The ability to inappropriately progress through S phase during drug treatment is a key determinant of tumor cell sensitivity to thymidylate synthase inhibitors such as 5-fluoro-2′-deoxyuridine (FdUrd). Previous studies suggest that SW620 cells, which are relatively resistant to FdUrd, have an intact early S-phase checkpoint that protects against FdUrd-induced DNA damage and cytotoxicity and that this checkpoint is defective in the relatively sensitive HT29 cells, which continue to progress through S phase during drug treatment. To test this hypothesis, we examined the expression and activation of known S-phase checkpoint mediators in FdUrd-treated SW620 and HT29 cells. FdUrd induced degradation of cdc25A in SW620, but not HT29 cells, in a manner that correlated with the previously described drug-induced S-phase arrest. This difference, however, could not be attributed to differences in either chk1 activation, which was similar in both cell lines, or chk2 activation, which only occurred in HT29 cells and correlated with uracil misincorporation/misrepair-induced DNA double-stranded breaks. These observations suggest that although FdUrd-induced S-phase arrest and associated cdc25A degradation are impaired in HT29 cells, signaling by ATM/ATR is intact upstream of chk1 and chk2. Finally, FdUrd induced premature mitotic entry, a phenomenon associated with deregulated cdc25A expression, in HT29 but not SW620 cells. Blocking cdc25A expression in HT29 cells with small interfering RNA attenuated FdUrd-induced premature mitotic entry, suggesting that progression of HT29 cells through S phase during drug treatment results in part from the inability of these cells to degrade cdc25A in response to FdUrd-induced DNA damage.
Fluoropyrimidines such as 5-fluorouracil and 5-fluoro-2′-deoxyuridine (FdUrd) have been used clinically as antineoplastic agents for decades, and they continue to form the cornerstone of front-line therapeutic regimens for the treatment of various solid tumors, especially those of gastrointestinal origin (1). The initial steps in the process by which these drugs kill tumor cells are well established: 5-fluorouracil and FdUrd are metabolized to the nucleoside monophosphate, FdUMP, which then forms a covalent ternary complex with thymidylate synthase and a reduced-folate cofactor, 5,10-methylenetetrathydrofolate, thus inhibiting the thymidylate synthase enzyme (2). Accordingly, cellular resistance to fluoropyrimidine cytotoxicity is observed under circumstances that subvert complete inhibition of thymidylate synthase activity, such as insufficient supply of reduced-folate cofactor (3) or elevated levels of thymidylate synthase protein (4). However, it is much less clear how regulation of events downstream of thymidylate synthase inhibition can affect the extent to which thymidylate synthase inhibition leads to cell death (5, 6).
One immediate consequence of thymidylate synthase inhibition is accumulation of the substrate of the blocked enzymatic reaction, dUMP. This leads to elevated levels of dUTP, misincorporation of uracil into DNA, and DNA double-stranded breaks (7, 8, 9, 10, 11, 12, 13). Results from our laboratory and others have shown that attenuation of this process, by elevation of dUTPase activity, can diminish DNA double-stranded breaks and cytotoxicity caused by thymidylate synthase inhibition (14, 15, 16, 17). The other immediate consequence of thymidylate synthase inhibition is depletion of the product of the blocked reaction, dTMP. Experiments using thymidylate synthase–deficient mammalian cells that depend upon exogenous thymidine (dThd) for survival have been used to characterize responses to dThd nucleotide starvation, independently from the effects of dUrd nucleotide pool expansion (18, 19, 20). These studies demonstrate that the prolonged absence of dThd nucleotides is also sufficient to cause cell death, presumably without involving dUrd nucleotides.
