1.

Introduction

Resonant transfer of optical excitation energy via dipole–dipole interaction was described by Förster about $60\text{years}$ ago.¹ Although this kind of energy transfer is well known in nature (e.g., in plants where optically excited antenna molecules transfer their energy to the reaction centers²), it was not used in cell biology and biotechnology until the 1980s.^{3, 4, 5} After successful cloning of the gene encoding for green fluorescent protein (naturally produced by the jellyfish Aequorea Victoria)⁶ and its variants emitting light in the blue, yellow, or red spectral ranges, Förster resonance energy transfer (FRET) from an optically excited donor to an acceptor molecule has become a well-established and broadly used analytical technique. This technique has also been applied for the detection of programmed cell death (apoptosis) by fusing cyan fluorescent protein (CFP) to yellow fluorescent protein (YFP) with a caspase-3 sensitive peptide Asp-Glu-Val-Asp (DEVD) used as a linker.^{7, 8} Upon induction of apoptosis, this linker was cleaved by caspase-3 activity, and FRET was interrupted. So far, FRET has been deduced from stationary measurements of donor (CFP) and acceptor (YFP) fluorescence, and a decrease of its efficiency up to a factor of five has been calculated.⁹ In view of further enhancement of FRET sensitivity as well as suppression of background fluorescence, enhanced CFP (ECFP) was anchored to the plasma membrane of HeLa cervical carcinoma cells, as depicted in Fig. 1, and membrane-associated fluorescence was excited selectively by an evanescent electromagnetic field.¹⁰ Energy transfer from donor (ECFP) to acceptor (enhanced YFP, EYFP) molecules was deduced from the ratio of acceptor/donor fluorescence as well as from the fluorescence lifetime of the donor, which, according to the equation

Fig. 1

Membrane-associated test system for apoptosis based on FRET and its interruption.

Thus, $k_{\mathrm{ET}}$ has been determined from measurements of individual cells, where apoptosis could be visualized by cell morphology. In contrast, larger cell collectives without any visual control were used in the present paper, i.e., mixed populations of apoptotic and nonapoptotic cells were examined, and apoptosis (of part of the cells) was deduced from changes in fluorescence lifetimes and intensities of the donor ECFP. In view of potential applications in high-content screening it was a main purpose of this paper to excite a larger number of individual samples simultaneously upon total internal reflection (TIR) of an incident laser beam with a penetration depth of the evanescent electromagnetic field around $100–150\mathrm{nm}$ . Therefore, an existing TIR reader system with multiple TIRs up to 96 individual wells of a microtiter plate¹¹ was modified for FRET measurements.

2.

Materials and Methods

2.3.

TIR Reader

For excitation of the membrane-associated donor ECFP or for direct excitation of the acceptor EYFP, we used pulsed laser diodes (pulse duration: $50–70\mathrm{ps}$ , repetition rate: up to $40\mathrm{MHz}$ , LDH-P-C-400 or LDH-P-C-470 with driver PDL 800-B, PicoQuant GmbH, Berlin, Germany) emitting light at $391\mathrm{nm}$ (average power: $65\mu \mathrm{W}$ ) or $470\mathrm{nm}$ (average power: $100\mu \mathrm{W}$ ), respectively. Collimated laser beams were coupled into a conventional microscope glass slide of $1\mathrm{mm}$ thickness (with superimposed cavities for individual samples) via polarization, maintaining single-mode fibers (kineFlex, Point Source, Southampton, UK) and a glass rod of rectangular shape, as depicted in Fig. 2 . In all cases, the electric-field vector was polarized perpendicular to the plane of incidence. Multiple TIRs occurred within the glass slide if the angle of incidence $\Theta$ was above the critical angle ${\Theta }_{\mathrm{c}}=63.9\mathrm{deg}$ (resulting from the refractive indices $n_1=1.525$ for the glass bottom and $n_2=1.37$ for the cells). In the present setup, an angle $\Theta =66\mathrm{deg}$ was chosen, resulting in a distance $s=4.5\mathrm{mm}$ between individual samples and a penetration depth $d=120–150\mathrm{nm}$ of the evanescent electromagnetic wave within the cells (depending on the excitation wavelength) according to

Eq. 2

$d=(\lambda ∕4\pi ){(n_1^2{\sin }^2\Theta -n_2^2)}^{-1∕2}.$

Due to the Gaussian-shaped laser-beam profile of $500\mu \mathrm{m}$ diam, each illumination spot on up to 12 samples (of about 1000 cells each) was of elliptical shape with an area around $0.5{\mathrm{mm}}^2$ . At the end of the glass slide, laser light was coupled with a second glass rod to avoid uncontrolled reflections.

Fig. 2

TIR fluorescence-lifetime reader ( $\mathrm{TCSPC}=\text{Time}$ -correlated single-photon counting).

