¹Leiden University Medical Center, Department of Hematology, Leiden, The Netherlands

TO THE EDITOR

Although the majority of patients with acute myeloid leukemia (AML) responds to initial treatment, relapse of the disease occurs in a significant percentage of these patients.¹ The cell cycle status of leukemic cells may play an important role in the response to treatment of leukemic cells. Especially, the broadly used cytotoxic agent Cytarabine (Ara-C) has been demonstrated to exert its action via intercalation into the DNA of cells specifically in the S phase of the cell cycle, and sensitivity to this agent is therefore described to be specific for cells in active cell cycle.² The role of the cell cycle in sensitivity of leukemic cells to anthracyclins like daunorubicin remains largely unknown.

We hypothesized that a fraction of noncycling leukemic (precursor) cells residing in dormancy might be unresponsive to chemotherapy. To test this hypothesis, we determined in a pilot study the in vivo cell cycle specificity of chemotherapeutic treatment by analysis of the cell cycle distribution of leukemic blasts freshly isolated from the peripheral blood of three patients with AML at diagnosis and at multiple consecutive days after in vivo treatment with Ara-C, daunorubicin, and etoposide. Detailed cell cycle analysis was performed by combined ki-67 staining and propidium iodide (PI) DNA staining. Ki-67 is a nuclear protein which is absent in G₀ phase and present in all other phases of the cell cycle,³ thereby facilitating the discrimination between cells in resting G₀ phase, activated G₁ phase, and cycling S or G₂/M phase. Table 1 shows the total nuclear cell counts, the percentages of leukemic blasts, and the cell cycle distribution of the cells measured in time during chemotherapy administration. Already after 1 day of treatment, a decrease in the number of cycling cells was noticed. In the following days, a further deletion of all activated ki-67 positive cells was observed, resulting in a profound enrichment for cells in the resting G₀ phase of the cell cycle. This relative enrichment for cells in G₀ phase could in part be explained by the enrichment for normal T and B cells. However, even the leukemic population in G₀ phase of the cell cycle appeared to be protected from the cytotoxic effects of the treatment.

Table 1 - Cell cycle distribution after in vivo chemotherapy.

Full table

To investigate in more detail the cell cycle specificity of daunorubicin, we used the human myeloid leukemic cell line AML-193 as a model, and studied the cell cycle distribution of viable cells within this cell line prior to and after 40 h of chemotherapy exposure. As shown in Figure 1a, AML-193 cells cultured in medium contained cells in resting G₀ phase (lower left quadrant), cells in activated G₁ phase (lower right quadrant), and cycling cells in S or G₂/M phase (upper right quadrant). As expected after 40 h of incubation, 10^-6 M Ara-C-induced complete deletion of the cycling S or G₂/M compartment was observed, whereas noncycling cells in resting G₀ phase and activated G₁ phase of the cell cycle were protected. In contrast, after 40 h of exposure to 10^-6 M daunorubicin, both cells from the cycling S/G₂/M compartment and the activated G₁ compartment were deleted. Only cells in the resting G₀ phase were not affected by daunorubicin-induced cell death. To analyze whether a similar cell cycle-specific sensitivity to Ara-C and daunorubicin was observed in other leukemic cell lines, we analyzed four acute lymphoblastic leukemia cell lines and one CML blast crisis cell line, which were generated in our laboratory by culturing of leukemic blasts from five different patients in serum-free medium at high cell concentrations until spontaneous sustained proliferation of the leukemic cells occurred. These cell lines cytogenetically and phenotypically resembled the primary leukemia. Figure 1b shows the median cell cycle distribution within these cell lines after incubation for 48 h in medium alone, or in medium containing 1 10^-6 M Ara-C or daunorubicin. As expected, Ara-C-induced cell death resulted in a deletion of cells from cycling S and G₂/M phases, resulting in relative enrichment for cells in G₀ and G₁ phases. In contrast, daunorubicin-induced cell death resulted in a deletion of cells from both activated G₁ phase and cycling S and G₂/M phases, resulting in a major enrichment of cells in G₀ phase. These data illustrated the specificity of daunorubicin not only for cells in active cell cycle, but also for noncycling cells in the activated G₁ phase of the cell cycle.

