Imaging Fluorescence Resonance Energy Transfer between Green Fluorescent Protein Variants in Live Cells

I. INTRODUCTION
In recent years, the measurement of fluorescence resonance energy transfer (FRET) between variants of fluorescent proteins has emerged as a powerful tool for intracellular measurements of protein reactions (Heim and Tsien, 1996; Matz et al., 1999; Tsien, 1998; Wouters et al., 2001). For FRET assays in living cells, the green fluorescent protein (GFP), cyan fluorescent protein (CFP), and yellow fluorescent protein (YFP) variants are used most frequently. These GFP variants are intrinsically fluorescent and do not require any exogenous cofactors or substrates, rendering them particularly useful as genetically encoded fluorescent tags to participate as a donor or acceptor fluorophore in a FRET pair. FRET is a photophysical effect where energy is transferred from an excited donor to an acceptor fluorophore; it is a process that is mediated by a direct electromagnetic interaction and does not involve the emission and subsequent absorption of a photon (Clegg, 1996). The efficiency of the transfer depends on the spectral properties of the donor and acceptor and on their relative orientation and distance. Most importantly, the energy transfer efficiency has an inverse sixth order dependency on the distance between the two fluorophores, and therefore FRET only occurs at distances that are typically less than 10 nm. As a result, FRET can be used to specifically image molecular interactions or conformational changes as these events may bring donor and acceptor fluorophores within this distance range, providing they are attached to the same macromolecule or to the interacting molecules.

FRET cannot be measured directly, but its occurrence is reflected by changes in the fluorescence kinetics of both the donor and the acceptor molecules that can be measured by optical means. Due to the direct transfer of energy from the donor to the acceptor, the rate at which the donor returns to its ground state after excitation increases and therefore its fluorescence lifetime decreases. Since the quantum yield of a fluorophore is proportional to its lifetime, the steadystate intensity of the donor molecule also decreases if FRET occurs. However, the steady-state fluorescence of the acceptor increases due to the sensitized emission induced by the energy transfer. Here we outline two methods to detect FRET that are based on the measurement of these effects: (1) fluorescent lifetime imaging microscopy (FLIM) and (2) sensitized emission measurements.

The fluorescent lifetime of a fluorophore is a measure of the time that the fluorophore spends in the excited state and is independent of probe concentration and light path length (Bastiaens and Squire, 1999). FRET reduces the fluorescence lifetime of the donor molecule since it depopulates its excited state. Thus, the measurement of the fluorescence lifetime of the donor can be used as an indicator for FRET. In such an assay, the donor-acceptor pair is chosen such that the donor fluorescence can be detected specifically without detecting any acceptor fluorescence. Therefore absolute specificity of the acceptor probe is not required, and experiments can be carried out with an excess of acceptor. The reduction of the fluorescence lifetime due to FRET should be judged in comparison to the fluorescence lifetime of the donor in the absence of FRET. This can be achieved using an internal control where at the end of the experiment the acceptor is photobleached (Bastiaens and Jovin, 1998). This may require illumination of the acceptor for an extended period during which relocation of the proteins may occur. However, because the fluorescence lifetime of the donor is independent of probe concentration, this poses no problem because only the average value of the lifetime after photobleaching is required. FLIM can be implemented in two conceptually different ways (Bastiaens and Squire, 1999). In the first approach, the sample is illuminated with a short laser pulse, and the resulting fluorescence decay is sampled using a fast, gated detector. The measured curve is a direct reflection of the fluorescence decay from which the fluorescent lifetime can be derived. The second approach uses sinusoidally modulated laser light to illuminate the sample at a high frequency (typically ~80MHz). The resulting fluorescence is also sinusoidally modulated but phase shifted and demodulated compared to the excitation. This phase shift and demodulation depend on the fluorescent lifetime, which can therefore be derived if these quantities can be measured. This can be achieved with a sinusoidally modulated detector whose phase is shifted systematically with respect to the phase of the illumination. The result is a sinusoidally shaped curve from which the phase shift and demodulation, and thus the lifetime, can be derived. Although this approach appears more complex, it requires a less expensive and complicated laser setup and has proven to be a reliable approach in biological applications.

