Data Analysis in Flow Cytometry

a Dynamic Approach

Michael G. Ormerod

7. Quantification of cell death

This chapter is included as a sample of the entire book. Data files for this chapter are included on this CD for demonstration purposes. They can be downloaded to your computer from the directory :4/data/viable. Download windmidi software on this CD from directory :3/progs/winmdi/wmdi.exe to analyse them.

7.1. Measurement of viability

  7.1.1 Propidium iodide/fluorescein diacetate (PI/FDA) assay

 The method is described by Ormerod (1994a). PI is excluded by cells with an intact plasma membrane and uptake of PI can be used to measure loss of viability. FDA is not fluorescent. It diffuses into cells where esterases (present in most cells) cleave the acetate groups to generate fluorescein. Fluorescein is charged and is trapped within the cell, diffusing out slowly. Examples are shown in files VIABLE.001-.003.

 File VIABLE.001 contains data from a human ovarian carcinoma cell line. The cells grow attached to the culture dish; they were harvested using trypsin/EDTA and labelled in suspension. The cells were prepared by Swee Sharp, Institute of Cancer Research, Sutton, and analysed by Jenny Titley on a Coulter Elite with an argon-ion laser giving 200 mW at 488 nm. Display green (PMT 2) versus red fluorescence (PMT 4) on logarithmic scales. (PMT 1 is the RALS). Viable cells stain green+ve, red-ve; dead cells green-ve, red+ve. The double-labelled cells are clumps. File VIABLE.002 contains similar data from another human ovarian cell line, CH1. The cells, which were prepared by Ciaran O'Neill, Institute of Cancer Research, had been incubated for 2 h with a trans-Pt(IV) dicarboxylate (JM335) 24 h previously. The cells had been over-labelled (by me) with FDA and I had not corrected the fluorescence for spectral overlap so that the green fluorescence spilled into the red channel. Despite that, there is a clear separation between the viable and dead cells. The dead cells gave a good DNA histogram showing a G2 block. Analysis of the cells which had detached from the culture dish is shown in file VIABLE.003.

7.2. Apoptosis

  Methods for quantifying apoptotic cells are given by Darzynkiewicz et al. (1994), Ormerod (1994a), Fraker et al. (1995) and Sherwood and Schimke (1995). The latter two references are in a volume which deals entirely with methods for studying cell death (Schwartz and Osborne, 1995). Files for this section are labelled APO.000.

 7.2.1 Introduction

 Apoptosis is an important mode of cell death, both in natural development and in response to cytotoxic insult. Apoptotic cells are distinguished by characteristic changes in the nuclear chromatin and, in the initial stage of any study, these cells should be identified morphologically by light or electron microscopy. In some cells, but not all, apoptosis is accompanied by internucleosomal degradation of DNA giving rise to a distinctive 'ladder' pattern on DNA gel electrophoresis. The introduction of double strand breaks at intervals of 30-50 kbp (possibly corresponding to the size of chromatin loops) is an earlier change in the DNA which can be observed by pulse field gel electrophoresis.

 These methods cannot be used to count the number of cells undergoing apoptosis. However, apoptotic cells can be quantified by flow cytometry and several methods for counting apoptotic cells have been developed. Most methods fall into two broad categories: those which detect degradation of DNA and those which rely on changes in the plasma membrane of apoptotic cells.

 After some methods of fixing or permeabilising cells, low molecular weight fragments of DNA are extracted and an extra peak ('sub-G1') of apoptotic cells can be seen in the DNA histogram. Alternatively, internucleosomal degradation of DNA can be visualised by labelling enzymatically the ends of the broken strands of DNA. In the flow cytometer, the cell cycle can be measured simultaneously.

