Intra-S-Phase Checkpoint Activation by Direct CDK2 Inhibition

Cell lines.A human ovary cancer cell line, A2780 cells (American Type Culture Collection, Manassas, Va.), was cultured in Dulbecco modified Eagle medium containing 10% fetal bovine serum plus streptomycin and penicillin. Two human lymphoblast cell lines, GM01526 and GM02254 (Coriell Institute, Camden, N.J.) were grown in RPMI 1640 medium containing 15% fetal bovine serum.

Chemical compounds.Roscovitine, olomoucine, SB202190, SB203580 (Calbiochem, San Diego, Calif.), PD98059, U0126 (Cell Signaling Technology, Beverly, Mass.), and aphidicolin (Sigma-Aldrich, St. Louis, Mo.) were dissolved in dimethyl sulfoxide (DMSO). Caffeine (Sigma-Aldrich) was dissolved in water. A specific CDK2 inhibitor, aminothiazole compound 25 (Bristol-Myers Squibb, Princeton, N.J.) was synthesized as previously described (23). The 90% inhibitory concentration (IC₉₀) of compound 25 was determined by 24-h 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. An IC₉₀ of 500 nM was routinely used in cell-based studies unless specified.

Antibodies, immunoprecipitation, and immunoblotting.Rabbit anti-CDK2, rabbit anti-cyclin E (Upstate Biotechnology, Lake Placid, N.Y.), rabbit anti-cyclin A, rabbit anti-p21, mouse anti-CHK2, goat anti-ATR, goat anti-replication protein A (RPA) p70 (Santa Cruz Biotechnology, Santa, Cruz, Calif.), mouse anti-CHK1 (Biomeda, Foster City, Calif.), mouse anti-p53, rabbit anti-ATR (Oncogene, San Diego, Calif.), rabbit anti-ATM (Oncogene and Santa Cruz Biotechnology), rabbit anti-ORC2, rabbit anti-minichromosome maintenance protein 2 (anti-MCM2), rabbit anti-MCM3, rabbit anti-MCM4, rabbit anti-MCM5, rabbit anti-MCM6, and rabbit anti-MCM7 (BD Biosciences, San Jose, Calif.) antibodies were used for immunoprecipitation or Western blot analysis. Rabbit anti-phosphorylated p53 (anti-phospho-p53) (phosphorylated at serine 15), rabbit anti-phospho-CHK1 (serine 345), rabbit anti-phospho-CHK2 (threonine 68), rabbit anti-phospho-NBS1 (serine 343) (Cell Signaling Technology), rabbit and mouse anti-phospho-H2AX (serine 139) (Upstate Biotechnology), and rabbit anti-phospho-ATM (serine 1981) (Abcam, Cambridge, Mass.) antibodies were used for detecting the specific phosphorylated proteins. Horseradish peroxidase-conjugated sheep anti-mouse, donkey anti-rabbit (Amersham Biosciences, Little Chalfont, England), and donkey anti-goat (Santa Cruz Biotechnology) were used as secondary antibodies. Phospho-serine/threonine ATM/ATR substrate antibody (Cell Signaling Technology) was used for immunoprecipitating phosphorylated ATM or ATR substrates. Cells were lysed in lysis buffer (50 mM HEPES [pH 8.0], 350 mM NaCl, 5 mM EDTA, 1 mM dithiothreitol, 0.1% Nonidet P-40, 1 mM Na₃VO₄, 1 mM NaF) with protease inhibitor cocktail (Roche, Mannheim, Germany) or modified radioimmunoprecipitation assay buffer (Cell Signaling Technology). For immunoprecipitation, 100 μg of cell lysate was incubated with 2 μg of antibody at 4°C overnight, 20 μl of protein A or G plus agarose beads (Santa Cruz Biotechnology) was added and incubated for an additional 2 h. Captured proteins were used for immunoblotting. To detect ATM and ATR, 60 μg of protein per sample was loaded and resolved on a sodium dodecyl sulfate-6% polyacrylamide gel and then electrophoretically transferred to a polyvinylidene difluoride membrane. To detect other proteins, 12% polyacrylamide gels were used.