Although uracil-mediated DNA double-stranded break formation and inhibition of replication fork progression due to lack of dTTP are both significant stresses, their impacts on survival can vary, depending on how a particular cell type responds to them. Responses that involve regulation of cell cycle progression and apoptosis are particularly important in this regard (5, 6, 21). In previous studies in which we compared the effects of FdUrd in two human colorectal tumor cell lines (HT29 and SW620), we found that the propensity of HT29 cells to try to advance through the cell cycle during profound thymidylate synthase inhibition correlated closely with their sensitivity to FdUrd-induced clonogenic death, relative to SW620 cells (22, 23, 24). Based on these observations, we hypothesized that the sensitivity of HT29 cells to thymidylate synthase inhibition may be due in part to a defect in their cell cycle checkpoint response(s) to FdUrd-induced DNA damage.
Fig. 1 shows a scheme of predicted responses following various insults to a cell with normal checkpoint function. This summary is a distillation of features that appear in a number of recent reviews (25, 26, 27, 28, 29), with emphasis on the pathway integration models that are contained in a series of papers from Bartek et al. (30, 31, 32, 33, 34, 35). According to this scheme, detection of DNA damage or replication inhibition leads first to activation of the ATM or ATR kinases, which then activate a second tier of kinases, chk1 and chk2. The substrates for these proteins include members of the cdc25 phosphatase family, which are required to remove inhibitory phosphates from cyclin/cdk complexes, thereby activating the cdks. When phosphorylated by chk1 or chk2, the cdc25 proteins are inactivated either by degradation (cdc25A) or nuclear export (cdc25C), resulting in a halt to their activation of cdks and therefore a halt to cell cycle progression.
Based on this scheme, we analyzed several elements involved in early S-phase checkpoint function, including cdc25A, chk1, and chk2, to determine whether differences in the responses of these proteins to thymidylate synthase inhibition account for the differences in cell cycle progression and sensitivity to FdUrd observed between HT29 and SW620 cells.
MATERIALS AND METHODS
Cell Culture and Drug Solutions.
SW620 cells were obtained from American Tissue Type Collection (Manassass, VA) and maintained in McCoy’s 5A medium supplemented with 10% fetal bovine serum (Life Technologies, Rockville, MD) and 2 mmol/L l-glutamine. The HT29 clones, HT.lac48 (expresses β-galactosidase) and HT.dut (expresses Escherichia coli dUTPase; ref. 15), were maintained in RPMI 1640 supplemented with 10% fetal bovine serum and 2 mmol/L l-glutamine. The HT.lac48 clone was used in place of parental HT29 cells in all experiments. Both SW620 and HT29 cell lines are homozygous mutant p53 (mut-p53; ref. 36). All drug media were supplemented with 10% dialyzed fetal bovine serum to minimize dThd salvage. All drugs except AG337 were obtained from Sigma Chemical (St. Louis, MO) and were dissolved in double-distilled water and stored at 4°C for up to 2 months. AG337 was a gift from John P. Montana (Agouron Pharmaceuticals, Inc., La Jolla, CA).
Western Blot Analyses.
Exponentially growing cells were treated with drug for 30 minutes to 24 hours, harvested with trypsin, washed with ice-cold PBS (2.7 mmol/L KCl, 138 mmol/L NaCl, and 8.1 mmol/L Na2HPO4; ICN Biochemicals, Aurora, OH) and resuspended in either Cell Lysis Buffer 1 [cd25A detection; 20 mmol/L Tris (pH 7.5), 250 mmol/L sodium chloride, 0.5% NP40, 0.1 mmol/L EDTA, 1 mmol/L sodium orthovanadate, 10 mmol/L sodium fluoride, and 1 mmol/L phenylmethylsulfonyl fluoride] and Complete Protease Inhibitor mixture (Roche, Gipf-Oberfrick, Switzerland; ref. 37) or Cell Lysis Buffer 2 [phospho-chk1 and phospho-chk2 detection and gel mobility shift; 20 mmol/L Tris (pH 7.5), 0.27 mol/L sucrose, 1% Triton X-100, 1 mmol/L EDTA, 1 mmol/L EGTA, 10 mmol/L sodium β-glycerophosphate, 5 mmol/L sodium PPI, 1 mmol/L sodium orthovanadate, 50 mmol/L sodium fluoride, 0.1% β-mercaptoethanol, and 1 mmol/L phenylmethylsulfonyl fluoride] and Complete Protease Inhibitor mixture (Roche; ref. 38). For experiments with the proteasome inhibitor calpain inhibitor I N-Acetyl-Leu-Leu-Norleu-al (LLnL), 25 μg/mL LLnL or vehicle control (0.1% DMSO) was added to the drug media for the last 60 minutes of treatment as described previously (33). Proteins were resolved on 10% gels for cdc25A detection (37.5:1 ratio acrylamide:bisacrylamide) or 8% gels for chk1 (13.8:0.4 ratio acrylamide:bisacrylamide) and chk2 mobility shift detection (10:0.1 ratio acrylamide:bisacrylamide) as described previously (38). Antibodies used are given in the figure legends.
