Propidium Iodide Quenches the Fluorescence of TdT-Incorporated FITC-Labeled dutp in Apoptotic Cells

Transcription

1 1998 Wiley-Liss, Inc. Cytometry 33: (1998) Propidium Iodide Quenches the Fluorescence of TdT-Incorporated FITC-Labeled dutp in Apoptotic Cells Trond Stokke, 1 * Kirsti Solberg, 1 Paula DeAngelis, 2 and Harald B. Steen 1 1 Department of Biophysics, Norwegian Radium Hospital, Montebello, Oslo, Norway 2 Institute of Pathology, Norwegian National Hospital, Oslo, Norway Received 4 December 1997; Revision Received 18 June 1998; Accepted 29 June 1998 Apoptotic cells with frequent DNA strand breaks may be detected by tagging with directly or indirectly labeled nucleotides incorporated by the use of terminal deoxynucleotidyl transferase (TdT). Propidium iodide (PI) is typically added for the simultaneous assessment of DNA content. PI was found to quench the specific in situ FITC-fluorescence of apoptotic cells which were labeled by TdT with FITC-conjugated dutp, biotin-dutp followed by streptavidin-fitc, or digoxigenin-dutp followed by FITClabeled anti-digoxigenin antibodies as measured by flow cytometry. The effect was concentration-dependent, with 50% quenching occurring at 0.8 g/ml, 1.5 g/ml, and 5 g/ml PI, respectively, at approximately cells/ml. Spectrofluorimetry in solution revealed that 15 g/ml PI was required to quench 50% of the fluorescence of ss FITC-labeled poly(du) 35. In contrast, the fluorescence of ds FITClabeled poly(du) 35 -poly(da) was quenched to 50% at 3 g/ml PI. The maximum of the fluorescence excitation spectrum of PI shifted from 490 nm to 535 nm upon binding to ds DNA as well as ss poly(du) 35, and the fluorescence yield of PI at 610 nm increased, but the binding required 10-fold higher concentrations of poly(du) 35 as compared to ds DNA. The spectroscopic properties of PI are therefore similar whether bound to poly(du) or to double-stranded DNA, but the binding to poly(du) is much weaker. The observed quenching in situ therefore cannot be explained by direct binding of PI to the poly(du) tails synthesized by TdT in situ in apoptotic cells, but may rather be due to radiationless energy transfer from FITC to PI bound to double-stranded DNA close to the nicks where TdT is known to start polymerization. Cytometry 33: , Wiley-Liss, Inc. Key terms: apoptosis; TdT; quenching of FITC; propidium iodide Discrimination of apoptotic cells from viable cells by flow cytometry is typically achieved by labeling the frequent DNA strand breaks in apoptotic cells via TdTmediated template-independent polymerization of labeled nucleotides from the free DNA ends (1,2). The nucleotides may be directly or indirectly labeled, and FITC is typically the fluorochrome of choice in the case of detection by fluorescence. In the indirect procedures, biotinylated or digoxigenin-labeled nucleotides are incorporated, and these may be detected by streptavidin-fitc or FITC-labeled anti-digoxigenin (anti-dig) antibodies, respectively. The advantage of the TdT-labeling method, in addition to its specificity, is that total DNA may be labeled with another dye for simultaneous assessment of DNA content (2,3). Because many flow cytometers are equipped only with a 488-nm argon laser, propidium iodide (PI) is typically the DNA dye of choice. We have observed that the discrimination between apoptotic and viable cells was much poorer when PI rather than Hoechst was used as the DNA stain. It is reported here that PI quenches the specific FITC fluorescence of apoptotic cells labeled by three different TdT procedures. MATERIALS AND METHODS Cells Human Reh (pre-b) and HL60 (promyelocytic) leukemic cells were grown in RPMI-1640 with 10% fetal calf serum (FCS). Reh cells were irradiated with 4-Gy X-rays and incubated for 24 h. At this time point, most cells are arrested in G2, but some are also arrested in G1, from which they undergo apoptosis. HL60 cells were treated with 0.5 µg/ml camptothecin for 4 h. S-phase cells die by apoptosis as a result of this treatment (2). Staining of Strand Breaks in Apoptotic Cells Cells were fixed with 1% paraformaldehyde for 10 min at 0 C, spun down, resuspended in 100% methanol at -20 C, and stored at this temperature. Fixed cells were washed once with phosphate-buffered saline (PBS), and Contract grant sponsor: Norwegian Cancer Society. *Correspondence to: Trond Stokke, Department of Biophysics, Norwegian Radium Hospital, Montebello, 0310 Oslo, Norway. tstokke@radium.uio.no

