Resistance to spontaneous apoptosis in acute myeloid leukaemia blasts is associated with p-glycoprotein expression and function, but not with the presence of FLT3 internal tandem duplications


Dr Monica Pallis, Academic Haematology, Clinical Sciences Building, Nottingham City Hospital, Nottingham NG5 1PB, UK. E-mail:


Summary. The ability of acute myeloid leukaemia (AML) blasts to survive in culture has been associated with poor patient response to chemotherapy. Other biological factors predicting an adverse outcome include p-glycoprotein (pgp) expression, which is associated with a reduced remission rate, and the presence of fms-like tyrosine kinase 3 gene (FLT3) internal tandem duplications (ITDs), predictive of a high rate of leukaemic relapse. Our previous work has indicated a drug efflux-independent role for pgp in apoptosis resistance. We measured spontaneous in vitro apoptosis in 58 primary AML samples to establish its relationship with functional and phenotypic pgp and with FLT3 ITDs. Cells were incubated for 48 h in a suspension culture, and the remaining viable cells were counted by flow cytometry. Median survival was 38% of baseline values. Resistance to spontaneous apoptosis was strongly associated with pgp (MRK-16 antibody) expression (P = 0·001) and with pgp functional activity (P < 0·001). FLT3 ITDs, found in 20 cases, were inversely associated with functional pgp activity: thus, the median pgp modulation ratio was 2·0 in FLT3 wild-type cases and 1·38 in ITD cases (P = 0·018). Also, the presence of FLT3 ITDs was not associated with in vitro apoptosis resistance. In conclusion, we have found that the presence of FLT3 ITDs is not related to AML blast survival in vitro, and is inversely associated with pgp activity, whereas pgp expression and activity are associated with resistance to spontaneous apoptosis. These results may help to explain the differing adverse effects of pgp (on remission induction) and FLT3 ITDs (on relapse) in AML.

According to recent trial reports, at least 50% of older patients and 15–18% of younger patients with acute myeloid leukaemia (AML) fail to respond to remission induction chemotherapy (Leith et al, 1997; Grimwade et al, 1998; Kottaridis et al, 2001). A high risk of relapse is observed in younger patients who do respond to such chemotherapy, for example a risk of 49% at 5 years was observed in the Medical Research Council (MRC) AML 10 trial (Grimwade et al, 1998). The cellular pathological features that contribute to poor outcome have been intensively investigated. Adverse cytogenetics (Leith et al, 1997; Grimwade et al, 1998), p-glycoprotein (pgp) expression (Campos et al, 1992; Del Poeta et al, 1996; Nussler et al, 1996; Hunault et al, 1997; Leith et al, 1997; van den Heuvel-Eibrink et al, 1997) and the fms-like tyrosine kinase 3 gene (FLT3) mutation (Kiyoi et al, 1999; Kottaridis et al, 2001) have all been identified as poor prognostic factors. Interestingly, although pgp expression is predictive for an impaired response to remission–induction chemotherapy (Campos et al, 1992; Del Poeta et al, 1996; Nussler et al, 1996; Hunault et al, 1997; Leith et al, 1997; van den Heuvel-Eibrink et al, 1997), the FLT3 mutation, which is found in ≈ 25% of AML patients, is associated with both poor overall survival and a high relapse rate but not an impaired initial response to chemotherapy (Kiyoi et al, 1999; Kottaridis et al, 2001).

Research into chemoresistant disease was initially based on the assessment of individual drug metabolism but, during the past decade, the possibility that chemoresistance is underpinned by altered apoptotic pathways has been intensively examined. Primary leukaemic blasts die in vitro by an apoptotic mechanism (Banker et al, 1997; Smith et al, 1998; Garrido et al, 2001; Milella et al, 2001). Smith et al (1998) were the first to show that the increased in vitro survival of primary AML blasts was predictive of a poor response to remission–induction chemotherapy. By incubating patient AML cells in suspension culture and calculating the percentage of spontaneous apoptosis in each sample, the researchers were able to determine a correlation between the in vitro apoptosis resistance of cells and the primary chemoresistance of the patient. Both our group and others have reported similar findings (Wuchter et al, 1999; Pallis et al, 2002). These findings pinpoint the need to study apoptosis-resistant primary AML clones in order to probe for the molecules and pathways that underpin prolonged in vitro cell survival. To this end, we have shown recently that AML blasts that are resistant to in vitro apoptosis are also resistant to mitochondrial membrane depolarization, suggesting that pathways either upstream of the mitochondrion or at the mitochondrial membrane itself play a major role in leukaemic cell survival. However, the resistance to mitochondrial membrane depolarization did not correlate with total cellular levels of bcl-2, bax or bcl-x expression, possibly reflecting the large number of bcl family proteins that may contribute pro- and antiapoptotic functions (Pallis et al, 2001).