Chk2 Kinase Assay.
Chk2 was immunoprecipitated from a 750-μg sample lysed in Cell Lysis Buffer 3 [50 mmol/L Tris-HCl (pH 7.4), 1% NP40, 0.25% sodium deoxycholate, 150 mmol/L sodium chloride, 1 mmol/L EDTA, 1 mmol/L sodium orthovanadate, 1 mmol/L sodium fluoride, 10 mmol/L sodium β-glycerophosphate, 5 mmol/L sodium PPI, and 1 mmol/L phenylmethylsulfonyl fluoride] and Complete Protease Inhibitor mixture (Roche) with preconjugated chk2 antibody-protein A agarose (sc-8813 AC; Santa Cruz Biotechnology, Santa Cruz, CA) overnight at 4°C. After washing three times with ice-cold PBS, samples were resuspended in 30 μL of assay dilution buffer [20 mmol/L 4-morpholinepropanesulfonic acid (pH 7.2), 25 mmol/L β-glycerophosphate, 5 mmol/L EGTA, 1 mmol/L sodium orthovanadate, and 1 mmol/L dl-dithiothreitol) and assayed for kinase activity using biotinylated CHKtide (no. 12-414; Upstate Biotechnology, Lake Placid, NY) as substrate, [γ-32P]ATP (3000 Ci/mmol; Amersham Biosciences, Piscataway, NJ) for labeling, and streptavidin-agarose (no. 16-126; Upstate Biotechnology) to isolate the 32P-labeled peptide, as described by Upstate Biotechnology. Results were compared with those from a purified chk2 (no. 14-347; Upstate Biotechnology) standard curve to determine milliunits of chk2 activity per sample.
Chk1 Kinase Assay.
Endogenous chk1 activity was measured as described previously (39). In brief, 500 μg of cellular protein was mixed with 1 μg of chk1 antibody (sc-7898; Santa Cruz Biotechnology) for 3 hours at 4°C. Samples were then incubated an additional 90 minutes at 4°C with 20 μL of 50% protein-A agarose (Invitrogen) to pull down the immune complex. Agarose pellets were washed twice with lysis buffer and once with chk1 kinase assay buffer [20 mmol/L Tris-Cl (pH 7.5), 0.1 mmol/L EGTA (pH 7.0), 10 mmol/L MgCl2, and 1 mmol/L dithiothreitol]. Samples were then incubated in 50 μL of total volume with 2 μmmol/L cold ATP, 3 μg of glutathione-S-transferase (GST)-cdc25A substrate, and 3 μCi of [γ-32P]ATP for 30 minutes at 37°C. One-half of each sample was spotted on squares of P81 phosphocellulose paper (Upstate Biotechnology), which were then washed three times in 0.75% phosphoric acid and once in acetone, dried, and counted in a scintillation counter. Data are presented as the fold change in immunoprecipitated chk1 activity and represent the mean ± SE from three separate experiments.
