Flavopiridol Induces Cell Cycle Arrest and p53-independent Apoptosis in Non-Small Cell Lung Cancer Cell Lines¹

Eukaryotic cell cycle progression is governed by cdks³, a family of enzymes that are regulated by phosphorylation and activated by their association with cyclins (1). In addition, two groups of inhibitors negatively regulate cdk activity. Universal inhibitors, including p21^Waf1 and p27^Kip1, inhibit all known cdks; ink4 family members, including p16^INK4A, inhibit only d-type cdk4 and cdk6 (2). In human cancers, including NSCLCs, cdk inhibitors are commonly absent or present at diminished levels. For example, p16^INK4A is frequently lost in tumors that retain a wild-type RB-susceptibility protein (3, 4). p21^Waf1 is transcriptionally regulated by p53-dependent and -independent pathways, (5, 6, 7, 8) and is difficult to detect in tumor cells carrying a p53 mutation. p27^Kip1, under control of the ubiquitin system (9), is also frequently present only at low levels in tumor cells, and patients whose tumors express the lowest levels tend to have a worse prognosis (10, 11, 12, 13).

Ectopic expression of cdk inhibitors in tumor cell lines often causes cell cycle arrest. For example, p21^Waf1 expression results in arrest in G₁ and G₂ (14, 15); high levels of p16^INK4A expression results in G₁ arrest in cells expressing wild-type RB (16, 17). The ability of cdk inhibitor expression to cause cell death has also been investigated, and the results are complex. For example, expression of p16^INK4A and p21^Waf1 can protect cells from apoptosis in some model systems (18, 19, 20). In contrast, infection of various cell types with adenoviruses expressing p16^INK4A, p18^INK4C, or p27^Kip1 can result in apoptosis, frequently in a delayed, p53-independent fashion (21, 22, 23).

These observations suggest that pharmacological cdk inhibition may inhibit tumor growth by causing cell cycle arrest and perhaps apoptosis; the latter may be cell type and context dependent (24). In this regard, the cdk inhibitor olomoucine can induce apoptosis in proliferating PC12 cells, whereas cdk inhibition can protect the same cells from apoptosis if they are differentiated (25). Another cdk inhibitor, butyrolactone I, has been shown to induce abrupt apoptosis in HL60 cells (26).

Flavopiridol, presently in clinical trials, is a novel flavone, structurally related to a natural alkaloid originally purified from the extract of the stem bark of Dysoxylum binectariferum, a plant native to India (27). The drug is a potent inhibitor of several cdks, including cdk1 (cdc2), cdk2, cdk4, and cdk7, because it has been shown to directly inhibit the kinase activity of cdks from exponentially growing cells with IC₅₀s of 100–400 nm(27, 28, 29). Structural studies have demonstrated that the aromatic portion of flavopiridol interferes with binding of ATP to the adenine binding pocket of the cdks (30). Furthermore, inhibition of cdk7 prevents the phosphorylation events necessary for activating the other cdks; therefore, the inhibition of cdk1, cdk2, and cdk4 is both direct and indirect.

Flavopiridol can cause arrest both in G₁ and G₂ phases of the cell cycle, consistent with its inhibition of cdk2, cdk4, and cdk1 (31). In addition, it is active against tumor xenografts in nude mice (27, 32, 33, 34). Recently, flavopiridol has been shown not only to be cytostatic but also cytotoxic to both hematopoietic and epithelial cell lines of diverse origin, including those derived from prostate, esophageal, lung, and head and neck carcinomas (32, 33, 34, 35, 36, 37, 38). In many cases, cytotoxicity is the result of apoptosis, based on cell morphology and DNA fragmentation. In other cells, including the one NSCLC cell line examined, morphological hallmarks of apoptosis have not been observed (35).

In clinical trials, flavopiridol is presently administered by a 72-h continuous infusion, resulting in median steady-state concentrations in the 200–300 nm range (39). To assess the potential activity of flavopiridol against NSCLC, we examined the effect of 72 h of treatment with 100–500 nm flavopiridol on several NSCLC cell lines. Like the majority of NSCLCs, these express wild-type RB and lack p16^INK4A, and most (except A549 cells) also lack functional p53. Here, we demonstrate that after prolonged exposure of 48–72 h, flavopiridol is cytotoxic to seven cell lines. Although some cell lines required a dose of 500 nm for significant cytotoxicity, most were sensitive at lower doses of flavopiridol, which are more easily achieved in vivo. At this dose range, shorter exposure to drug (i.e., 24 h) was insufficient to achieve equivalent cytotoxicity. In the majority of cell lines examined, cell death was preceded or accompanied by cell cycle arrest and occurred by p53-independent apoptosis.