We have shown previously that pgp, conventionally considered to be primarily a drug-efflux pump in AML cells, is directly involved in the resistance to drug-independent apoptosis (Pallis & Russell, 2000). In suspension culture, the serum withdrawal-induced apoptosis of pgp-positive AML blasts was enhanced by the presence of the pgp-blocking antibody UIC2. Pgp is also known to be a primary apoptosis resistance molecule in other cell types (Robinson et al, 1997; Smyth et al, 1998; Johnstone et al, 1999). In the current study, we have investigated whether there is a statistically significant association between the expression of pgp and the apoptosis-resistant phenotype of AML patient blasts.

The FLT3 ligand has significant antiapoptotic and growth-promoting effects on primary AML cells in vitro (Lisovsky et al, 1996; Drexler et al, 1999). Moreover, transfection of FLT3 cDNA containing internal tandem duplications (ITDs) into myeloid cell lines induces an increased resistance to apoptosis, as well as constitutive activation of ras and STAT-5 and enhanced clonogenicity (Hayakawa et al, 2000; Mizuki et al, 2000). Taken together, these studies suggest that the FLT3 mutation may also be associated with an apoptosis-resistant phenotype in AML.

In the current study, we have measured spontaneous in vitro apoptosis in AML blasts with the aim of establishing its relationship with both functional and phenotypic pgp and also with FLT3 ITDs.

Materials and methods

Cells.  Blood or bone marrow samples from patients with untreated AML were studied. Mononuclear cells were isolated using a standard density gradient/centrifugation method with Histopaque (Sigma) and were cryopreserved in liquid nitrogen. Cryopreserved samples were thawed and rested in culture medium enriched with 20% fetal calf serum (FCS; First Link, Wolverhampton, UK) for 90 min before experimental procedures were begun. FCS was heat inactivated for 60 min at 56°C. Only samples with a post-rest viability of > 85% were used. Samples were depleted of T and natural killer cells using CD2-Dynabeads (Dynal) at a concentration of 25 µl beads per 107 cells in order to eliminate a potential source of cytokines that may have affected in vitro blast survival.

Fixed stained cell preparation for cell enumeration.  Mononuclear cells from 20 ml of peripheral venous blood from a healthy volunteer were isolated using the standard density gradient separation technique and incubated with 20 µl of CD45 fluorescein isothiocyanate (FITC) antibody (Becton Dickinson) for 20 min in the dark. The suspension was washed in phosphate-buffered saline (PBS), resuspended in 2% paraformaldehyde and stored at 4°C until required. The fixed, stained cell concentration was determined manually using a haemocytometer. Fixed, stained cells were mixed with patient blasts and used as an internal standard for cell enumeration. They were distinguished from patient blasts by their CD45-FITC positivity, as described previously (Pallis et al, 1999a).

Flow cytometric measurement of apoptosis resistance. Primary AML cells were incubated at 37°C in 5% CO2 at a concentration of 5 × 105 cells/ml in suspension culture with Roswell Park Memorial Institute (RPMI)-1640 medium, 10% FCS and 2 mmol/l l-glutamine for 48 h, using sterile, lidded Falcon tubes. At the end of this time, cells were mixed thoroughly before the removal of 200 µl. An aliquot of 50 µl of 7-amino-actinomycin D (7-AAD, Sigma) at 50 µg/ml was added. Cultures were incubated for 20 min in the dark at room temperature before processing for flow cytometry. The fixed, stained cell sample (50 µl) was added to the patient cells, and flow cytometric analysis was performed using gating, described in detail elsewhere (Pallis et al, 1999a) and illustrated in Fig 1B, to select for both fixed, stained cells and viable (7-AAD low/forward scatter high) leukaemic populations. The flow cytometer used was a facscalibur, and analysis was performed using cellquest software (Becton Dickinson, Cowley, UK). Instrument performance was monitored weekly using the Becton Dickinson Calibrite/Autocomp system. The concentration of viable cells was calculated from the formula: the number of events in Gate 4 × correction factor × 100 ÷ number of events in Gate 3.