Exponentially growing cells were treated with drug, trypsinized, washed with ice-cold PBS, and fixed at a concentration of 2 × 106 cells/mL in ice-cold 70% ethanol. Washed, fixed cells were incubated with a rabbit anti-phospho-histone H3-specific antibody (no. 06-570; Upstate Biotechnology), followed by a FITC-conjugated antirabbit secondary antibody (F-0382; Sigma Chemical) as described previously (40). Cells were then stained with propidium iodide to measure total DNA content and analyzed on a FACScan flow cytometer (Becton-Dickinson, Palo Alto, CA) with FlowJo software (Tree Star, Inc., Ashland, OR). Samples were analyzed within 72 hours of collection.
RNA Interference Experiments.
RNA interference was performed in 12-well dishes with TransIT-TKO transfection reagent using a serum-free protocol supplied by Mirus Corporation (Madison, WI). The previously described small interfering RNA oligonucleotide corresponding to nucleotides 82 to 102 of human cdc25A (41) was supplied by Qiagen (Valencia, CA). Cells were treated with drug 24 hours post-transfection and analyzed 36 hours later.
Thymidylate Synthase Inhibition Causes cdc25A Degradation in SW620 but not HT29 Cells.
Results from several studies support the hypothesis that S-phase progression during drug treatment is an important determinant of fluoropyrimidine-mediated cytotoxicity and radiosensitization (22, 23, 24, 42, 43). Mechanistically, the progression of the relatively sensitive HT29 cells into S phase during periods of thymidylate synthase inhibition has been linked to a drug-induced increase in cyclin E–associated cdk2 activity (44). In contrast, cyclin E-cdk2 activity is unchanged in FdUrd-treated SW620 cells, which are relatively resistant to FdUrd (44). Because cdc25A is a phosphatase that can activate cdk2 through dephosphorylation of residues Thr-14 and Tyr-15, we hypothesized that drug-induced changes in cdc25A might correlate with FdUrd-induced S-phase arrest in these cell lines. In SW620 cells, cdc25A protein levels decreased beginning at about 6 hours of treatment with 100 nm FdUrd (Fig. 2,A). This decrease correlates temporally with the previously described drug-induced early S-phase arrest observed in these cells (24, 42, 44). In addition, the FdUrd-induced decrease in cdc25A was prevented by cotreatment with the proteasome inhibitor LLnL, supporting the hypothesis that FdUrd stimulated degradation of cdc25A (Fig. 2 B).
In contrast, cdc25A levels were maintained or slightly elevated during periods of thymidylate synthase inhibition in HT29 cells. This result is consistent with the incomplete drug-induced S-phase arrest previously observed in HT29 cells (24, 42, 44). The genomic cdc25A sequence from both SW620 and HT29 cells matches that reported for wild-type human cdc25A in the National Center for Biotechnology Information database (accession no. BC007401),3 indicating that the failure of FdUrd to induce cd25A degradation in HT29 cells is not due to the expression of a mutant, stabilized protein. Similar cdc25A response patterns were observed in SW620 and HT29 cells treated with other thymidylate synthase inhibitors, including 5-fluorouracil, AG337, and methotrexate (Fig. 2 A).
Chk2 Responses Are Intact in HT29 and SW620 Cells, but They Do Not Explain cdc25A Behavior.
Considering the checkpoint pathway model proposed by Falck et al. (31), we predicted that the cdc25A degradation observed in FdUrd-treated SW620 cells might be a consequence of DNA damage-induced, ATM-mediated chk2 activation. To test this hypothesis, we first established that the ATM-chk2-cdc25A checkpoint pathway is activated by ionizing radiation in our cell lines. Activation of chk2 [as determined by phosphorylation of Thr-68 and mobility shift of total chk2 protein (45, 46, 47)] occurred within 15 to 30 minutes of irradiation in both cell lines, although the response was slightly stronger in HT29 cells (Fig. 3). The onset of cdc25A degradation coincided with chk2 activation in SW620 cells, although it was delayed and less extensive in HT29 cells. Exposure of either cell line to 5 mmol/L hydroxyurea for as long as 24 hours resulted in negligible activation of chk2 (data not shown). These results indicate that both cell lines partially conform to the prevailing model, in that chk2 is activated by DNA double-stranded breaks (but not replication fork stalling). However, the relatively small effect of ionizing radiation on cdc25A in HT29 cells suggests there may be other pathways operating in these cells that oppose cdc25A degradation.