Fluorescence intensities of all samples were detected simultaneously by an integrating digital charge-coupled device (CCD) camera (ProgRes C10, Jenoptik GmbH, Germany) equipped with a wide-angular objective lens (Cinegon $1.4∕12$ —0515, Schneider GmbH, Germany). ECFP fluorescence (excited at $391\mathrm{nm}$ ) was measured with a bandpass filter at $475\pm 20\mathrm{nm}$ , whereas EYFP fluorescence (excited directly at $470\mathrm{nm}$ ) was detected with a long pass filter at $\lambda ⩾515\mathrm{nm}$ . Exposure times were $10\mathrm{s}$ at $470\mathrm{nm}$ and $20\mathrm{s}$ at $391\mathrm{nm}$ . Measurements of EYFP fluorescence at an excitation wavelength of $391\mathrm{nm}$ (ECFP excitation with subsequent energy transfer $\mathrm{ECFP}\to \mathrm{EYFP}$ ) were omitted, because in this case a spectral overlap by ECFP was unavoidable. After measurement of fluorescence intensities¹¹ prior to $\left(I_{\mathrm{p}}\right)$ and after $\left(I_{\mathrm{a}}\right)$ incubation with staurosporine, the ratio $I_{\mathrm{a}}∕I_{\mathrm{p}}$ of all samples was calculated, and median $\text{values}\pm \text{median}$ absolute deviations (MADs) of this ratio were determined for 20–30 individual measurements of HeLa-MemECFP-DEVD-EYFP as well as HeLa-MemECFP-DEVG-EYFP cells. Control measurements were carried out with cells prior to and after a time interval of $4\mathrm{h}$ but without application of staurosporine. Because of possible errors due to light absorption in the glass slide, variations in the number of cells per sample or optical aberrations within the detection path affected $I_{\mathrm{a}}$ and $I_{\mathrm{p}}$ in the same way, these errors were eliminated by calculation of the ratio $I_{\mathrm{a}}∕I_{\mathrm{p}}$ . Therefore, no further correction was needed.

In addition to fluorescence intensities, we determined fluorescence lifetimes of ECFP using the same optical setup as reported above with the following changes: the repetition rate of the $391\mathrm{nm}$ laser diode was reduced to $20\mathrm{MHz}$ , and the CCD camera was replaced by a microscope objective lens $(20\times ∕0.50)$ , a photomultiplier tube (H5783-01, Hamamatsu Photonics Deutschland GmbH, Germany), and a time-correlated single-photon counting device (TimeHarp 200, PicoQuant GmbH, Berlin, Germany), as depicted in Fig. 2. Because in this case simultaneous measurements of all samples were not possible, the whole detection unit was moved on a programmable scanning table and positioned below each sample of interest. Fluorescence decay curves $I\left(t\right)$ were fitted as a sum of exponential terms corresponding to

Eq. 3

$I\left(t\right)=\sum _{\mathrm{i}}{A}_{\mathrm{i}}{\exp }^{-\mathrm{t}∕{\tau }_{\mathrm{i}}}$

with fluorescence lifetimes ${\tau }_{\mathrm{i}}$ and pre-exponential factors $A_{\mathrm{i}}$ representing the fractional contributions of the component $i$ . The decay parameters were iteratively recovered with a nonlinear least-squares error minimization based on the Levenberg–Marquardt algorithm.¹³ The reduced ${\chi }^2$ ratios and their autocorrelation functions were used to facilitate the assessment of the fit quality. A ${\chi }^2$ value below 1.2 is regarded to be an indicator of a good fit.¹⁴

3.

Results and Discussion

TIR images of HeLa-MemECFP-DEVD-EYFP cells under the fluorescence microscope are depicted in Fig. 3 upon optical excitation of donor (a) or acceptor (b) molecules. Donor fluorescence in Fig. 3a was selected by an interference filter at $475\pm 20\mathrm{nm}$ , whereas acceptor fluorescence in Fig. 3b was measured exclusively at $\lambda ⩾515\mathrm{nm}$ . Similarity of both images proves colocalization of ECFP and EYFP molecules within the MemECFP-DEVD-EYFP complex.

Fig. 3

TIR images of HeLa cells transiently transfected with the MemECFP-DEVD-EYFP plasmid upon excitation of the donor (ECFP) at $391\mathrm{nm}$ and its detection at $475\pm 20\mathrm{nm}$ (a) as well as upon excitation of the acceptor (EYFP) at $470\mathrm{nm}$ and its detection at $\lambda ⩾515\mathrm{nm}$ (b). Image size: $140\times 140\mu \mathrm{m}$ .

