Characterization of cells with different mitochondrial membrane potential during apoptosis




Until now, the simultaneous analysis of several parameters during apoptosis, including DNA content and mitochondrial membrane potential (ΔΨ), has not been possible because of the spectral characteristics of the commonly used dyes. Using polychromatic flow cytometry based upon multiple laser and UV lamp excitation, we have characterized cells with different ΔΨ during apoptosis.


U937 cells were treated with the flavonoid quercetin (Qu) and stained with JC-1 to detect ΔΨ, propidium iodide (PI) for cell viability, Hoechst 33342 for DNA content, Annexin V conjugated with Alexa Fluor-647 for detection of phosphatidilserine (PS) exposure, marker of early apoptosis, or Mitotracker Deep Red for the determination of mitochondrial mass.


Treatment with Qu provoked the onset of three cell populations with different ΔΨ: (1) healthy cells, with normal ΔΨ, DNA content and physical parameters, high mitochondrial mass, PI- and Annexin V-negative; (2) cells with intermediate ΔΨ and normal DNA content, but with physical parameters typical of apoptotic cells and low mitochondrial mass; most of them were PI+ and Annexin V+; (3) cells with collapsed ΔΨ that had low mitochondrial mass and were Annexin-V+, PI+; half of them showed diminished DNA content. Similar results, i.e. the presence of cells with intermediate ΔΨ, were observed in other models of apoptosis.


During Qu-induced apoptosis, loss of ΔΨ, PS exposure, and decrease of mitochondrial mass are early events that precede permeability to PI and loss of DNA. Populations of cells with different ΔΨ, as revealed by flow cytometry after JC-1 staining, differed also for other parameters associated to apoptosis. Thus, the simultaneous analysis of several parameters by polychromatic flow cytometry permits a better identification of many stages of cell death, and, more in general, allows to evaluate the eventual heterogenic sensibility of the population under study to a given compound. © 2005 Wiley-Liss, Inc.

Characterizing functional changes that occur in different cellular compartments during programmed cell death/apoptosis are of universal interest. Many techniques have been developed to analyze this process, and indeed, the quantification of cells undergoing apoptosis is also important to determine the efficacy of new drugs, such as anticancer agents or antiviral compounds. Mitochondria represent key organelles for the cell survival, and their role in programmed cell death is known since several years (1–3). Mitochondrial alterations during cell death have been widely described (4, 5); even if cytochrome c release or caspase activation, and consequent DNA degradation can be independent of loss of mitochondrial membrane potential (ΔΨ) (6), we accept that this event is a hallmark of apoptosis (7).

As far as ΔΨ is concerned, taking advantage of the possibility to use different sensitive dyes, it is nowadays simple to identify cells with high or low ΔΨ. The dye JC-1, excited by the argon laser present in all flow cytometers emits in FL-1 (monomers, typical of low ΔΨ) and FL-2 (aggregates, whose formation is due to a high ΔΨ) (8, 9). However, its spectral characteristics can create some difficulties when used simultaneously with other dyes excitable by such a laser. Thus, until now, the study of changes in DNA content and ΔΨ in the same cell during apoptosis have been difficult if not impossible, and all data, including ours (10), have been obtained by separate measures.

We describe here a 5-color staining that allows the simultaneous determination of four independent parameters that can change during apoptosis: ΔΨ, exposure of phosphatidilserine (PS) on plasma membrane's external leaflet, cell viability (in terms of lack of permeability to propidium iodide, PI), and DNA content. In addition, we used a 4-color staining where detection of PS exposure was replaced by analysis of mitochondrial mass. We took advantage of a flow cytometer equipped with three lasers and a UV lamp, which can collect up to 12 fluorescences. By inducing apoptosis in the myeloblastic cell line U937 with flavonoid quercetin (Qu) (11), we found that apoptotic populations with different ΔΨ were also discordant for other parameters, such as DNA content and PI permeability, but not for PS exposure. This was well distinguishable after short time of Qu incubation, while, at prolonged times, cells with different ΔΨ were mostly similar. This indicates that apoptosis proceeds independently from loss of ΔΨ, which is however, the first, irreversible event detectable by the methodology used in this article. Moreover, by the simultaneous analysis of several parameters in the same cell, it is possible to characterize the kinetics of programmed cell death and to determine early and late alterations of various relevant phenomena.