Figure 1.

(A) Spontaneous apoptosis of primary leukaemic clones in suspension culture. After 48 h culture, the number of viable cells remaining was counted using flow cytometry. All samples were originally plated at 5 × 105/ml, and percentage apoptosis was calculated in relation to the starting concentrations. (B) Scatter plots showing flow cytometric analysis. Plots i and ii illustrate a case with low spontaneous apoptosis, in contrast to plots iii and iv. In plots i and iii, leukaemic events are shown in R2 (low FL1), and internal standard cells are shown in R3 (CD45-FITC stained, giving high FL1). Plots ii and iv are gated on R2 only. R4 represents viable (7-AAD-excluding, forward scatter high) cells. The number of viable cells is calculated from the R4:R3 ratio and the predetermined correction factor (Pallis et al, 1999a).

Determination of multidrug resistance (MDR) status.  The pgp status of patient blasts was determined by previously described methods (Pallis et al, 1999b,c). First, pgp was measured on unfixed cells using an unconjugated MRK16 antibody (Kamiya, Seattle, WA, USA) or a concentration- and isotype-matched control antibody and a FITC-conjugated second layer (Dako, High Wycombe, UK). Pgp values were considered to be positive when the Kolmogoroff–Smirnoff D-value, calculated by the cellquest software from a comparison of test and control distributions, was > 0·1.

Pgp substrate efflux modulation by valspodar (PSC833) was determined in an accumulation assay using rhodamine 123 (R123, Sigma). Thawed and rested patient blasts were pelleted, resuspended at 106 cells/ml and incubated, in duplicate, in medium containing 10% FCS at 37°C,with 200 ng/ml R123, with or without 2 µmol/l PSC833 or solvent control. PSC833, a gift from Novartis, was dissolved in ethanol to produce a stock solution and stored at 4°C. Control tubes with R123 incubated at 4°C were also set up. After 75 min incubation, the cells were pelleted at 4°C, rinsed, resuspended and analysed flow cytometrically as described previously (Pallis et al, 1999b). The mean fluorescence intensity (MFI) of each sample in the FL1 channel was recorded. Results were expressed as: (MFI with modulator–cold control MFI)/(MFI with diluent control–cold control MFI). This was referred to as the R123 retention ratio. Values greater than 1·7 were considered to be positive, based on a previously published comparison with drug retention (Pallis et al, 1999b).

Determination of FLT3 mutation.  Seven samples from the MRC AML 12 trial were included in our study, and the presence of FLT3 ITDs had been determined previously for five of these samples as described earlier (Kottaridis et al, 2001). Genomic DNA was extracted from the remaining samples using a QIAamp blood DNA isolation kit (Qiagen, Crawley, UK) according to the manufacturer's protocol. Approximately 100 ng of genomic DNA was used as template in a polymerase chain reaction (PCR) to amplify exon 11, 12 and the intervening intron of the FLT3 gene using previously described primers (Kiyoi et al, 1999). The 25 µl reaction also consisted of 150 µmol/l each deoxyribonucleoside triphosphate (dNTP; Amersham Biosciences UK, Little Chalfont, UK), 2 µmol/l each primer, 1·5 mmol/l/l MgCl2 and 1 unit of Amplitaq Gold (PE Applied Biosystems, Warrington, UK) in the manufacturer's buffer. After an initial heat activation step at 95°C for 10 min, amplification was performed in a PTC-100TM programmable thermal controller (MJ Research, Watertown, MA, USA) using the following conditions: denaturation at 94°C for 1 min, annealing at 52°C for 1 min and extension at 72°C for 1 min for a total of 35 cycles ending with a final extension at 72°C for 10 min. Amplified products were separated using 2% agarose gel electrophoresis (Helena Biosciences, Sunderland, UK), stained with ethidium bromide and analysed under ultraviolet light.