NSCLC cell lines, including A549, Calu-1, SK-Lu-1, Calu-6, NCI-H520, NCI-H661, and SW900, were obtained from the American Type Culture Collection (Rockville, MD). Cells were maintained in DMEM supplemented with 10% bovine calf serum. All express RB and lack functional p16^INK4A (3, 40, 41, 42, 43, 44). A549 cells express wild-type p53 (45). The p53 status of SW900 cells is not documented in the literature; in the remaining five cell lines, p53 is deleted or inactivated by mutation (45, 46, 47, 48).

PA317 packaging cell lines expressing the LXSN and LXSN16E6 amphotropic viruses were provided by Erik Espling and Denise Galloway (49, 50). Filtered viral supernatants, containing Polybrene at a final concentration of 4 μg/ml, were used to infect subconfluent A549 cells. After 48 h, cells were trypsinized and selected in G418 at 400 μg/ml; several colonies were isolated individually, and others were subsequently pooled and maintained in G418.

Flavopiridol was provided by Dr. Edward A. Sausville of the Developmental Therapeutics Program, National Cancer Institute. A 50 mm stock of flavopiridol was prepared in DMSO and stored at −70°C. Drug was diluted in medium and used at final concentrations ranging from 100 to 2000 nm. Twenty-four to 30 h prior to treatment, cells were plated at a subconfluent density of ∼1 × 10⁶/100-mm dish (A549, NCI-H520, and SW900), 2 × 10⁶/100-mm dish (Calu-6), or 8 × 10⁵/100-mm dish (Calu-1, NCI-H661). In all cases, untreated cells behaved identically to those treated with DMSO alone.

Nonadherent and adherent cells (collected by trypsinization) were pooled or kept separated, depending on the experimental goals. Cell lysates were prepared in NP40 lysis buffer [50 mm Tris-HCl (pH 8.0), 150 mm NaCl, 1.0% NP40, and 1 mm phenylmethylsulfonyl fluoride]. Extracts were clarified by centrifugation, and protein concentrations were determined using the Bradford assay (Bio-Rad, Richmond, CA). Equivalent amounts (usually 100 μg) were subjected to SDS-PAGE. Gels were soaked in transfer buffer [10 mm 3-(cyclohexylamino)-1-propanesulfonic acid (pH 11.0) and 15% methanol], and protein was electrophoretically transferred to Immobilon-P membranes (Millipore, Danvers, MA). Nonspecific binding sites were blocked by incubation with Tris-buffered saline (TBS)/10% nonfat dried milk. Primary antibody incubations were carried out in TBS/1% dried milk using the following antibodies: anti-PARP C2-10 (PharMingen, San Diego, CA), anti-p53 DO-1, anti-p27 N-20, and anti-p27 C-19 (all from Santa Cruz Biotechnology, Santa Cruz, CA), and anti-p21^Waf1 (Oncogene Research Products, Cambridge, MA or Transduction Laboratories, Lexington, Kentucky). Primary antibody dilutions were 1:1000 for all antibodies except anti-p21^Waf1 (Oncogene Research Products), which was diluted at 1:200. Filters were washed with TBS/0.1% Tween 20, incubated with horseradish peroxidase-conjugated secondary antibody (Amersham, Arlington Heights, IL), washed again, developed with enhanced chemiluminescence substrate (Amersham) according to the manufacturer’s instructions, and exposed to X-ray film. Equal loading of extracts was confirmed using anti-tubulin (Sigma Chemical Co., St. Louis, MO) or anti-PTP1D (Transduction Laboratories).

At the appropriate time after drug exposure, attached cells were released by trypsinization and combined with nonadherent cells. After centrifugation, cells were resuspended in PBS and treated with 0.2% trypan blue, and trypan blue-excluding cells were counted using a hemocytometer. Viable cell counts were always compared with the number of cells/dish at time 0 of flavopiridol treatment.