 Apoptotic cells have altered membranes and new entities may be exposed on their surface. One of these, phosphatidyl serine, can be detected by incubating the cells with fluorescein-labelled Annexin V (Homburg et al., 1995). Apoptotic cells may also show changes in the uptake and retention of some dyes. In some cells, apoptosis can be measured in viable cell preparations by staining with the dye, Hoechst 33342, and propidium iodide. The latter dye identifies loss of membrane integrity. Hoechst 33342, which on binding to DNA fluoresces blue under UV light, may be taken up more rapidly by apoptotic cells. Recently, a similar method using the dye, YOPRO-1, has been published (Idziorek et al., 1995). This dye has the advantage that it can be excited with blue light.

 Other methods of distinguishing apoptotic cells include following changes in the emission spectrum of DNA-binding dyes, such as Hoechst 33342 and 33258. Finally, apoptotic cells shrink and forward light scatter can sometimes be used to identify them.

 7.2.2 Measuring a 'sub-G1 peak' in the DNA histogram

 Frequently, at a late stage in the apoptotic process, the DNA is cut at the linker regions between the nucleosomes. During fixation and subsequent rehydration of the cells, some of the lower molecular weight DNA leaches out, lowering the DNA content. These cells can be observed as a hypodiploid or 'sub-G1' peak.

 Files APO.001 & .002, contain data from a murine haemopoetic cell line, BAF3 which were fixed in 70% ethanol, resuspended in PBS and then stained with PI. The control cells (APO.001) were grown with the growth factor, IL-3, added; about 3% of the cells are apoptotic. For the second file, IL3 was omitted for 16 h; about 47% cells became apoptotic. Gate on a cytogram of DNA-peak versus DNA-area to exclude clumps (see Chapter 4.2), be careful to include the cells with a DNA content less than G1. Use a cytogram of RALS versus DNA to gate out some of the debris (low RALS). Display a gated histogram of DNA and note the presence of a 'sub-G1' peak. The cells incubated without growth factor are mainly in G1 of the cell cycle; there is also a greatly enhanced 'sub-G1' peak in the DNA histogram due to apoptotic cells (Figure 7.1).


Figure 7.1. A murine haemopoetic cell line, BAF3, incubated for 16 h without IL3. Cells were fixed in 70% ethanol, rehydrated and stained with 20 g/ml PI. Data gated on a cytogram of DNA-peak versus DNA-area and also a cytogram of DNA versus RALS. Coulter Elite with argon-ion laser giving 200 mW at 488 nm. File: APO.002


The amount of DNA extracted - and hence the position of the sub-G1 peak - depends on the type of cell being studied and the buffer in which the cells are resuspended. File APO.003 contains data from a sub-line of the human lymphoblastoid cells, W1L2, incubated with KCN. This treatment triggers apoptosis in these cells but there is no 'sub-G1' peak. The cells were fixed in 70% ethanol and resuspended in PBS. Inspection of a cytogram of DNA-peak versus DNA-area reveals two populations of the same DNA content but one has smaller nuclei (higher DNA-peak for the same DNA-area) (see Figure 7.2). Cell sorting revealed that these cells were apoptotic (Ormerod, M.G., Titley, J.C. and Kimbell, R., unpublished work). A cytogram of DNA versus time of flight shows that the apoptotic cells are only slightly smaller than the non-apoptotic cells; there is no significant difference in the light scatter. Normal W1L2 cells show a single population on a cytogram of DNA-peak versus DNA-area and slightly greater light scatter (File APO.004).

Figure 7.2. A sub-line of W1L2 incubated for 18 h with 2 mM KCN. Cells were fixed in 70% ethanol, resuspended in PBS and stained with 20 g/ml PI. The green dots are from apoptotic cells, the red from normal and the black from clumped cells. Coulter Elite; argon-ion laser giving 200 mW at 488 nm. File: APO.003.


After ethanol fixation, if the cells are resuspended in a phosphate-citrate buffer (0.2 M Na2HPO4, 4 mM citric acid, pH 7.8), more DNA is extracted from the apoptotic cells (Gong et al., 1994). With BAF3 cells, the sub-G1 peak may disappear off the end of the cytogram - the DNA can be over-extracted (File APO.005). In the W1L2 cells, a sub G1 peak is formed, albeit rather smeared out (File APO.006).

