plasma membrane. Within the membrane, the close proximity of the two dyes produces a high level of FRET. Photobleaching Cy5 in a selected region of the cell (white box in Figure 6(e)) increases the donor dequenching (increase in green fluorescence in Figure 6(f)) in a corresponding area when viewing fluorescence in the donor channel only.

Fluorescence Lifetime Imaging Microscopy (FLIM)

Lifetime measurements are by far the most rigorous method for determining FRET; furthermore, they are also less prone to crosstalk artifacts due to the fact that only the donor fluorescence is monitored. All fluorescent molecules exhibit an exponential decay in their fluorescence on a nanosecond timescale, and the rate of this decay is sensitive to environmental variables that quench the fluorescence. Thus, the basic concept of FLIM is somewhat related to that of acceptor photobleaching. The donor fluorescence is quenched by the FRET interaction, and the amount of quenching can be determined by measuring the decrease in fluorescence decay time of the donor in the presence of FRET. In this manner, FLIM provides an unambiguous value of the FRET efficiency. Among the advantages of combined FLIM-FRET measurements is their insensitivity to direct acceptor excitation artifacts as well as the fact that fluorescent donors can be coupled to acceptors that are not themselves fluorescent. Both of these aspects serve to expand the number of useful fluorescent protein FRET pairs available to investigators.

FLIM has several limitations that prevent it from being the dominant approach in FRET imaging. Primarily, measurements in the nanosecond lifetime region are complex and the instrumentation is expensive to obtain and maintain. Also, this type of sophisticated equipment is not widely available. Additionally, FLIM is usually among the slower imaging methodologies, potentially requiring several minutes to acquire each image, which limits its utility in many FRET experiments. These constraints might be lifted in the future as more user-friendly and faster turnkey commercial systems are developed by the manufacturers. Another significant downside is that the lifetimes of fluorescent proteins in live cells often display multi-exponential decays that require more comprehensive data analysis for quantitative FRET assays. Furthermore, localized environmental factors, such as autofluorescence or a change in pH, can also shorten the measured fluorescence lifetime, leading to artifacts. Thus, a great deal of care must be taken in the interpretation of FLIM-FRET data in living cells.

Spectral Imaging

The technique of spectral imaging is a variation on the sensitized emission FRET detection method, but instead of acquiring data through two individual channels, the entire emission spectrum containing both donor and acceptor fluorescence is collected upon excitation of the donor. Recording of the entire spectrum is a typical approach used for spectroscopy experiments, but is a relatively recent addition to the tool palette in widefield and confocal microscopy. The concept centers on the premise that collection of the entire fluorescence spectrum enables overlapping spectra to be separated by using not just the emission peaks but also the distinct shapes of the spectral tails. In gathering the spectrum from both the donor and acceptor fluorophore, it is possible to determine the relative levels of donor and acceptor fluorescence.

The spectral imaging technique requires specialized equipment, but excellent systems are readily available on many commercial confocal microscopes and can be added onto a conventional fluorescence microscope at modest cost. Conducting a quantitative analysis of the level of crosstalk due to direct excitation of the acceptor, or the use of two excitation wavelengths in confocal microscopy, permits an accurate determination of the amount of FRET. The principal drawback of this approach is the reduced signal-to-noise ratio associated with acquiring the complete spectrum rather than collecting it through two channels with a filter-based system. As more commercial systems are being developed and installed, however, the application of spectral imaging in FRET assays is increasing. In the near future, it is quite possible that spectral imaging will become one of the primary methods for performing FRET imaging experiments.

