The present study introduces fluorescence anisotropy decay (FAD) and its associated parameters as a means for monitoring cellular activities, to achieve a more profound interpretation of events occurring in individual cells before and after biological manipulations.
One of the most powerful tools in cell research is fluorescence based cytometry.1 Measurement of cell functionality is among its most significant and productive branches. The first functional parameter used was fluorescence anisotropy (FA, also denoted by r).1 It was initially proposed by the Cerceks as an indicator of lymphocyte activation.1 FA is considered to be an indicator of “long term biological alterations” as compared to “fast indicators” that monitor transient effects such as change in pH and membrane potential.2 3
Generally, FA of a fluorescent probe provides—in an extent much greater than its fluorescence intensity (FI)—valuable information about the environmental conditions of the media hosting that probe. According to the Perrin equation4 and its more convenient representation using anisotropy instead of polarization,5 the steady-state FA of a distinct fluorescent molecular species undergoing rotational diffusion, is related to the fluorescence lifetime (FLT)— τF and to the rotational correlation time τR as follows:r0 is the FA of a “frozen gas like system” and is solely determined by the angle between the absorption and the emission transition dipole moments of the probe. The ratio τF/τR is defined as σ. Generally, τR is defined as the ratio ηV/RT (where η is the microenvironmental viscosity, V the molar volume of a spherical probe, R the gas constant, and T the absolute temperature). Thus, τR is the most dominant variable, reflecting environmental changes as monitored by FA. The more limited the rotational motion of the fluorescent probe, the higher the FA, and vice versa.
When monitoring cell functionality, cytometrists more frequently utilize FI rather than FA measurements. Nevertheless, unlike direct FI measurements, which are strongly dependent on nonrelevant variables (optical variables, light source instability, etc.), FA and the related parameters, τF and τR, are intrinsic, which in itself is a fortunate circumstance since it permits more reliable comparisons between data obtained from different experiments and different laboratories. Much more importantly, FAD allows for a deeper and more differential insight into steady state FA results. The emission of cellular probes, even of the same type, is heterogeneous due to the heterogeneous chemo-physical nature of the hosting cellular media, where the probes are distributed. The weight of each emitting group, its τF and τR values, may be best assessed via FAD measurements. Knowledge of these three variables provides a powerful means for understanding the mechanisms involved in FA alterations in general, and cellular/membranal changes in particular.
Briefly, the time dependences of FI(t) and r(t) areFI(t) is the instantaneous emission intensity comprised of n different decay components, r(t) is the instantaneous FAD comprised of m different components and r0 is the FAD immediately after excitation, at t=0. The steady state FA is obtained by integrating the intensity-weighted r(t) over time:
In the present study, an apparatus was designed and constructed, and a methodology was developed to permit measurements of steady state FA, as well as FLT and FAD, in individual cells in a population, before, during and after biological manipulations.
Materials and Methods
Phytohemagglutinin (PHA) (Wellcome Ltd., London, UK) was reconstituted in 5 mL of double-distilled water and further diluted 1:10. For stimulation, 0.1 mL of this solution was added to 1 mL cell suspension (final concentration 90 μg/mL). Fluorescein solutions (1, 5, 10, and 100 μM) were prepared by dissolving fluorescein (Sigma, St. Louis, MO, USA) in phosphate buffered saline (PBS) solutions having different pH (5, 6. 5, 7. 4, 8) and viscosity [different glycerol (Gly) PBS concentrations: 0, 30%, 60%, and 80% Gly]. Polyscience fluorescent beads of fluoresbrite (fluorescein), 5.7±0.1 μm in diameter, with low fluorescent intensity [2000–50 000 molecules of equivalent soluble fluorochromes (MESF)] were used, from Bangs Laboratories, Inc. (Fishers, IN, USA).