Plasmid and siRNA transfection.A2780 cells were plated in 6-cm-diameter tissue culture dishes. Transfection was performed when cells reached 80% confluence. The control plasmid (2 μg) and dominant-negative CDK2 (CDK2DN) (43) expression plasmid (2 μg) was transfected with Lipofectamine 2000 reagent (Invitrogen, Carlsbad, Calif.) for 48 h before harvest. Cells were split when reaching confluence. Synthesized small interfering RNA (siRNA) duplexes (Dharmacon, Lafayette, Colo.) were transfected with Lipofectamine 2000 reagent at 50 nM. After 48 h, cells were harvested or treated with drugs for protein analysis. The targeting sequences follow: for luciferase, CAT TCT ATC CTC TAG AGG ATG; for p53, AAG ACT CCA GTG GTA ATC TAC; for ATM, TAG AGC TAC AGA ACG AAA G, GAA TGT GAA CAC CAC CAA A, CTA CAC AAA TAT TGA GGA T, and CTG TAC TTC CAT ACT TGA T; for ATR, ATR-1 (AAG CCA AGA CAA ATT CTG TGT) and ATR-2 (AAC CTC CGT GAT GTT GCT TGA), and for CDK2, CAA AGC CAG AAA CAA GTT G, AAA TAA ACT CTA CCT GGT T, AAA CCT CAG AAT CTG CTT A, and GTT ACT TCT ATG CCT GAT T.

Retrovirus production.A p53 siRNA expression cassette with puromycin selective marker or an empty cassette was cloned into pLXSN vector (BD Biosciences). Viral stock was made by transfecting the constructed plasmid into Phoenix amphotropic packaging cell line (provided by Gary Nolan). After transduction, A2780 cells were grown in medium containing puromycin.

Synchronization and cell cycle analysis.A2780 cells were treated with aphidicolin (6 μg/ml) for 16 h. After release, either DMSO or compound 25 was added. Cells were collected at different time points, fixed in 70% ethanol, and stained with propidium iodide. Flow cytometric analysis was performed on a FACSCalibur system (BD Biosciences).

Isolation of chromatin-enriched cellular fraction.A2780 cells were lysed in cytoskeleton extraction (CSK) buffer at 4°C for 20 min. CSK buffer consists of 10 mM piperazine-N,N′-bis(2-ethanesulfonic acid) (pH 7.0), 100 mM NaCl, 300 mM sucrose, 3 mM MgCl₂, and 0.1% Nonidet P-40 plus protease inhibitor cocktail (Roche). Lysates were centrifuged at 300 × g for 5 min in the cold. Chromatin-enriched pellets were washed with CSK buffer and further lysed in high-salt radioimmunoprecipitation assay buffer (20 mM Tris-HCl [pH 7.5], 350 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton, 2.5 mM sodium pyrophosphate, 1 mM β-glycerophosphate, 1 mM Na₃VO₄, 1 mM NaF) at 4°C for additional 20 min. The cleared supernatants were used for detecting chromatin-bound protein.

Rapid CDK2 inhibition triggers an intra-S-phase delay.To better understand the precise role of CDK2 in S-phase regulation, we wished to determine the impact of rapid CDK2 inhibition on synchronized cells. We chose a relatively selective CDK small-molecule inhibitor, aminothiazole compound 25 (Bristol-Myers Squibb) (23) to study its effect on checkpoint regulation. Compound 25 is a highly potent CDK2 inhibitor. It is 9 times more potent against CDK2 than CDC2 and more than 100 times less potent against CDK4 (23) (Table 1). We examined the cell cycle effect of compound 25 on S-phase progression in A2780 cells, an ovarian cancer cell line with functional p53 and pRb. A2780 cells were synchronized with aphidicolin at the G₁/S boundary (Fig. 1). After release into S phase, cells were treated with compound 25 or DMSO. Compared to DMSO-treated cells, cells incubated with compound 25 progressed through S phase in an obviously slower manner, as judged by the kinetics of DNA content shift from 2N-rich to 4N-rich population. This pattern is characteristic of an intra-S-phase delay and is also associated with increased apoptosis in these cells (Fig. 1). Our observation is consistent with findings on flavopiridol (22), which inhibits CDK1, CDK2, and CDK4 (17) (Table 1).

• Open in new tab
• Download powerpoint

FIG. 1.

CDK inhibition by compound 25 delays S-phase progression. A2780 cells were synchronized with aphidicolin for 16 h. After release, cells were treated with either compound 25 (CPD25) or DMSO. Cells were collected at the indicated time points with propidium iodide staining for flow cytometric analysis.