2 incubated for 30 min at 37 C in 50 µl TdT solution containing 5 units TdT (Boehringer Mannheim, Mannheim, Germany), 5 µl 5 reaction buffer (supplied with the TdT), 1.5 mm CoCl 2, 0.5 nmol labeled dutp (see below), 0.1 mm dithiothreitol, and distilled water. Three different labeled dutps were employed: FITC-12-dUTP (DuPont, Boston, MA; FITC-dUTP method); biotin-16- dutp (Boehringer Mannheim; biotin-dutp method); or digoxigenin-11-dutp (Boehringer Mannheim; dig-dutp method). Staining controls received similar treatment, except that the TdT was omitted. The cells were thereafter washed once with PBS. Cells labeled with biotin-dutp or dig-dutp were incubated in 50 µl 1/50 streptavidin-fitc (Amersham, Buckinghamshire, UK) or 8 µg/ml FITClabeled anti-dig antibody in PBS with 0.1% Triton X-100 and 5% dry milk for 30 min at 0 C, respectively, and washed once in PBS with 0.1% Triton X-100. All solutions containing dry milk were centrifuged at 1,000g to get rid of particulate matter. The pellet was resuspended in PBS containing 1.5 µg/ml Hoechst and 100 µg/ml RNase A (Pharmacia, Uppsala, Sweden) to a final concentration of approximately cells/ml. Propidium iodide (Calbiochem, La Jolla, CA; molecular weight, 668) was added stepwise from concentrated stock solutions to give final concentrations of 0.2, 0.5, 1, 2, 5, 10, 20, 50, and 100 µg/ml. Flow Cytometry Forward scatter, side scatter, FITC fluorescence ( nm), and PI fluorescence ( 630 nm) were measured upon excitation with 200-mW, 488-nm laser light (argon laser; Spectra Physics, Mountain View, CA) in a FACStar flow cytometer (Becton Dickinson, San Jose, CA). Hoechst fluorescence ( nm) was excited with 50 mw UV (351/356 nm) at the second laser intercept (krypton laser; Spectra Physics). Acquisition was triggered on the forward-scatter signal, and the threshold was set to a low level to include apoptotic bodies. The area and width of the Hoechst signal were also measured and used to gate away aggregates. It is very important that the UV intensity at the 488-nm focus be very low, since UV-excited green Hoechst fluorescence will interfere with FITC fluorescence. This has occurred in our flow cytometers, and is due to the diffraction maxima outside the main Gaussian peak of the UV laser caused by a number of apertures in the optics. This artifact is tested for by checking whether the FITC fluorescence decreases when the UV laser is blocked, and can be avoided by increasing the power of the 488-nm laser, decreasing the power of the UV laser, or both. In the case of Reh cells, the median FITC intensities of apoptotic and viable G 2 cells were calculated. The median intensity of apoptotic early S-phase cells and viable G 1 cells was calculated for HL60 cells. The specific apoptosis-associated FITC fluorescence was calculated by subtracting the fluorescence of the viable cells from the fluorescence of apoptotic cells. Hoechst (pulse area) and PI (pulse height) intensities were calculated as the mean fluorescence of the G 1 peak. The Hoechst fluorescence is given relative to PI QUENCHES FITC FLUORESCENCE OF APOPTOTIC CELLS 429 the intensity observed in the absence of PI, and the PI fluorescence is given relative to the fluorescence observed at a PI concentration of 100 µg/ml PI. Absorption and Fluorescence Spectrometry The absorbances at 480 and 520 nm at different PI concentrations were measured in a Perkin Elmer Lambda 15 absorption spectrophotometer (Buckinghamshire, UK) to correct for inner-filter effects in fluorescence measurements. For calculation of the distance for 50% energy transfer (critical radius; R 0 ) from FITC to free or DNAbound PI, the molar extinction coefficient of PI (1.5 µm ) in the region of nm was measured in the absence or presence of a large excess of double-stranded human genomic DNA (150 µm), respectively. The fluorescence intensity of 0.25 µm FITC-12-dUTP, 0.25 µm (base concentration) poly(du) 35 labeled with FITC at the 3 end (Eurogentec, Seraing, Belgium), or 0.25 µm FITC-labeled poly(du) 35 in the presence of µm poly(da), was measured in a Perkin Elmer LS 5 fluorescence spectrometer at increasing PI concentrations. Excitation was at 480 nm and emission was measured at 520 nm. Inner-filter