Flavopiridol circumvents Bcl-2 family mediated inhibition of apoptosis and drug resistance in B-cell chronic lymphocytic leukaemia


Dr Paul Bentley, Department of Haematology, Llandough Hospital, Penlan Road, Penarth, South Glamorgan CF64 2XX, UK. E-mail: paul.bentley@llanhaem.demon.co.uk


Flavopiridol, a synthetic flavone, is currently under clinical investigation for the treatment of B-cell chronic lymphocytic leukaemia (B-CLL). In this study, we examined the in vitro effects of flavopiridol and fludarabine on B-CLL cells from 64 patients (36 treated and 28 untreated) in terms of apoptosis induction and Bcl-2 family expression. Both flavopiridol and fludarabine induced apoptosis in all the samples tested with mean LD50 values (± SD) of 59·7 nmol/l (± 36·5) and 6·2 μmol/l (± 7·5) respectively. Mean flavopiridol LD50 values were not significantly different between the treated and untreated patient groups (P = 0·35), whereas the fludarabine LD50 values were significantly higher in the previously treated patient group (P = 0·01). Bcl-2 and Mcl-1 expression were downregulated in both flavopiridol and fludarabine-induced apoptotic cells, but the increase in Bax expression that accompanied fludarabine-induced apoptosis was not evident in flavopiridol-treated cells. In addition, Bcl-2:Bax ratios were not predictive of flavopiridol cytotoxicity (P = 0·82), whereas they were highly predictive of in vitro responsiveness to fludarabine (P = 0·001). Overall, these findings suggest that flavopiridol exerts its cytotoxic effect through a novel cell-death pathway that is not subject to the Bcl-2 family mediated resistance mechanisms that reduce the efficacy of many conventional chemotherapeutic drugs.

B-cell chronic lymphocytic leukaemia (B-CLL) is currently an incurable disease characterized by the progressive accumulation of mature-looking B-lymphocytes, most of which are in G0/G1 of the cell cycle. Even in cases of clinically aggressive or advanced disease, most B-CLL cells are quiescent and increase in number primarily because of failed apoptosis rather than as a result of increased proliferation (Reed, 1997). The underlying defect in the apoptotic machinery in B-CLL cells is elusive, but constitutively high expression of the anti-apoptotic protein Bcl-2 is found in most cases (Hanada et al, 1993; Robertson et al, 1996). In addition, the ratio of Bcl-2 to Bax (a homologous pro-apoptotic protein) has been shown to be critical in determining the relative sensitivity of B-CLL cells to chemotherapeutic drugs in vitro (Aguilar-Santelises et al, 1996; Pepper et al, 1996, 1997). Furthermore, the highest Bcl-2:Bax ratios have been found in those patients who had previously received treatment, suggesting that the development of resistance may be mediated through selection of subclones in which high Bcl-2:Bax ratios are found (Pepper et al, 1999). Mcl-1, another anti-apoptotic Bcl-2 family member, is thought to be upregulated in response to DNA damage (Zhan et al, 1997), and in another study of 58 B-CLL patients Mcl-1 expression was inversely correlated with in vivo responsiveness to chemotherapy (Kitada et al, 1998). In contrast, Bcl-XL and BAD, two other Bcl-2 family members, are not usually expressed in B-CLL (Kitada et al, 1998) and are therefore not likely to play a role in either the development of drug resistance or disease progression.

Drug resistance is the major obstacle to the successful management of B-CLL. Although the causes of drug resistance in B-CLL are probably complex, it seems likely that dysregulation of the apoptotic pathway(s) may play a central role (Reed, 1998). In this context, the tumour-suppresser gene p53 and its accessory genes may be key determinants in the development of resistance; several studies have correlated p53 mutation with poor survival and non-responsiveness to therapy (El Rouby et al, 1993; Wattel et al, 1994; Dohner et al, 1995). The first-line therapies of choice in B-CLL, fludarabine and chlorambucil, are thought to induce their effects predominantly through a common apoptotic pathway that is reliant on p53 upregulation in response to DNA damage. Consequently, when drug resistance to one of these agents is established it is likely to have a major impact on the relative effectiveness of other similar agents. Novel therapeutic agents that do not mediate their effects, at least not exclusively, through a p53-dependent route, offer the possibility of effective treatment in resistant patients.