We next examined chk2 phosphorylation and activation in FdUrd-treated SW620 and HT29 cells. Western blot analysis revealed that FdUrd treatment caused chk2 phosphorylation and mobility shift in HT29 cells, but not SW620 cells (Fig. 4,A). Similar phosphorylation patterns were observed in cells treated with 5-fluorouracil, AG337, or methotrexate (data not shown). To confirm that the presence of Thr-68–phosphorylated, mobility-shifted chk2 corresponded to activation of that protein, we measured the activity of endogenous chk2 immunoprecipitated from drug-treated cells (Fig. 4 B). Chk2 activity was induced after 6 to 12 hours of drug treatment in HT29 cells, but not SW620 cells, and corresponded with the appearance of mobility-shifted, Thr-68–phosphorylated chk2 on Western blots.
Because SW620 cells are capable of responding to DNA double-stranded breaks by activating chk2 (Fig. 3), it seems plausible that their failure to do so when exposed to 100 nm FdUrd reflects the formation of fewer DNA double-stranded breaks in SW620 cells than in HT29 cells. This explanation is supported by our previous finding that SW620 cells contain severalfold higher levels of dUTPase than HT29 cells, which should attenuate dUTP accumulation and uracil-mediated DNA double-stranded break formation (14). We previously described an HT29 derivative, HT.dut, which expresses E. coli dUTPase and is thus partially protected from the DNA damaging and cytotoxic effects of uracil misincorporation/misrepair associated with thymidylate synthase inhibition (15). To further test the hypothesis that uracil-mediated DNA double-stranded breaks cause chk2 activation in HT29 cells, we examined chk2 phosphorylation in HT.dut cells treated with FdUrd (Fig. 4). As expected, we found that chk2 phosphorylation was delayed and attenuated in HT.dut cells in a manner that correlated temporally with the previously reported delay in the accumulation of FdUrd-induced DNA double-stranded break formation in these cells (16).
Chk1 Responses to DNA Damage Are Similar in HT29 and SW620 Cells.
Because chk1 is also a key effector in DNA damage response pathways, with documented ability to regulate cdc25A levels (33, 35), we investigated the potential role of this kinase in our system. Chk1 activation was assessed by measurement of phosphorylation at Ser-317 and Ser-345 (48) and the ability of immunoprecipitated endogenous chk1 to phosphorylate GST-cdc25A in an in vitro assay (39). When SW620, HT29, or HT.dut cells were exposed to 5 mmol/L hydroxyurea, a strong, rapid signal was obtained for chk1 phosphorylation that correlated with an initial decrease in cdc25A protein levels (Fig. 5). Hydroxyurea-induced cdc25A degradation in these cells was blocked by LLnL (data not shown). Despite the continued presence of hydroxyurea and sustained phosphorylation of chk1, cdc25A protein levels recovered within 6 to 12 hours but fell again at 16 to 24 hours, a biphasic response similar to that observed in irradiated SW620 cells (Fig. 3). This biphasic cdc25A response pattern may indicate that multiple DNA-damage response pathways are stimulated by hydroxyurea in these cells, i.e., an early, transient response to stalled replication forks, which correlates with the initial phosphorylation of chk1, and a secondary, delayed response to subsequent DNA damage. Biphasic responses to ionizing radiation have been described previously (41, 49) and may represent activation of separate ATM- and ATR-mediated checkpoint pathways.