effects were corrected for, assuming that both the excitation path and the emission path were 5 mm (10 10 mm cuvette). The fluorescence excitation spectrum of 1 µm PI was measured at an emission wavelength of 610 nm with increasing concentrations of poly(du) 35 (Eurogentec, Seraing, Belgium), or double-stranded human genomic DNA. The LS 5 fluorescence spectrometer contains a quantum counter which corrects for the quantum intensity of the excitation light. The excitation and emission slits were set to 5 nm in all experiments, and a UV exclusion filter was used on the excitation side. PBS was used as the buffer and source of salts, i.e., the same as in the flow cytometry experiments. Calculation of Energy Transfer The efficiency of energy dissipation by radiationless Förster -type energy transfer (see Stryer (4) for a thorough discussion of the subject) is: Q 1/(1 R 6 /R 6 0 ), where the critical distance R 0 is the distance for 50% energy transfer. R 0 is a function of the overlap between the emission spectrum of the donor and the absorption spectrum of the acceptor (J), the fluorescence quantum yield of the donor ( ), the relative orientation of the transition moments of the two chromophores ( 2 ), and the index of refraction of the medium between them (n): R 0 ( J 2 /n 4 ) Å. J was calculated from the corrected emission spectrum of FITC and the extinction coefficients of PI measured in the absence (free PI) or presence (DNA-bound PI) of an excess of double-stranded DNA. was assumed to be 0.25, which is in middle of the range given by Waggoner (5). 2 was assumed be 2/3 (random orientation). n was assumed to be 1.33 in the case of energy transfer to free PI (i.e., water between the chromophores), and 1.5 in the case of energy transfer to DNA-bound PI (DNA between the chromophores). In the preceding paragraph, the distance between the chromophores was assumed to be the same for all pairs.

3 430 STOKKE ET AL. FIG. 1. Quenching of apoptosis-associated FITC fluorescence in Reh cells. Reh cells were X-irradiated and further incubated for 24 h to produce cells arrested in G 1 and G 2, as well as apoptotic G 1 and G 2 cells. Cells were fixed and stained with the FITC-dUTP (A, D, G), biotindutp (B, E, H), or dig-dutp (C, F, I) method. Cells were counterstained with 1.5 µg/ml Hoechst and measured for FITC fluorescence and Hoechst fluorescence (A C). PI was added to a final concentration of 2 µg/ml (D F), or 20 µg/ml (G I), and the measurements were repeated. The high voltage for the Hoechst signal was adjusted to position the (viable) G 1 peak in channel 200 for each histogram. This is obviously not the case if PI is free in solution, where the molecules would be randomly distributed. An expression for the quenching of FITC by free PI was obtained for low acceptor concentrations (i.e., concentrations such that the chance of finding two or more PI molecules within approximately the volume of the sphere with radius R 0 is very low). Under these conditions, the quenching efficiency is: Q mean 4 R 2 dr n mean 1/(1 R 6 /R 0 6 ), where n mean is the average density of PI molecules per Å 3, and the integration is performed from zero to infinity. This integral is solved by Rottmann (6), and it can be shown that Q mean M R 0 3 where M is the molar concentration of the acceptor. RESULTS Preliminary observations indicated that the distinction of apoptotic Reh cells from viable Reh cells with the FITC-dUTP method was seriously impaired when PI was used as the DNA counterstain rather than Hoechst This was due to a decrease in the FITC fluorescence in the presence of PI, excluding the possibility that the effect was caused by the PI fluorescence contribution in the FITC channel. All three labeling procedures (see Materials and Methods) gave satisfactory discrimination between apoptotic and viable cells when Hoechst was used as the counterstain (Fig. 1A C). Addition of low concentrations of Hoechst (1.5 µg/ml; see Stokke and Steen (7) for a discussion of the concentration dependence of the fluorescence of DNA-bound Hoechst 33258) did not have any effect on the FITC fluorescence distributions of apoptotic and viable cells (data not shown). In spite of the high signal in apoptotic cells stained by the dig-dutp method (Fig. 1C), the separation between apoptotic and viable cells was no better than with the FITC-dUTP