Illustrated in Figure 7(a) are changes in the donor lifetime decay (mCerulean fluorescent protein) of a pseudo-FRET biosensor consisting of mCerulean and mVenus fluorescent proteins fused together with a 10-amino acid linker. The blue decay curve shows the lifetime observed in cells expressing mCerulean alone, whereas the red decay curve presents the mCerulean lifetime obtained when cells express the concatenated proteins. Note the decrease in mCerulean lifetime when the protein is involved in resonance energy transfer. The area between the curves represents the energy that is transferred through FRET from mCerulean (donor) to mVenus (acceptor) in the FRET pairing. The emission profile from 450 to 650 nanometers of mCerulean-mVenus in the same pseudo-biosensor when excited at 405 nanometers in live cells is depicted by the red curve in Figure 7(b). Transfer of energy from mCerulean to mVenus results in a substantial emission peak at 529 nanometers (the mVenus emission maximum), with a much lower value (approximately 25 percent) at 475 nanometers, the peak emission wavelength of mCerulean. After photobleaching mVenus with a 514-nanometer laser and repeating the spectral scan, the emission profile shifts to lower wavelengths and closely resembles the spectrum of mCerulean in the absence of a FRET partner. The difference in intensities at 475 and 529 nanometers of these spectral profiles is related to the FRET efficiency between the coupled proteins.

Polarization Anisotropy Imaging

Measurements of fluorescence polarization offer particular advantages for high-contrast discrimination of fluorescent protein FRET. The concept is based on the fact that excitation with polarized light selects a population of fluorescent molecules whose absorption vectors are aligned parallel to the polarization vector of the exciting light. Immediately after excitation, most of the fluorescence emission will remain polarized parallel to the excitation so that the fluorescence can be considered anisotropic in terms of polarization. The anisotropy will disappear if the molecules rotate during the nanosecond fluorescent lifetime. However, because fluorescent proteins are large and rotate slowly, their fluorescence does not depolarize to any great degree during the measurement time course. If FRET occurs between two fluorescent proteins that are slightly misaligned, then the polarized fluorescence emission will emerge at a different angle (from the excitation vector), which simulates a rotation of the fluorescent protein.

The primary strength of this approach is the ease of measuring fluorescence polarization parallel and perpendicular to the excitation vector with high signal-to-noise. Because polarization anisotropy data can be acquired rapidly and with minimal image processing requirement, the technique is well-suited for applications in high-content screening. Direct excitation of the acceptor must be avoided, however, because it can decrease the donor signal and reduce the signal-to-noise ratio of the measurement. In addition, although this technique is superb in discriminating between the presence and absence of FRET, it is not a good approach for differentiating between strong and weak FRET. Finally, polarization can be degraded in high numerical aperture objectives, so polarized FRET experiments should be limited to imaging with objectives having a numerical aperture of 1.0 or less.

Presented in Figure 8 is a graphic illustration of polarization anisotropy using fluorescent proteins as a model system. When a randomly oriented population of fluorescent proteins (Figure 8(a)) is excited with linearly polarized light (cyan wave), only those molecules whose absorption dipole vector is oriented parallel to the polarization azimuth are preferentially excited. Emission from properly oriented fluorescent proteins can be observed as a signal using an analyzer that is also parallel to the excitation light polarization vector (green wave). The resulting anisotropy, which is an indicator of the degree of orientation, can be determined by measuring and comparing the emission intensity through the vertically and horizontally oriented analyzers. The anisotropy signal level will decrease if the fluorescent protein rotates in the timescale of the experiment (Figure 8(b)) or if it transfers excitation energy due to FRET to a neighboring protein (Figure 8(c)) having a different orientation. As described above, due to the fact that resonance energy transfer can occur far more rapidly than molecular rotation for large fluorescent protein molecules, depolarization due to FRET can be readily distinguished from the loss of anisotropy that occurs during rotation.