Figure 1.

Cell cycle specificity of daunorubicin and Ara-C. (a) Cell cycle analysis of viable AML-193 cells by PI/ki-67 analysis after 40 h of incubation in medium alone, or in medium containing 10^-6 M daunorubicin, or 10^-6 M Ara-C. (b) Cell cycle analysis of viable cells of five ALL/CML cell lines after 40 h of incubation in medium alone, or in medium containing 10^-6 M Ara-C or 10^-6 M daunorubicin. The percentages of cells in G₀ phase (), G₁ phase (), S phase (), and G₂/M phase () were determined. The mean percentages of cells in every cell cycle fractions.d. are listed.

Full figure and legend (128K)

The AML-193 cell line model was used to further investigate the influence of manipulation of the cell cycle status of leukemic cells on chemotherapy sensitivity. As described previously, it is possible to separately manipulate the cell cycle status and the proliferative capacity of these AML-193 cells using GM-CSF and interferon (IFN) treatment.⁴ Three AML-193 subcell lines were generated by culturing of cells either in the absence of cytokines (control), in the presence of 1000 U/ml IFN-, or in the presence of 500 U/ml IFN-, as described previously.⁴ GM-CSF was used to induce proliferation in these cells. None of these cell cycle manipulations did induce apoptosis. The long-term culture period in the presence of IFN- did not affect the proliferation rate of the cells in the presence or absence of GM-CSF (data not shown). However, IFN- treatment induced a 2.5-fold decrease in both baseline proliferation and proliferation in the presence of GM-CSF. Detailed cell cycle analysis of cells cultured in the presence or absence of GM-CSF and interferons was performed.

As shown in Figure 2, in control AML-193 cells cultured in the absence of GM-CSF as well as interferons, most cells were in resting G₀ phase (66%), 19% of cells were in G₁ phase, and 15% in S or G₂/M phase (n=20). Long-term interferon treatment of control cells induced a significant increase in activated G₁ cells to 31% in IFN-- and 27% in IFN--treated cells (P<0.0001). Small but significant higher percentages of cells in S or G₂/M phase were observed in IFN--treated cells (21%) and IFN--treated cells (18%), as compared to control cells (P<0.01). These increases coincided with a decrease in the number of cells in resting G₀ phase to 48% in IFN--treated cells, and to 55% in IFN--treated cells. Induction of proliferation with GM-CSF resulted in increased percentages of cells in activated G₁ phase (45%) and in the cycling S/G₂/M phase (34%). In the presence of GM-CSF, IFN treatment did not exert an additive effect on the cell cycle distribution. These results illustrate that long-term interferon treatment of AML-193 cells resulted in activation of resting cells without induction of proliferation, facilitating separate analysis of the effects of cell cycle manipulation and proliferative status manipulation on chemotherapy sensitivity. Flow cytometric analysis of control cells and IFN-- and IFN-- treated cells in the absence and presence of GM-CSF revealed no differences in the expression levels of the apoptosis-related proteins Fas, Bcl-2, Bax, p53, TNF-R1, and P-glycoprotein.⁴ Figure 3 illustrates the sensitivity of AML-193 cells cultured in the six different culture conditions to daunorubicin-induced apoptosis. The results of all individual experiments are plotted as percentage-specific lysis after 24 (Figure 3a) and 40 h (Figure 3b). Medians are indicated with bold lines. Incubation of control cells with 10^-6 M daunorubicin resulted in a median lysis of 23% after 24 h (n=40) and 49% after 40 h (n=30). Activation of control cells by IFN- or IFN- without induction of proliferation induced a two-fold increase in median sensitivity to daunorubicin after 24 h, resulting in 48% lysis (n=20; P<0.0001) and 36% lysis (n=15; P=0.02) in IFN-- and IFN--treated cells, respectively. The same increase in sensitivity to daunorubicin was found after induction of proliferation with GM-CSF (n=40; P<0.0001). IFN- treatment of GM-CSF-treated cells resulted in a minor additional increase in the sensitivity to daunorubicin. Despite inhibition of proliferation by IFN- treatment, no decrease in the sensitivity to daunorubicin of GM-CSF-treated cells was found (n=20; P=0.6). After 40 h of exposure to daunorubicin, all GM-CSF- or IFN-treated activated cells were equally well lysed. Only resting control cells showed a lower sensitivity to daunorubicin (P<0.001). These results illustrated that the proliferation rate itself, as determined by complete progression through the cell cycle, did not coincide with the sensitivity to daunorubicin-induced apoptosis, since the slowly proliferating IFN- cell line was equally sensitive as the GM-CSF- or IFN--treated cell lines. In contrast, the sensitivity to Ara-C was directly related to the percentages of cycling cells (data not shown).