4 PI QUENCHES FITC FLUORESCENCE OF APOPTOTIC CELLS 431 FIG. 2. Quenching of apoptosis-associated FITC fluorescence in HL60 cells. HL60 cells were treated with camptothecin for 4 h to produce viable G 1 and G 2 cells, as well as apoptotic S-phase cells. The cells were fixed and stained with the FITC-dUTP (A, D, G), biotin-dutp (B, E, H), or dig-dutp (C, F, I) method. Cells were counterstained with 1.5 µg/ml Hoechst and measured for FITC fluorescence and Hoechst fluorescence (A C). PI was added to a final concentration of 2 µg/ml (D F), or 20 µg/ml (G I), and the measurements were repeated. The high voltage for the Hoechst signal was adjusted to position the (viable) G 1 peak in channel 200 for each histogram. method (Fig. 1A), due to a much higher background in the viable cells. The biotin-dutp method gave the best discrimination with the reagents we used (Fig. 1B). At 2 µg/ml PI (Fig. 1D F), and particularly at 20 µg/ml PI (Fig. 1G I), the discrimination between apoptotic and viable cells was significantly reduced. This effect was most pronounced with the FITC-dUTP method; the apoptotic cells could not be resolved at PI concentrations above 5 µg/ml. Although the FITC fluorescence of apoptotic cells stained by the dig-dutp method decreased considerably with increasing PI, the background fluorescence also decreased, and the apoptotic cells could be resolved even at 100 µg/ml PI. HL60 cells treated with camptothecin were also stained by the three different TdT procedures. Similar results were obtained as for X-irradiated Reh cells (Fig. 2). However, the resolution between apoptotic and viable cells was poor for the dig-dutp method, even in the absence of PI, because the FITC fluorescence distribution of the apoptotic cells was wide (Fig. 2C,F,I). Cells stained in the absence of TdT (staining control) showed FITC fluorescence intensities similar to those of the viable cells in the samples receiving TdT (data not shown). The background was mostly due to nonspecific binding of the nucleotides and detection reagents, since the autofluorescence intensities of these two cell lines in the FITC channel were approximately 5 units (compare with the intensity of nonapoptotic cells in Figs. 1 and 2). The results suggest that Hoechst should be used as the DNA counterstain. It could also be asked whether the incorporated FITC-labeled nucleotides in apoptotic cells quench the Hoechst fluorescence of apoptotic cells selectively. Figure 1 shows that the Hoechst fluorescence intensities of the apoptotic G 1 and G 2 cells were lower than for the viable G 1 and G 2 cells, respectively. However, this reduced DNA stain fluorescence in

5 432 STOKKE ET AL. the apoptotic cells was also observed in the control cells which did not receive TdT, as well as when PI was used as the counterstain (data not shown). The reduced DNA stain fluorescence, compared to that of viable cells, must therefore be due to loss of DNA or a reduced stain accessibility to DNA in apoptotic cells. This also means that no significant increase ( 10%) was observed in the PI fluorescence of apoptotic cells compared to viable cells in TdT-stained samples. X-irradiated Reh cells and camptothecin-treated HL60 cells were stained with the three TdT procedures and titrated with PI at concentrations from µg/ml PI for a quantitative view of the concentration-dependence of the quenching. Figure 3A shows specific apoptosisassociated FITC fluorescence as a function of PI concentration. The apoptosis-associated FITC fluorescence with the FITC-dUTP method was reduced to 50% at 0.8 µg/ml PI. The corresponding 50% quenching concentrations were 1.5 µg/ml and 5 µg/ml PI for the biotin-dutp and the dig-dutp methods, respectively. There were no obvious differences in the shape of the three curves, except for the translation along the PI concentration axis (Fig. 3A). In contrast, the Hoechst fluorescence, which decreased in a similar manner to the FITC fluorescence of the apoptotic cells by the biotin-dutp method at the higher PI concentrations, was less quenched at the lowest PI concentrations (Fig. 3B). Figure 3B also shows the PI fluorescence intensity normalized to the value measured at 100 µg/ml PI. In practice, the (apoptotic) signal to (viable) background ratio is more important than the absolute intensities, because the former