Treatment of each cell line with 100 nm FdUrd produced a strong signal for chk1 phosphorylation by the 6-hour time point, which was accompanied by a modest increase in chk1 activity (Fig. 6 A and B). Although chk1 remained phosphorylated throughout the drug treatment period, chk1 activity had returned to control levels by the 24-hour time point. This concentration of FdUrd is sufficient to almost completely inhibit thymidylate synthase (and, presumably, replication fork progression) in both cell lines within 2 hours (44), so we anticipated that a chk1 response would be seen. It is not clear, however, why FdUrd-induced chk1 phosphorylation was delayed relative to the hydroxyurea-mediated chk1 response, which occurred almost immediately after hydroxyurea treatment. One possibility is that hydroxyurea stalls replication forks more effectively than FdUrd, because it depletes all four deoxyribonucleotides, instead of just dTTP. In any case, the observed time course for chk1 activation is consistent with chk1 being responsible for FdUrd-induced cdc25A degradation in SW620 cells. It is as yet unclear why FdUrd-induced chk1 activation does not lead to cdc25A degradation in HT29 cells, but it may be due to a functional defect in one or more of the recently described cofactors for in vivo chk1 activity, such as Nbs-1 (49) or claspin (50).
5-Fluoro-2′-Deoxyuridine Causes Premature Mitotic Entry in HT29, but not SW620 Cells.
The above data demonstrate that cdc25A levels remain high during periods of thymidylate synthase inhibition in HT29 cells, despite activation of both chk1 and chk2. Because deregulated cdc25A expression causes not only accelerated G1-S–phase transition (51, 52) but may also prematurely force cells into mitosis (34, 52), we used a flow cytometric assay to characterize the fraction of SW620, HT29, and HT.dut cell populations that entered mitosis (as indicated by positive staining for phospho-histone H3) during FdUrd treatment (Fig. 7). The mitotic fractions of the three cell lines were similar at the beginning of drug treatment (2.3–2.9%) and in all cases, mitotic cells were depleted by 16 hours of drug treatment. SW620 cells failed to enter mitosis to a significant extent through 36 hours of drug treatment, consistent with the notion that this line has a relatively effective G2-M checkpoint system. In contrast, after a 24-hour exposure to FdUrd, some HT29 cells could be seen to enter mitosis. We were surprised to see that phospho-histone H3-positive cells arose not only with a G2-M complement of DNA but also with a G1-early S complement of DNA, indicating that some cells attempted to enter mitosis after having replicated little or none of their genome. The response of HT.dut cells was qualitatively similar to that of HT29 cells, except that it was delayed by about 12 hours, a result consistent with the delayed accumulation of FdUrd-induced DNA double-stranded breaks, delayed activation of chk2, and partial protection from FdUrd-induced cytotoxicity observed in these cells. Neither SW620 nor HT29 cells entered mitosis after a 24-hour treatment with a similarly toxic dose of hydroxyurea (data not shown).
To better understand the role of cd25A in FdUrd-induced premature mitotic entry, we transfected HT29 cells with a previously described double-stranded small interfering RNA specific for cdc25A (41) and determined the fraction of phosphohistone H3-positive cells in the absence or presence of drug. FdUrd-induced premature mitotic entry was partially blocked in HT29 cells with decreased cdc25A expression (Fig. 8), suggesting that the continued presence of cdc25A in mock-transfected HT29 cells contributes to their inability to completely arrest in early S phase in response to FdUrd-induced DNA damage. The increase in the mitotic fraction of HT29 cells transfected with cdc25A small interfering RNA (in the absence of drug treatment) appears to be cell line specific because there was no change in the mitotic fractions of either HeLa or SW620 cells transfected with cdc25A small interfering RNA (ref. 41; data not shown).