Dyes and Chemicals

JC-1 (5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazolcarbocyanine iodide), Hoechst 33342 and Alexa 647-conjugated Annexin V, and Mitotracker Deep Red were from Molecular Probes (Eugene, OR); PI, Qu (3,3′,4′,5,7-pentahydroxyflavone), stavudine, doxorubicin, and daunomycin were from Sigma Immunochemicals (St. Louis, MO).

Cell Culture and Apoptosis Induction

Cell lines of different origins (monocytic: U937 and HL60, lymphocytic: A301 and CEM) and peripheral blood mononuclear cells (PBMCs) were plated at a density of 0.5 × 106 cell/ml in RPMI-1640 supplemented with 10% heat-inactivated (56°C for 30 min) fetal bovine serum (FBS), 2 mM L-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin (all from Gibco Laboratories, Carlsbad, CA,) in humidified atmosphere (5% CO2) at 37°C. The same conditions were used to perform a kinetic analysis where U937 cells were cultured for different times (2, 4, 6, 12, 16, and 24 h). Apoptosis was induced by incubating cells in multi 6-well plates or in 25 cm2 flasks with various agents such as Qu (10, 50, and 100 μM), doxorubicin (1 μM for 24 h), daunomycin (1 μM for 24 h), stavudine (10 μM, for 5 days), staurosporine (STS, 5 μM for 4 h). After different times of incubation, cells were collected, washed with PBS, and stained as described later.

Morphological Analysis of Qu-Treated U937 Cells

U937 cells were collected after 16 h of incubation with 100 μM Qu, spun down on a glass slide and stained with May–Grünwald Giemsa, according to standard methods.

Analysis of Mitochondrial Membrane Potential

0.5 × 106 cells were resuspended in RMPI 1640 supplemented with 10% FBS and stained with 2.5 μg/mL JC-1, as previously described (3, 8–10). An incubation of 10 min at room temperature was followed. Cells were then washed and the pellet was resuspended in PBS for flow cytometric analysis.

Flow Cytometric Analysis of Apoptosis and Cell Viability

U937 (0.5 × 106) cells were resuspended in RPMI 1640 and stained with 5 μM Hoechst 33342 for 30 min at 37°C. Cells were then centrifuged at 200g for 5 min at room temperature to remove unbound Hoechst, resuspended in RPMI 1640 supplemented with 10% of foetal bovine serum (FBS), and stained with JC-1 as mentioned earlier. Cells were then washed with PBS, resuspended in 195 μL of Annexin binding buffer (10 mM HEPES, 140 mM NaCl, and 2.5 mM CaCl2, pH 7.4); 5 μL of Alexa 647-conjugated Annexin-V were added. Cells were then incubated for 10 min at room temperature, washed with 800 μL of Annexin binding buffer as mentioned earlier and resuspended in 1.5 mL of Annexin binding buffer. To analyze cell viability in cells treated as described earlier, 2 min before acquisition, PI was added at a low concentration (0.25 μg/mL) to avoid resonance energy transfer with Hoechst 33342 (12). In the case of the simultaneous analysis of mitochondrial mass, DNA content, and ΔΨ, Alexa 647-conjugated Annexin-V was replaced by 100 nM Mitotracker Deep Red, whose staining was performed in RPMI 1640 for 30 min at 37°C, together with Hoechst.

All samples were analyzed using a Cyflow ML (Partec GmbH, Münster, Germany), equipped with a solid state laser (emitting at 488 nm, 200 mW, kept at 50 mW), a UV Mercury lamp HBO (100 long life, 100 W), a red diode laser (635 nm, 25 mW), a Nd:YAG laser (532 nm, 50 mW, not utilized in this study), and a CCD camera. Emission of JC-1 monomers was detected in FL1 using a 520/30 nm bandpass filter; that of JC-1 aggregates in FL2 using a 590/30 nm bandpass filter; emission of PI was collected in FL12 using a 770 long pass filter; that of Hoechst 33342 was collected in FL8 using a 455/40 bandpass filter; that of Alexa 647-conjugated Annexin-V or Mitotracker Deep Red was collected in FL4 using a 670/40 band pass filter. Data were analyzed by using Partec Flomax 3.0 and WinMDI softwares, both under Windows XP.