Sensitized emission measurements can be achieved using standard wide-field or confocal microscopes if the appropriate excitation sources are available and the fluorescence emissions of the donor and acceptor can be detected. Generally it is not possible to image the sensitized emission directly, as excitation of the donor cannot be achieved without also directly exciting the acceptor. In addition, bleed through from the donor fluorescence also contributes to the signal that is detected in the acceptor channel upon excitation of the donor. Thus, the signal that is measured upon excitation of the donor consists of three components: bleed through from donor emission, acceptor emission due to direct excitation of the acceptor, and sensitized emission from the acceptor due to energy transfer. Thus the measurement must be corrected for bleed through and direct excitation, which can be achieved using an additional specific measurement of the donor fluorescence upon donor excitation and a measurement of the acceptor fluorescence upon specific excitation of the acceptor. These corrections require scalar calibration factors that can be found using two reference samples that contain either donor or acceptor molecules alone.

The use of green fluorescent protein variants as partners in a FRET pair requires consideration of the relative expression levels of the two protein-GFP fusions in the light of the measurement method that is being used. Using FLIM implies that only the donor fluorophore is being used to measure FRET, which makes this method eminently useful in the case that the acceptor is expressed in a saturating excess over the donor. This is an advantage if the acceptor-tagged protein is used as a sensor to report on the activity or covalent state of the donor-tagged protein, as it is then ensured that any donor-tagged protein can bind to at least one sensor molecule. In sensitized emission measurements the use of excess amounts of acceptor may lead to problems, as the contribution of sensitized emission may then be low compared to the direct excitation of the acceptor. Thus, care should then be taken that the expression levels of acceptor and donor are comparable.

This article uses an example of an assay for the detection of the phosphorylation state of the epidermal growth factor receptor (EGFR). YFP was fused to a phosphotyrosine-binding domain (PTB-YFP) from Shc, which recognizes three high-affinity binding sites on the intracellular part of the EGFR upon phosphorylation (Zhou et al., 1996). We used the citrine variant of YFP that, compared to previous variants, has a reduced sensitivity to pH due to a much lower pKa (5.7), has twice the photostability, and exhibits much better expression at 37°C (Griesbeck et al., 2001). CFP was fused to EGFR (EGFR-CFP) and FRET between CFP and YFP was used to detect the binding of PTB to EGFR and thereby the phosphorylation of EGFR.

II. MATERIALS AND INSTRUMENTATION
A. Materials
Dulbecco's modified Eagle's medium
CO2-independent imaging medium, containing no components that are autofluorescent. Available commercially from Life Technologies or prepared using the standard formulation of Dulbecco's modified Eagle's medium by omitting pH indicator phenol red, penicillin, streptomycin, folic acid, and riboflavin.
Glass-bottomed tissue culture dishes (35-mm, MatTek Corporation)
Transfection reagent, e.g., Superfect, Qiagen; Lipofectamine, Invitrogen; FuGENE-4, Roche
Plasmids encoding the CFP- and YFP-tagged proteins of interest

B. Instrumentation
Fluorescence lifetime imaging microscope suitable for measuring the lifetime of CFP and equipped with a filter set suitable for specifically exciting and imaging YFP. For the experiments described here, we used a home-built wide-field system (Squire and Bastiaens, 1999). CFP excitation: Ar laser at 457.9nm, CFP filter set (dichroic beam splitter 467; DELTA, Lyngby, Denmark, emission filter HQ480/20; Chroma). YFP excitation: 100-W mercury lamp, YFP filter set (dichroic filter 530 long pass, excitation 510/25, emission filter HQ560/ 50).
Leica SP2 confocal microscope, equipped with Ar laser with 457.9-nm laser line for CFP excitation and a 514-nm line for YFP excitation
Wide-field fluorescence microscope with appropiate filter sets. For example (from Sorkin et al., 2000): YFP filter set (excitation 500/20nm, emission 535/ 30 nm), CFP filter set (excitation 436/10nm, emission 470/30nm), FRET filter set (excitation 436/10nm, emission 535/30 nm).
Computer with image processing software, such as IPLab (Scanalytics)

III. PROCEDURES
A. Cell Preparation
To prepare cells for FRET experiments with either FLIM or sensitized emission, the donor and acceptor plasmid DNA must be introduced into the cells. This is accomplished using standard transfection methods or, alternatively, using nuclear microinjection. The following steps are used to prepare adherent cells for imaging using transient transfection:
  1. Day 1: Seed the cells onto glass-bottomed dishes.
  2. Day 2: Cell transfection.
    1. Transfect a number of dishes with both plasmids encoding CFP- and YFP-tagged protein, according to the protocol provide by the supplier of the transfection agent.
    2. For sensitized emission measurements only: transfect a number of dishes with plasmids encoding the CFP-tagged protein.
    3. For sensitized emission measurements only: transfect a number of dishes with plasmids encoding the YFP-tagged protein.