For separation of peripheral blood mononuclear cells (PBMC), 20 mL of heparinized blood, obtained from healthy volunteers, was diluted 1:1 in PBS, pH 7.2. Blood was then layered in a 10 mL density gradient solution (1.077 g/cm3, Ficol Paque, Pharmacia, Upsala, Sweden), and centrifuged at 550 g for 30 min. Cells accumulated at the interphase, between the plasma and ficol layers, were collected and washed three times with complete PBS and resuspended in PBS at a final concentration of 6×106 cells/mL. More than 70% of the cells were defined as T lymphocytes, and viability, determined by trypan blue exclusion, was always higher than 90%. Human Jurkat T-lymphoblast cell line was grown in a humidified atmosphere containing 5% CO 2, in a RPMI 1640 medium (Biological Industries, Beit Haemek, Israel), supplemented with 10% (v/v) heat inactivated fetal calf serum (Biological Industries, Beit Haemek, Israel), 2 mM L-glutamine, 10 mM HEPES buffer solution, 1 mM sodium pyruvate, 50 U/mL penicillin, and 100 μg/mL streptomycin.
Mitogenic Stimulation and Apoptosis Induction
Freshly prepared PBMC (107 cells/mL) were incubated at 37 °C with 90 μg/mL PHA for 45 min for mitogenic activation.
Apoptosis was induced by incubating Jurkat T-lymphoblast cells in complete RPMI 1640 medium in the presence of 50 μM H 2 O 2 at 37 °C for various time periods, followed by a quick rinsing of the H 2 O 2, to determine the initial time of the apoptotic process.
Fluorescein Diacetate Staining
An aliquot of 100 μL of unstained cell suspension (5×106 cells/mL) was loaded on the picoliter well-per-cell array. Initial scanning was then performed in order to detect background scattering and auto-fluorescence. This undesired signal was recorded per measurement location and subtracted from the total emission signal (after staining) in order to obtain the correct fluorescence signal.
Cells were washed twice with incomplete RPMI 1640 medium without phenol red, containing 10 mM HEPES buffer solution. An aliquot of 100 μL of cell suspension (5×106 cells/mL) was added to 50 μL of fluorescein diacetate staining solution (Sigma, St. Louis, MO, USA, F7378) in PBS, and incubated at room temperature for 5 min. At the end of incubation, cells were loaded onto the picoliter well-per-cell array (see the following explanation), washed three times with fresh buffer to remove excess staining solution, and measured.
Figure 1 shows (a) light and (b) fluorescent micrographs of the same individual living PBMC stained with fluorescein diacetate, positioned in the picoliter well-per-cell array. As it can be seen in Fig. 1, the diameter of cells, as observed by light and fluorescent microscopy, is the same. This signifies that the dye uniformly stains the entire cell.
Plasma membrane integrity of fluorescein diacetate stained cells was checked by re-staining the same cells with propidium iodide (PI). At the end of fluorescein diacetate measurement, the cells were washed twice with fresh buffer and a solution containing PI (2.5 μg/mL) was added on top of the pretested localized cells for 5 min. Cells were then washed twice and remeasured. Positive PI cells were excluded from the analysis.
To confirm apoptosis, Annexin V staining was performed. One million cells were washed with PBS and resuspended in binding buffer (Genzime, Cambridge, MA, USA). FITC Annexin V was added at a final concentration of 1 μg/mL. After 10 min of incubation in the dark, at room temperature, cells were analyzed by FACS (Beckton-Dickenson, FACS-Calibur, Mountainview, CA, USA).
Measurements of FAD and τF were performed on live cells by microscopy, based on the frequency domain methodology.
Schematic depiction of the frequency domain time resolved microscope (FDTRM) is shown in Fig. 2. The FDTRM comprises two main stand-alone components: the optical system (an upright, epi-fluorescent Zeiss microscope Z-1) and a modified frequency domain FLT and FAD ISS K-2 spectrofluorimeter.
Briefly, an excitation of a 488 nm polarized beam from a continuous wave (cw) argon ionic laser (model 161B-070, Spectra-Physics, Mountain View, CA, USA) is directed to the light source entrance of the spectrofluorimeter, passing through an electro-optic Pockel’s cell (Laser Metrics KD*P 1044, Winterpark, FL, USA) which modulates the beam at the frequency range from 1 kHz to 500 MHz. In most experiments, ten frequencies were used, ranging from 2 to 200 MHz in logarithmic scale. A small portion of the modulated excitation beam is sampled by a reference detector (Hamamatsu R928 PMT, Hamamatsu City, Japan), while the rest of it exits the spectrofluorimeter. Then it is doubly refracted by a mirror stirrer, impinges on the microscope illuminator (the epifluorescent arrangement) which directs the beam through the objective lens [×40, numerical aperture (NA)=0.60, LWD CDPlan 40 PL, Olympus, Japan] that focuses the beam to, at most, a 15 μm diameter spot on the interrogation plane, where it illuminates the microscopic fluorescent sample under investigation. The Olympus objective was fitted to the Zeiss optical arrangement.