Detection of DNA Fragmentation by Agarose Gel

Detection of DNA fragmentation was performed as described (13). In brief, 2 × 106 cells were resuspended in a buffer containing 10 mM EDTA, 50 mM TrisHCl pH 8.0, 0.5% sodium lauril sarcosinate, and 0.5 mg/ml Proteinase K; an incubation at 50°C for 1 h is followed. Cells were then treated with for the same time with 10 μL of a solution containing 0.5 mg/ml RNAse A. After incubation, 10 μL of 70°C melted loading solution (1% low melting agarose, 40% sucrose, 0.25% bromophenol blue, 10 mM EDTA) were added to cell extracts that were loaded in a 2% agarose gel.

Statistical Analysis

Non-parametric ANOVA for repeated measures was used to compare treated versus control samples. Mean differences were considered significant when P < 0.05.


Quercetin Induces Apoptosis in U937 Cell Line

Qu can induce apoptosis in a wide variety of cancer cells and cell lines (14–16), including the U937 myeloblastic cell line (11). Our first aim was to investigate which alterations occurred after exposure at different doses of Qu, ranging from 10 to 100 μM. Preliminary experiments showed that a 16-h incubation was a good choice to have a detectable amount of cells in early and late apoptosis. Fig. 1 shows that treatment of U937 cells with 50 and 100 μM quercetin induced loss of ΔΨ and PS exposure in numerous cells (about 50%), while membrane permeability to PI and DNA degradation occurred only in a smaller amount of cells. Differently, 10 μM quercetin was not able to induce any significant alteration. Confirmation of the apoptotic type of cell death derived from the analysis of DNA fragmentation (Fig. 1b), and from the morphology of the cells, that revealed the typical presence of cell shrinking, chromatin condensation, and cell fragmentation (Fig 1c).

Figure 1.

Quercetin induces apoptosis in U937 cell line. (a) Percentages of U937 cells with depolarized mitochondria, exposed PS, altered plasma membrane permeability (PI+), and decreased DNA content after treatment with different doses of Quercetin (10–100 μM) for 16 h. Values are indicated as mean ± SD of four independent experiments. * = P < 0.05; ** = P < 0.01, compared with that of control (CTRL). (b) Fragmentation of DNA in cells treated for 16 h with different doses of Qu. (c) Upper panel: control cells; lower panel: cells treated for 16 h with 100 μM Qu, showing the typical morphological alterations and nuclear fragmentation occurring during apoptosis. Bar = 10 μm.

Characterization of Cells with Different ΔΨ

JC-1 staining of Qu-treated U937 cells allowed us to identify three populations with different ΔΨ (Fig. 2a–2c): a first population with high ΔΨ that we have painted in red, a second with intermediate ΔΨ indicated in blue, and a third indicated in green, where ΔΨ was collapsed.

Figure 2.

Simultaneous detection of several functional parameters in living cells by polychromatic flow cytometry. Control, untreated U937 cells (2a), and cells treated with 50 μM Quercetin for 16 h (2b). Cells with high ΔΨ (H) are indicated in red, cells with intermediate ΔΨ (I) are indicated in blue, cells with low ΔΨ (L) are indicated in green; numbers refer to the percentage of cells. Cellular debris was eliminated by an electronic gate on normal and apoptotic cells in FSC vs. SSC. Three different gates were subsequently done on JC-1 stained cells to separately analyze the characteristics of cells with different ΔΨ. Dot plots combining Annexin V staining, PI permeability, and DNA content illustrate single populations. Black arrows indicate cell populations with decreased DNA content and not permeable to PI. (2c) analysis of mitochondrial mass with Mitotracker Deep Red, mitochondrial membrane potential (JC-1), and DNA content (Hoechst) in control (CTRL) or cells treated with 50 μM Qu for 16 h (Qu50). Dot plots combining mitochondrial mass and DNA content illustrate single populations with high, intermediate, and low ΔΨ.

