The Role of Apoptosis in 2′,2′-Difluoro-2′-deoxycytidine (Gemcitabine)-mediated Radiosensitization¹

The nucleoside analogue dFdCyd³has shown promising clinical effectiveness against a range of solid tumors, most importantly non-small cell lung cancer and pancreatic cancer, and has been shown both in laboratory and clinical studies to be a potent radiation sensitizer (reviewed in Refs. 1, 2, 3, 4). We have shown that sensitization in log phase cells occurs under treatment conditions that produce simultaneous cell cycle redistribution of cells into S phase and depletion of intracellular dATP pools. However, the mechanism of sensitization remains unclear.

We wished to determine whether dFdCyd affected radiation-induced apoptosis. There is substantial evidence that chemotherapeutic drugs,including dFdCyd (5), can activate the cellular apoptotic machinery, and that alterations in the expression of pro- and antiapoptotic proteins directly affect the sensitivity of cancer cells to chemotherapy (6, 7, 8). Activation of the apoptotic machinery in response to a variety of cellular stresses and physiological stimuli culminates in the activation of a family of cysteine proteases with specificity for aspartic acid residues(Caspases). Upon activation, caspases initiate cleavage of a number of essential cellular polypeptides (e.g., lamins, (poly)ADP ribose polymerase, actin, and so forth) resulting in cell death. Cleavage of these “death substrates” as well as cleavage of chromosomal DNA results in morphological changes, such as nuclear condensation, loss of cell shape, and cytoplasmic shrinkage and fragmentation of the nucleus, which are characteristic of apoptotic cells (9, 10).

To begin to determine the role of apoptosis in dFdCyd-mediated radiosensitization, we assessed the effect of dFdCyd on the radiation sensitivity of three cell lines that differ substantially in their propensity to undergo radiation-induced apoptosis: HT29 human colon cancer cells, UMSCC-6 (head and neck) squamous cancer cells, and A549 lung cancer cells. Under the conditions used in this study, a relatively high fraction of clonogenic death from radiation is attributable to apoptosis in HT29 cells, whereas apoptosis plays less of a role in the overall clonogenic death of UMSCC-6 and A549 cells. If dFdCyd radiosensitized solely by increasing apoptosis, we would predict that all three of the cell types would show increased apoptosis in proportion to radiosensitization. In contrast, if dFdCyd produced radiosensitization by potentiating preexisting death pathways,then we would predict that only the apoptosis-prone HT29 cells would show a substantial increase in apoptosis when treated with the combination of dFdCyd and radiation, and that UMSCC-6 cells and A549 cells would be radiosensitized through nonapoptotic mechanisms. When we found that the latter was the case, we determined whether this increased apoptosis in HT29 cells was accompanied by an increase in caspase activation and whether inhibition of this activation could decrease radiosensitization.

Cell lines were cultured under standard conditions using RPMI medium,as described previously (11). Two days prior to drug addition, cells were released from the flasks with PBS containing 0.03% trypsin and 0.27 mm EDTA and plated into 100-mm culture dishes. This procedure results in log-phase cells at the time of drug addition. Cells were checked for Mycoplasma every 3 months. Clonogenic survival was assessed using a standard clonogenic assay, as described previously (11). Radiation survival data from drug-treated cells were corrected for plating efficiency using an unirradiated plate treated with drug under the same conditions. Cell survival curves were fitted using the linear-quadratic equation, and the mean inactivation dose was calculated according to the method of Fertil et al. (12). The cell survival enhancement ratio was calculated as the ratio of the mean inactivation dose under control conditions divided by the mean inactivation dose after drug exposure.

Caspase 3 activity was measured fluorometrically in cell lysates, as recommended by the manufacturer (Enzyme Systems Products, Livermore,CA) using an artificial substrate (AFC-138) that fluoresces when it is cleaved (excitation at 400 nm, emission at 505 nm).

Cells were irradiated using ⁶⁰Co at 1–2 Gy/min. Dosimetry was carried out using an ionization chamber connected to an electrometer system that is directly traceable to a National Institute of Standards and Technology calibration.

