Beyond genetics: surprising determinants of cell fate in anti-tumor drugs

Summary

In this issue of Cancer Cell, Gascoigne and Taylor report their findings of following 10,000 single cells incubated with three classes of anti-mitotic drugs, including paclitaxel. This extensive analysis reveals an unappreciated complexity in response to such drugs and demonstrates that it's more than genetics that determine life or death.

To maintain its chromosome content, each dividing cell must partition its replicated genome such that one copy of each chromosome is deposited into each new daughter cell. Errors in this process lead to the production of aneuploid daughter cells that contain an abnormal chromosome content. Aneuploidy can have drastic consequences for both the cell and the organism (Kops et al., 2005). Indeed, while most aneuploid human embryos are inviable, those that do survive develop severe birth defects such as Down's syndrome (Jacobs and Hassold, 1995). Furthermore, aneuploidy is a common characteristic of tumor cells and has been shown in some instances to play a causative role in the development of cancer (Weaver et al., 2007).

A major safeguard for protecting cells against chromosome missegregation is the mitotic checkpoint (also know as the spindle assembly checkpoint) (Kops et al., 2005; Musacchio and Salmon, 2007). This cell cycle control mechanism operates in every mitosis to ensure the faithful partitioning of the duplicated chromosomes between the two daughter cells. Chromosomes are sorted in mitosis by the microtubule spindle to which they attach at a specialized protein structure known as the kinetochore. Unattached kinetochores are the source of the mitotic checkpoint signal and act by producing an inhibitor that delays cyclin B1 degradation and chromosome segregation until all chromosomes have stably attached to the spindle.

Some drugs that target the mitotic checkpoint have a long history as successful anti-cancer agents and are commonly used in the treatment of a variety of human cancers (Jackson et al., 2007). These agents, known as anti-mitotic drugs, generally act by chronically activating the mitotic checkpoint leading to a prolonged mitotic delay (Rieder and Maiato, 2004). There are several possible outcomes of this drug induced mitotic delay. Cells may undergo death while arrested in mitosis or exit mitosis in the presence of the drug and transition into the subsequent interphase, a process known as “adaptation”. Filming of cyclin B1 levels in living cells has suggested cyclin B1 is progressively lost during a mitotic delay (Brito and Rieder, 2006). Eventually, cyclin B1 levels fall below the threshold required to maintain a mitotic arrest and adaptation occurs. Adapted cells may then continue to cycle, senescence or execute death pathways from interphase. Despite the clinical importance of predicting which tumors will show favorable responses to anti-mitotic drugs, the pathways that determine whether a cell dies, arrests, or proliferates following treatment with anti-mitotic drugs are not known (Rieder and Maiato, 2004).

Determining cell fate

In this issue, Gascoigne and Taylor provide the most in depth analysis to date of the response of cancer cells to anti-mitotic agents (Gascoigne and Taylor, 2008). Utilizing automated, high-throughput, time-lapse microscopy, the authors determine the response of individual cells from a large panel of tumor cell lines treated with three classes of anti-mitotic drugs. As anticipated from earlier work, different tumor cell lines vary widely in their response to anti-mitotic agents. Surprisingly, however, cells within the same line often also displayed substantial variation in their response to drug treatment. The authors elegantly and persuasively demonstrate that the variation in the fate of individual cells within a population cannot be accounted for by initial genetic variation. Indeed, even presumably genetically identical daughter cells frequently behave differently in the next mitosis, varying in their response as much as two cells chosen at random from the population. Thus, cell fate in response to chronic mitotic checkpoint activation is not genetically pre-determined. Instead, Gascoigne and Taylor's observations implicate a highly unexpected level of stochastic variation in the signaling pathways activated by anti-mitotic agents.

Having determined a tremendous variation in the behavior of cells to anti-mitotic agents, Gascoigne and Taylor formulate and test a model in which mitotic cell fate is determined by two competing networks, one involving the slow destruction of cyclin B1 and the other, the progressive activation of a putative caspase-dependent death pathway. By establishing the threshold level of cyclin B1 that is required for mitotic exit, the authors demonstrate that cells which typically die directly from mitosis only slowly degrade cyclin B1, thereby failing to breach the threshold required for mitotic exit prior to death (Figure 1B). By contrast, cells that undergo adaptation progressively destroy cyclin B1 and surpass this threshold, leading to mitotic exit (Figure 1A). For each cell line tested, inhibiting caspase activation with a pan-caspase inhibitor, not only reduced the proportion of those cells that died in mitosis, but also prolonged the duration of the mitotic arrest. This suggests that by inhibiting caspase-dependent death signals produced by chronic mitotic checkpoint signaling, cells are provided with more time to degrade cyclin B1 to a level sufficient to promote mitotic exit (Figure 1C).