Considerations for Using Fluorescent Proteins in FRET

The choice of suitable probes for examining FRET in living cells is limited. Synthetic fluorophores, ideal for resonance energy transfer investigations in fixed cells, are difficult to administer and target in live cells. Likewise, quantum dots can be utilized to label membrane components for examination of phenomena on the exterior of a cell, but they too are unable to penetrate the membrane and, consequently, of little use in intracellular compartments such as the nucleus, mitochondria, or endoplasmic reticulum. Genetically encoded fluorescent proteins currently represent the best candidates for high-resolution imaging of FRET in live cells, as evidenced by the volume of literature that is published in this arena on a yearly basis. However, many of the typical artifacts that are encountered in measuring FRET with synthetic fluorophores and quantum dots are particularly acute when applied to fluorescent proteins. For example, contrary to the 30-40 nanometer bandwidth of emission spectral profiles in synthetics, those in fluorescent proteins range from approximately 60 nanometers to 100 nanometers, often leading to significant overlap when attempting to segregate donor and acceptor fluorescence. The broad spectra of fluorescent proteins also limit the number of probes that can be used together in FRET and other types of imaging experiments. Furthermore, fluorescent proteins exhibit a wide variation in brightness levels. For example, one of the most popular donor proteins, ECFP, has fivefold less brightness than its common yellow acceptor partner, EYFP.

Surrounding the fluorescent protein chromophore is a 220+ amino acid polypeptide wound into a three-dimensional cylindrical structure approximately 2.4 by 4.2 nanometers in dimension (termed a beta-barrel or beta-can), and composed of extensively hydrogen-bonded polypeptide beta-sheets that surround and protect a central alpha-helix containing the chromophore (see Figure 9). The ends of the barrel are capped with semi-helical peptide regions that serve to block entry of ions and small molecules. The interior of the protein is so tightly packed with amino acid side chains and water molecules that there is little room for diffusion of oxygen, ions, or other intruding small molecules that manage to pass through the ends of the barrel. These favorable structural parameters, which are partially responsible for the resilient photostability and excellent performance of fluorescent proteins, also contribute to a reduction in FRET efficiency. The large size of the barrel effectively shields adjacent fluorescent protein chromophores with peptide residues (to a limiting close approach distance of 2 to 3 nanometers; indicated by the red line in Figure 9), resulting in a reduction of the maximum FRET efficiency to approximately 40 percent of the theoretical value. Regardless, the numerous benefits of using fluorescent proteins for live cell FRET imaging far outweigh the costs.

Compounding the high degree of spectral bandwidth overlap and size problems that occur with fluorescent proteins is their tendency to oligomerize. Almost all of the fluorescent proteins discovered to date display at least a limited degree of quaternary structure, as exemplified by the weak tendency of native Aequorea victoria green fluorescent protein and its derivatives to dimerize when immobilized at high concentrations. This tendency is also verified by the strict tetramerization motif of the native yellow, orange, and red fluorescent proteins isolated in reef corals and anemones. Oligomerization can be a significant problem for many applications in cell biology, particularly in cases where the fluorescent protein is fused to a host protein that is targeted at a specific subcellular location. Once expressed, the formation of dimers and higher order oligomers induced by the fluorescent protein portion of the chimera can produce atypical localization, disrupt normal function, interfere with signaling cascades, or restrict the fusion product to aggregation within a specific organelle or the cytoplasm. This effect is particularly marked when the fluorescent protein is fused to partners which themselves participate in natural oligomer formation. Fusion products with proteins that form only weak dimers (in effect, most Aequorea victoria variants) may not exhibit aggregation or improper targeting, provided the localized concentration remains low. However, when weakly dimeric fluorescent proteins are targeted to specific cellular compartments, such as the plasma membrane, the localized protein concentration can, in some circumstances, become high enough to permit dimerization. This can be a particular concern when conducting intermolecular FRET experiments, which can yield complex data sets that are sometimes compromised by dimerization artifacts. On the other hand, the naturally occurring weak dimerization in Aequorea proteins can be, in some cases, utilized to increase the FRET signal in biosensors that otherwise would exhibit limited dynamic range.