Intensitiy ratios $I_{\mathrm{a}}∕I_{\mathrm{p}}$ of the donor ECFP in TIR reader experiments after $\left(I_{\mathrm{a}}\right)$ and prior to $\left(I_{\mathrm{p}}\right)$ the application of staurosporine are given in Table 1 for HeLa-MemECFP-DEVD-EYFP as well as for HeLa-MemECFP-DEVG-EYFP cells (the latter containing the noncleavable linker DEVG). This table shows that the intensity of ECFP fluorescence in HeLa-MemECFP-DEVD-EYFP cells increased upon apoptosis, when energy transfer to EYFP was disrupted, but remained constant in HeLa-MemECFP-DEVG-EYFP cells as well as in control cells (without incubation with staurosporine). In contrast to ECFP, EYFP fluorescence decreased in HeLa-MemECFP-DEVD-EYFP as well as in HeLa-MemECFP-DEVG-EYFP cells. In the case of the DEVD construct, one would expect that this was due to cleavage of EYFP from ECFP and subsequent EYFP movement out of the evanescent field of illumination. However, EYFP decreased also in cells containing the DEVG control construct, where cleavage was not expected to occur¹⁰ (and where, consequently, ECFP fluorescence did not increase). Looking for an explanation of this finding, we observed in microscopic experiments that part of the cells were detached from the surface upon incubation with staurosporine. Their escape from the evanescent electromagnetic field might be a dominating effect in the TIR reader for both cell lines and could explain a general decrease of fluorescence, which appeared to be more pronounced for EYFP than for ECFP. Evaluation of EYFP fluorescence may be further complicated by some fluorescence fading observed in control measurements of HeLa-MemECFP-DEVG-EYFP cells (without incubation with staurosporine). Therefore, until now only the fluorescence intensity of MemECFP, but not that of EYFP, can be used as an indicator of apoptosis.

Table 1

Intensity ratio of ECFP and EYFP fluorescence after (Ia) and prior to (Ip) incubation with staurosporine (4h) in two transfected HeLa cell lines (medians±MADs) . ECFP was excited at 391nm and measured at 475±20nm , whereas EYFP was excited at 470nm and measured at λ⩾515nm . In control experiments, Ia was also determined after 4h but without staurosporine. The number of measurements is indicated in brackets.

Decay curves of ECFP fluorescence in HeLa-MemECFP-DEVD-EYFP cells after picosecond laser pulse excitation are depicted in Fig. 4 prior to and after application of staurosporine. The fluorescence decrease appears slightly longer after incubation with staurosporine. This is confirmed by the median $\text{values}\pm \mathrm{MADs}$ of 21 individual experiments depicted in Table 2 . All decay curves were fairly tri-exponential with fluorescence lifetimes around 0.65, 2.6, and $10.6\mathrm{ns}$ , relative amplitudes of about 0.52, 0.41, and 0.07, and ${\chi }^2$ values around 1.15. Upon addition of staurosporine, almost no changes of relative amplitudes were observed, whereas all fluorescence lifetimes increased in HeLa-MemECFP-DEVD-EYFP and remained constant in HeLa-MemECFP-DEVG-EYFP cells. This indicates that according to Eq. 1, energy transfer was interrupted or lowered when the cleavable linker DEVD was used, but not when the noncleavable linker DEVG was used, and shows that apoptosis can be evaluated in principle on the basis of fluorescence lifetimes of a membrane-associated fluorescent protein (ECFP).

Fig. 4

Fluorescence decay curves of ECFP in HeLa-MemECFP-DEVD-EYFP cells prior to and after application of staurosporine (individual samples). Excitation wavelength: $391\mathrm{nm}$ ; detection range: $475\pm 20\mathrm{nm}$ .

Table 2

Decay times, relevant amplitudes (normalized to 1), and χ2 values of ECFP fluorescence after and prior to incubation with staurosporine (4h) in two transfected HeLa cell lines (medians+MADs) . ECFP was excited at 391nm and measured at 475±20nm . The number of measurements is indicated in brackets.

A prolongation of the fluorescence lifetime ${\tau }_2$ of ECFP upon interruption of energy transfer $\mathrm{ECFP}\to \mathrm{EYFP}$ has already been detected by fluorescence microscopy of individual apoptotic cells.¹⁰ Because the induction of apoptosis by staurosporine after $4\mathrm{h}$ is never homogeneous in a cell population,¹⁷ it should be emphasized that a similar prolongation is now measured for a mixed population of apoptotic and nonapoptotic cells without any visual control. This prolongation now includes the fluorescence lifetimes ${\tau }_1$ and ${\tau }_3$ , as depicted in Table 2 and predicted by Eq. 1. Only the two shorter components $({\tau }_1,{\tau }_2)$ , but not the comparably weak long-lived component $\left({\tau }_3\right)$ , can be correlated with literature data of (E)CFP fluorescence.^{18, 19}

Commonly, fluorescence readers in clinical and pharmaceutical laboratories include parallel detection of larger cell collectives in microtiter plates, e.g., for evaluation of drug effects on cells. Those plate readers have previously also been used for stationary measurements of FRET, including its application to apoptosis.^{20, 21} The main advantages of the present reader system over existing systems are the involvement of fluorescence-lifetime measurements as well as sample excitation by an evanescent electromagnetic field with multiple TIRs of an incident laser beam. This permits versatile measurements of membrane-associated fluorophores within nanometer ranges, offering perfect suppression of background fluorescence from inner parts of the cells as well as from the surrounding medium. Therefore, the present system appears to be a sensitive device not only for detection of apoptosis but, more generally, for drug screening and in vitro diagnostics in a nanometer scale.