Sequential events of apoptosis involving docetaxel, a microtubule-interfering agent: A cytometric study

Apoptosis is the most widespread form of cell death, characterised by distinguished biochemical and morphological events triggered by signalling cascades and finely regulated by the products of oncogenes, tumour suppressor genes, and apoptosis-specific genes. It has proven useful to distinguish two main phases in apoptotic cell death: induction and execution. Induction is controlled by a wide range of mechanisms that involve either surface (extrinsic) signalling and subsequent signal transduction mechanisms, or internal (intrinsic) pathways leading to the release of mitochondrial components. Both pathways converge in common events during the execution phase. These events specifically characterise apoptotic cell death and lead to the successful phagocytosis of cell corpses, still surrounded by leak-proof membranes [1, 2].

Both basic and clinically oriented studies require reliable evaluation and quantification of apoptosis. In particular, most chemotherapeutic drugs trigger apoptosis in actively dividing cells. A good deal of information about their efficacy and their therapeutic index may be obtained by the identification of cell death mode (apoptosis vs. necrosis), quantification of dead cells, and by understanding the effect of the drug on the cell cycle.

The unique changes that characterise the apoptotic process provide several features that permit the recognition and quantitation of apoptotic cell death by cytometric methods [3]. These methods have two important advantages in that they are quantitative and allow the possibility of evaluating multiple parameters, including cell cycle status. Some of them exploit specific features that are a hallmark of apoptosis, such as DNA degradation – as revealed by reduced binding of the intercalating fluorescent dye, propidium iodide [4], by the labelling of DNA strand breaks [5], or by the redistribution of membrane phospholipids, which can be detected by tagged fluorochromes [6]. A common parameter for distinguishing between apoptosis and necrosis is the integrity of the cell membrane, which blocks the entry of propidium iodide in non-permeabilised cells [7]. In the course of signal transduction that leads to apoptosis, mitochondria may release pro-apoptotic molecules, a process accompanied by loss of the transmembrane potential (Δψm) that can be detected by specific fluorescent markers [8]. The centralrole played by mitochondria in the control of both intrinsic and extrinsic apoptotic pathways is well known [9].

The aim of the present study was to determine the temporal succession of the main events that characterise taxane-induced apoptosis in a cancer cell line. Taxanes induce apoptosis by interfering with microtubule dynamics, in particular by preventing tubulin depolymerisation. Actively dividing cells are thus induced to undergo apoptosis, triggered by bcl-2 phosphorylation and involving the activation of the c-Raf-1/Ras or the p53/p21WAF1/CIP1 signalling pathway [10]. In epithelial cells, taxol induces a cyclosporin A (CsA)-sensitive decrease in ψm and a Ca²⁺ efflux from the mitochondria [11]. It has also been suggested that changes in the microtubule structure induced by taxanes may be responsible for opening the mitochondrial permeability transition pores [12].

In this and in most drug-induced apoptotic processes, the release of mitochondrial components is a requisite for the activation of caspases, which are apoptosis-specific proteases [13]. DNA fragmentation and most of the biochemical and morphological changes that characterise apoptosis then take place under the control of effector caspases. However, the order of events has seldom been investigated. We therefore evaluated a number of supposedly independent apoptotic events and their temporal succession in an attempt to determine which are actually independent from the other.

A secondary aim of the study was to compare multiple cytometric methods to study the apoptotic process and acquire information that can be used as a guideline for the choice of the most suitable parameter(s) for different experimental systems. In particular, since some methods actually assess the same event, we will discuss possible discrepancies in results. These parameters can be employed to assess the execution phase of apoptosis, which is believed to take place in a stereotyped way, independently of the apoptotic trigger [14].

Despite the considerable amount of information on apoptosis in tumour cell lines acquired from cytometry-based studies, little attention seems to have been paid to the time course of the main events that characterise the apoptotic process, and in particular those induced by taxanes.

Earlier studies established that phosphatidylserine exposure to the outer membrane leaflet occurs simultaneously with chromatin condensation [16] and precedes DNA fragmentation [17]. Other membrane changes, as shown by an increase in Merocyanin 540 binding of apoptotic cells [18–20], may not be independent of the phosphatidylserine contribution to the membrane lipid composition [21]. On the other hand, AnnV staining itself may be lacking in certain apoptotic cell populations [20]. These findings underline the need to evaluate multiple parameters when assessing cell death. A number of studies have addressed the issue of mitochondrial involvement in the apoptotic process by evaluating Δψm decrease with different dyes [22–24] inan attempt to establish a temporal – and possibly – causal relationship within the events that characterise apoptosis. It appears that the bcl-2 content of the cells may be related to the timing of mitochondrial involvement [25]. This consideration is particularly important in taxane-induced apoptosis, where bcl-2 is involved in the cell death induction mechanisms as well as in drug resistance [26–28].