The apoptotic fraction was calculated by taking the ratio of floating cells:total cell number (floating plus adherent). This assay was validated by demonstrating that >90% of the floating cells were morphologically apoptotic by propidium iodide staining (see“Results”). The scoring of floating cells as apoptotic was further validated by gel electrophoresis. For this assay, cells were lysed, and treated with RNase and proteinase K. The DNA was precipitated with isopropanol, dried, resuspended in Tris/EDTA, and electrophoresed in a 0.75% agarose gel at 120 V for 10 min. followed by 90 V for 60 min. Bands were visualized using ethidium bromide staining. This method of DNA preparation does not recover total cellular DNA; rather it detects soluble (fragmented) DNA (see “Results”). A 1-kb DNA ladder was used as a standard.

Unless otherwise indicated, all of the data are presented as the mean ± SE of at least three experiments. Student’s ttest was used to compare two means. Statistical significance was defined at the level of P < 0.05.

We began these studies by determining whether the fraction of floating cells could be used as a marker for apoptosis. HT29 cells were exposed to 10 nm dFdCyd for 24 h, the drug was washed away, and the cellular morphology of floating and adherent cells was assessed 3 days later using propidium iodide staining (Fig. 1). We found that essentially all of the adherent cells had normal nuclear morphology, whereas virtually all of the floating cells showed nuclear fragmentation (reviewed in Refs. 13 and 14). Identical results were obtained with UMSCC-6 and A549 cells (data not shown), although in the latter cell line, very few floating cells were observed. To confirm that these morphological changes reflected apoptosis, we used agarose gel electrophoresis, as described in “Materials and Methods,” to detect DNA fragmentation in cells treated with dFdCyd. We found substantial DNA fragmentation in HT29 cells, moderate fragmentation in UMSCC-6 cells, and little fragmentation in A549 cells (Fig. 2). These findings suggested that we could use the fraction of floating cells to quantify apoptosis and began to suggest that dFdCyd treatment increased radiation-induced apoptosis in HT29 cells, but less so in UMSCC-6 cells and A549 cells.

We then quantified the effect of radiation, in the absence and presence of dFdCyd, on the extent of apoptosis. When HT29 cells, UMSCC-6 cells,and A549 cells were exposed to 10 nm dFdCyd for 24 h,followed by radiation, all of the three cell lines showed radiosensitization (Fig. 3; Table 1) with an enhancement ratio of 1.81 ± 0.16, 1.47 ± 0.24, and 1.31 ± 0.04, respectively. Radiosensitization of HT29 cells was significantly greater than that of A549 cells (P <0.05) and tended to be greater than that of UMSCC-6 cells(P = 0.10), although the latter comparison did not reach statistical significance. Radiosensitization in all of the three cell lines was related more to a change in the α component (the low dose or “shoulder” region) of the survival curve than to the βcomponent.

We then quantified the fraction of apoptotic cells after varying doses of radiation in the presence or absence of dFdCyd under the same conditions under which we measured radiosensitization. We chose doses of radiation that produced approximately the same clonogenic survival(10 and 1%) in the three cell lines. We found that apoptosis after radiation alone, dFdCyd alone, and the combination, was greatest in HT29 cells, intermediate in UMSCC-6 cells, and least in A549 cells(Fig. 4). To quantify the contribution of apoptosis to overall clonogenic death, we determined the fraction of cell death attributable to apoptosis under these conditions for these three cell lines (Table 2). This analysis confirmed that the order of importance of apoptosis in clonogenic survival is HT29 > UMSCC-6 > A549.

To better characterize the effect of radiation and dFdCyd on apoptotic pathways, we measured caspase 3 activation beginning 1–2 days after treatment using the dose of radiation that produced 10% surviving fraction. (These conditions permitted us to determine activation prior to the morphological appearance of apoptosis.) We found a dramatic increase in caspase activation by radiation (in the presence or absence of dFdCyd) in HT29 cells, whereas there were only marginal increases in caspase activation in UMSCC-6 and A549 cells (Fig. 5).