Statistics.  Statistics were performed using the Statistical Package for Social Sciences (SPSS) software. Non-parametric tests were used throughout. Mann–Whitney tests were used to analyse differences between two independent variables, and Spearman's rho was used to assess correlations.


Definition of resistance to spontaneous apoptosis

T cell-depleted samples from 58 AML patients were cultured at 37°C in a 5% CO2 atmosphere in liquid medium containing serum but without added cytokines or stromal cells. The concentration of cells remaining after 48 h was measured flow cytometrically, using 7-AAD with fixed, stained cells as an internal standard. We had previously determined that the viability of primary AML cells determined with 7-AAD is similar to that determined with annexin V after suspension culture (Pallis & Russell, 2000). 7-AAD is the probe of choice in the current study, because its use minimizes the number of processing steps that could give rise to an erroneous count. The rationale for this research was that the study of apoptosis resistance in AML blasts might help in the understanding of chemoresistance in patients. Therefore, it is important to note that the methodology for measuring apoptosis resistance was used in a previous study, in which a significant correlation between in vitro cell survival and chemoresistant disease was observed (Pallis et al, 2002).

The median survival of AML clones after 48 h was 38% of baseline values (Fig 1). We defined samples as being apoptosis resistant if their in vitro survival was above the median, and as being apoptosis sensitive if their survival was below the median. Examples of flow cytometric 7-AAD and forward scatter distributions of both a sensitive and a highly resistant clone are also illustrated in Fig 1.

Pgp expression and high R123 retention ratios are associated with in vitro blast survival in AML

Previous work has shown that in vitro apoptosis of pgp-positive AML blasts, in the absence of cytotoxic drugs, is increased by culture with the pgp-blocking antibody UIC2 (Pallis & Russell, 2000), which suggests that pgp confers a survival advantage upon AML blasts that is independent of its classic drug efflux function. To determine the prevalence of an apoptosis-resistant phenotype in pgp-positive compared with pgp-negative primary AML samples, the percentage of spontaneous apoptosis in pgp positive and negative samples was compared using Mann–Whitney analysis. Some 76% of apoptosis-resistant clones were positive for the pgp protein (Table I), compared with 31% of apoptosis-sensitive cases (P = 0·001). The R123 retention ratio is a sensitive (Broxterman et al, 1996), although not necessarily specific (Sincock & Ashman, 1997), functional assay for pgp. In our cohort, 37% (20/54) of samples were pgp positive with a high PSC833-mediated R123 retention ratio, and 43% (23/54) were pgp negative with a low R123 retention ratio. However, the remaining 20% had discordant results: 11% (6/54) were pgp positive with a low retention ratio, and 9% (5/54) were pgp negative with a high retention ratio. We determined the prevalence of an apoptosis-resistant phenotype in samples with a high compared with a low R123 retention ratio. Some 74% of apoptosis-resistant clones had a high R123 retention ratio (Table II) compared with 19% of apoptosis-sensitive cases (P < 0·001).

Table I.  Co-distribution of pgp protein and apoptosis phenotypes among the primary AML clones.
 Apoptosis sensitive
(> 62%), n = 29
Apoptosis resistant
(< 62%), n = 29
  1. An MRK-16 D-value of ≥0·1 was considered positive, based on previous work in the laboratory. The median MRK-16 D-value for this cohort was 0·105. Significant differences in resistance to spontaneous apoptosis were noted in Mann–Whitney analysis of MRK-16-positive versus -negative clones (P = 0·001). n, number of clones.

MRK-16 positive (n = 31) 922
MRK-16 negative (n = 27)20 7
Table II.  Co-distribution of R123 retention ratios and apoptosis phenotypes among the primary AML clones.
 Apoptosis sensitive
(> 62%), n = 27
Apoptosis resistant
(< 62%), n = 27
  1. An R123 retention ratio of ≥1·7 was considered positive, based on a previous study in which R123 modulation above this value was associated with decreased daunorubicin uptake (Pallis et al, 1999b). Significant differences in resistance to spontaneous apoptosis were obtained in Mann–Whitney analysis of functional pgp-positive versus -negative clones (P < 0·001). n, number of clones.