For analysis of DNA content, adherent cells were collected by trypsinization, combined with nonadherent cells, washed, and resuspended in 1 ml of PBS. An additional 1 ml of 80% ethanol was added, and cells were fixed overnight at 4°C. Fixed cells were centrifuged and resuspended in 0.5 ml of 500 μg/ml RNase A and incubated for 45 min at 37°C. Cells were centrifuged and resuspended in 1 ml of 69 μm PI in 38 mm sodium citrate and incubated at room temperature for a minimum of 30 min. Cells were analyzed for DNA content by flow cytometry. Both ModFit (Verity Software House, Topsham, ME) and CellQuest (Becton Dickinson, San Jose, CA) programs were used.

A fluorescein apoptosis detection system was used (Promega, Madison, WI). Briefly, after drug treatment, adherent cells were released by trypsinization, combined with nonadherent cells, collected by centrifugation, rinsed, and resuspended in 0.5 ml of PBS. Cells were fixed by adding 5 ml of 1% methanol-free formaldehyde for 20 min on ice. Cells were collected by centrifugation, rinsed twice, and resuspended in 0.5 ml of PBS and added to 70% ice-cold ethanol. After ethanol fixation, cells were rinsed and resuspended in PBS, and 5 × 10⁵ cells were transferred to a 15-ml conical tube. After centrifugation, the pellet was equilibrated in the manufacturer’s buffer and incubated with fluorescein-12 dUTP in the presence or absence of TdT for 90 min at 37°C, with protection from direct light exposure. The reaction was terminated by adding 1 ml of 20 mm EDTA. Cells were washed twice in PBS containing 5 mg/ml BSA and 0.1% Triton X-100, and the cell pellet was finally resuspended in 0.5 ml of PBS containing 5 μg/ml PI and 500 μg/ml RNase A. After a 30-min room temperature incubation in the dark, cells were analyzed by two-color flow cytometry.

Cells were lysed for 1 h at 4°C in 50 mm HEPES (pH 7.2), 150 mm NaCl, 1 mm EDTA, 2.5 mm EGTA, 1 mm DTT, and 0.1% Tween 20 supplemented with 10% glycerol, 1 mm NaF, 0.5 mm sodium orthovanadate, 1 μg/ml aprotinin, 1 μg/ml leupeptin, 10 mm β-glycerophosphate, and 100 μg/ml phenylmethylsulfonyl fluoride. Lysates were clarified by centrifugation, and protein concentration was determined by Bradford assay. One mg of protein was precleared with rabbit IgG prior to immune precipitation using 1 μg of rabbit antiserum against cdk2. Immune precipitates were collected by using protein A-Sepharose beads equilibrated with lysis buffer containing 4% BSA. Beads were washed four times with lysis buffer and then twice with kinase buffer containing 50 mm HEPES (pH 7.2), 10 mm MgCl₂, 5 mm MnCl₂, and 1 mm DTT. After the final wash, 25 μl of kinase mix were added, consisting of kinase buffer with 20 μm ATP, 10 μCi of [γ-³³P]ATP, and 1 μg of histone H1. Samples were incubated at 37°C for 30 min with occasional mixing, boiled in SDS-PAGE sample buffer, and fractionated by electrophoresis through 10% polyacrylamide gels. Proteins were electrophoretically transferred to Immobilon-P, and phosphorylated Histone H1 was visualized by autoradiography. Bands were quantitated using phosphorescence image analysis.

Seventy-two h of continuous exposure to 200–500 nm flavopiridol resulted in cessation of A549 cell growth and a small amount of cell death during the first 48 h of treatment. This was followed by significant cell death during the last 24 h, when analyzed by trypan blue exclusion (Fig. 1), confirming an earlier report of the cytotoxic response of these cells. In contrast, cells treated with 100 nm flavopiridol continued to grow but at a slower rate than untreated cells. The IC₅₀s at 24 and 72 h were 212 ± 100 nm and 126 ± 46 nm, respectively (mean ± SD of three experiments).