Toxicity is an issue that occurs due to excessive concentrations of synthetic fluorophores and the over-expression or aggregation of poorly localized fluorescent proteins. Furthermore, the health and longevity of optimally labeled mammalian cells in microscope imaging chambers can also suffer from a number of other deleterious factors. Foremost among these is the light-induced damage (phototoxicity) that occurs upon repeated exposure of fluorescently labeled cells to illumination from lasers and high-intensity arc-discharge lamps. In their excited state, fluorescent molecules tend to react with molecular oxygen to produce free radicals that can damage subcellular components and compromise the entire cell. Fluorescent proteins, due to the fact that their fluorophores are buried deep within a protective polypeptide envelope, are generally not phototoxic to cells. In designing FRET experiments, fluorescent protein combinations that exhibit the longest possible excitation wavelengths should be chosen in order to minimize damage to cells by short wavelength illumination, especially in long-term imaging experiments. Thus, rather than creating fusion products and biosensors with blue or cyan fluorescent proteins (excited by ultraviolet and blue illumination, respectively), variants that emit in the yellow, orange, and red regions of the spectrum would be far more ideal.

Investigators should take care to perform the necessary control experiments when using new fluorescent protein biosensors and cell lines to ensure that cytotoxicity and phototoxicity artifacts do not obscure FRET results or other important biological phenomena. In some cases, lipophilic reagents induce deleterious effects that may be confused with fluorescent protein toxicity during imaging in cell lines following transient transfections. Oligomeric fluorescent proteins (discussed above) from reef corals have a far greater tendency to form aggregates (combined with poor subcellular localization) than do the monomeric jellyfish proteins, but improperly folded fusion products can occur with any variant. Recently, a fluorescent protein capable of generating reactive oxygen species (ROS) upon illumination with green light has been reported as an effective agent for inactivation of specific proteins by chromophore-assisted light inactivation (CALI). Appropriately named KillerRed, this genetically encoded photosensitizer is capable of killing both bacteria and eukaryotic cells upon illumination in the microscope. Previous studies on EGFP phototoxicity indicate that even through the chromophore is capable of generating singlet oxygen, the fluorescent protein is relatively inefficient as a photosensitizer. However, prolonged illumination of cells expressing EGFP and its variants can result in physiological alterations and eventual cell death, a definite signal of the potential for phototoxicity in long-term imaging experiments.

In live-cell experiments, fluorescent proteins are highly advantageous for extended imaging due to their reduced rate of photobleaching when compared to synthetic fluorophores. Although there is a high degree of uncorrelated variability between fluorescent proteins in terms of photostability, most variants are useful for short-term imaging (from 1 to 25 captures), while several of the more photostable proteins can be employed in time-lapse sequences that span periods of 24 hours or longer (in which hundreds to thousands of images are gathered). The long term stability of any particular protein, however, must be investigated for every illumination scenario (widefield, confocal, multiphoton, swept-field, etc.) because differences in photostability are often observed with the same protein when illumination is produced by an arc-discharge lamp versus a laser system. Thus, in terms of photostability, the selection of fluorescent proteins is dictated by numerous parameters, including the illumination conditions, the expression system, and the effectiveness of the imaging setup.

Potential Fluorescent Protein FRET Partners

Over the past few years, a wide variety of new fluorescent protein variants have been developed and refined to feature emission profiles spanning a 200-nanometer range (from approximately 450 nanometers to 650 nanometers), thus filling many gaps to provide potentially useful FRET partners in every color class. Recent advances in developing proteins in the blue (440 nanometers to 470 nanometers) and cyan (470 nanometers to 500 nanometers) spectral regions have yielded several new probes that may be of use for imaging and FRET investigations. Three protein engineering groups have reported improved blue Aequorea fluorescent protein variants that feature significantly higher brightness and photostability compared to EBFP. Named Azurite, SBFP2 (strongly enhanced blue FP), and EBFP2 (see Table 1), these proteins offer the first real hope for successful long-term imaging of live cells in the blue spectral region, and all have significant applicability to combination with EGFP and derivatives in FRET biosensors. The brightest and most photostable of the new blue reporters, EBFP2, exhibits typical GFP-like behavior in fusions and has been demonstrated to be an excellent FRET donor for proteins in the green spectral class. All of the blue fluorescent proteins can be readily imaged in a fluorescence microscope using standard DAPI filter sets or proprietary BFP sets available from aftermarket manufacturers.