R123 retention ratio > 1·7 (n = 25) 520
R123 retention ratio < 1·7 (n = 29)22 7

As significant associations have been reported between pgp and age (Leith, 1998) and also between apoptosis resistance and age in AML (Garrido et al, 2001), we determined whether the ages of the patients in our study might help to explain the association of pgp with apoptosis resistance. The median age in our study was 61 years (interquartile range 44–71 years). There were trends towards correlations between age and pgp expression (P = 0·06), and between age and R123 retention (P = 0·15). However, there was no correlation between age and in vitro cell survival (P = 0·26). Therefore, age could not account for the strong associations between pgp, R123 modulation and in vitro cell survival.

Pgp activity is lower in FLT3 ITD samples than in FLT3 wild-type samples

FLT3 ITD mutations were determined for all cases in which DNA was available (55/58 cases) and were found in 20 cases (36%). The incidence of FLT3 ITDs in this cohort is high, but it should be borne in mind that FLT3 mutations are associated with a high white cell count, and that these experiments could not be performed on samples with a low cell count: there is therefore a probable bias towards the selection of FLT3 mutant samples. As FLT3 ITDs, pgp and R123 modulation are all predictors of poor outcome in AML, it was hypothesized that pgp might be overexpressed in FLT3 ITD cells. Surprisingly, Mann–Whitney analysis of R123 retention ratios in wild-type compared with FLT3 ITD cases showed that pgp activity was lower in the latter sample group (P = 0·018): the median R123 modulation ratio in FLT3 wild-type cases was 2·00, whereas in FLT3 ITD cases, it was 1·38 (Fig 2). A comparison of pgp expression measured by the MRK16 antibody in the FLT3 wild type compared with ITD cases also demonstrated a higher median fluorescence in the FLT3 wild type (1·2 vs 0·75), but this difference failed to reach statistical significance (P = 0·09).

Figure 2.

Distribution of (A) R123 retention ratios and (B) pgp (p-glycoprotein), according to FLT3 status. The dotted lines indicate the positive cut-off point as explained in the Materials and methods section. The solid black cross-bars indicate medians.

The presence of FLT3 ITDs is not associated with resistance to spontaneous apoptosis

Some 33% of apoptosis-resistant samples had the FLT3 ITD, as did 39% of the apoptosis-sensitive samples (Table III). No significant association of FLT3 mutation with apoptosis resistance was determined using the Mann–Whitney test (P = 0·86).

Table III.  Co-distribution of FLT3 mutation and apoptosis phenotypes among the primary AML clones.
 Apoptosis sensitive
(> 62%), n = 28
Apoptosis resistant
(< 62%), n = 27
  1. n, number of clones.

FLT3 ITD (n = 20)11 9
FLT3 wild type (n = 35)1718

High R123 retention ratios are associated with a comparatively low peripheral blood white cell count

Low R123 retention in bone marrow cells has previously been associated with quiescence (Smeets et al, 1999). Quiescent cells may be more resistant to certain inducers of apoptosis, as they avoid certain cell cycle checkpoints that trigger apoptotic responses. Because we noted an inverse relationship between R123 retention ratios and FLT3 ITDs, we wondered whether this relationship would be reflected in the peripheral blood white counts at the time of presentation in our patient cohort. Indeed, using Mann–Whitney analysis, we found that R123 retention ratio was negatively associated with the peripheral blood white count (Table IV). In agreement with previous results, we also found that FLT3 ITDs were associated with higher white counts (Kottaridis et al, 2001). Resistance to spontaneous apoptosis was not significantly associated with white cell count.

Table IV.  Association of peripheral blood white cell counts with R123 retention ratios and FLT3 ITDs.
 Median WCC (× 109/l)P-value
  1. WCC, white cell count.