 Instead of fixing cells in ethanol, they can be suspended in a buffer containing detergent, PI and RNase (see Chapter 4.4). A 'sub-G1' peak from apoptotic cells can be observed (File APO.007). This method should be used with caution since, if nuclei are fully released by the action of detergent, fragmented nuclei will give more than one particle. The membrane of apoptotic cells are altered and the cells may be more refractory to lysis compared to normal cells. In which case, the apoptotic cells can be distinguished on a cytogram of RALS versus DNA by their increased RALS and decreased DNA content (unpublished observation, an FCS listed data file not available).

 7.2.3 In situ end labelling (ISEL) of strand breaks

 Scission of the chromatin at the sites of nucleosomal linkage creates a large number of strand breaks in the DNA of apoptotic cells. The broken ends can be labelled enymatically in situ (In Situ End Labelling - ISEL). The enzyme used is either the Klenow fragment of Escherichia coli polymerase or terminal deoxynucleotidyl transferase (Tdt), the latter now being used almost universally. The ends of DNA can be labelled directly using fluorescein-deoxyuridine triphosphate (dUTP) or biotin-dUTP followed by fluorescein-streptavidin or digoxygenin-dUTP followed by fluorescein-anti-digoxygenin. The last two methods are more sensitive. The cells are fixed in ice cold 1% paraformaldehyde followed by 70% ethanol. The paraformaldehyde fixation crosslinks the DNA into the cell and prevents the low molecular weight DNA from being extracted. The cells are counter-stained with PI so that the position in cell cycle from which the cells committed apoptosis can be observed (Figure 7.3). Files APO.008 & .009 show data from BAF3 cells.


Figure 7.3. A murine haemopoetic cell line, BAF3, incubated for 16 h without IL3. Cells were fixed in 1% paraformaldehyde at 0C for 15 min followed 70% ethanol. After washing in Tris-buffered saline, they were incubated with Tdt and biotin-dUTP followed by fluorescein-streptavidin and counter-stained with 20 g/ml PI. Data gated on a cytogram of DNA-peak versus DNA-area. Green fluorescence on PMT2 LOG, red fluorescence (PI/DNA) on PMT4. Cells prepared by Simone Detre, Royal Marsden Hospital NHS Trust, London. Coulter Elite with argon-ion laser giving 200 mW at 488 nm. File: APO.009.


Data obtained by Kelvin Cain from nuclei from rat hepatocytes are shown in files APO.010 - APO.015. These nuclei from normal cells are a good model for studying the DNA degradation observed in apoptotic cells (Cain et al., 1995). When isolated nuclei are incubated in culture, depending on the concentrations of Ca++ and Mg++ ions present, different nucleases are activated. In the presence of Mg++ breaks are introduced into the DNA at 50 kbp intervals. If Ca++ is added, the DNA is degraded at the internucleosomal linkers to give oligomers of 200 bp. The nuclei were fixed in paraformaldehyde and ethanol. They were incubated with Tdt in the presence of digoxygenin-dUTP and then with fluorescein-anti-digoxygenin. The data from the negative control are in file APO.010. The data in files APO.011 and 12 are from nuclei which had been incubated with 4 mM and 8 mM Mg++ respectively. Upon addition of 200 M Ca++, there was a dramatic increase in green fluorescence due to increased DNA strand breakage (files APO.013 and 14). Ca++ on its own does not induce nuclease action (file APO.015). Note that the technique is sensitive enough to detect strand breaks at 50 kbp intervals (files APO.011 and .012).

 The Tdt method has been used to study the induction of apoptosis in tumours from patients undergoing chemotherapy (Gorczyca et al., 1993). Files APO.016 - 019, which were supplied by Frank Traganos, The Cancer Research Institute, New York Medical College, contain data from patients with acute myeloid leukaemia. They were recorded on a Becton-Dickinson FACScan. The data were pre-gated on light scatter and only two parameters (DNA on FL3 and green fluorescence on FL1) were listed.