These data indirectly supported our hypothesis that apoptosis plays an important role in radiosensitization of HT29 cells, as opposed to UMSCC-6 and A549 cells. To assess directly the role of apoptosis in radiosensitization, we treated HT29 cells with the caspase 3 inhibitor DEVD (2 μm) beginning 4 h prior to the initiation of dFdCyd treatment and continuing after combined treatment with dFdCyd(10 nm) and radiation (7 Gy). We found that DEVD significantly reduced apoptosis produced by the combination of dFdCyd and radiation (by 24 ± 8%). Furthermore, DEVD, used under the same conditions that decreased apoptosis, significantly decreased dFdCyd-mediated radiosensitization in HT29 cells, with a resulting enhancement ratio of 1.36 ± 0.24. In contrast, neither 2μ m zVAD (a more general inhibitor of caspases) nor DEVD,at concentrations of between 0.5 and 4 μm, had an effect on the radiation sensitivity (in the presence or absence of dFdCyd) of UMSCC-6 cells and A549 cells (data not shown).

In this study, we have found that apoptosis plays an important role in dFdCyd-mediated radiosensitization of HT29 cells, but not of UMSCC-6 and A549 cells. The importance of apoptosis in dFdCyd-mediated radiosensitization of HT29 cells is suggested by our findings that apoptosis is responsible for a high fraction of overall clonogenic cell death, and that the inhibition of caspase activation significantly reduces radiosensitization. However, our findings also demonstrate that the stimulation of apoptosis is not the only mechanism responsible for radiosensitization, based on the findings that UMSCC-6 and A549 cells are sensitized with modest-to-minimal apoptosis, and that zVAD reduces,but does not eliminate, sensitization in HT29 cells.

The significance of apoptosis in the response to radiation therapy(with or without radiation sensitizers) is controversial. Although some investigators have correlated the ability to undergo apoptosis with response, several recent reviews have summarized findings that suggest that apoptosis is, in some cases, only a minor component of overall clonogenic death (15, 16). Our findings suggest that the role of apoptosis in dFdCyd-mediated radiosensitization depends on the cell line and does not represent a general property of the drug. We also found that dFdCyd alone causes apoptosis to varying degrees in these three cell lines derived from solid tumors. Gemcitabine alone causes substantial apoptosis in leukemia cells (17),further supporting the cell type dependence of the response.

A limitation of this study is that we investigated only the terminal parts of the apoptotic pathway (caspase 3 activation and morphological changes). Apoptosis can result from the activation of several upstream pathways, including ceramide production, tumor necrosis factor receptor simulation (18, 19), and mitochondrial release of cytochrome C (20). We chose to investigate only the final common pathways to assess the overall potential importance of apoptosis. Our findings suggest that it might be worthwhile to determine which of these pathways is activated,because it remains possible that selective potentiation of the active pathway could increase the effectiveness of dFdCyd as a radiation sensitizer in some cell types.

Previous studies have suggested that dFdCyd-mediated radiosensitization depends on the simultaneous depletion of dATP pools and the redistribution of cells into S phase (21, 22). This condition affects neither radiation-induced DNA damage nor the repair of damage, as assessed by pulsed field gel electrophoresis (22). It seems possible that dFdCyd-perturbed nucleotide pools produce misrepair after radiation that is not detectable at the level of pulsed-field gel electrophoresis. Present studies are directed toward elucidating this potential defect.

An initial goal of this work was to develop a predictive assay for determining which patients might benefit from dFdCyd-mediated radiosensitization. Our findings suggest that, although apoptosis may play a role in radiosensitization, it might not be possible to base the assessment on this single factor. Likewise, p53 status alone does not seem to predict radiosensitization in RKO colon cancer cells (23), further supporting the concept that single gene or pathway alterations are unlikely to explain radiosensitization across multiple cell lines. It is hoped that developments in microarray technology will permit simultaneous assessment of multiple pathways that will permit a broader view of cellular responses than can be determined using more traditional approaches. In addition, the usefulness of dFdCyd as a clinical radiosensitizer depends not only on the tumor response but also on the normal tissue response. For instance, clinical trials have demonstrated that doses of dFdCyd as low as 50 mg/m² administered weekly through a course of fractionated radiation for patients with head and neck cancer can produce unacceptable toxicity in the oral mucosa (24),whereas a much higher dose of dFdCyd (600 mg/m²)is tolerable for patients undergoing a standard course of radiation for pancreatic cancer (25). There is a critical need for improved models for both tumor control and normal tissue toxicity to help improve our ability to predict clinical usefulness from preclinical data.