An original 5-color flow cytometric assay allowed us to measure ΔΨ, DNA content, PS exposure, and viability in the same cell. Fig. 2a (control) and 2b (Qu-treated) show that U937 cells with high ΔΨ represent healthy cells, with normal morphology (as indicated by FSC and SSC) and plasma membrane integrity (negativity to PI); they do not expose PS and have normal DNA content. Cells with intermediate ΔΨ displayed altered physical parameters (those typical of apoptotic cells, i.e. low FSC and high SSC), PS exposure, altered plasma membrane permeability (as detected by entry of PI into cells), and some of them, the minority, condensed/fragmented DNA, as indicated by a decrease in Hoechst fluorescence. Concerning cells with low ΔΨ, they showed altered physical parameters, PS exposure, and most of them were permeable to PI. These also showed decreased DNA content. Inside this population, we detected certain cells with lower DNA content, but negative for PI (see DNA content vs. PI permeability, Figs. 2a and 2b, indicated by arrow). Loss of ΔΨ was also accompanied by a decrease in mitochondrial mass (Fig. 2c), but we did not detect any difference between cells with intermediate and low ΔΨ.

The presence of a population with intermediate ΔΨ was found also in some models of apoptosis, such as HL60 cells treated with STS, U937 cells treated with stavudine, human lymphocytes treated with DNA damage-inducing agents such as doxorubicin and daunomycin (Figs. 3a–d), but not in others (Figs. 3e and 3f). This suggests that the formation of a cellular population with intermediate ΔΨ is not a general feature of apoptosis, but can occur in some widely used models of cell death.

Figure 3.

Detection of cells with intermediate ΔΨ in different models of apoptosis. A population with intermediate ΔΨ can appear in some models of apoptosis, and in particular: (a) HL60 treated with staurosporine, (b) U937 treated with stavudine, (c) human PBMCs treated with doxorubicin or (d) daunomycin, but not in others: (e) human CEM cell line treated with stavudine, (f) PBMCs treated with Qu. Note that in the latter cases ΔΨ was completely collapsed. Bars separate populations with high, intermediate, and low ΔΨ (a–d), or with high and low ΔΨ (e and f).

Analysis of ΔΨ Allows the Identification of Early and Late Apoptotic Cells After Short-Time Exposure to Quercetin

To better characterize cells with different ΔΨ during apoptosis, we performed a kinetic analysis where U937 cells were cultured for different periods with or without Qu. Two or 4 h of treatment were not sufficient to trigger apoptosis in U937 cells, even at high concentrations of flavonoid (not shown). Apoptosis was detected 6 h after treatment (Fig. 4, row b), as assessed by the appearance of AnxV+ cells with decreased ΔΨ (blue and green), while there were no changes in plasma membrane permeability and DNA content compared with those of the control. At this time, cells with intermediate membrane potential were extremely different from those cells with collapsed ΔΨ. In fact, cells with intermediate ΔΨ (indicated in blue) were all positive for Annexin V staining and displayed altered physical parameters, but most of them were negative for PI and had a normal DNA content. In contrast, cells with collapsed ΔΨ (indicated in green) showed positivity for Annexin V and were PI negative, but about 50% of them had hypodiploid DNA. At 12 h (Fig. 4, row c), we found an increase in cells with collapsed ΔΨ (indicated in green in JC-1 monomers-JC-1 aggregates dot plot); half of them incorporated PI, but most of them had fragmented DNA. At 24 h (Fig. 4, row d), populations with intermediate and low ΔΨ differed only on the basis of DNA content. In fact, at this time, cells with intermediate ΔΨ maintained PS exposure and, like those with collapsed ΔΨ, became permeable to PI; however, they were still characterized by a normal DNA content.

Figure 4.

Kinetic analysis of Qu-induced apoptosis. Representative example of a 5-colour – 7 parameter kinetic analysis of apoptosis after incubation with 50 μM Quercetin for 6 h (row b), 12 h (row c), and 24 h (row d). Control is shown in row a. Cells with high ΔΨ are coloured in red, those with intermediate ΔΨ are colored in blue and those with collapsed ΔΨ are colored in green. Numbers indicate the percentages of cells belonging to subpopulations indicated above.


In this article, we show that, in some models of apoptosis, a population of cells with intermediate ΔΨ can appear. Multiparametric analysis of this population during Qu-induced apoptosis revealed that they have different characteristics from those with high or low ΔΨ. This was well detectable at short times after apoptotic engagement, when Qu induces a relevant oxidative stress (17).