parameter, together with the widths of the distributions, determines whether the apoptotic cells can be discriminated from the viable cells. Figure 3C shows these signal-to-background ratios for the three different methods as a function of PI concentration. The signal-to-background ratio decreased with increasing PI concentrations with the FITC-dUTP and the biotindutp methods, but was not as pronounced as the specific apoptosis-associated FITC fluorescence (Fig. 3A). The reason is that the background also decreased slightly with PI concentration. The signal-to-background ratio decreased less with the dig-dutp method than with the other two methods (Fig. 3C). The quenching is not caused by perturbation of the (photoelectron) wave functions of FITC, because PI does not bind, or binds very weakly, to FITC-dUTP (see below). The quenching also occurs at concentrations far too low to be explained by collision between the chromophores, or by absorption of the excitation light/emission within the cell (see Discussion). To our knowledge, the only remaining mechanism causing quenching of fluorescence is radiationless energy transfer ( Förster -type energy transfer), because PI absorbs in the spectral region where FITC emits fluorescence. However, certain requirements for the geometrical arrangement of the transition moments must also be satisfied for energy transfer to occur, the most important being that the separation between the chromophores is on the order of or less than the critical radius R 0. FIG. 3. Quenching of FITC fluorescence by PI. Reh cells (two separate experiments) and HL60 cells (two separate experiments) were treated and measured as described for Figures 1 and 2 in the presence of a wide range of PI concentrations. The specific apoptosis-associated fluorescence was calculated as the difference between the intensity of the apoptotic G 2 (Reh) or early S-phase (HL60) cells and the viable G 2 (Reh) or G 1 (HL60) cells. The intensities, normalized to the one measured in the absence of PI, and the standard deviations, are given (A) for the FITC-dUTP method (, four experiments), the biotin-dutp method (, four experiments), and the dig-dutp method (, four experiments). B: Hoechst fluorescence intensity ( ) and the PI fluorescence intensity ( ) normalized to the intensity in the absence of PI and with 100 µg/ml PI, respectively (four experiments with an average of the three TdT procedures in each experiment). C: Signal-to-background ratio, i.e. the ratio between apoptotic and viable cells, is plotted for the three different labeling procedures. Symbols are the same as in A. A also shows the (normalized) fluorescence intensity at 520 nm (excitation, 480 nm) measured by spectrofluorimetry of 0.25 µm FITC-dUTP (, three experiments), 0.25 µm (base concentration) FITC-poly(dU) 35 (, three experiments), or 0.25 µm (base concentration) FITC-poly(dU) 35 in the presence of an excess of poly(da) ( ), three experiments employing 0.5, 1.0, or 2.5 µm (base concentration) poly(da). These three experiments gave similar results, and the average is given.) Bars represent standard deviations. Based on spectroscopic data, R 0 was calculated to be 33 Å and 31 Å for energy transfer from FITC to free and DNA-bound PI, respectively. Hence, energy transfer may occur if PI binds up to bases away from FITC in

6 FITC-labeled polynucleotides. Possible binding of PI to polynucleotides was studied by fluorescence spectroscopy, which revealed that the FITC fluorescence of FITC-labeled poly(du) 35 was quenched to 50% at approximately 15 µg/ml PI (Fig. 3A). Fivefold higher concentrations of PI were required for the quenching of the fluorescence of equimolar concentrations of unpolymerized FITC-dUTP (Fig. 3A), indicating that polymerization increased the affinity of PI for dutp. The quenching caused by randomly distributed, fixed PI molecules at a concentration of 50 µg/ml was calculated to be 2.1%, which is considerably lower than observed with FITCdUTP. However, we have not taken into account the diffusion of PI during the lifetime of the excited state of FITC. Since the mean diffusion length of molecules the size of PI is on the order of tens of angstroms in 4 ns (e.g., see Birks and Georghiou (8)), this means that the quenching by free PI may be higher than calculated. Hence, the quenching of FITC-dUTP