One such agent is flavopiridol, a synthetic flavone that has been shown to inhibit certain protein kinases by interacting with their ATP binding sites (Parker et al, 1998). Although the relevant targets for flavopiridol in B-CLL remain to be established, this agent has been demonstrated to have cytotoxicity in B-CLL, which is not thought to be dependent upon functional p53 (Byrd et al, 1998). In the present study, we assessed the ex vivo expression of Bcl-2, Bax and Mcl-1, as well as changes in these proteins after in vitro culture with both fludarabine and flavopiridol. Furthermore, we assessed whether p53 mRNA expression was modulated in B-CLL cells after exposure to flavopiridol.

Patients and methods

Isolation of leukaemic and normal lymphocytes Peripheral blood samples from 64 patients with B-CLL (36 treated and 28 untreated) were obtained with the patients' informed consent. The treated group of patients had previously received a number of alternative therapeutic regimes, as shown in Table I. Clinical staging was based on the Binet system (Binet et al, 1981). Peripheral blood lymphocytes were isolated using density centrifugation on Ficoll-Hypaque (Sigma, UK) prior to ex vivo analysis of intracellular Bcl-2 family protein expression or dilution in culture medium.

Table I.   Clinical characteristics of the B-CLL patients in this study.
 Ex vivo LD50 values  

  1. CHOP, cyclophosphamide, adriamycin, vincristine and prednisolone; CLB, chlorambucil; E/C, epirubicin, chlorambucil; FLU, fludarabine.

1621910128617·0106372·15·0CE/C CLB
1730358168918·0163220·91·7CE/C FLU

Ex vivo protein expression of Bcl-2, Bax and Mcl-1 All 64 samples were analysed for Bcl-2 family protein expression prior to culture to establish whether baseline expression of these proteins could be used as a predictor of in vitro responsiveness to fludarabine and flavopiridol. The flow cytometric method used for quantification of these proteins is detailed below.

Cell culture conditions Freshly isolated peripheral blood lymphocytes (1 × 106/ml) were cultured in Eagle's medium (Gibco, UK) supplemented with penicillin, streptomycin and 10% fetal calf serum (FCS). All cultures were incubated for 48 h at 37°C in a 5% carbon dioxide atmosphere in the presence of either fludarabine (0·5, 1·0, 2·5, 5·0, 10·0, 15·0 μmol/l) or flavopiridol (5, 10, 25, 50, 75, 100 nmol/l). Control cultures were also performed in which no drug was added to cultured B-CLL cells. Cells were subsequently harvested using centrifugation and analysed by flow cytometry using the methods outlined below. All experiments were performed in triplicate.

Flow cytometric analysis of Bcl-2, Bax and Mcl-1 protein expression Samples from all 64 patients were analysed using triple-colour immunofluorescent staining using combinations of Bcl-2, Bax and Mcl-1 in conjunction with CD19 (pan B-cell marker). Briefly, 1 × 106 cells were incubated with anti-CD19, Cy5 phycoerythrin (PE)-conjugated antibody or an isotype negative control (DAKO, UK). The cells were then fixed using a commercially available kit (Fix & Perm, Caltag, USA), then resuspended in permeabilization solution together with titration-determined volumes of each antibody or isotype-matched negative control, i.e. Bcl-2 fluorescein isothiocyanate (FITC) (DAKO, UK), Bax and Mcl-1 (Santa Cruz Biotechnology, USA). Subsequently, a PE-labelled secondary antibody was added to the Bax and Mcl-1-labelled cells (Sigma, UK). The cells were then resuspended in 0·5 ml of 1% paraformaldehyde prior to flow cytometric analysis using a FACScan flow cytometer (Becton Dickinson, USA). From each sample, 10 000 cells were analysed and non-specific binding was excluded by gating using isotype-negative control antibodies. Gating of the CD19-positive B cells was performed in all of the analyses to ensure that apoptosis and Bcl-2 family protein expression were quantified in the B-lymphocyte subpopulation of cells. The mean fluorescent intensity (MFI) was calculated for each individual protein using WinMDI software (J. Trotter, Scripps Research Institute, USA) and these values were then converted to molecules of equivalent soluble fluorochrome (MESF) using a calibration curve to standardize the data. The calibration curve was constructed by operating the flow cytometer under identical conditions and monitoring a mixture of beads labelled with known amounts of fluorochrome. In this way inter-experimental variation in fluorescence detection was controlled for and, in addition, differences in fluorescence intensity between FITC and PE were normalized.