The diverging cone of the emitted fluorescence is then collected and collimated by the same objective, filtered by an emission filter (530±5 nm) and thereafter impinges on a Glan-Tomphson polarization analyzer, which splits the emission into two polarization (FA) components—parallel (I∥) and orthogonal (I+) emission field vectors—relative to the excitation field. Two modulated emission detectors (Hamamatsu R928 PMT) then simultaneously detect the 2 FA components, from which FAD can be calculated, as well as τR. For τF measurements, the same setup is used, but whereas one of the emission detectors is used for the measurement of the fluorescence signal, the other serves to measure the reference scattered excitation signal.
The induced photocurrent is then fed back to the ISS spectrofluorimeter electronic and computer units for calculation of the requested parameters. During measurements, the modulation frequency is changed automatically according to a predesigned menu. The integrated system is fully computer controlled. For FLT measurements, the polarizers of the system were configured to measure at the magic angle.
Individual Cell Measurement
50 μL of cell suspension are loaded on top of the specially designed picoliter well-per-cell array (Patent No. WO 03/035824). Single cells are then settled into the single cell wells (SCW) array by sedimentation. After sedimentation, the excess unsettled cells are washed out of the picoliter well-per-cell array. The bottom of a SCW possesses high optical quality. As a result, the transparent light image of a cell within a well can be recorded and correlated with its fluorescence signal. Each cell being held by a well, biological manipulations on the cell are possible (introduction of staining solutions, drugs, etc.) while preserving its identity. The excitation beam individually and sequentially illuminates the retained cells, and FLT and FAD of an individual fluorescent cell are measured.
The model function was represented by a sum of exponentials as shown in Eq. (2), where I(t) is the fluorescence intensity at time t and αi is the amplitude of the i’th lifetime τi, such that ∑iαi=1.
I(t)=I∥(t)+2MI+(t), and M is the microscope polarization correction factor which compensates for any optical and/or electrical distortion of the polarization components.
FAD was analyzed by the global fitting of I∥(t) and I+(t) to the following equations:r(t) is then represented as described by Eq. (2), where r(t) is the FA at time t and βj is the amplitude of the j’th rotational correlation time τRj, such that ∑jβj=r0.
The time resolved measurements were compared to steady state anisotropy (average over time) by utilizing both equalities in Eq. (2):
The quality of fits was evaluated by chi-square (χ2) values based on weighted residuals.6 A fit was considered to be within acceptable limits when χ2<2.
Decay curves of FI were analyzed by the standard multi-exponential MATLAB (MathWorks version 5.3 R11 Natick, MA, USA) fitting software and compared to ISS data analysis as well as to the software of the Center for Fluorescence Spectroscopy (CFS, University Maryland, College Park, MD, USA).
System Performance Validation
All measurements were conducted at room temperature. Checkups and calibration runs of the entire system were performed daily, and included calculation of the relative modulation depth, phase shift, and the correction factor M=I∥/I+. The latter ratio was determined by measuring the FI of a nonviscose fluorescein solution. For all practical purposes, the measurements yield the M value of unity.
Calibration and validation runs were performed by comparing two types of measurements by two different systems (ISS-K2 and FDTRM) for the same sample: two steady-state FA measurements, and two TRM measurements. The TRM results were compared to the steady-state results using Eq. (5). For that purpose, 10 μM fluorescein-glycerol solutions with varying viscosities were measured. The results and their standard deviations (based on 10–50 repetitions) are listed in Table 1. They indicate a high level of correlation and thus the fitness of FDTRM.
|Steady state anisotropy of 10 μM fluorescein–glycerol solutions as measured by the FDTRM in two modes (steady-state and TRM) and the ISS-K2 spectrofluorimeter.|
in buffer aqua
|Steady-state fluorescence anisotropy|
|20%||0.047 ± 0.037||0.054 ± 0.01||0.058 ± 0.025|
|40%||0.103 ± 0.014||0.105 ± 0.05||0.097 ± 0.012|
|60%||0.120 ± 0.026||0.121 ± 0.013||0.118 ± 0.009|
|80%||0.271 ± 0.028||0.281 ± 0.02||0.275 ± 0.014|
The repeatability of the FDTRM performance was checked on fluorescent beads having 2000–50 000 MESF. One lifetime component equal to 4±0.1 ns was found. The FLT value of 4 ns for fluorescein is known to be as that in water solution (see Table 5, row 1, and Refs. 7 and 8), i.e., the dye labeling the beads experiences the same environmental conditions as in water solution. In addition, in FAD measurements (more than 50 measurements were conducted), only one component of τR equal to 3.1±0.85 ns was observed.