Flow cytometry was used to analyze the cell cycle distribution of flavopiridol-treated cells. Consistent with the initial inhibition of cell growth, 200–500 nm flavopiridol resulted in both G₁ and G₂ arrest, manifested by an increased G₁ and G₂-M content, decreased S phase content (data for 500 nm flavopiridol are shown in Fig. 2,A), and an inhibition of cdk2 activity (Fig. 2,B). This pattern was evident after 24 h and persisted over the 72 h of treatment. However, at later times, cells with a sub-G₁ content appeared, correlating with cell death evaluated by trypan blue exclusion. A TUNEL assay indicated that the response of A549 cells to flavopiridol was apoptotic (Fig. 3,A). The percentage of TUNEL-positive cells was dose dependent and reached 48 ± 13% (mean ± SD of six experiments) after 72 h of treatment with 500 nm flavopiridol. In addition, time- and dose-dependent cleavage of PARP occurred, also diagnostic of apoptosis (51) in flavopiridol-treated cells (Fig. 3, B and C). Although detection of the M_r 85,000 PARP cleavage product in these cells required preparation of extracts made only from the nonadherent population after flavopiridol treatment, TUNEL assays performed on adherent cells confirmed that this population was also undergoing apoptotic cell death (Fig. 3 D).

Unlike the majority of NSCLC cell lines, A549 cells carry a wild-type p53 gene. Interestingly, flavopiridol exposure resulted in accumulation of p53 in treated cells (Fig. 4,A). In some experiments, p53 accumulation was maximal within the first 24 h. After treatment with 300 nm flavopiridol, a concomitant small increase in p21^Waf1 was also seen (Fig. 4,B); however, at 500 nm, p21^Waf1 levels fell dramatically with time (Fig. 4, A and B). This is most likely the result of cleavage of p21^Waf1 during apoptosis. With antibodies directed against the NH₂-terminal domain of p21^Waf1, the expected cleavage product (52, 53, 54) was seen only in the nonadherent population (Fig. 4,C). p27^Kip1 cleavage during flavopiridol-mediated apoptosis was also demonstrated in these cells (Fig. 4 D).

To determine whether the accumulation of p53 was required for flavopiridol-mediated apoptosis, the HPV16E6 oncoprotein was introduced into A549 cells by retroviral transfer. E6-expressing cells were incapable of G₁ arrest after exposure to DNA-damaging agents, including γ-irradiation and Adriamycin (Fig. 5, A and B). After γ-irradiation, some cells were treated with nocodazole so that G₁ content could be analyzed without the exit of cells from mitosis (50). E6 expression prevented p53 accumulation after DNA damage, although a small p53-independent increase in p21^Waf1 occurred (Fig. 5,C). Taken together, the results are consistent with the abrogation of p53 function in E6-expressing cells. Although flavopiridol does not result in significant p53 accumulation in E6-expressing cells, they readily undergo apoptosis (Fig. 6), indicating that cell death is indeed p53 independent.

To confirm that flavopiridol-mediated apoptosis is p53 independent, we analyzed the response of several lung cancer cell lines lacking wild-type p53 to 72 h of treatment with 100–500 nm drug. In Calu-1 (Fig. 7) and SK-LU-1 cells (not shown), as in A549 cells, cell cycle arrest clearly preceded cell death. Fig. 7, A and B, shows that over the first 48 h of treatment at doses of 400 and 500 nm, there is cessation of cell growth, followed by cell death during the last 24 h of treatment. Consistent with this, Fig. 7,C shows that in Calu-1 cells over the first 24 h, prior to significant cell death and at doses between 300 and 500 nm, there is an increase in cells with G₁ and G₂ contents and marked diminution of cells in S phase, consistent with G₁ and G₂ arrest. Cell cycle arrest persists and is followed by cell death at 72 h, as indicated by a dose-dependent increase in percentage of cells with a sub-G₁ DNA content. The apoptotic response of these cells was confirmed by TUNEL assay and by demonstration of PARP cleavage (Fig. 7, D and E).

In contrast, in NCI-H661, Calu-6 (Fig. 8), and NCI-H520 cells (data not shown), cell death was more marked and apparent as early as 24 h at the highest doses (Fig. 8, A and B). In these cells, diminution in S phase accompanied apoptosis, without the same separation in time between cell cycle arrest and cell death, as was observed in Calu-1 cells (Fig. 8,C). Once again, apoptosis was confirmed by TUNEL assay and by demonstration of PARP cleavage (Fig. 8, D–F); in these cell lines, the M_r 85,000 PARP cleavage product was readily apparent in total and adherent cell extracts as well as nonadherent cell extracts. Finally, cleavage of p27^Kip1 was also demonstrated, another target of caspases during apoptosis (Fig. 8 G).