When the current studies were undertaken, we made the general prediction that checkpoint responses to thymidylate synthase inhibition would be less effective in HT29 than in SW620 cells. At the level of chk1 and chk2, this does not appear to be the case. The time course and magnitude of chk1 activation were indistinguishable among the cell lines studied (Figs. 5 and 6). It was somewhat surprising to find that chk1 phosphorylation did not occur sooner than 6 hours after FdUrd treatment, considering that thymidylate synthase inhibition is maximal within 2 hours (44) and that chk1 phosphorylation was evident in hydroxyurea-treated cells after as little as 30 minutes (Fig. 5). It may be the case that a small residual pool of dThd nucleotides persisted for a short while after FdUrd treatment that was sufficient to keep replication from immediately coming to a complete stop.
A very different situation was observed with regard to FdUrd-induced chk2 activation, which occurred to a significant degree in HT29 cells, but not in SW620 cells (Fig. 4). At least three lines of evidence suggest that chk2 activation may be attributed to uracil-mediated DNA double-stranded breaks in this situation. First, the time course of chk2 activation in HT29 cells corresponds to the time course with which FdUrd-induced DNA double-stranded breaks were observed in a previous study (16). Second, we found earlier that FdUrd-induced accumulation of dUTP is significantly lower in SW620 cells than in HT29 cells, because of a severalfold higher level of dUTPase expression in the SW620 line (14). Third, FdUrd-induced activation of chk2 was delayed by about 12 hours in HT.dut cells compared with the HT29 cells (Fig. 4). This corresponds closely to the delay in appearance of DNA double-stranded breaks in HT.dut cells upon FdUrd treatment, relative to HT29 cells (15).
In human colon cancer specimens, dUTPase expression varies widely and is negatively correlated with a favorable prognosis (53, 54). Our findings suggest that low dUTPase expression might also predict for activation of chk2 signaling in response to thymidylate synthase inhibition. Considering the number of known and potential substrates for chk2 in various signaling pathways (55), dUTPase expression might therefore have an important impact on cellular response to thymidylate synthase inhibition through chk2, as well as through modulation of the underlying lesion (DNA double-stranded breaks).
Among the key processes shown to be regulated by chk1 and chk2 is cdc25A degradation (29, 35, 41, 56). If phosphorylation by these kinases is the major determinant of changes in cdc25A stability in response to DNA damage, then we would expect to see a consistent pattern in which chk1 or chk2 activation corresponds to loss of cdc25A protein. Instead, in the present study, there were many circumstances in which changes in cdc25A protein levels cannot be explained simply by activation of chk1 or chk2. For example, in SW620 cells exposed to 10 Gy of ionizing radiation, cdc25A vanished within 15 minutes, although activated (gel-shifted) chk2 was not detected until the 30-minute time point (Fig. 3). Cdc25A subsequently recovered at the 6- and 10-hour time points, when chk2 activation was still maximal, and then cdc25A declined again at the 24-hour point, when chk2 activation was waning. When SW620 cells were treated with 5 mmol/L hydroxyurea, the initial disappearance of cdc25A did coincide with chk1 activation (Fig. 5); however, cdc25A protein was restored to control levels at later time points, even though chk1 remained phosphorylated. Only after FdUrd treatment in SW620 cells did chk1 activation track with cdc25A degradation (Figs. 2 and 6).
In HT29 cells, the behavior of cdc25A was even more discrepant with activation of chk kinases. In general, cdc25A was more resistant to down-regulation in HT29 cells than in SW620, even though the baseline levels of cdc25A were 2- to 3-fold higher in SW620 cells (data not shown). Factors other than chk1 and chk2 are known to regulate cdc25A expression, and our results indicate that some of these may be of greater importance in HT29 cells. At the level of protein degradation, it has been shown that cyclin B/cdk1 can phosphorylate cdc25A at Ser-17 and Ser-115, stabilizing the cdc25A protein against proteasomal destruction (34). Cdk2 has also been implicated recently in regulating cdc25A levels in HeLa cells through a destabilizing phosphorylation, although the site of that modification was not defined (57). On the input side of the process, transcription of the cdc25A gene is responsive to E2F-1 (58, 59). Because E2F-1 can be stabilized by chk2 (60), it is plausible that chk2 activation could lead to an increase in cdc25A protein content through E2F-mediated up-regulation. The contributions of all of these mechanisms will need to be analyzed to understand why cdc25A levels are maintained in HT29 cells after thymidylate synthase inhibition.