Measurement of in vitro apoptosis Changes in forward scatter (FSC) and right-angle light scatter (SSC) characteristics were used to quantify apoptotic and viable cell populations as described previously (Pepper et al, 1999). Typically lymphocytes show a reduction in FSC (caused by cytoplasmic shrinkage) and an increase in SSC (owing to increased granularity) when they undergo apoptosis (Ferlini et al, 1996). The measurement of apoptosis using an FSC/SSC gating strategy in conjunction with back gating of CD19-positive lymphocytes allowed simultaneous acquisition of Bcl-2 family protein expression data in viable and apoptotic B-lymphocyte subpopulations. All LD50 values (the concentration of fludarabine or flavopiridol required to kill 50% of cells) were derived from dose–response curves. Duplicate samples were assessed using FITC-labelled annexin V to confirm the presence of apoptotic cells in the cell cultures, and to validate the FSC/SSC quantification method (Vermes et al, 1995). The annexin V assay was performed on CD19-positive B-lymphocytes isolated using CD19-labelled magnetic beads (Dynal, UK).

Detection of p53 messenger RNA Lymphocytes from six B-CLL patients were pelleted at 0, 4, and 24 h after exposure to either flavopiridol or fludarabine in in vitro culture. The mRNA was isolated from 1 × 106 cells using a Dynal mRNA direct kit according to the manufacturers instructions (Dynal, UK) and was subsequently reverse transcribed using an Advantage RT-for-PCR kit (Clontech, UK) to obtain 20 μl of cDNA. This was diluted with 80 μl of diethylprocarbonate (DEPC)-treated water and 5 μl of cDNA was amplified in a 50 μl polymerase chain reaction (PCR) mixture using the following p53 and G3PDH-specific primers: sense primer (p53) 5′-TCTGTGACTTGCACGTAC TC-3′; antisense primer (p53) 5′-CACGGATCTGAAGGGTGA AA-3′; sense primer (G3PDH) 5′-TGAAGGTCGGAGTCAAC GGATTTGGT-3′; antisense primer (G3PDH) 5′-CATGTGGGC CATGAGGTCCACCAC-3′.

Cycling conditions were optimized for 30 amplification cycles using an annealing temperature of 57°C (Aguilar-Santelises et al, 1996). G3PDH-specific primers were used to determine whether the quantity of template cDNA was approximately the same in all PCR reactions. The PCR products were resolved using electrophoresis on a 2% agarose gel and were subsequently photographed after staining of the PCR products with ethidium bromide.

Statistical analysis The data obtained in these experiments were evaluated using the equal variance t-test and correlation coefficients were calculated using least-squares linear regression plots. LD50 values were calculated from line-of-best-fit analysis of the dose–response curves.


Fludarabine and flavopiridol in vitro cytotoxicity

Peripheral blood mononuclear cells from all 64 patients were cultured for 48 h in the presence of a range of concentrations of fludarabine or flavopiridol. Control cultures were also performed in which cells were incubated without the addition of drug. Apoptotic cell death was quantified using changes in FSC/SSC in CD19-gated B lymphocytes. Figure 1 shows the gating strategy used in these experiments, and also illustrates the close correlation between changes in FSC/SSC and annexin V-labelling of apoptotic cells. LD50 values were calculated from the dose–response curves derived from this data. The results are summarized in Table I. All the samples demonstrated in vitro apoptosis in response to fludarabine and flavopiridol with mean LD50 values (± SD) of 6·2 μmol/l (± 7·5) and 59·7 nmol/l (± 36·5) respectively.

Figure 1.

 Comparison of the measurement of apoptosis in CD19-positive (R3-gated) B-CLL lymphocytes using changes in forward light scatter (FSC) and right angle light scatter (SSC) characteristics, and labelling with annexin V and propidium iodide. The annexin V assay was performed on CD19-positive B-lymphocytes isolated using magnetic beads.

Flavopiridol is equally cytotoxic to previously treated and untreated B-CLL cells

Drug resistance is commonly encountered in B-CLL and this becomes more marked in previously treated patients, irrespective of which standard therapy the patient has received previously. In this study, we compared mean LD50 values (± SD) derived from the previously treated (56·9 ± 36·6 nmol/l; n = 36) and the untreated (63·5 ± 36·4 nmol/l; n = 28) patient groups for the novel agent flavopiridol. The mean LD50 values were not significantly different between the two groups (P = 0·35) as analysed by the equal variance t-test. This indicates that in vitro flavopiridol cytotoxicity is not modulated by prior exposure to other therapeutic drugs. These results support the findings of Byrd et al (1998) who previously reported this phenomenon in 27 B-CLL patients (16 untreated and 11 treated). In contrast, in vitro fludarabine LD50 values (± SD) were markedly higher (P = 0·01) in the previously treated patient group (9·8 ± 9·9 μmol/l) compared with the untreated patient group (3·5 ± 2·3 μmol/l).