fluorescence may be compatible with no binding of PI, or with very weak binding. When the FITC-labeled poly(du) 35 was allowed to hybridize with an excess of poly(da), the quenching occurred at much lower concentrations, resembling the results obtained in situ (Fig. 3A; 50% quenching at 3 µg/ml PI). However, the fluorescence decreased much less at the higher concentrations of PI than was the case for FITC-poly(dU) 35 in the absence of poly(da). The possible binding of PI to single-stranded poly(du), indicated by the results shown in Figure 3A, was investigated by titrating a low concentration of PI with polynucleotides (Fig. 4). As expected, the corrected fluorescence excitation maximum of PI shifted from 490 nm to approximately 535 nm in the presence of double-stranded DNA, and the fluorescence yield at 610 nm increased 11-fold compared to the free dye (Fig. 4A; excitation at 520 nm, which is the isobestic point of the absorption spectra, data not shown). Dye binding was apparently complete at µm DNA. The fluorescence excitation maximum was shifted to 535 nm, also in the presence of poly(du) 35, and the fluorescence yield at 610 nm increased threefold in the presence of 100 µm poly(du) 35 (Fig. 4B; excitation, 520 nm). Much higher concentrations of poly(du) were required to observe the spectroscopic changes of PI, and there was no indication that all dye was bound even at the highest poly(du) concentrations. PI QUENCHES FITC FLUORESCENCE OF APOPTOTIC CELLS 433 FIG. 4. Binding of PI to DNA and poly(du). The corrected fluorescence excitation spectrum of 1 µm PI (emission, 610 nm) was measured at 0 ( ), 5µM( ), 10 µm ( ), 20 µm ( ), 50 µm ( ), and 100 µm ( ) (base concentration) of double-stranded genomic DNA (A), or single-stranded poly(du) 35 (B). DISCUSSION Detection of apoptotic cells by flow cytometry has become a popular exercise in cell and cancer biology. It is often of interest to measure the DNA content simultaneously with TdT-incorporated (FITC-) tagged nucleotides for identification of apoptotic cells. The results reported here suggest that PI should not be used in conjunction with the FITC-dUTP and biotin-dutp methods, because the fluorescence of the apoptotic cells is dramatically reduced at or above 2 µg/ml PI, the lowest concentration which gives decent DNA histograms (data not shown). PI can be used with the dig-dutp method, but the concentration should be kept as low as possible. Preferably, controls without addition of PI should be run. We have observed that the quenching of FITC fluorescence is slightly less pronounced when using 7-aminoactinomycin D as the DNA stain (data not shown). Hence, this dye may provide an alternative to PI in instruments with 488-nm excitation only. Our results with apoptotic cells suggest that PI may also quench the fluorescence of FITC-labeled DNA probes in fluorescence in situ hybridization experiments, as well as the fluorescence of FITC-labeled anti-brdu antibodies. Indeed, preliminary experiments employing laser scanning cytometry to assess the fluorescence of hybridized FITC-dUTP-labeled whole genomic probes to interphase cells show that the FITC fluorescence is decreased approximately 40% in the presence of 2 µg/ml PI. Crude calculations show that inner-filter effects are too small to explain the observed quenching of FITC fluorescence in apoptotic cells, even if it is assumed that one PI molecule is bound for every base pair throughout the genome. Collision quenching requires millimolar to molar concentrations of the quencher. (Our unpublished results show that approximately 750 mm trichloracetic acid, a strong fluorescence quencher, is required to decrease the

7 434 STOKKE ET AL. fluorescence lifetime of fluorescein from 4.4 ns to 2.2 ns). Since PI binds very weakly, if at all, to FITC-dUTP, long-range radiationless energy transfer is, to our knowledge, the only mechanism which can explain the quenching of FITC in situ. The quenching of the fluorescence in situ occurs at 10-fold lower concentrations of PI than those required for the quenching of fluorescence of FITC-labeled poly(du), indicating that PI binding to poly (du) is not responsible for the quenching of fluorescence in situ. If, however, the length of the poly(du) tails synthesized in situ is on the order of or shorter than R 0 (i.e., about 10 bases), excited