 The method can also be used to investigate apoptosis in conventionally prepared tissue from the histopathology laboratory, that is, tissue which has been fixed in formalin and embedded in paraffin wax. File APO.020, show data from a transplantable murine tumour, PC6. The nuclei were extracted from 50 m thick sections according to the method first described by Hedley et al. (198) (See Chapter 4.4). The extracted nuclei were incubated with Tdt and biotin-dUTP as described above. The low percentage of labelled cells accorded with the number of apoptotic cells scored by examining a conventionally stained section of the tumour. A fine needle aspirate (FNA) taken from a different transplant of the PC6 tumour showed a similar percentage of labelled cells (File APO.021; cells prepared by Simone Detre, Royal Marsden NHS Trust, London). In the cells obtained by FNA, the unstained cells had two levels of green fluorescence; I have no idea why.

 Files APO.022 and .023, show data from xenografts of a human breast carcinoma cell line, MCF7. Analysis of file APO.020, will show that the Tdt assay labelled a high percentage of nuclei. These tumours contain large necrotic areas and, by using ISEL to label tissue sections, it was clear that the bulk of these cells came from necrotic areas. Heating the isolated nuclei at 75C for about 30 min before running the ISEL procedure selectively removed most of the labelled cells (File APO.23). It is possible that the necrotic cells were being removed by heat leaving the non-necrotic nuclei intact. This possibility was under active investigation at the time this section was written.

 7.3.4 Methods relying on a change in membrane permeability

 The plasma membrane shows a variety of changes during apoptosis. One such change is an increase in the permeability of the membrane, Two methods are described below which exploit this change to quantify apoptotic cells. These methods will work with some cells but not others, The reason for this is not understood.

 The advantage to these methods is that they are applied to unfixed cells. Cells with a damaged plasma membrane can be excluded by either gating on light scatter or by adding PI and gating for negative red fluorescence. Apoptotic and pre-apoptotic cells can be sorted for further study. The Hoechst/PI method

 Cells are incubated at 37C with 1 g/ml Hoechst 33342 for 5-10 min depending on the type of cell under study. The apoptotic cells take up the dye faster and fluoresce bright blue compared to the dim blue fluorescence of the pre-apoptotic cells. Figure 7.4 shows an example of a murine haemopoetic cell line, BAF3, incubated for 16 h without IL3 (File APO.024). A histogram of blue fluorescence gated on blue positive, red negative fluorescence (R2 in the Figure) will show that about 10% of the 'viable' cells are apoptotic (viable in this context is referring to cells with an intact plasma membrane). Inspection of a cytogram of RALS versus FALS will reveal that the cells taking up PI can be distinguished by light scatter alone and that the apoptotic cells have the same light scatter as the pre-apoptotic population. Analysis of cells grown in the presence of IL3 is shown in File APO.025.

Figure 7.4 A murine haemopoetic cell line, BAF3, incubated for 16 h without IL3. Labelled for 10 min at 37C with 1 g/ml Hoechst 33342. 5 g/ml PI added just before analysis. Display of blue (Hoechst) versus red (PI) fluorescence. Region 2 defines the 'viable' cells. The apoptotic cells have been coloured blue in the display. Coulter Elite with argon-ion laser tuned to give 100 mW UV. File: APO.024. The YOPRO-1 method