Positive R123 retention ratio340·046
Negative R123 retention ratio64·5 
FLT3 ITD cases630·019
FLT3 wild-type cases32·9 
Apoptosis-sensitive cases580·23 (not significant)
Apoptosis-resistant cases50 


The finding of a significant association between pgp overexpression and prolonged in vitro survival of primary AML blasts, together with our previous finding that the UIC2 antibody increases in vitro apoptosis in pgp-positive AML blasts (Pallis & Russell, 2000), provides strong evidence that pgp is an important cellular survival-related molecule in AML. These findings are compatible with evidence of the survival-promoting functions of pgp in other systems (Robinson et al, 1997; Smyth et al, 1998). The mechanisms through which pgp may promote survival have been discussed previously elsewhere (Johnstone et al, 2000; Pallis et al, 2002), but include the modulation of both ion channels (Idriss et al, 2000) and the translocation of signalling lipids (Borst et al, 2000). Our laboratory has previously determined that pgp-positive primary leukaemic cells are comparatively resistant to apoptosis mediated by the exogenous application of the phospholipid sphingomyelin (Pallis & Russell, 2000).

We found that pgp activity was lower in FLT3 ITD samples than in FLT3 wild-type samples. The reasons for this finding are unclear, but it may be because pgp overexpression and the mutation of the FLT3 gene are both class I leukaemogenic events (i.e. events that confer either a proliferative or a survival advantage; Gilliland, 2001) and are therefore not likely to be selected for in the same clones. To date, there is no evidence that MDR1 overexpression is associated with leukaemogenesis. However, as proof of principle, Bunting et al (1998, 2000) have demonstrated that MDR1 overexpression in murine bone marrow cells induces both an increase in the number of progenitor cells and excessive myeloproliferation. An alternative explanation as to why pgp activity is lower in FLT3 ITD samples than in FLT3 wild-type samples might be that FLT3 ITDs are associated with a relatively high rate of leukaemic cell growth, whereas pgp activity is associated with relative quiescence (Smeets et al, 1999). We have shown that the FLT3 ITD samples in our cohort were from patients with a relatively high white cell count at the time of presentation. In addition, pgp was found to be active in samples from patients with relatively low white cell counts, providing tentative, but as yet highly speculative, support for the concept that the cell growth rate may underpin the inverse association between FLT3 ITDs and pgp function in patient samples.

Based upon reports that the transfection of FLT3 ITDs into myeloid cell lines enhanced survival as well as growth (Mizuki et al, 2000), we hypothesized in the current study that AML blast survival would also be enhanced by FLT3 ITDs. However, our examination of 58 cases, including 20 with FLT3 ITDs, has not shown an in vitro survival advantage for FLT3 mutant samples, compared with wild-type cells. It is important to note that we compared AML clones with each other and not with normal haematopoietic cells: all AML clones have a survival advantage compared with their normal haematopoietic counterparts, yet some AML clones resist apoptosis more than others. In this context, our study neither contradicts the finding that transfection of FLT3 ITDs enhances cell survival, nor does it contradict the finding that FLT3 ITD cells can be selectively induced to undergo apoptosis by FLT3 tyrosine kinase inhibitors (Levis et al, 2002). However, the results of this study do demonstrate that clones with FLT3 ITDs are not more resistant to apoptosis than other leukaemic clones.

In the introduction to this report, the association between FLT3 ITDs and adverse survival despite normal remission rates was contrasted with the association between pgp and primary chemoresistant disease. We have found that the primary resistance to chemotherapy of pgp-positive AML patients was associated with apoptosis resistance of the blast cells, whereas the poor overall survival of FLT3 ITD AML patients did not appear to be mediated by this mechanism. Although individual cases in our study were found to be both FLT3 mutant and pgp positive, we have now demonstrated differences in the biology of R123 effluxing and FLT3 mutant cases that might underpin differences in the patient response to therapy. The fact that FLT3 ITD-positive cells are not particularly resistant to apoptosis highlights the possibility that the use of FLT3 tyrosine kinase inhibitors will be effective inducers of apoptosis in mutated cells.


We thank the MRC AML adult working party for access to their material, and Steve Langabeer and the Kay Kendall Leukaemia Fund for sending us stored samples. We thank Panos Kottaridis who determined the FLT3 ITD status of MRC 12 trial patients and kindly gave us his results. We thank Ian Carter, who established the FLT3 ITD assays in Nottingham, and Melanie Brown for expert technical assistance.