Regardless of the mechanism by which it occurs, the persistence of cdc25A activity in HT29 cells should promote cdk activation and, therefore, cell cycle progression. This is consistent with previous results showing that cyclin E/cdk2 activity rises in HT29 cells but not SW620 cells during thymidylate synthase inhibition (44). Because it has been found that ectopic expression of cdc25A can drive cells into premature mitosis (34, 52), it was not surprising to find that the continued presence of cdc25A in HT29 cells also permitted mitotic entry during drug treatment (Figs. 7 and 8). We did not predict, however, that a population of cells would arise that stains positively for phospho-histone H3 while having DNA content equivalent to G1-early S–phase cells. As far as we are aware, premature mitotic entry of this nature has previously only been seen in cells exposed to drugs that directly affect checkpoint proteins. Presumably, this prereplicative mitotic entry required not only activity of one or more of the cdc25 phosphatases but also accumulation of cyclin B1 and its nuclear localization, along with cdk1. We plan to investigate this phenomenon further to determine whether the cells staining positive for histone H3 phosphorylation show other signs of having entered mitosis prematurely (such as chromosome condensation and nuclear membrane breakdown) and whether FdUrd treatment has any relevant effects on cyclin B1 expression or localization.
We were surprised by the observation that depletion of cdc25A with small interfering RNA caused an increase in the mitotic fraction of HT29 cells in the absence of drug treatment. Previous studies (41), which we were able to replicate (data not shown), found no change in the mitotic fraction of cdc25A small interfering RNA-transfected HeLa or SW620 cells. These data further support the hypothesis that cdc25A-mediated checkpoint pathways in HT29 cells may differ from other cell lines in multiple ways. Presumably the increased mitotic fraction in these circumstances is caused by accelerated mitotic entry, diminished mitotic exit, or some combination of these two processes. Because cd25A is known to contribute to activation of cyclin B/cdk1, it is hard to see how removing cdc25A activity could enhance mitotic entry. However, the process of mitotic exit is controlled in a complex manner by another protein phosphatase, cdc14 (61), which can both activate and antagonize the mitotic exit network (62). Although we are not aware of any known overlap between substrates for cdc25A and cdc14, one could speculate that cdc25A function in HT29 cells might be aberrant in a way that confers upon it a positive role in mitotic exit. The basis for this phenomenon will need to be elucidated in future work.
In summary, it is clear that HT29 and SW620 cells differ markedly in their early S-phase checkpoint responses when stressed by thymidylate synthase inhibition, particularly with regard to regulation of cdc25A and mitotic entry. It remains to be determined whether the apparent lack of stringent S-phase checkpoint control in HT29 cells constitutes a basis for sensitivity to this class of drugs that can be exploited for therapeutic advantage.
Grant support: NIH grant R01-CA77391. L. Parsels was a recipient of an Upjohn L.N. Clinical Pharmacology Research fellowship.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Requests for reprints: Jonathan Maybaum, 4701 Upjohn Center for Clinical Pharmacology, University of Michigan Medical School, Ann Arbor, MI 48109-0504; Phone: 734-647-1436; Fax: 734-763-3438; E-mail: firstname.lastname@example.org
D. C-H. Tai, L. A. Parsels, J. Maybaum, unpublished data.
We thank Dr. John Lazo (Department of Pharmacology, University of Pittsburgh, Pittsburgh, PA) for providing the GST-cdc25A plasmid, Jessica O’Konek for producing the GST-cdc25A protein, and Drs. Christine Canman and James L. Park for their thoughtful comments on the manuscript.