  3. Incubate the transfected cells under the appropriate conditions for 15-24h (e.g., 37°C, 5% CO2) to allow the cells to express the proteins.
  4. Day 3: Before imaging, proceed optionally with any necessary cell-handling protocols such as starvation in medium without growth factors.
  5. Immediately before imaging replace the culture medium with CO2-independent imaging medium.


B. Fluorescence Lifetime Microscopy
Fluorescence lifetime microscopy is well suited for FRET experiments where the acceptor is present in excess, as only the fluorescence of the donor needs to be measured. Figure 1 shows an example where FRET between EGFR-CFP and PTB-YFP was measured with FLIM before and after stimulation with epidermal growth factor (EGF). An experiment on living cells requires following the response of the cells to a stimulus in time. At selected time points, a FLIM data set is acquired that contains fluorescence intensity and lifetime information of the donor. Additionally, a fluorescence intensity image of the distribution of the acceptor is acquired. Finally, a control measurement is performed to obtain the lifetime of the donor in the absence of FRET. A detailed description of the FLIM data acquisition itself falls outside the scope of this article. Acquisition of a single data set that captures the spatially resolved lifetimes at a given time is therefore assumed to be a single step in our protocol. More details can be found in the relevant literature (Clegg and Schneider, 1996; Gadella et al., 1993; Squire and Bastiaens, 1999). An experiment involves the following steps.
  1. Acquire a FLIM data set using the 457.9-nm line to excite the CFP donor.
  2. Acquire an image of the YFP fluorescence using the 100-W mercury lamp and the YFP filter set.
  3. Optionally, repeat steps 1 and 2 for several time points to require a time-lapse series.
  4. Illuminate the acceptor extensively with the YFP filter set until the fluorescence from YFP is abolished completely.
  5. Acquire a final FLIM data set to find the lifetime of the donor in the absence of acceptor.


FIGURE 1 Fluorescence lifetime imaging microscopy of CFP-tagged epidermal growth factor receptor 
(EGFR-CFP) cotransfected with YFP-tagged phosphotyrosine-binding domain (PTB-YFP) stimulated with epidermal growth factor (EGF). Measurements are made before and after stimulation, and finally after photobleaching of the acceptor. The lifetime of the donor in the absence of FRET is calculated from the average lifetime after photobleaching and is used to calculate the energy transfer efficiencies for the other measurements as described in the text. From left to right: CFP fluorescence (donor), YFP fluorescence (acceptor), fluorescence lifetime, and apparent energy transfer efficiency. (a) Results before stimulation with EGF. (b) Results after stimulation with EGF for 18 min. (c) Results after destruction of the acceptor fluorophores by photobleaching. The average apparent lifetime of the donor after photobleaching of the acceptor was 1.9ns.
FIGURE 1 Fluorescence lifetime imaging microscopy of CFP-tagged epidermal growth factor receptor
(EGFR-CFP) cotransfected with YFP-tagged phosphotyrosine-binding domain (PTB-YFP) stimulated with epidermal growth factor (EGF). Measurements are made before and after stimulation, and finally after photobleaching of the acceptor. The lifetime of the donor in the absence of FRET is calculated from the average lifetime after photobleaching and is used to calculate the energy transfer efficiencies for the other measurements as described in the text. From left to right: CFP fluorescence (donor), YFP fluorescence (acceptor), fluorescence lifetime, and apparent energy transfer efficiency. (a) Results before stimulation with EGF. (b) Results after stimulation with EGF for 18 min. (c) Results after destruction of the acceptor fluorophores by photobleaching. The average apparent lifetime of the donor after photobleaching of the acceptor was 1.9ns.

C. Fluorescence Lifetime Imaging Microscopy Data Analysis

Analysis of fluorescence lifetime imaging microscopy data is an extended subject that is described elsewhere (Gadella et al., 1994; Verveer et al., 2001). We will therefore assume that the FLIM data analysis software is provided as a part of the instrumental setup and calculates a lifetime value at each pixel position from the measured FLIM data. The result found at a given pixel location is a weighted summation of the fluorescence lifetimes of the individual molecular components present, and if a significant proportion of those components have a short lifetime due to FRET, this is reflected in a decrease in the measured lifetime. This measured value should be compared to the lifetime of the donor fluorophore in the absence of FRET, which can be found from the sample itself by photobleaching the acceptor. Given the measured lifetime at each pixel τDA(i), and the donor lifetime τD, an apparent energy transfer efficiency can be calculated in each pixel:
E(i) = 1- τDA(i) .
τD

The analysis consist of the following steps.
  1. Calculated the fluorescence lifetime image for the FLIM data set that was obtained after acceptor photobleaching. Calculate the average value to obtain the lifetime of the donor τD in the absence of acceptor.
  2. Calculate the fluorescent lifetime images for all the other measured FLIM data sets.
  3. Divide each pixel of the fluorescent lifetime images by τD.
  4. Multiply the resulting pixel values by-1 and add 1 to obtain E(i).