Fluorescent proteins in the cyan spectral region have been widely applied as FRET donors when paired with yellow-emitting proteins, and were dominated by variants of the original Aequorea ECFP until the introduction of a monomeric teal-colored reporter, known as mTFP1. Teal fluorescent protein exhibits higher brightness and acid stability compared to Aequorea CFPs, and is far more photostable. The high emission quantum yield of mTFP1 (see Table 1) provides an excellent alternative to the cyan derivatives, mECFP and mCerulean, as a FRET donor when combined with either yellow or orange fluorescent proteins. Additional investigations have produced useful proteins in the cyan spectral class. Among the improved cyan fluorescent proteins that have recently been introduced, CyPet and the enhanced cyan variant termed Cerulean show the most promise as candidates for fusion tags, FRET biosensors, and multicolor imaging. Cerulean is at least 2-fold brighter than ECFP and has been demonstrated to significantly increase contrast as well as the signal-to-noise ratio when coupled with yellow-emitting fluorescent proteins, such as Venus (see below), in FRET investigations. The CFP variant named CyPet (from the acronym: Cyan fluorescent Protein for energy transfer) was derived through a unique strategy utilizing fluorescence-activated cell sorting (FACS) to optimize the cyan and yellow pairing for FRET. CyPet is about half as bright as EGFP and two-thirds as bright as Cerulean, but expresses relatively poorly at 37 degrees Celsius. However, CyPet has a more blue-shifted and narrower fluorescence emission peak than CFP, which greatly increases its potential for multicolor imaging.

The introduction of beneficial folding mutations into monomeric variants of ECFP has resulted in the production of new variants featuring enhanced brightness, folding efficient, solubility, and FRET performance. Termed super CFPs (SCFPs), the new reporters are significantly brighter than the parent protein when expressed in bacteria and almost two-fold brighter in mammalian cells. These high-performance probes should be useful both for routine fusion tags and in creating new CFP-YFP FRET biosensors exhibiting high dynamic range. Another new monomeric cyan reporter, TagCFP, was derived from a GFP-like protein from the jellyfish Aequorea macrodactyla. Specific details about the protein are unavailable in the literature, but it is commercially available as mammalian cloning vectors and fusions from Evrogen. TagCFP is reported to be brighter than ECFP and Cerulean, but of similar acid resistance. Another cyan-emitting protein, Midoriishi-Cyan (abbreviated MiCy) was originally designed as the donor in a new FRET combination with the monomeric Kusabira Orange (mKO; see Table 1) to generate a biosensor with high spectral overlap (Förster distance of 5.3). This protein features the longest absorption and emission wavelength profiles (472 and 495 nanometers, respectively) reported for any probe in the cyan spectral region. The high molar extinction coefficient and quantum yield exhibited by MiCy render the protein of equal brightness to Cerulean.

Properties of Selected Fluorescent Protein FRET Pairs
Protein Pair Donor Excitation
Acceptor Emission
Donor Quantum
Acceptor Molar
Förster Distance
EBFP2-mEGFP 383 507 0.56 57,500 4.8 1:2
ECFP-EYFP 440 527 0.40 83,400 4.9 1:4
Cerulean-Venus 440 528 0.62 92,200 5.4 1:2
MiCy-mKO 472 559 0.90 51,600 5.3 1:2
TFP1-mVenus 492 528 0.85 92,200 5.1 1:1
CyPet-YPet 477 530 0.51 104,000 5.1 1:4.5
EGFP-mCherry 507 610 0.60 72,000 5.1 2.5:1
Venus-mCherry 528 610 0.57 72,000 5.7 3:1
Venus-tdTomato 528 581 0.57 138,000 5.9 1:2
Venus-mPlum 528 649 0.57 41,000 5.2 13:1
Table 1