We took advantage of a multilaser flow cytometer that allowed us to detect several independent apoptotic parameters in the same cell: cells with different ΔΨ were investigated for Annexin V positivity, plasma membrane permeability to PI and DNA content. Following JC-1 staining, we identified three populations with different ΔΨ that belong to different stages of the apoptotic process, because they are discordant for other parameters associated to apoptosis. Short times after Qu treatment (6 h), cells with high ΔΨ did not display alterations in physical parameters, exposure of PS, cell viability, and DNA fragmentation/condensation. On the contrary, cells with intermediate ΔΨ were Annexin V+ and had low FSC, but other parameters were still similar to those of the control; this suggests that they could be early apoptotic. Moreover, in these cells, loss of ΔΨ was not accompanied by changes in DNA content (measured by Hoechst fluorescence); thus, changes in ΔΨ seem to precede DNA fragmentation. Hypodiploid DNA content and PI permeability, in addition to PS exposure, were found in cells with fully collapsed ΔΨ. Inside these cells, we also detected a subpopulation of cells that, despite the presence of fragmented DNA, was PI negative. These could be either cellular fragments that are not able to retain PI, or apoptotic bodies that have a very low DNA content, and thus are not fluorescent.

Prolonging the time of treatment with Qu resulted in an increase in cells with intermediate or collapsed ΔΨ. However, despite differences in ΔΨ, cells with intermediate ΔΨ became permeable to PI and they were mostly similar to those with collapsed ΔΨ, except for DNA content. This indicates that cells with intermediate ΔΨ switched to a more advanced apoptotic stage and it suggests that, in this model of apoptosis, loss of ΔΨ seems to be independent from other events associated to apoptosis, because PI permeability and, in part, DNA degradation occurred also in cells with incomplete collapse in ΔΨ. This raises many questions about the role of ΔΨ during apoptosis. It is known that maintaining of ΔΨ is indispensable for the production of ATP which is also required for cell death induction. Recent data suggest that a tight link between glycolysis and apoptosis exists because glucokinase and BAD reside in a mitochondrial macromolecular complex (18). Thus, alterations, but not collapse, of ΔΨ could be important to maintain, at least in part, ATP production that is necessary for the formation of the apoptotsome and for the activation of caspases (19, 20). Nevertheless, more studies are needed to characterize this population with intermediate ΔΨ.

The simultaneous analysis of several parameters has a relevant interest for a better understanding of complex phenomena such as apoptosis. In particular, the role of changes in ΔΨ are under investigations for several years (3, 21, 22). An impairment of mitochondrial function leading to decrease in ΔΨ and the consequent permeability transition are considered hallmarks of apoptosis (7), even if it has been demonstrated that the release of cytochrome c can be independent of mitochondrial depolarization both in a cell free system (23), and in intact cells (24). More recently, it has been shown that decrease in ΔΨ or permeability transition phenomena are per se incapable of triggering caspase activation and nuclear apoptosis (25, 26). Thus, it is possible to conceive that, even in the models in which mitochondrial function is involved, decrease in ΔΨ would be an ancillary event of the apoptotic process. Furthermore, it is not known if all mitochondria are impaired in a similar manner and to what extent, nor if alteration of all mitochondria is required to trigger cell death.

Recently, we found that certain cells do not dissipate completely their ΔΨ after staurosporine treatment (10). We observed the same phenomenon in Qu treated cells. The presence of the intermediate population could be explained by the fact that, during mitochondrial depolarization, the simultaneous presence of a low number of aggregates (orange) and a high number of monomers (green) can be present at the same time. Another possibility exists, which does not contradict what has been previously discussed, i.e. that cells do not depolarize all organelles simultaneously. In any case, further characterization of these cells is of interest for understanding their possible fate.

On the whole, our data indicate that, in this model of apoptosis, (i) loss of ΔΨ, PS exposure, and decrease of mitochondrial mass are early events and they precede permeability to PI and DNA condensation/fragmentation; (ii) populations with different ΔΨ, as revealed by flow cytometry after JC-1 staining, also differ for other parameters associated to apoptosis; (iii) the simultaneous analysis of multiple parameters with polychromatic flow cytometry allows the identification of many stages of the same phenomenon, i.e. apoptosis, and the consequent different sensibility of populations under study with the drug tested. This polychromatic approach thus allows more sophisticated analyses that can reveal the heterogeneity of the response to a given compound or stimulus.


We acknowledge Professor W. Göhde (Univ. of Münster, Germany) for helpful discussions and precious advices, and Luca Cicchetti (Space Import Export s.r.l., Milan, Italy) for continuous support.