 This method is similar to the Hoechst/PI method. YOPRO-1 is a cyanine based dye which binds to DNA. It is excited by blue light and fluoresces green and is produced by Molecular Probes (Oregon, USA). Further information can be obtained from their catalogue. This method is again illustrated using the cell line, BAF3. The green fluorescence of the YO-PRO dyes was recorded using a logarithmic amplifier. Files APO.26 & 28 contain data from cells incubated without IL3 for16h; files APO.27 & 29 data from control cells. The cells in APO.26 & 27 were incubated with YOPRO-1 alone; in APO.28 & 29, PI was added before analysis. Display a cytogram of RALS versus FALS. Two clusters are evident; that with the lower FALS, higher RALS is from cells with a damaged plasma membrane. Gate on the cluster of viable cells and then display a histogram of log green fluorescence. The apoptotic cells have higher green fluorescence. Note that there are a group of cells which have high RALS, low FALS and have not stained with either PI or YOPRO-1. These are presumably dead cells whose DNA has been nearly completely degraded. A word of caution

 Damage to the plasma membrane, which falls short of permitting ingress of PI, can allow dyes such as Hoechst 33342 and YOPRO-1 to be taken up more rapidly. This phenomenon is illustrated in File APO.30, which contains data from a murine leukaemia cell line, L1210, incubated with glyceryl trinitrate and then incubated with 1 g/ml Hoechst 33342 for 8 min. The treated cells took up the dye more rapidly than untreated cells (file APO.31); the blue fluorescence gave a good DNA histogram. A cytogram of blue versus red fluorescence demonstrates that the bright blue cells are dying as evidenced by the movement of cells towards the blueweak, red+ve cluster. Other methods (DNA gel electrophoresis, DNA histogram, morphology) failed to produced evidence of apoptosis (data not shown). These data underline the importance of using morphology to identify apoptotic cells before selecting a flow cytometric method of quantifying them. This file was supplied by Andrew Webb, Institute of Cancer Research, Sutton, England.

 7.3.6. Spectral shift analysis using Hoechst 33342

 When cells are incubated with Hoechst 33342, as the amount of dye bound to the DNA increases, the emission spectrum shifts to longer wavelengths. This phenomenon is best observed by recording the ratio between blue and orange fluorescence (for the orange, use the filter set for PE) (see Chapter 4). In a mixture of normal and apoptotic cells, the apoptotic cells may show a greater shift. This difference is probably caused by the greater uptake of Hoechst 33342 into the apoptotic cells. The effect is demonstrated in Files APO.032 - .037. BAF3 cells were incubated with 10 g/ml Hoechst 33342 for different times. Display a cytogram of orange versus blue fluorescence and observe how the cluster of apoptotic cells moves relatively further towards the orange as the time of incubation increases. A change in the emission spectrum in both normal and apoptotic cells can also be observed by displaying a histogram of the blue/orange ratio gated on the viable cells.


Figure 7.5. A murine haemopoetic cell line, BAF3, incubated for 16 h without IL3. Labelled for 5 min at 37C with 10 g/ml Hoechst 33342. Display of blue versus orange fluorescence. The normal cells have been coloured red, the apoptotic cells blue, the 'dead' cells green and the debris pink. Coulter Elite with argon-ion laser tuned to give 100 mW UV. File: APO.032.

7.3.5 Light scatter

 Cells generally shrink during apoptosis and will, in some cases, scatter less light in a forward direction. Analysis of the data from BAF3 cells will confirm that there is little change in the light scatter of apoptotic cells with intact plasma membranes. When the cells undergo secondary necrosis at a late stage in apoptosis(that is, the plasma membrane becomes ruptured), they show marked changes in both RALS and FALS.

 Apoptotic rat thymocytes scatter less light over a narrow angle in the forward direction but the amount of light scattered relative to normal cells is highly dependent on the angle of light collection. We have investigated this effect using UV. On the old Ortho Cytofluorograph, we could obtain go separation between normal and apoptotic cells. On the Becton-Dickinson FACS Vantage and the Coulter Elite, using the standard configuration, little separation was achieved. However, the two types of cell could be resolved if the narrowest angle of collection of FALS was used (Ormerod et al., 1995). Typical data are shown in files APO.038 & .039. Gate out the PI +ve cells and then display blue fluorescence versus forward scatter. Using the FALS mask which selected light of the smallest angle (I estimated that the light was collected over 1.5 to 4) gave good separation (file APO.038) as opposed to the normal FALS mask (angle of collection, 10 to 18) (file APO.039) (Figure 7.6).