The best current choice for live-cell imaging of FRET reporters in the green color class (500 nanometers to 525 nanometers) is the GFP derivative Emerald, which has properties similar to its EGFP parent. Emerald contains the F64L and S65T mutations featured in EGFP, but the variant also has four additional point mutations that improve folding, expression at 37 degrees Celsius, and brightness. Recently, a new addition to the green spectral region has been coined superfolder GFP, which is brighter and more acid resistant than either EGFP or Emerald and has similar photostability. Therefore, the superfolder variant should be an excellent candidate for fusions with mammalian proteins and the construction of FRET biosensors, especially those that demonstrate folding problems with standard GFP derivatives. Another brightly fluorescent reporter, which may be a good FRET candidate, is termed Azami Green and has been isolated from the stony coral Galaxeidae and demonstrated to mature rapidly during expression in mammalian cell lines. In addition, two bright, monomeric GFP reporters obtained through site-directed and random mutagenesis in combination with library screening in cyan proteins have been reported. Derived from the Clavularia coral genus, mWasabi is a potential alternative green-emitting FRET partner for blue fluorescent proteins due to negligible absorbance at 400 nanometers and lower where blue variants are often excited. The new green reporter is commercially available (Allele Biotechnology) and should be particularly useful for two-color imaging in conjunction with long Stokes shift proteins (such as T-Sapphire) as well as a localization tag in fusions with targeting proteins. A derivative of TagCFP, named TagGFP, is a bright and monomeric green variant having an absorption maximum at 482 nanometers and emission at 505 nanometers. TagGFP, which is only slightly brighter than EGFP, is available as cloning vectors and fusion tags from Evrogen, but has not been thoroughly characterized in literature reports.

Yellow fluorescent proteins (525 nanometers to 555 nanometers) are among the most versatile genetically-encoded probes yet developed and should provide candidates acting as both donors and acceptors in FRET pairings. The variants known as Citrine and Venus are currently the most useful proteins in this spectral class (see Table 1), but neither is commercially available. Another variant, named after the birthstone Topaz, is available from Invitrogen and has been of service in fusion tag localization, intracellular signaling, and FRET investigations. A new member of the Evrogen “Tag” commercial series of localization reporter proteins, TagYFP, is a monomeric jellyfish (Aequorea macrodactyla) derivative that is slightly less bright than EYFP, but an order of magnitude more photostable. Similar to its partners, TagYFP (emission peak at 524 nanometers) has not been characterized in the literature, but can be purchased as mammalian cloning vectors or fusion tags.

During the same fluorescence-activated cell sorting investigation that led to the generation of CyPet (discussed above), the evolutionary optimized complementary FRET acceptor, termed YPet, was also obtained. Named after its proficiency in FRET (YFP for energy transfer), YPet is the brightest yellow variant yet developed and demonstrates reasonable photostability. The resistance to acidic environments afforded by YPet is superior to Venus and other YFP derivatives, which will enhance the utility of this probe in biosensor combinations targeted at acidic organelles. However, although the optimized CyPet-YPet combination should be the preferred starting point in the development of new FRET biosensors, there remains a serious doubt as to the origin of YPet's increased performance, which is likely due simply to enhanced dimerization with its co-evolved partner, CyPet. Likewise, the suitability of CyPet and YPet in fusion tags for localization experiments, bimolecular complementation analysis, and other applications has yet to be established. Both proteins exist in solution as weak dimers, but presumably can be converted to true monomers using the A206K mutation that has worked so well with other Aequorea variants (although this apparently destroys their advantages in FRET).