CDK2 inhibition induces p53 and p21.We speculated that the delayed S-phase progression induced by compound 25 may link to the activation of an intra-S-phase checkpoint. Given the profound roles of p53 and p21 in enforcing cell cycle arrest and preventing DNA rereplication during checkpoint activation (44, 45), we initially studied whether p53 and p21 were involved. After A2780 cells were treated with compound 25 for 24 h, we examined any possible changes in two CDK2 complexes, CDK2/cyclin A and CDK2/cyclin E. With compound 25 treatment, a large amount of p21 was coimmunoprecipitated with CDK2, cyclin A, and cyclin E antibodies (Fig. 2A). The increased association of p21 with CDK2 complexes correlated with an increase in total levels of p21 (Fig. 2A). Since it is well-known that p21 is a p53-induced CDK inhibitor (15), we next examined the time course response of p53 and p21 in asynchronously growing A2780 cells after CDK2 inhibition. We observed an increase of p53 at the protein level as early as 1 h after compound 25 treatment, accompanied by a steady accumulation at later time points. The response of p21 followed the pattern of p53 induction, though it lagged behind the p53 response, suggesting a p53-regulated mode (Fig. 2B). Transfection of p53 siRNA efficiently knocked down not only p53 but also the p21 response triggered by compound 25, further confirming that the increase in p21 is dependent on p53 (Fig. 2C). p53 induction by compound 25 was also seen in mouse embryo fibroblasts and human WS1 fibroblast cells (data not shown), indicating it is a general effect, rather than a cell type-specific effect.

• Open in new tab
• Download powerpoint

FIG. 2.

CDK inhibition activates p53 and p21. (A) Western blot analysis of p21 in total cell lysates and samples immunoprecipitated (IP) by anti-cyclin A, anti-cyclin E, or anti-CDK2 antibody (Ab) after treatment with compound 25 (CPD25). Ig, immunoglobulin. (B) Time course of p53 and p21 induction by compound 25 in asynchronously growing A2780 cells. (C) p53-dependent p21 response induced by compound 25. A2780 cells were transfected with p53 siRNA or luciferase (Luc) siRNA and then treated with 0.5 μM compound 25. Samples were collected for immunoblotting. (D) Expression of phospho-p53 (phosphorylated at serine 15), p53, and p21 induced by roscovitine, olomoucine, and compound 25. (E) CDK2 siRNA induces p53 and p21 response. Luciferase siRNA or CDK2 siRNA was transfected into A2780 cells, and samples were harvested after 48 h. The protein levels of CDK2, phospho-p53 (serine 15), p53, p21, cleaved PARP (p89), and loading control were shown. (F) A2780 cells were transfected with either a control plasmid or plasmid expressing a hemagglutinin-tagged CDK2DN. At 48 h, phospho-p53 (serine 15) and phospho-H2AX at serine 139 (γ-H2AX) were blotted.

Induction of p53 phosphorylation at serine 15 by CDK2 inhibition.We next addressed the mechanism of p53 induction. Phosphorylation is important for stabilization of p53 under certain stress conditions. In response to different stimuli, p53 is phosphorylated at multiple sites, thus becoming less sensitive to MDM2-mediated and ubiquitin-dependent degradation (45). CDK inhibition by compound 25 induced accumulation of p53 phosphorylation at serine 15, which correlated with an increase in total p53 level (Fig. 2D). In addition, two structurally different general CDK inhibitors, roscovitine and olomoucine, which target CDK1 and CDK2 but not CDK4 and CDK6 (17) (Table 1), induced both p53 and p21 significantly in A2780 cells, suggesting that p53 induction is not dependent on inhibiting CDK4 activity (Fig. 2D).