Figure 7.6. Thymocytes from an immature rat stained with 1 g/ml Hoechst 33342 for 7 min. 5 g/ml PI was added before analysis. FALS collected over 1.5 - 4. Cells prepared by Xiao-Ming Sun, MRC Toxicology Unit, Leicester and analysed by Mark Cheetham, Coulter Ltd, Luton, UK, on a Coulter Elite with a He-Cd laser producing 15 mW UV. File: APO.038.

 7.4 References

 Cain, K., Inayat-Hussain, S.H., Wolfe, J.T., Cohen, G.M. (1994). DNA fragmentation into 200-250 and/or 30-50 kilobase pair fragments in rat liver nuclei is stimulated by Mg2+ alone and Ca2+/Mg2+ but not by Ca2+ alone. FEBS Lett. 349, 385-91.

Darnzynkiewicz, Z., Li, X. and Gong, J. (1994). Assays of cell viability: discrimination of cells dying by apoptosis. In Flow Cytometry. Darzynkiewicz, Z. Robinson, J.P. and Crissman, H.A. (eds.) Methods in Cell Biology, 41, Academic Press, Inc., San Diego. pp15-38.

 Fraker, P.J., King, L.E., Lill-Elghanian, D. and Telford, W.G. (1995). Quantification of apoptotic events in pure and heterogeneous populations of cells using flow cytometry. In Cell Death. Schwartz, L.M. and Osborne, B.A. (eds.) (1995). Methods in Cell Biology, 46, Academic Press, San Diego. pp 57-76.

 Gorczyca, W., Bigman, K., Mittelman, A., Ahmed, T., Gong, J., Melamed, M.R. and Darnzynkiewicz, Z. (1993). Induction of DNA strand breaks associated with apoptosis during treatment of leukaemia. Leukaemia 7, 659-670.

 Homburg, C.H., de Haas, M., von dem Borne, A.E., Verhoeven, A.J., Reutelingsperger, C.P., Roos, D. (1995). Human neutrophils lose their surface Fc gamma RIII and acquire Annexin V binding sites during apoptosis in vitro. Blood. 85, 532-40.

 Idziorek, T., Estaquier, J., De Bels, F. Ameisen, J-C. (1995) YOPRO-1 permits cytofluorometric analysis of programmed cell death (apoptosis) without interfering with cell viability. J. Immun. Meth. 185, 249-258.

 Ormerod, M.G. (1994). Further applications to cell biology. In Flow Cytometry. A Practical Approach. Ormerod, M.G. (ed.). 2nd edition. IRL Press at Oxford University Press, Oxford. pp. 261-273.

 Ormerod, M.G., Paul, F., Cheetham, M. and Sun, X-M. (1995). Discrimination of apoptotic thymocytes by forward light scatter. Cytometry 21, 300-304.

 Schwartz, L.M. and Osborne, B.A. (eds.) (1995). Cell Death. Methods in Cell Biology, 46, Academic Press, San Diego.

 Sherwood, S.W. and Schimke, R.T. (1995). Cell cycle analysis of apoptosis using flow cytometry. In Cell Death. Schwartz, L.M. and Osborne, B.A. (eds.) (1995). Methods in Cell Biology, 46, Academic Press, San Diego. pp 77-98.

Back to teaching materials

Home PageTable of ContentsSponsorsE-Mail ArchiveWeb Sites

CD-ROM Vol 3 was produced by Monica M. Shivelyand other staff at the Purdue University Cytometry Laboratoriesand distributed free of charge as an educational service to thecytometry community. If you have any comments please direct themto Dr. J. Paul Robinson, Professor & Director, PUCL, PurdueUniversity, West Lafayette, IN 47907. Phone:(765) 494-0757; FAX(765) 494-0517; Web,EMAIL