In summary, cell death after flavopiridol treatment was a common outcome in p53-deficient NSCLC cell lines, with a dose and time-dependent decrease in viable cell number and increase in the percentage of apoptotic cells. IC₅₀s are shown in Table 1.

In SW900 cells, both apoptosis and cell cycle arrest occurred but appeared unrelated (Fig. 9). For example, at 48 h, there is evidence of apoptosis at 500 nm flavopiridol, without apparent cell cycle arrest. The same is true at 72 h at 300 nm flavopiridol. These results demonstrate that flavopiridol-mediated cell cycle arrest need not occur prior to a cytotoxic response in lung cancer cell lines.

To determine whether prolonged continuous treatment was necessary to achieve apoptosis in NSCLC cell lines, cells were treated for 24 h only and collected after an additional 48 h (i.e., at the 72-h time point) and compared with those treated continuously for 72 h. Table 2 demonstrates that in A549 cells, after 24 h of treatment at doses <500 nm, recovery occurs such that cytotoxicity is markedly diminished compared with a 72-h continuous exposure to drug. At higher doses, including 1000 nm and above, cell death at the 72-h time point is similar whether cells have been treated continuously or only for 24 h. In contrast, in NCI-H661 (Table 3) and Calu-1 cells (data not shown), cytotoxicity was greater at all doses after 72 h of continuous exposure.

These experiments were extended to analyze even longer exposures to flavopiridol. At 72 h, flavopiridol was either continued, or drug was removed and replaced with drug-free medium for an additional 48 h. Analysis of samples at 120 h indicated that removal of drug at 72 h did not reverse the lethal effect of flavopiridol, and cells continued to die. Overall, 72 and 120 h of exposure produced comparable cytotoxicity when cells were analyzed at 120 h.

In some cell lines, apoptosis clearly followed cell cycle arrest, suggesting that noncycling cells could undergo apoptosis. To further investigate this, A549 cells were arrested by starvation in 0.1% bovine calf serum and were either left untreated or treated with flavopiridol for an additional 72 h in low serum. Fig. 10 demonstrates that cell death can occur in cells arrested in G₁ prior to treatment. The IC₅₀ at 72 h in starved A549 cells was 195 nm, representing a 1.5-fold increase over that seen in exponentially growing cells.

We have shown that flavopiridol causes apoptotic cytotoxicity in a panel of NSCLC cell lines. This was demonstrated by time- and dose-dependent decreases in viable cell count by trypan blue exclusion and concomitant increases in the percentage of cells with a sub-G₁ DNA content. Apoptosis was confirmed by TUNEL assays and by the demonstration of cleavage of known caspase targets, including PARP, p21^Waf1, and p27^Kip1. Flavopiridol-mediated cytotoxicity had been demonstrated previously in A549 cells. In that study, morphological hallmarks of apoptosis were not observed (35). In contrast, our data clearly indicate that both DNA fragmentation and caspase-mediated events occur in response to flavopiridol, indicating activation of an apoptotic pathway in these cells.

In contrast to flavopiridol, the flavonoid flavone does not result in p53 accumulation in A549 cells, although G₁ arrest occurs because p21^Waf1 is transcriptionally induced in a p53-independent manner (55). Other flavonoids have been shown to activate wild-type p53 in nontransformed cells, including apigenin, luteolin, and quercetin (56), and relatively low doses of apigenin result in p53-dependent increases in p21^Waf1 (57). p53 activation by flavonoids may be secondary to the formation of oxygen radicals or the inhibition of topoisomerase activity (56). Although flavopiridol does not inhibit topoisomerase activity (35, 36), p53 reproducibly accumulated after treatment of A549 cells. In addition, at 300 nm, a small increase in the level of p21^Waf1 was observed. The accumulation of p53 after flavopiridol treatment may be cell type specific, because, for example, it was not observed in SUDHL-4 B-cell lymphoma cells, although this cell line had high baseline levels and was only treated for 24 h (36).

At high doses of other flavonoids, apoptosis was observed during 72 h of treatment; in the nontransformed cell types studied, apoptosis appeared to be p53 dependent (56), but among transformed cells, this was not necessarily the case (58). Despite the accumulation of p53 observed in A549 cells, the apoptosis mediated by flavopiridol is independent of p53 by two criteria: (a) HPV16E6-expressing and control wild-type p53-expressing A549 cells have a similar apoptotic response; and (b) apoptosis readily occurs in cell lines in which p53 is deleted or inactivated by mutation.