Collectively, our data imply that the activation of p53 by compound 25 requires the involvement of rapid CDK inhibition. To exclude the possibility that our observation was only a nonspecific off-target effect of the synthetic small-molecule inhibitors and to discriminate between the effects of CDK2 versus non-CDK2 inhibition, we used two other approaches to validate our findings. First, we used siRNA specifically targeting CDK2. CDK2 siRNA transfection not only efficiently decreased CDK2 protein level but also induced p53 response with increased serine 15 phosphorylation (Fig. 2E). After CDK2 siRNA transfection, we also noticed a modest increase in p89 PARP, a cleavage product of caspases 3 and 7. This is consistent with our observation that compound 25 triggers caspase 3 activation in A2780 cells (data not shown). Second, expression of a CDK2DN (D145N) (43) induced a similar increase in p53 phosphorylation at serine 15 and p21 (Fig. 2F). It is noteworthy that CDK2DN expression induces a pattern of delayed S-phase progression (19, 41), bearing a close resemblance to the effect of compound 25. Taken together, CDK2 small-molecule inhibitors, CDK2 siRNA, and CDK2DN all affect the steady-state level of p53.

Phosphorylation of ATM and ATR substrates by CDK2 inhibition.Serine 15 of p53 is a target for multiple serine/threonine kinases (4, 9, 37, 38, 42). To identify the kinase responsible for this phosphorylation, we used different kinase inhibitors to screen the possible involvement of candidate kinase(s) during CDK2 inhibition. Caffeine and LY294002, both PI3K inhibitors, effectively blocked the induction of p53 phosphorylation by compound 25, whereas neither p38 inhibitors (SB202190 and SB203580) nor MEK inhibitors (PB98059 and U0126) had any apparent inhibitory effects (Fig. 3A).

• Open in new tab
• Download powerpoint

FIG. 3.

ATM and ATR are involved in p53 phosphorylation at serine 15 by CDK2 inhibition. (A) A2780 cells were treated with the indicated inhibitors for 4 h. Compound 25 was subsequently added to the medium for an additional 4 h. PI3K inhibitors caffeine and LY294002 effectively blocked compound 25-induced p53 phosphorylation and accumulation, while p38 inhibitors SB202190 (SB1), SB203580 (SB2), and MEK inhibitors (PD98059 and U0126) had no effect. (B and C) After either treatment with compound 25 or transfection of CDK2 siRNA, cell lysates were incubated with a specific phospho-serine/threonine ATM/ATR substrate antibody (Ab) and subject to immunoprecipitation (IP). p53, CHK1, and NBS1 were phosphorylated and immunoprecipitated by this antibody. Luciferase (Luc) siRNA transfection was used as a control for CDK2 siRNA transfection. (D) H2AX phosphorylation at serine 139 (γ-H2AX) was induced by compound 25, roscovitine, and olomoucine. (E) A2780 cells were treated with DMSO and compound 25 (CPD25) for 4 h. Samples were blotted with phospho-CHK1 (serine 345) or phospho-CHK2 (threonine 68) antibody. Total levels of CHK1 and CHK2 were also shown. (F) A2780 cells were treated with DMSO or compound 25 for 4 h. Samples were blotted with phospho-ATM (serine 1981) or ATM antibody.

Since ATR and ATM are two PI3K family members that phosphorylate p53 at serine 15 upon activation (6, 16, 36, 49), we reasoned that ATM and ATR activation may follow CDK2 inhibition. If this were true, concomitantly we would expect an increased pool of phosphorylated ATM and ATR substrates after CDK2 suppression. To verify this hypothesis, we used a specific phospho-serine/threonine ATM/ATR substrate antibody to immunoprecipitate phosphorylated proteins. Afterwards we screened the pulled-down proteins with different antibodies against possible ATM and ATR targets. We identified that p53, CHK1, and NBS1, which are all ATM and ATR targets involved in cell cycle checkpoint activation (1), were phosphorylated in both inhibitor-treated and CDK2 siRNA-transfected A2780 cells (Fig. 3B and C). As further evidence of ATM and ATR activation, we also detected the increased phosphorylation of CHK1 at serine 345 and of CHK2 at threonine 68 by direct Western blotting (Fig. 3E). As more direct evidence, autophosphorylation of ATM at serine 1981 was detected within 4 h of CDK2 inhibition (Fig. 3F), an event recently shown as an early marker of ATM activation (3). Last, a histone H2A variant, H2AX, which is a critical regulator in DNA damage response and a well-established ATM or ATR substrate (8, 35, 47), was phosphorylated at serine 139 (γ-H2AX) after either treatment with CDK inhibitors (compound 25, roscovitine, and olomoucine) (Fig. 3D) or CDK2DN expression (Fig. 2F).