D. Sensitized Emission Measurements
Sensitized emission measurements are well suited for cases where donor and acceptor are expressed at similar levels. Detecting FRET by sensitized emission requires correction for fluorescence bleed through and unmixing of direct and sensitized acceptor emission (Gordon et al., 1998; Nagy et al., 1998). For illustration purposes we will assume the use of a Leica SP2 microscope, but the discussion generalizes to any microscope equipped with the proper excitation and emission filters. Figure 2 shows an example where FRET between EGFR-CFP and PTB-YFP was measured using sensitized emission before and after stimulation with EGF. This article limits itself to donor and acceptor pairs with the following spectral properties: (1) upon donor excitation the donor fluorescence can be detected specifically; and (2) the acceptor can be excited specifically (i.e., no donor fluorescence upon acceptor excitation). The CFP-YFP FRET pair that we use in this article meets these requirements. The CFP was excited at 457.9nm, which yields also significant direct excitation of YFP, and the YFP was excited specifically at 514nm. The Leica SP2 microscope allows flexible settings for measurements of the emission with up to four detectors simultaneously. We simultaneously detected fluorescence at 460-495nm (CFP channel) and at 520-570nm (YFP channel). In the CFP channel, no significant YFP emission is detected, but in the YFP channel, a significant contribution of CFP fluorescence is present upon excitation of CFP. Measurement of the CFP channel upon CFP excitation allows for estimation of the amount of bleed through if it is known how much CFP emission is detected in the YFP channel. This can be calibrated by excitation of CFP in a reference sample transfected with CFPtagged protein only. Dividing the total signal in the YFP channel by the total signal in the CFP channel gives the scalar bleed-through correction factor. To calculate the bleed-through image in a sample with both CFP and YFP, each pixel in the image from the CFP channel is multiplied with the scalar bleed-through correction factor obtained from a reference sample. In a similar fashion, the amount of direct excitation is calibrated from a reference sample that expresses only the YFP-tagged protein. The total signal detected in the YFP signal upon CFP excitation is divided by the total signal upon YFP excitation to obtain the scalar direct excitation correction factor. To calculate the direct excitation image in a sample with both CFP and YFP, each pixel in the image measured in the YFP channel upon specific YFP excitation is multiplied with the scalar direct excitation correction factor. An image of the sensitized emission is then calculated by pixel-wise subtraction of the bleed-through and direct excitation images from the image that is measured in the YFP channel upon excitation of CFP.

FIGURE 2 Sensitized emission measurement of FRET between CFP-tagged epidermal growth factor receptor (EGFR-CFP) cotransfected with YFP-tagged phosphotyrosine-binding domain (PTB-YFP) stimulated with epidermal growth factor (EGF). Measurements were made before stimulation and after 90s of stimulation with EGF. Different images and the calculated result before (left) after (right) stimulation. (a) Images measured in the acceptor channel upon excitation of the donor. (b) Images measured in the donor channel upon donor excitation. (c) Images measured in the acceptor channel upon acceptor excitation. (d) Sensitized emission calculated by subtracting the images in b and c from the images in a after scaling with the appropriate scalar correction factors for bleed through and direct excitation.
FIGURE 2 Sensitized emission measurement of FRET between CFP-tagged epidermal growth factor receptor
(EGFR-CFP) cotransfected with YFP-tagged phosphotyrosine-binding domain (PTB-YFP) stimulated with
epidermal growth factor (EGF). Measurements were made before stimulation and after 90s of stimulation with EGF. Different images and the calculated result before (left) after (right) stimulation. (a) Images measured in the acceptor channel upon excitation of the donor. (b) Images measured in the donor channel upon donor excitation. (c) Images measured in the acceptor channel upon acceptor excitation. (d) Sensitized emission calculated by subtracting the images in b and c from the images in a after scaling with the appropriate scalar correction factors for bleed through and direct excitation.