Computation of the steady state FA via Eq. (5), utilizing the measured values of τF and τR, yielded r=0.18±0.02, in agreement with the steady state FA measured by both systems: the ISS-K2 and the FDTRM in a steady state mode.
Measurements of FLT and FAD in Cellular Models
Two cellular models were considered: the mitogenic stimulation of peripheral blood mononuclear cells (PBMC), and apoptosis induction in Jurkat T cells (JTC). Previous studies9 10 11 have indicated that in mitogenic activation the steady state FA decreases, whereas in apoptosis the FA increases. These two findings were the main reason for choosing these models for the present study.
Monitoring of mitogenic stimulation in PBMC via FA
Table 2 shows the measured FLT and FAD of fluorescein diacetate stained PBMC, treated with the mitogen PHA (defined as As condition) or untreated (defined as A0 condition). As it can be seen, following mitogenic stimulation, the steady-state FA (r) decreases. In addition, the marker’s FLT increases by more than 12%, from 4±0.05 ns in untreated cells to 4.5±0.11 ns in cells incubated with PHA. Moreover, measurement of the rotational correlation time clearly indicates that there exists only one decaying component with τR decreasing by 20%, from 2.5±0.09 (in untreated cells) to 2±0.25 ns in PHA treated cells. In agreement with Eq. (1), this behavior of τF and τR explains the decrease in the steady state FA following mitogenic stimulation of fluorescein diacetate stained cells.
|Measurements of FLT and FAD in individual cells incubated with (As) and without (A0) PHA.*|
|State||τ (ns)||β||τR (ns)||r||r steady-state mode|
|* The steady-state FA was calculated and compared to the steady-state mode of FDTRM and ISS-K2 apparatuses (r steady state mode: average by the two systems).|
Monitoring of apoptosis in Jurkat T cells via FA
The FLT and FAD were measured in fluorescein diacetate stained JTC, either untreated or treated for apoptosis induction for 3 h with 50 μM H 2 O 2. The results, presented in Table 3, show that τF values in apoptotic Jurkat T cells are similar to those of untreated cells (4.1 ns). However, τR increases from 2.1±0.09 in untreated cells to 3.3±0.3 ns in cells in which apoptosis was induced. The apoptotic process was confirmed by measuring the steady state FA and by Annexin staining.11
|FLT and FAD as measured by FDTRM in individual cells at t=0 (A0) and t=180 min (A H 2 O 2 ) after the onset of apoptosis.|
|State||τ (ns)||β||τR (ns)||r||r steady-state mode|
|A H 2 O 2||4.05±0.25||0.395±0.08||3.3±0.3||0.177±0.03||0.17±0.02|
Contrary to the mitogenic stimulation described previously, in the apoptotic model, changes in the steady-state FA seem to be due to changes in the rotational correlation time rather than FLT.
Generally, cellular events involve several chemo-physical alterations that may affect the steady-state FA. Among these are pH, viscosity, polarity, concentrations of various intracellular ions, etc.8 12 In the following experiments we investigated the possibility of a direct influence of some of the above variables upon the FLT and FAD of a hosted probe.
FLT and FAD dependence on pH
It might be suspected that changes in steady-state FA only reflect a variation in pH occurring during cell activation.13 To examine the dependence of FLT and FAD on pH, four solutions of fluorescein (2 μM)-glycerol (40%) in PBS were prepared, having pH values of 5, 6.5, 7.4, and 8.