Flavopiridol cytotoxicity does not correlate with fludarabine cytotoxicity

All 64 patient samples were assessed for in vitro sensitivity to both flavopiridol and fludarabine. Consequently, it was possible to determine whether responsiveness to one of these agents could predict responsiveness to the other. In this study, we found no significant correlation between the LD50 values derived for both drugs (P = 0·078). Because seven of the previously treated patients were considered to be clinically resistant to fludarabine, the analysis was repeated without those patients to ensure that no statistical bias was introduced. There was still no correlation between flavopiridol and fludarabine sensitivities (P = 0·067), indicating that the lack of correlation between the responsiveness to the two drugs was not attributable to the fact that a subgroup of patients were fludarabine-resistant.

Ex vivo Bcl-2 family protein expression

All 64 patient samples were assessed for ex vivo Bcl-2 family expression (i.e. prior to culture). Bcl-2 protein expression was found to be significantly higher (P = 0·008) and Bax expression significantly lower (P = 0·001) in the treated patient group. Hence, Bcl-2:Bax ratios were very significantly higher in the previously treated patient group (P < 0·0001). However, mean Mcl-1 protein expression was not significantly different in the treated and untreated patient groups (P = 0·43). In accordance with our previous work (Pepper et al, 1997), fludarabine LD50 values were shown to correlate with Bcl-2:Bax ratios. In contrast, flavopiridol LD50 values were not correlated with the ex vivo expression of any of the Bcl-2 family members investigated in this study (see Table II).

Table II.   Comparison of drug sensitivity (LD50 values) and ex vivo Bcl-2 family protein expression in 64 B-CLL patients assessed using linear regression analyses and expressed as P values.
 Flavopiridol LD50Fludarabine LD50
Bax0·63 0·018

Bcl-2 family modulation in viable and apoptotic subpopulations

Bcl-2 family protein expression was measured after 48 h culture in the presence of either fludarabine or flavopiridol. Figure 2A shows how apoptotic and viable cells were discriminated using changes in FSC and SSC, and these gates were subsequently used to determine Bcl-2 and Bax protein expression in the two separate populations of cells. The histograms for expression of both Bcl-2 and Bax in viable (solid histograms) and apoptotic cells (open histograms) were overlaid to illustrate the expression differences in the two populations. Exposure to fludarabine led to a 50% reduction, approximately, in Bcl-2 expression and a two- to threefold increase in Bax expression in those cells undergoing apoptosis. These changes were not observed in cells that resisted apoptosis. Figure 2B shows the Bcl-2 and Bax expression in cells exposed to flavopiridol. These cells exhibited a similar reduction in Bcl-2 expression, but showed no evidence of increased Bax expression. Again, these changes were only observed in those cells committed to apoptosis. In addition, Mcl-1 expression was also reduced in the apoptotic cells of cultures exposed to either fludarabine or flavopiridol (data not shown).

Figure 2.

 Dotplots showing the characteristic changes in forward scatter (FSC) and side scatter (SSC) in cells undergoing apoptotic cell death from a B-CLL patient after 48 h in vitro culture with fludarabine (A) and flavopiridol (B). Gates R1 (viable cells) and R2 (apoptotic cells) were then applied to simultaneously derived Bcl-2 and Bax protein expression data, and the separate histograms were overlaid (R1, shaded histograms; R2, open histograms).

Flavopiridol induces apoptosis through a p53-independent pathway

It has previously been suggested that flavopiridol may not require functional p53 in order to induce its cytotoxic effects (Byrd et al, 1998). The results obtained in this study indirectly support this hypothesis because flavopiridol did not induce Bax protein expression. Although post-transcriptional modification of p53 has been shown to be an important mechanism for stabilizing p53 and, hence, increasing its expression in some cell types (Kubbutat et al, 1997), it has not yet been established whether this mechanism is used in B-CLL cells during apoptotic cell death. Therefore, we decided to compare the p53 mRNA expression in B-CLL cells at 0, 4 and 24 h post-incubation with flavopiridol or fludarabine. Although no attempt was made to quantify the results, they indicated that p53 mRNA was poorly induced in flavopiridol-treated cells compared with fludarabine-treated cells at all of the time points studied (data not shown).