Orange fluorescent proteins, all of which have all been isolated from coral reef species, have the potential to be useful in a variety of FRET imaging scenarios. Perhaps the most versatile of these is monomeric Kusabira Orange, a protein originally derived as a tetramer from the mushroom coral Fungia concinna (known in Japanese as Kusabira-Ishi). A monomeric version of Kusabira Orange (abbreviated mKO) was created by introducing over 20 mutations through site-directed and random mutagenesis. The monomer (commercially available from MBL International) exhibits similar spectral properties to the tetramer and has a brightness value similar to EGFP, but is slightly more sensitive to acidic environments than its parent. The photostability of this reporter, however, is among the best of any protein in all of the spectral classes, making mKO an excellent choice for long-term imaging experiments. Furthermore, the emission spectral profile is sufficiently well separated from cyan fluorescent proteins to increase the FRET efficiency in biosensors incorporating mKO, and the probe is useful in multicolor investigations with a combination of cyan, green, yellow, and red probes.

Illustrated in Figure 10 are spectral profiles of ECFP (Figure 10(a)), EGFP (Figure 10(b)), EYFP (Figure 10(c)), and mOrange (Figure 10(d)), each acting as a FRET donor to mPlum, a far-red emitting fluorescent protein acceptor. As the emission spectral profiles of the donors shift to longer wavelengths (from cyan to orange), the spectral overlap (filled gray region) and calculated Förster distance (R(0)) increases correspondingly. Similarly, the excitation and emission crosstalk (red and blue hatched regions, respectively) also increases as the wavelength separation between the donor and acceptor emission peaks decreases. Note that ECFP and mPlum exhibit only a limited degree of overlap in the excitation spectra and virtually none in the emission spectra. In contrast, there is a high level of both excitation and emission crosstalk when mOrange is paired with mPlum. As the fluorescent protein color palette continues to expand, a wide spectrum of new FRET pairs should become readily available to investigators.

The mRFP1 derivative, mOrange, is slightly brighter than mKO, but has less than 10 percent the photostability, thus severely compromising its application for experiments that require repeated imaging. However, mOrange remains one of the brightest proteins in the orange spectral class and is still an excellent choice where intensity is more critical than long-term photostability. In addition, combined with the green-emitting T-Sapphire, mOrange is a suitable alternative to CFP-YFP proteins as a FRET pair to generate longer wavelength biosensors, and can be coupled with proteins in other spectral regions for multicolor investigations. An improved version of mOrange (named mOrange2) featuring dramatically increased photostability is now available. A bright new monomeric orange protein, named TagRFP has recently been introduced as a candidate for localization and FRET studies and may prove to be effective in a wide number of biosensor constructs. The brightest fluorescent protein in any spectral class is the tandem version of dimeric Tomato (dTomato), an orange derivative that was one of the original Fruit proteins. The Tomato protein contains the first and last seven amino acids from GFP on the N- and C- termini in an effort to increase the tolerance to fusion proteins and reduce potential artifacts in localization as well as enhance the possibility of its use in FRET biosensors. A tandem-dimer version (effectively a monomer) was created by fusing two copies, head-to-tail, of dTomato with a 23-amino acid linker. Due to the presence of twin chromophores, the resulting tdTomato is extremely bright and has exceptional photostability. The major drawback in the use of this protein is the larger size (twice that of a monomeric protein), which may interfere with fusion protein packing in some biopolymers.

The search for an ideal red-emitting fluorescent protein has long been the goal for live-cell and whole animal imaging using FRET biosensors and fusions, primarily due to the requirement for probes in this spectral region in multicolor imaging experiments as well as the fact that longer excitation wavelengths generate less phototoxicity and can probe deeper into biological tissues. To date, a wide spectrum of potentially useful red probes has been reported (emission at 590 nanometers to 650 nanometers), many of which still suffer from some degree of the obligatory quaternary structure bestowed by their species of origin. Unlike the jellyfish proteins, most of the native and genetically engineered variants of coral reef proteins mature very efficiently at 37 degrees Celsius. Extensive mutagenesis research efforts, including newly introduced methodology, have successfully been applied in the search for yellow, orange, red, and far-red florescent protein variants that further reduce the tendency of these potentially efficacious biological probes to self-associate while simultaneously pushing emission maxima toward longer wavelengths. The result has been improved monomeric proteins that feature increased extinction coefficients, quantum yields, and photostability, although no single variant has yet been optimized by all criteria.