In all measurements, only one lifetime and one rotational correlation time component were found in both ISS-K2 and FDTRM systems. Further, the results by both apparatuses were similar (Table 4). By the FDTRM, τF, τR, and r 0 were found to be 3.78±0.1 ns, 1.38±0.3 ns and 0.37±0.02, respectively, whereas with the ISS-K2 system they were 3.8±0.2 ns, 1.3±0.42 ns, and 0.38±0.1, for all pH values. Applying Eq. (5) to these results, the steady-state FA was calculated and found to be 0.108±0.001, similar to the directly measured value (Table 1, second row).
|Dependence of FLT and FAD on pH.*|
|τ (ns)||τR (ns)||r0||FA||τ (ns)||τR (ns)||r0||FA|
|* FLT and FAD of fluorescein in 40% glycerol solutions, having pH levels of 5, 6.5, 7.4, and 8, were measured by the FDTRM and ISS-K2. χ2<1.48.|
FLT and FAD dependence on viscosity
One of the effects of viscosity upon fluorescein is the shortening of its fluorescence decay time.13 We examined the possibility that the shortening of FLT as a function of η might enhance the obvious dependence of r on η. Four fluorescein (2 μM) solutions with different glycerol concentrations in PBS (0, 30%, 60%, and 80%) were prepared, all having pH value of 7.4.
The results shown in Table 5 indicate a strong dependence of FLT on glycerol concentration (a decrease from 4.01 ns at 0% Gly to 3.6 ns at 80% Gly), and an increase of calculated r value from 0.04 to 0.27, correspondingly (the latter value is similar to that obtained by direct r steady state measurements, see Table 1, last row). These results are in agreement with recent publications.14 Measurements of FAD show an even greater dependence on glycerol concentration, an increase of τR from 0.5 to 35 ns, correspondingly.
|Influence of viscosity on FLT and FAD in 2 μM fluorescein solutions at pH=7.4.*|
|0||4.014 ± 0.2||0.4 ± 0.1||0.5 ± 0.09||0.04||0.4|
|30||3.85 ± 0.35||0.37 ± 0.15||1.5 ± 0.17||0.1||0.37|
|60||3.7 ± 0.17||0.33 ± 0.09||2.2 ± 0.3||0.12||0.33|
|80||3.6 ± 0.1||0.3 ± 0.11||35 ± 5.4||0.27||0.3|
|* Viscosity was altered by changing glycerol concentrations. χ2<1.429.|
We further examined the cause of FLT decrease, namely, whether it depends on viscosity changes or interaction with glycerol molecules. To test this, FLT of three fluorescein (2 μM) glycerol solutions in PBS (0, 40%, and 60%) were measured at the temperature range from 15 to 30 °C. The pH levels of the solutions were 6.12, 7.3, and 8.3. The results, shown in Table 6, indicate almost no change in FLT, whereas the steady-state FA clearly decreases. This signifies that FAD is the dominant parameter affecting the steady-state FA. Further, glycerol seems to alter the fluorescein molecule conformation or the angle between the absorption and emission dipole moments as evidenced by the changes in r0 values listed in Table 5.
|Dependence of FLT and FAD on glycerol concentrations and temperature.|
|Gly %||T (°C)||pH=8.3||pH=7.3||pH=6.12|
|τ (ns)||r||τ (ns)||r||τ (ns)||r|
FLT dependence on H2O2
One of the best-known collisional quenchers of fluorescence is molecular oxygen, a product of H 2 O 2. Hence, in the presence of H 2 O 2, FLT might be changed, and consequently FA may change as well.6 In order to negate the possibility that during H 2 O 2 apoptosis induction the measured fluorescence characteristics were directly influenced by the presence of H 2 O 2 and/or its products, the following experiment was preformed in cell free solutions with known viscosity. Fluorescein (10 μM) glycerol (80%) solutions, in the presence of different concentrations of H 2 O 2 (0, 50, 1000, and 16 000 μM) were prepared and FA and FLT were measured.
The results are shown in Fig. 3. The FLT did not change for the H 2 O 2 range of 0–16 000 μM, and remained at the value of 3.6±0.15 ns. However, FA increased by more than 10% from 0.28±0.05 at 0 μM, to 0.312±0.06 at 16 000 μM H 2 O 2, which indicates that FA is not influenced by FLT.
The present study offers FLT and FAD measurements as differential indicators of cellular activity.