Flavopiridol consistently induced apoptosis in vitro in B-CLL cells. Furthermore, it was equally potent against the leukaemic cells of both the untreated and previously treated patients in this study. Even those patients who were clinically refractory to fludarabine showed in vitro responsiveness to flavopiridol. This suggested that flavopiridol was inducing its cytotoxic effects through a different mechanism to fludarabine. Because pleiotropic drug resistance is a commonly observed phenomenon in previously treated B-CLL patients, flavopiridol may provide a useful therapeutic option in this type of scenario.

Bcl-2 family protein expression has been associated with the sensitivity or resistance of cells to a wide range of anti-cancer drugs (Aguilar-Santelises et al, 1996; Pepper et al, 1996; Thomas et al, 1996). In accordance with previous findings, the ex vivo Bcl-2:Bax ratios in the current study were inversely correlated with in vitro responsiveness to fludarabine (P = 0·001). However, no such relationship was demonstrated between Bcl-2:Bax ratios and flavopiridol LD50 values (P = 0·82), indicating that flavopiridol responsiveness was independent of constitutive Bcl-2 and Bax expression. In the light of these findings, we set out to determine whether flavopiridol modulated Bcl-2 family expression in a similar fashion to that observed in fludarabine-treated B-CLL cells. Flavopiridol-treated B-CLL cells showed a consistent downregulation in the expression of the anti-apoptotic protein Bcl-2 in those cells undergoing apoptosis; a phenomenon also observed in fludarabine-treated cells. However, flavopiridol did not induce the marked upregulation in expression of the pro-apoptotic protein Bax, consistently seen in fludarabine-treated cells. These differences were observed in cells exposed to flavopiridol and fludarabine doses of equivalent toxicity, i.e. cell cultures showing equal proportions of apoptosis after exposure to either drug. Therefore, the observed differences in Bax modulation could not be attributed to differences in the percentage of apoptotic cells in each culture. Mcl-1 expression was also measured in the study and was found to be downregulated in apoptotic cells after exposure to either drug in a similar fashion to Bcl-2. However, the downregulation in Bcl-2 and Mcl-1 protein expression was observed only in those cells that exhibited signs of undergoing apoptosis (changes in light scatter), indicating that these changes were the result of apoptosis rather than initiating events in the apoptotic process. These observations are in contrast to those of Kitada et al (2000) who suggested that decreases in both Bcl-2 and Mcl-1 preceded apoptotic changes in B-CLL cells. Although the flow cytometric technique used in these experiments allowed discrimination of Bcl-2 family protein expression in viable and apoptotic subpopulations, more detailed time-course experiments would be required to elucidate the exact role that Bcl-2 and Mcl-1 play in the process of apoptosis because protein expression was only measured at one time point (48 h) in this study.

Previous studies have indicated that flavopiridol may not require functional p53 in order to exert its apoptotic effects (Byrd et al, 1998); the present data supports that notion in two ways. First, the absence of Bax induction after exposure to flavopiridol is consistent with a p53-independent pathway of apoptosis induction; Bax protein expression is a thought to be regulated by p53 binding to the bax promoter region leading to increased transcription and, hence, elevated protein expression (Selvakumaran et al, 1994; Zhan et al, 1994; Miyashita & Reed, 1995). Second, p53 mRNA expression was measured at time intervals after exposure to either flavopiridol or fludarabine; the fludarabine-treated cells showed an apparent increase in p53 mRNA after 4 h compared with the flavopiridol-treated cells. The same pattern was observed in samples taken at 24 h, although the difference was less pronounced; this may be because, although alternative apoptotic pathways exist, these are not mutually exclusive and there is likely to be ‘cross-talk’, with apoptotic signals being amplified by more than one pathway after the commitment to apoptosis.

In conclusion, this study indicates that flavopiridol should be an effective drug in the treatment of B-CLL. It may be particularly useful in the treatment of patients who are resistant to standard chemotherapy as it appears not only to mediate its effect through a p53-independent pathway but also seems to circumvent the Bcl-2 family mediated inhibition of apoptosis commonly seen in B-CLL. In the context of previous findings that have demonstrated a link between p53 mutation, failed apoptosis, non-responsiveness to therapy and poor prognosis in B-CLL (El Rouby et al, 1993; Wattel et al, 1994; Dohner et al, 1995), flavopiridol represents a relevant and timely addition to the chemotherapeutic arsenal for the treatment of this condition.


This work was supported in part by grants from the Leukaemia Research Appeal for Wales and the Welsh Bone Marrow Transplant Research Fund.