Several studies have now demonstrated the association of caspase-mediated cleavage of p21^Waf1 and p27^Kip1 with apoptosis (52, 53, 54). In several systems, it has been proposed that this would facilitate apoptosis by releasing cells from growth arrest or allowing cdk2 activation. Such a mechanism is most likely not operating after flavopiridol treatment, because cell death can follow cell cycle arrest, and continued cdk2 inhibition occurs during flavopiridol exposure. It may be that cleavage of p21^Waf1 and p27^Kip1 is a consequence, rather than a cause, of apoptosis.

The mechanism of p53-independent apoptosis mediated by flavopiridol remains to be elucidated. Although cdks are the known primary targets at the doses studied, other kinases, including protein kinase C and epidermal growth factor receptor tyrosine kinase, are inhibited at higher doses of flavopiridol (27). Although inhibition of cdks clearly mediates cell cycle arrest by flavopiridol, it is not yet proved whether cdk inhibition is the cause of the apoptotic response. In this regard, our data in SW900 cells suggest that cell cycle arrest and apoptosis need not be intimately linked, as has been reported in a B-cell lymphoma cell line (36). In addition, if cdks are the critical targets, it will be important to determine whether inhibition of one or several is necessary for apoptosis. The answer to these questions will require the generation of flavopiridol-resistant cell lines as well as cell lines overexpressing various wild-type or mutant cdks.

We have confirmed the ability of flavopiridol to induce cytotoxicity in noncycling cells. The ability of arrested cells to undergo cell death during flavopiridol exposure was evident in experiments in which cell cycle arrest (both in G₁ and G₂) clearly preceded the time at which maximal cell death occurred. In addition, TUNEL assays suggested that cell death was not confined to a particular phase of the cell cycle. Flavopiridol-mediated cytotoxicity of noncycling cells has been investigated previously in density-arrested A549 cells by two groups of investigators (35, 59). Although both groups found that density-arrested cells were not resistant to flavopiridol, one group reported that the IC₅₀ in arrested cells was significantly higher. In our experiments, serum-starved cells could also undergo cell death in response to flavopiridol, with a 1.5-fold higher IC₅₀. Experiments in both serum-starved NCI-H661 cells and density-arrested Calu-1 cells yielded similar results. Although these cells were clearly killed by flavopiridol, there was a trend suggesting decreased apoptosis (up to 2-fold) in the noncycling populations (data not shown).

In other cells, including NCI-H661, NCI-H520, and Calu-6 cells, a diminution in S phase accompanied the onset of apoptosis. Although cell cycle arrest and apoptosis could be occurring concomitantly, our data do not rule out the possibility that S-phase cells are preferentially dying and are more sensitive to flavopiridol-mediated apoptosis. Additional experiments, including simultaneous bromodeoxyuridine/TUNEL analyses, will be necessary to determine whether this is the case. In this regard, one lung cancer cell line, NCI-H23, has been reported to undergo a delay in S-phase traversal after flavopiridol, with more abrupt apoptosis, evident earlier than 24 h (60). The sensitivity of S-phase cells might explain why flavopiridol potentiates the activity of chemotherapeutic drugs capable of causing S-phase delay (61, 62).

Our data have implications for the clinical use of flavopiridol. The apoptotic response of NSCLC cells suggests promise for the efficacy of flavopiridol in lung cancer. However, it is noteworthy that significant apoptosis in vitro frequently requires doses of 300–500 nm, which are higher than median levels achieved in vivo when a 72-h continuous infusion is used at the maximal tolerated dose (200–300 nm; Ref.39). Our data suggest that if higher levels are achieved (i.e., >500 nm), a 24-h treatment may be acceptable. It is possible that higher doses of flavopiridol may be tolerated if the exposure is shorter. However, the data indicate that if levels >500 nm are not easily achieved, the 72-h continuous infusion is preferable to a shorter period of treatment. This schedule is presently in use in a Phase II trial for patients with advanced NSCLC.

We thank Edward Sausville for helpful discussions and for the supply of flavopiridol used in these experiments. We also thank members of the Dana-Farber Cancer Institute Cell Sorter Facility, including Maris Handley, Joseph O’Brien, Jon Michael Vore, and Michael Price.