FITC molecules along these tails should be efficiently quenched by PI molecules bound to doublestranded DNA close to the nick in genomic DNA where the TdT started polymerizing the poly(du) tail. The length of the poly(du) tails can in principle be calculated if we know the number of FITC molecules bound and the number of nicks in apoptotic cells. The median intensity of FITC fluorescence in apoptotic cells corresponded to approximately FITC molecules (measured against beads with known FITC content). This implies that for the poly(du) tails to be longer than 10 bases, the frequency of nicks in apoptotic cells must be less than 1 per 120 kilo-base pairs. This seems highly unlikely, since from gel electrophoresis experiments we know that a considerable fraction of the DNA in apoptotic cells has lengths of a few hundred to a few kilo-base pairs. Hence, the in situ synthesized poly(du) tails are probably so short ( 10 bases) that PI bound to the double-stranded genomic DNA adjacent to the nick is close enough to the FITC for efficient energy transfer to occur. It might be argued that energy transfer to a fluorescent molecule should result in enhanced acceptor fluorescence, which should lead to a relatively higher PI signal from apoptotic cells compared to viable cells in the presence of TdT-incorporated FITC. This was not observed within the limits of resolution ( 10%). However, considering the large amount of PI fluorescence excited directly by the 488-nm laser, calculations (not shown) indicate that any PI fluorescence excited via FITC accounts for no more than 1% of the total PI fluorescence, making such an enhancement impossible to observe. The quenching of (nonspecific) FITC fluorescence in viable cells and cells stained in the absence of TdT may also be explained by energy transfer if the nucleotides bind nonspecifically to double-stranded DNA. It might be speculated that the different degrees of quenching observed with the three methods reflect the processivity of TdT when FITC-dUTP, biotin-dutp, and dig-dutp are used as substrates. However, the different quenching may also reflect differences in the spatial distribution of the FITC molecules with respect to the poly(du) backbone with the different methods. If it is correct that the quenching depends on the length of the poly(du) tails synthesized in situ, it may be worthwhile to try to increase the processivity of the TdT to avoid extensive quenching by PI. For practical purposes, as far as the discrimination between viable and apoptotic cells in the absence of PI is satisfactory, the best alternative is to use Hoechst as the DNA stain if an instrument with dual-laser excitation is available. Hoechst does not affect the FITC fluorescence of apoptotic cells, and this DNA stain typically gives better resolution than PI. A further advantage of using Hoechst as the DNA counterstain is that phycoerythrin-labeled antibodies may then be used for identification, for instance, of immunophenotype (9). LITERATURE CITED 1. Gavrieli Y, Sherman Y, Ben-Sasson SA: Identification of programmed cell death in situ via specific labeling of nuclear DNA fragmentation. J Cell Biol 119: , Gorczyca W, Bruno S, Darzynkiewicz RJ, Gong J, Darzynkiewicz Z: DNA strandbreaks during apoptosis: Their early in situ detection by the terminal deoxynucleotidyl transferase and nick translation assays and prevention by serine protease inhibitors. Int J Oncol 1: , Gorczyca W, Bigman K, Mittelman A, Ahmed T, Gong J, Melamed MR, Darzynkiewicz Z: Induction of DNA strand breaks associated with apoptosis during treatment of leukemias. Leukemia 7: , Stryer L: Fluorescence energy transfer as a spectroscopic ruler. Annu Rev Biochem 47: , Waggoner AS: Fluorescent probes for cytometry. In: Flow Cytometry and Sorting, Ed 2, Melamed MR, Lindmo T, Mendelsohn ML (eds). John Wiley & Sons, Inc., New York, 1990, pp Rottmann K: Mathematische Formelsammlung. Bibliographisches Institut AG, Mannheim, 1962, p Stokke T, Steen HB: Binding of Hoechst to chromatin in situ. Cytometry 7: , Birks JB, Georghiou S: Energy transfer in organic systems. VII. Effect of diffusion on fluorescence decay. J Phys Biol 1: , Stokke T, Holte H, Smedshammer L, Smeland EB, Kaalhus O, Steen HB: Proliferation and apoptosis in malignant and normal cells in B-cell non-hodgkins lymphomas. Br J Cancer 77: , 1998.