To investigate the role of ATM during CDK2 inhibition, we used two lymphoblast cell lines, one carrying an ATM mutation (GM01526), and the other with functional ATM (GM02254). Treatment of GM02254 with compound 25 led to a rapid accumulation of p53 phosphorylation at serine 15 and an increase in p21 (Fig. 4A and B). GM02254 cells also had elevated levels of NBS1 phosphorylation at serine 343 (Fig. 4B),and serine 139 phosphorylation of H2AX (Fig. 4C). In contrast, GM01526 cells had a dramatically delayed and diminished p53-p21 response (Fig. 4A and B), and phosphorylation of NBS1 and H2AX was not detected in these cells (Fig. 4B and C). These data strongly implicate ATM as a critical mediator in CDK2 inhibition-induced checkpoint activation.

• Open in new tab
• Download powerpoint

FIG. 4.

ATM and ATR modulate CDK2 inhibition-induced checkpoint activation. (A) GM01526 cells (ATM deficiency) and GM02254 cells (wild-type ATM) were incubated with compound 25 (CPD25) at the indicated concentrations for 24 h. Levels of phospho-p53 (phosphorylated at serine 15), total p53, and p21 were detected by Western blotting. (B) Phosphorylation of p53 at serine 15, H2AX at serine 139 (γ-H2AX), and NBS1 at serine 343 in GM01526 and GM02254 cells induced by compound 25 at the indicated time points. (C) γ-H2AX induction was present in GM02254 cells but absent in GM01526 cells. (D) A2780 cells were transfected with siRNA targeting luciferase, ATM, or ATR. After 48 h, the protein levels of ATM and ATR were examined. (E) Forty-eight hours after siRNA transfection, A2780 cells were treated with compound 25 for an additional 4 h, phospho-p53 (serine 15) and γ-H2AX were examined by Western blotting with specific antibodies.

However, in ATM-deficient GM01526 cells, the presence of a weak, delayed but steady accumulation of p53 and p21 (Fig. 4A and B) suggested that other mechanisms may be also involved, such as the ATM-related kinase ATR. To more directly explore the roles of ATM and ATR, we used siRNAs targeting ATM or ATR. The transfection of ATM siRNA into A2780 cells effectively decreased the protein level of ATM (Fig. 4D). ATR-2 duplex is more potent than ATR-1 duplex in ATR knock-down efficiency (Fig. 4D), which is consistent with the results of the study by Casper et al. (11). After treatment with compound 25, A2780 cells expressing ATM siRNA showed less phosphorylation of p53 and H2AX than cells transfected with luciferase siRNA (Fig. 4E). Interestingly, knocking down ATR decreased phosphorylation of p53 at serine 15, while it had minimal effect on H2AX phosphorylation (Fig. 4E). These findings are consistent with the idea that ATM and ATR are both activated by CDK2 inhibition, stimulating the phosphorylation of several distinct substrates with potential substrate preferences.

CDK2 and p53 cooperate to inhibit rereplication during cell cycle S phase.In an effort to dissect the role of p53 in the CDK2 inhibition-induced checkpoint response, A2780 p53⁻ cells were generated by stable expression of p53 siRNA through retroviral integration, while isogenic A2780 p53⁺ cells were transduced with a control retrovirus. Compound 25 induced a dose-dependent p53-p21 response in A2780 p53⁺ cells that was apparently absent in A2780 p53⁻ cells (Fig. 5A). Next, cells were synchronized at the G₁/S boundary with aphidicolin and later were released into medium containing compound 25. A2780 p53⁻ cells had similar cell cycle kinetics as A2780 p53⁺ cells for entry into S phase, with the exception of a slightly faster progression at the 8-h time point, the late stage of S phase (Fig. 5B). In addition, less apoptosis was observed in these cells, as shown in Fig. 5B (gate M1). Surprisingly, in the presence of CDK2 inhibitor, these A2780 p53⁻ cells progressed through S phase with an increased >4N DNA content (Fig. 5B). Thus, although intra-S-phase delay by CDK2 inhibition may not solely depend on the status of p53 and p21, the induction of p53 seems a necessary step for preventing the emergence of cells with more than 4N DNA content during low-CDK2-activity S phase, possibly via the selective deletion of cells with unresolved rereplication.

• Open in new tab
• Download powerpoint

FIG.5.