Figure 1.

Model for two competing networks that together dictate mitotic cell fate. A. Cyclin B1 (blue) is progressively destroyed during a mitotic delay and at the same time a putative death signal (possibly also produced by the mitotic checkpoint) accumulates (red). Eventually cyclin B1 levels fall bellow a threshold required to maintain mitotic arrest and the cell (e.g., a retinal pigmented epithelial [RPE] cell) undergoes adaptation. B. Cyclin B1 levels fall during mitotic delay, but fail to decline to the threshold required to allow mitotic exit. Instead, the apoptotic signal reaches a level sufficient to promote cell death in mitosis (e.g., HT29 cell). C. Slowing the activation of apoptotic effectors by treatment with a caspase inhibitor provides (e.g., the HT29 cell) with more time to destroy cyclin B1. This allows cyclin B1 levels to reach the threshold required to promote mitotic exit before apoptosis is initiated.

Chromosomal instability revisited

Aneuploidy is a remarkably common characteristic of tumor cells often resulting from an underlying chromosomal instability (CIN), characterized by the frequent gains and losses of chromosomes during cell division (Kops et al., 2005). A longstanding, widely held view is that chromosomally unstable cancer cells possess a weakened mitotic checkpoint that permits the onset of anaphase in the presence of misaligned chromosomes. This dogma was challenged in a recent study, which showed CIN cells do not prematurely initiate anaphase, but rather exhibit an increased incidence of lagging anaphase chromosomes (Thompson and Compton, 2008). Gascoigne and Taylor confirm these observations and further demonstrate that CIN cells arrest in mitosis for as long as chromosomally stable cell lines. Taken together, the evidence is now compelling that the mitotic checkpoint is not compromised in chromosomally unstable cancer cells. Moreover, the detailed single-cell analysis dispels the proposal that CIN cells are more sensitive to the killing action of anti-mitotic drugs.

Utilizing their extensive data set, Gascoigne and Taylor further explore the widely held view that the duration of the mitotic delay may dictate the subsequent cellular response. Importantly, however, no correlation was found between the length of the delay and the resulting cell fate. This concurs with a recent report which showed the ability of the mitotic checkpoint to delay cells in mitosis is a poor predictor of cellular response to anti-mitotic drugs (Shi et al., 2008). Rather, cell fate was found to correlate much more closely with the sensitivity of cell lines to apoptosis induced by other factors, such as the broad-spectrum kinase inhibitor staurosporine (Shi et al., 2008). This implies that sensitivity to apoptotic signaling may be an important source of variation in the response of cancer cells to anti-mitotics and suggests more favorable drug responses may be achieved by combining anti-mitotic drugs with agents that restore apoptotic pathways in death resistant cells.

Future directions

Previous studies investigating cell fate in response to anti-mitotic agents have predominantly relied upon indirect assays that average the responses of populations of cells. The enhanced resolution afforded by direct single-cell assays has now revealed that these prior studies have dramatically underestimated the variation in cell fate in response to anti-mitotic drugs. Central questions now raised, but still left unanswered, are what is the identity of the putative death signal that accumulates during a mitotic delay? What are the relative thresholds of cyclin B1 and apoptotic signaling required to promote adaptation and death in various cell lines? What are the pathways responsible for determining cell fate following adaptation and why do some cell lines show dramatically different responses in different anti-mitotic drugs that all chronically activate the mitotic checkpoint? Finally, it is important to bear in mind that the in vitro conditions used here fail to recapitulate aspects of the tumor microenvironment that may have important influences on cell fate including, for example, drug clearance by the circulation, which would lead to renewed cell cycling. Indeed, knowing that stochastic processes can be central to the cells decision to live or die leaves open the question as to how closely the response of cell lines will model the fate of a heterogeneous population of tumor cells.