The combination of the frequency mode time resolved measurements and the picoliter well-per-cell array is designed to allow FLT and FAD measurements in individual cells under varying biological conditions. There are three considerable advantages to the suggested combined methodology. First, the analysis of fluorescence decay is self-referential since the characteristics of the decay are indicated by the way in which the fluorescence intensity of the sample decreases relative to its intensities at earlier time points. The second is the ability to characterize a nonadhesive and heterogeneous population, based on functional time resolved parameters. The third advantage lies in the measured parameters themselves. In addition to their bioactivity indicative capacity, FLT, FAD, r0 and β∞ may provide valuable information about cellular mechanisms that may involve complex molecular diffusion dynamics, as well as information about structural changes which the reporter molecule experiences in the course of cell activation.
The exemplary experimental models of this study demonstrate these capabilities. Stimulation of PBMC with PHA was found to alter both the FLT (from 4±0.05 before to 4.5±0.11 ns after stimulation) and FAD (2.5±0.09 and 2±0.25 ns, correspondingly) of intracellular fluorescein (see Table 2). In stimulated PBMC, both parameters change, meaning that not only the viscous environment changed, but the fluorescein molecular quantum parameters changed as well. In other cases, only changes in viscosity affect the FA. The experiment on JTC gives evidence that the FAD is the dominant determinative parameter and not the FLT. In JTC undergoing H 2 O 2 induced apoptosis, only FAD changed (from 2.1±0.09 ns at t=0 after the treatment with H 2 O 2 to 3.3±0.3 ns at t=180 min) and the FLT remained constant (4.1 ns)—see Table 3. This finding is supported by the fact that, in fluorescein solutions, the same behavior is observed (Fig. 3). When measuring fluorescein in low viscosity buffer, the steady-state FA is low (0.04 in 20% glycerol solution—Table 1), and in a more viscous solution the steady-state FA is high (0.281 in 80% glycerol solution—Table 1). The reason for the difference is not the FLT, which remained unchanged in solutions with different viscosity (3.9 ns in 40% glycerol—Table 6), but the FAD which was changed. In solutions with low viscosity, the FAD is very short, whereas in the more viscous solution the FAD is much longer (Table 5).
High FA values in cells may point out to a significant restriction of the probe mobility, imposed by its intracellular microenvironment. There may be several causes for such restriction: structural, e.g., caging and “wobbling-in-cone,”15 binding properties (covalent bonding, dipole–dipole interactions, or interactions of higher orders of charge arrangement), or simply the solvent viscosity. The present results indicate that the FA alterations during stimulation and apoptotic process may appear due to both the FLT and fluorescence correlation time changes, and in some cases due to the fluorescence correlation time rather than FLT (FLT is indicative of changes in the fluorescence quantum state). The same interpretation can be made of our previous findings that indicated a clear correlation with respect to FA between the processes of PBMC response to mitogens,16 as well as to antigens17 and phorobol esters,18 the interaction between killer and target cells,8 and the contraction cycle of cardiac cells10—in all, the FA decreases during 20 min following the stimulation/conjugation.
In addition to the FLT and the rotational correlation time, there are two other parameters which may determine FAD. The first is r0, which is in fact the FA limiting value. The second parameter is β∞ which indicates that even at the time→∞, FA differs from 0. The r0 indicates an intramolecular structural aspect, i.e., the angle between the absorption and emission transition dipole moments. A change in r0 of a fluorescent solute reporter hosted in a live cell, may reflect the micro-environmental forces acting on the reporter under different intracellular physiological conditions.
In the present study, r0 of fluorescein was changed from 0.4 to 0.3 when dissolved in 0% glycerol and 80% glycerol solutions, respectively (Table 5). Our results for fluorescein in nonviscous solutions are in agreement with previous studies.19 Lowering of the viscosity of fluorescein–glycerol solution by mild heating decreased the FA (Table 6), which is exponentially dependent on temperature, after Andrade20 and others,12 but did not alter r0. Since the gentle heating does not alter the glycerol concentration, the fluorescent probe should experience, on the average, the same amount of influence by glycerol molecules. This may suggest that the decrease of r 0 from 0.4 to 0.3, indicative of conformational changes, may be due to glycerol–fluorophore interaction rather than to the viscosity per se.
This research was supported by the Horowitz Foundation and by the U.S. Army Medical Research and Materiel Command Grant No. DAMD17-01-1-0131