Figure 2.

Cell cycle distribution after interferon treatment in the absence and presence of GM-CSF. Cell cycle distribution was analyzed by combined ki-67/propidium iodide DNA staining. The percentages of cells in G₀ phase (), G₁ phase (), S phase (), and G₂/M phase () were determined.

Full figure and legend (80K)

Figure 3.

The effect of cell cycle manipulation by GM-CSF and interferon treatment on sensitivity to daunorubicin. To determine the chemotherapy sensitivity, the cells were exposed to 10^-6 M daunorubicin for 24 (a) and 40 h (b). The percentage-specific lysis was measured in ⁵¹Cr release assays. Each experiment is plotted as a different dot. Medians are visualized as bars.

Full figure and legend (32K)

To illustrate that the cell cycle specificity of daunorubicin-induced cell death was independent of the cell cycle distribution of the cells at the onset of exposure, we analyzed AML-193 cells cultured under three different culture conditions in time. Table 2 demonstrates the constant cell cycle distribution in AML-193 cells cultured in the absence of GM-CSF in time. Ara-C induced a diminishment only of cells in the cycling S/G₂/M compartment. After 48 h of daunorubicin-induced cell death, activated cells in G₁, S, and G₂/M phases were deleted, resulting in relative enrichment for cells in the resting G₀ phase. Similar results were obtained in the proliferating AML-193 cells cultured in the presence of GM-CSF.

Table 2 - Cell cycle specificity of chemotherapy-induced apoptosis in AML-193 cells in the absence of the proliferation-inducing growth factor GM-CSF.

Full table

In conclusion, we demonstrated that exposure of leukemic cells to the anthracyclin daunorubicin specifically preserves leukemic (precursor) cells in the resting G₀ phase of the cell cycle, illustrating the relative specificity of daunorubicin for leukemic cells in the activated phases of the cell cycle. This is in contrast to the clearly proliferation-dependent sensitivity for Ara-C. These results may favor a therapeutic strategy using cell cycle modulators like interferons in an attempt to decrease the percentage leukemic (precursor) cells in dormancy, without inducing leukemic cell proliferation. In the clinical studies performed until now, interferons were used as maintenance therapy after chemotherapy intervention.⁵ Based on our data, it may be rational to use interferon as treatment before and during chemotherapy. However, in combination with conventional untargeted chemotherapy, the interferon-induced recruitment of normal, nonleukemic cells from the resting G₀ phase into active state might result in severe toxicity within normal tissues. Therefore, exploration of new agents with relative specificity for leukemic cells, but with similar ability to recruit these cells from the resting G₀ phase, will be needed. Alternatively, interferon combined with leukemia-targeted chemotherapeutic agents might be a feasible approach to increase the effectiveness of the chemotherapy with relative specificity for the leukemic cells.