The reference measurements should be taken at the same instrumental settings (detector gains and offsets, illumination power) as used for the sensitized emission measurements itself. Results of the reference measurements for a given setting should be reproducible and it is advisable to check this by repeating the reference measurement procedure a few times. Reference measurements are done using the following steps.
  1. A sample with only donor molecules is used to calibrate the contribution of bleed through.
    1. Excite the donor and acquire an image of the donor channel, yielding image DDD(i), where i is the pixel index.
    2. Excite the donor and acquire an image of the acceptor channel, yielding image DDA(i).


    Care should be taken that the sample is not moved between acquisitions. In live cells the images should be taken in quick succession to avoid artifacts due to relocation of the donor-tagged protein or movement or deformation of the cell. In some instruments (e.g., the Leica SP2 used by us) it is possible to acquire the two images simultaneously, which is preferable.
  2. The contributions of direct excitation are calibrated using a sample that has only acceptor molecules.
    1. Excite the donor and acquire an image of the acceptor channel, yielding image ADA(i).
    2. Excite the acceptor and acquire an image of the acceptor channel, yielding image AAA(i).


    These images must be taken separately, but the same concerns about sample movement, as described in point 1, are valid. The Leica SP2 microscope allows switching between donor and acceptor excitation sources on a line-by-line basis or on a frame-by-frame basis. The former is preferable, as less movement will occur between acquisitions of successive lines than between acquisitions of successive images.


For sensitized emission measurements with samples that have both donor and acceptor molecules, the following steps are taken.
  1. Excite the donor and acquire an image of the donor, yielding image FDD(i).
  2. Excite the donor and acquire an image of the acceptor, yielding image FDA(i).
  3. Excite the acceptor and acquire an image of the acceptor channel, yielding image FAA(i).


Again, care should be taken that the sample is not moved between image acquisition or that significant relocation of the proteins of interest or the cell as a whole takes place. Using a Leica SP2 microscope, steps 1 and 2 can be done simultaneously and switching between illumination sources should be done on a line-by-line basis.

When possible, the detector used should have a high dynamic range for all measurements. For instance, modern confocal microscopes allow recording data with 12-bit precision.

E. Sensitized Emission Data Analysis
The data analysis steps for sensitized emission measurements are fairly simple and can be performed with many commercially available image processing packages. The following steps are used for calculating the sensitized emission image from the measured images.
  1. Background correction, applied to all images:
    1. Select a small region of interest in the background of the image.
    2. Calculate the mean intensity with the region of interest to obtain the background value.
    3. Subtract the background value from each pixel the image.


  2. Calculate the correction factor for the bleed through from the reference measurements by taking the ratio between the signal in the acceptor channel to the donor channel.
    1. Create a mask from the pixels with high signal within image DDD(i), by thresholding, or interactive selection. It is important to exclude saturated pixels from the mask, as the calculated result from such pixels will be incorrect.
    2. Calculate the sum of the pixel intensities in image DDD(i), within the mask:


    3. Calculate the sum of the pixel intensities in image DDA(i), within the mask:


    4. Calculate the scalar bleed-through correction factor

      Cbleed-through = sum(DDA) .
      sum(DDD)



  3. Calculate the correction factor for direct excitation from the reference measurements by taking the ratio of the signals in the acceptor channel upon exciation of the donor and acceptor, respectively.
    1. Create a mask from the pixels with high signal within image AAA(i) by thresholding or interactive selection. Exclude saturated pixels.
    2. Calculate the sum of the pixel intensities in image ADA(i), within the mask:


    3. Calculate the sum of the pixel intensities in image AAA(i) within the mask:


    4. Calculate the scalar direct excitation correction factor:

      Cdirect-excitation = sum(ADA) .
      sum(AAA)



  4. Calculate the bleed-through correction in each pixel by multiplying the image in the donor channel upon donor excitation with the correction factor obtained in step 2:

    BT(i) = Cbleed-through · FDD(i)

  5. Calculate the correction for direct excitation of the acceptor in each pixel by multiplying the image in the acceptor channel upon acceptor excitation with the correction factor obtained in step 3:

    DE(i) = Cdirect-excitation· FAA(i)

  6. Calculate the sensitized emission image by subtracting the corrections for bleed-through and direct excitation calculated in steps 4 and 5 from the image that was measured in the acceptor channel upon donor excitation:

    S(i) = FDA (i) - BT(i) - DE(i).