Taxanes and their derivatives inhibit microtubule depolymerisation, thus preventing cell cycle completion [29] and thus causing apoptotic cell death. A similar effect is also caused by drugs that inhibit microtubule polymerisation, such as vinblastine and vincristine. Apoptosis triggered by agents that disrupt microtubule organisation is associated with bcl-2 phosphorylation and is independent of p53 involvement, at variance with apoptosis due to DNA-damaging agents [30]. Despite these differences, both microtubule disruption and DNA damage activate the intrinsic pathway of cell death, causing the release of mitochondrial components and the formation of apoptosomes. In both cases, the triggering mechanism is believed to involve changes in the amount and/or activity of anti- and pro-apoptotic members of the bcl-2 family.

Conversely, it has been shown that administration of paclitaxel – the prototype for taxane compounds – causes an almost immediate increase in cytosolic Ca²⁺, associated with a rapid decline in mitochondrial membrane potential, presumably due to the opening of the permeability transition pores (mPTPs) [11, 31]. The rapidity of this event apparently rules out the participation of mechanisms controlling microtubule assembly status, thus suggesting a pathway of death induction different from that previously reported [30]. However, it is recognised that loss of Δψm and variations in [Ca²⁺]i may be transient and do not necessarily lead to apoptosis. In particular, mPTPs have two conformations, one characterised by low- and the other by high-conductance state [32]. The opening of the former is reversible, allows the release of small molecules only – such as calcium ions – and may be involved in the rapid [Ca²⁺]i changes described by other authors [11]. In contrast, late changes in Δψm, such as those observed by us 16 hours after drug treatment, are likely to involve the high-conductance mPTP, which is responsible for the release of apoptosis-inducing compounds. The delayed change in Δψm is not apparently affected by prior exposure to CsA (table 3). However, it is obvious that cell cycle is already deeply affected as early as 6 hours after the end of drug treatment (Fig. 2). Thus, a more realistic view puts the mitochondrial involvement in apoptosis downstream of paclitaxel-induced G₂/M arrest. Within this cell-cycle phase, microtubule impairment may activate mTOR kinase, thus affecting bcl-2 phosphorylation and its subsequent proteasome-dependent degradation [33]. The data on the action/activity of CsA presented in the present study are only preliminary and the role played by Ca²⁺ variation in this process will be the object of future investigation.

Whilst the time course of early events has yet to be defined, its eventual clarification will provide new insight into the mechanism whereby taxanes induce cell death. Moreover, other well-studied events that belong to the execution phase of apoptosis have not yet been placed in a precise time setting. These include phosphatidylserine exposure to the outer cell surface, DNA cleavage, and loss of membrane integrity. Changes in surface glycosylation are less well understood and have been included here because they currently being studied by the authors (Marini et al., manuscript in preparation). Our results suggest that these changes are not peculiar to a given cell type and potentially represent a new marker of apoptosis. They would seem to occur about the same time as PS exposure to the outer surface and may be related to it.

As described in the present work, apoptotic events in the HT1376 bladder cancer cell line treated with DOC would appear to occur in the following order: 1) cell cycle arrest; 2) loss of mitochondrial membrane potential; 3) cell shrinkage and decrease in granularity; 4), DNA fragmentation (hypodyploid peak and electrophoretic laddering); 5) and 6) change in syalilation of surface proteins and loss of surface phospholipid asymmetry; 7) loss of membrane integrity (cells become PI+); 8) further DNA fragmentation (TUNEL). The above outlined sequence of events takes into account intrinsic limitations in cytometric methods, which include a lesser degree of sensitivity in comparison with biochemical determinations and the variability in drug response within an asynchronous cell population. In fact, it must be stressed that all the cells do not die simultaneously after drug treatment because death is dependent on factors such as the cell cycle status and the individual energy level. Moreover, in most artificial systems, e.g. culture flasks, cells dying by apoptosis are not phagocyted by competent cells and eventually lose their membrane potential and undergo secondary necrosis. Thus, within an asynchronously growing cell population, cycling cells, cycle-arrested cells, cells undergoing apoptosis and cells already undergoing secondary necrosis can all be found and will be sending out conflicting signals that require interpretation.

Data on DOC-treated U937 myelomonocytic cell lines were reported in the present work for comparison purposes and were not as complete as those obtained on HT 1376 bladder cancer cell line. Nevertheless, they suggest that Δψm is an early event in both cell types, that bear though different they may be, and that MAA marks the same cell subset as AnnV.