The red mFruit proteins, mApple, mCherry and mStrawberry (emission peaks at 592, 610, and 596 nanometers, respectively), have brightness levels ranging from 50 percent to 110 percent of EGFP, but mApple and mCherry are far more photostable than mStrawberry and are the best probe choices to replace mRFP1 for long-term imaging experiments. Further extension of the mFruit protein spectral class through iterative somatic hypermutation has yielded two new fluorescent proteins with emission wavelength maxima of 625 and 649 nanometers, representing the first true far-red genetically engineered probes. The most potentially useful probe in this pair was named mPlum, which has a rather limited brightness value (10 percent of EGFP), but excellent photostability. This monomeric probe should be useful in combination with variants emitting in the cyan, green, yellow, and orange regions for multicolor imaging experiments and as a biosensor FRET partner with green and yellow proteins, such as Emerald and Citrine (see Figure 10). Several additional red fluorescent proteins showing varying degrees of promise have been isolated from the reef coral organisms. The application of site-specific and random mutagenesis to TurboRFP variants, followed by screening for mutations exhibiting far-red fluorescence, resulted in a dimeric protein named Katushka (emission maxima of 635 nanometers). Although only two-thirds as bright as EGFP, Katushka exhibits the highest brightness levels of any fluorescent protein in spectral window encompassing 650 to 800 nanometers, a region that is important for deep tissue imaging. Introduction of the four principal Katushka mutations into TagRFP generated a monomeric, far-red protein named mKate that has similar spectral characteristics. The photostability of mKate is reported to be exceptional and the protein displays brightness similar to that of mCherry, which makes it an excellent candidate for localization and FRET experiments in the far-red portion of the spectrum.

Despite significant advances in expanding the fluorescent color palette into the orange, red, and far-red regions of the spectrum, cyan and yellow Aequorea derivatives remain the most useful pairing scenario for developing useful biosensors. This unforeseen discrepancy occurs because most of the orange and red coral-derived proteins exhibit a relatively broad absorption spectral profile having a long excitation tail that extends into the violet and cyan regions, thus producing direct acceptor excitation. Another factor that might come into play is the relative maturation rates of the fused fluorescent protein partners. In most cases, variants derived from Aequorea proteins mature far more rapidly than those obtained from reef corals so it is possible that immature acceptors contribute to the poor sensitized emission exhibited by many of the coral-derived proteins. In addition, the broad adsorption spectra of the orange and red proteins, combined with the reduced quantum yields of the monomeric versions, likely render them difficult for use in FRET. Future success in fluorescent protein FRET experimental design will focus on, among other factors, matching the maturation rates of the paired proteins.


Although FRET experiments based on the ubiquitous fluorescent protein family offer tremendous potential to reveal molecular dynamics in living cellular systems, as yet there is not a perfect FRET pair. The optimized versions of CFP and YFP still provide the most effective pair for general use, although better combinations may loom over the horizon. Likewise, there is no perfect technique with which to measure FRET, although the approaches described above all have strengths that can be leveraged depending on the particular experimental situation under investigation. As more optimized fluorescent proteins become available, including bright red variants that might be appropriate as acceptors for GFP or YFP donors, FRET using fluorescent proteins should become even more useful for protein-protein interaction investigations in live cells. As discussed, the broad absorption spectra of the current palette of red fluorescent proteins, in addition to the lower quantum yields of the monomeric versions, make these candidates difficult to employ in FRET. However, the rapid pace of fluorescent protein improvements lends optimism that such proteins will be available in the near future and will help to further revolutionize this new approach to elucidating intracellular biochemical mechanisms.


Contributing Authors

Gert-Jan Kremers and David W. Piston - Department of Molecular Physiology and Biophysics, Vanderbilt University, 702 Light Hall, Nashville, Tennessee, 37232.

Michael W. Davidson - National High Magnetic Field Laboratory, 1800 East Paul Dirac Dr., The Florida State University, Tallahassee, Florida, 32310.