Checkpoint activation by CDK2 inhibition is associated with rereplication. (A and B) Role of p53 in intra-S-phase checkpoint response. A2780 p53⁻ cells were stably transduced by a retrovirus expressing p53 siRNA. A2780 p53⁺ cells were infected with a control virus. (A) Cells were treated with compound 25 (CPD25) for 24 h. p53 and p21 were detected by Western blotting. (B) A2780 p53⁺ and A2780 p53⁻ cells were synchronized by aphidicolin and then released into medium containing 250 nM compound 25 (cpd25). Cell cycle profiles were examined at different time points. The percentages of rereplicating cells (with more than 4N DNA content [gate M2]) and sub-G₁ cells (gate M1) are shown. (C) Dynamic change of MCMs during S phase. A2780 cells were synchronized with aphidicolin (Aphi) and subsequently released into medium containing DMSO or compound 25. At the indicated time points, chromatin-enriched fractions were collected. Protein levels of chromatin-bound MCMs, RPA p70 subunit, and p53 were probed with specific antibodies. DNA-bound ORC2 levels were used as loading control.

Replication of genomic DNA requires the preassembly of MCM complex onto chromatin, an essential component for replication initiation machinery (7, 25). Since MCM proteins are CDK2 substrates in mammalian cells, as well as the DNA helicase responsible for unwinding the DNA helix at replication initiation (20, 21), we hypothesized that inhibition of CDK2 activity during S phase may affect the loading of the MCM complex onto chromatin, setting a cellular state that is suitable for replication reinitiation. We monitored the dynamic change of MCMs on chromatin by using a similar synchronization or release procedure as described in earlier experiments. With CDK2 inhibition, we detected an unexpected increase in binding of MCMs (MCM2, -3, -4, -5, -6, and -7) onto chromatin during S phase (Fig. 5C). The increased loading was not caused by change in total levels of MCM proteins (data not shown). ORC2 levels were used as loading control for chromatin-binding proteins, as the levels did not change during a replication cycle (Fig. 5C). In addition, we observed enhanced RPA loading onto chromatin (Fig. 5C), which is consistent with the scenario that an intra-S-phase checkpoint is activated by CDK2 inhibition, as RPA-bound single-stranded DNA is a structural requirement for recruiting ATR (51). These events occurred simultaneously with p53 accumulation (Fig. 5C) and with up-regulated activity of MRE11-RAD50-NBS1 DNA damage repair complex, as evidenced by increased NBS1 phosphorylation (data not shown). This apparent increased loading of the MCM complex onto chromatin in the absence of CDK2 activity may create a cellular environment favoring DNA rereplication, which in turn could lead to the amplification of the ATM and ATR cascade.

Impact of checkpoint activation by CDK2 inhibition during DNA replication.Our experiments show that CDK2 inhibition leads to the activation of ATM- or ATR-mediated checkpoint response. To determine whether checkpoint activation affects the cellular response to CDK2 inhibitor, we used caffeine to alleviate the checkpoint response and studied its effect on cell cycle progression in combination with CDK2 inhibition. A2780 cells were synchronized by aphidicolin in the presence or absence of caffeine and subsequently released into S phase. Caffeine- and compound 25-treated A2780 cells successfully entered early S phase by 2 h after release, as shown by DNA staining and bromodeoxyuridine labeling (Fig. 6). However, most of these cells failed to progress further at later time points, with significantly increased sub-G₁ population (Fig. 6). Neither drug alone induced this level of apoptosis (Fig. 6). Thus, it appears that cells entering S phase with suppressed CDK2 activity and deficient checkpoint response are prone to apoptosis. To summarize, checkpoint activation by compound 25, which leads to an S-phase delay, may provide an overall survival mechanism to protect cells proceeding through late S phase with inhibited CDK2.

• Open in new tab
• Download powerpoint

FIG. 6.

Inhibition of intra-S-phase checkpoint activation by caffeine sensitizes A2780 cells to compound 25-induced apoptosis. A2780 cells were incubated with caffeine or mock treated for 3 h before aphidicolin synchronization. Cells were released into medium containing DMSO, DMSO plus caffeine, compound 25 (250 nM) alone, or compound 25 (250 nM) with caffeine. The cell cycle profile was examined by propidium iodide staining as well as bromodeoxyuridine (BrdU) labeling. In caffeine-treated samples, caffeine was in the medium during the entire experiment. The percentages of cells in S phase (R3), G₂ phase (R4), and sub-G₁ (R5) are shown.