In permeabilised cells, DNA fragmentation results in decreased PI staining (i.e. cells appear to have a hypodiploid DNA content) [7]. Accordingly, this event appears to occur between 24 and 48 hours after drug administration. However, the onset of DNA fragmentation may occur earlier, as cells with duplicated DNA join the sub-G_0/1 peak after passing through the S and G_0/1 peaks. On the other hand, it is possible that, over time, the progressive increase in DNA fragmentation causes a marked loss in PI stainability, causing the hypodiploid peak to be underestimated, as suggested by comparing data with results obtained from unpermeabilised cells and from TUNEL staining, another index of DNA fragmentation. This technique, where 3' OH termini – resulting from DNA cleavage – are labeled with FITC-conjugated nucleotides in a reaction utilising exogenous terminal deoxynucleotide transferase (TdT), is apparently less sensitive than a decrease in PI stainability, possibly because it requires more extensive DNA cleavage. However, for longer culture times, TUNEL staining appears to give a better estimate of the percentage of apoptotic cells with respect to the evaluation of the hypodiploid peak. In fact, 96 hours after DOC treatment, the percentage of TUNEL+ cells was much higher than that of cells in the sub-G_0/1 peak, and approximated that of AnnV+ cells.

The characteristic ladder revealed by gel electrophoresis of DNA from DOC-treated cells demonstrates that chromatin fragmentation did not occur randomly, but rather at internucleosomal sites. Neither high molecular weight DNA cleavage (as occurs during the initial stages of apoptosis), nor aspecific DNA fragmentation (following necrotic cell death), produce such a specific pattern of electrophoretic migration. The same applies to the formation of the hypodiploid peak [34].

Cell surface modifications were observed 24 hours after DOC treatment and over time extended to other cells (table 1). Due to its strong affinity to negatively charged phospholipids, AnnV binds phosphatidylserine as it becomes exposed to the outer cell surface, a common feature of cells undergoing apoptosis. Variations in cell surface carbohydrate composition can be identified by changes in lectin-mediated recognition. Apoptotic cells may change the oligosaccharide composition of their glycocalix, thus displaying a differential binding to some lectins. In particular, Maakia Amurensis lectin (MAA), which preferentially recognises galactose residues bound to α 2,3 sialylic acid [35], was found to bind to apoptotic cells rather than to their "healthy" counterpart (Marini et al, manuscript in preparation). The percentage of cells showing variations in the glycosylation of surface molecules initially increased more sharply than that of cells displaying phosphatidylserine on the outer membrane leaflet. However, over time the percentage of MAA+ cells did not reach that of AnnV+ cells. Studies are ongoing to further the understanding of these variations in surface syalilation and of their relevance in apoptosis. As already pointed out, the cell population appeared to acquire both MAA and AnnV markers before losing its membrane integrity (i.e. before becoming permeable to PI). In addition to such information, the percentages of apoptotic PI- and PI+ cells observed shows that the increase in TUNEL positivity (i.e. progression in DNA degradation) also occurred in cells that had lost membrane integrity.

Ninety-six hours after drug treatment, the near totality of cells displayed a decrease in transmembrane mitochondrial potential Δψ_m that accompanied the efflux of pro-apoptotic molecules from the mitochondrion, as evaluated by the cationic fluorochrome JC-1 (table 1). This suggests that almost all of the DOC-treated cells eventually undergo apoptosis. Moreover, it supports the view hypothesis that mitochondrial events occur very early during cell culture and pave the way for both DNA cleavage and surface modifications. Furthermore, cytoskeleton modifications that cause cell shrinkage and loss of granularity (Fig. 4) may be subsequent to a loss of Δψ_m, as suggested elsewhere [36].

In the present paper we compared cytofluorimetric with non cytofluorimetric methods of evaluation of cell death. The same conclusions about the cytotoxic and cytostatic effects of the drug were reached using the SRB assay, the dye-exclusion test and the cytofluorimetric evaluation of cell cycle. Apoptosis-induced cell surface and chromatin modifications can be assessed by both morphological evaluation and cytometry. However, cytometry appears to be better suited for quantification purposes due to its simplicity of use and to the wide range of applications that are made possible by technical improvements.

Our data show that the taxane-induced cell cycle block begins in the first few hours after DOC treatment, whereas at that time only a minority of cells display a decrease in Δψ_m. These events suggest that the actual impairment in the mitochondrial function, which initiates the apoptotic process, has a later onset and follows the cell cycle block. However, as already highlighted, this impairment is of great importance because, over time, the near totality of DOC-treated cells that presented a loss in Δψ_meventually undergoes apoptosis.

By evaluating different cytometric methods concomitantly, we were able to draw up a chronological table of apoptotic events in taxane-induced cell death, where G₂/M arrest appears to precede mitochondrial involvement, which occurs before DNA fragmentation and cell surface modifications. The present study also provides a sort of time-based guideline for the choice of the most appropriate parameters to detect specific apoptotic modifications, such as cell surface changes or early/initial vs. late/massive DNA fragmentation.

In conclusion, we believe that the effects of a particular drug on a cell culture can best be investigated by using multiple parameters to assess apoptosis, by following them up for a relatively long time, and by comparing the dynamics of treated and untreated cell populations. Having said that, some information may still be missing, such as the absolute number of disintegrated cells, as there is no way of relating the amount of cell debris to that of the cells which originated them. On the other hand, a fairly simple assay such as cell cycle distribution and its evolution over time may be sufficiently informative to determine whether or not a drug warrants further investigation.