    This image represents the fluorescent light due to energy transfer that is directly proportional to the concentration of donor-acceptor complexes.
  7. Optional: Calculate an apparent energy transfer efficiency using

    EA = S(i) .
    FAA (i)

    This equation normalize the sensitized emission with the emission of the acceptor only, and therefore the result is proportional to the fraction of interacting molecules with respect to the total amount of acceptortagged molecules (Wouters et al., 2001). It is also possible to normalize the sensitized emission with the donor fluorescence; however, because the donor emission is quenched by FRET, this is not proportional to the relative fraction of donor-tagged molecules that is in an interacting pair.


References
Bastiaens, P. I. H., and Jovin, T. M. (1998). Fluorescence resonance energy transfer microscopy. In "Cell Biology: A Laboratory Handbook" (J. E. Celis, ed.), pp. 136-146. Academic Press, New York.

Bastiaens, P. I. H., and Squire, A. (1999). Fluorescence lifetime imaging microscopy: Spatial resolution of biochemical processes in the cell. Trends Cell Biol. 9, 48-52.

Clegg, R. M. (1996). Fluorescence resonance energy transfer. In "Fluorescence Imaging Spectroscopy and Microscopy" (X. F. Wang and B. Herman, eds.), pp. 179-252. Wiley, New York.

Clegg, R. M., and Schneider, P. C. (1996). Fluorescence time-resolved imaging microscopy: A general description of lifetime-resolved imaging measurements. In "Fluorescence Microscopy and Fluorescence Probes" (J. Slavik, ed.), pp. 15-33. Plenum Press, New York.

Gadella, T. W. J., Jr., Clegg, R. M., and Jovin, T. M. (1994). Fluorescence lifetime imaging microscopy: Pixel-by-pixel analysis of phase-modulation data. Bioimaging 2, 139-159.

Gadella, T. W. J., Jr., Jovin, T. M., and Clegg, R. M. (1993). Fluorescence lifetime imaging microscopy (FLIM). Spatial resolution of microstructures on the nanosecond time-scale. Biophys. Chem. 48, 221-239.

Gordon, G. W., Berry, G., Liang, X. H., Levine, B., and Herman, B. (1998). Quantitative fluorescence energy transfer measurements using fluorescence microscopy. Biophys. J. 74, 2702-2713.

Griesbeck, O., Baird, G. S., Campbell, R. E., Zacharias, D. A., and Tsien, R. Y. (2001). Reducing the environmental sensitivity of yellow fluorescent protein: Mechanism and applications. J. Biol. Chem. 276, 29188-29194.

Heim, R., and Tsien, R. Y. (1996). Engineering green fluorescent protein for improved brightness, longer wavelengths and fluorescence resonance energy transfer. Curr. Biol. 6, 178-182.

Matz, M. V., Fradkov, A. F., Labas, Y. A., Savitsky, A. P., Zaraisky, A. G., Markelov, M. L., and Lukyanov, S. A. (1999). Fluorescent proteins from nonbioluminescent Anthozoa species. Nature Biotechnol. 17, 969-973.

Nagy, P., Vfimosi, G., Bodnfir, A., Locket, S. J., and Sz6116si, J. (1998). Intensity-based energy transfer measurements in digital imaging microscopy. Eur. Biophys. J. 27, 377-389.

Sorkin, A., McClure, M., Huang, F., and Carter, R. (2000). Interaction of EGF receptor and Grb2 in living cells visualized by fluorescence resonance energy transfer (FRET) microscopy. Curr. Biol. 10, 1395-1398.

Squire, A., and Bastiaens, P. I. H. (1999). Three dimensional image restoration in fluorescence lifetime imaging microscopy. J. Microsc. 193, 36-49.

Tsien, R. Y. (1998). The green fluorescent protein. Annu. Rev. Biochem. 67, 509-544.

Verveer, P. J., Squire, A., and Bastiaens, P. I. H. (2001). Frequency-domain fluorescence lifetime imaging microscopy: A window on the biochemical landscape of the cell. In "Methods in Cellular Imaging" (A. Periasamy; ed.), pp. 273-292. Oxford Univ. Press, Oxford.

Wouters, F. S., Verveer, P. J., and Bastiaens, P. I. H. (2001). Imaging biochemistry inside cells. Trends Cell Biol. 11, 203-211.

Zhou, M. M., Harlan, J. E., Wade, J. S., Crosby, S., Ravichandran, K. S., Burakoff, S. J., and Fesik, S. W. (1996). Binding affinities of tyrosine-phosphorylated peptides to the COOH-terminal SH2 and NH2-terminal phosphotyrosine binding domains of Shc. J. Biol. Chem. 271, 31119-31123.