|Home | About | Journals | Submit | Contact Us | Français|
The development of leukemia is a complex, multi-step process which is not fully understood. While many leukemias harbor chromosomal translocations and additional mutations which drive oncogenic transformation, non-genetic factors also contribute to leukemia development. We have demonstrated that the relative fitness of hematopoietic progenitor cells (HPC) plays a major role in leukemogenesis (1, 2). That is, the presence of normally functioning HPC can slow or eliminate leukemia development from a small pool of cells harboring an oncogene, while the presence of HPC which have impaired fitness enhances leukemogenesis. However, the importance of these findings seems to be under-appreciated, perhaps because a clinically relevant model has not yet been demonstrated. We propose that a clinical correlate of our previous work is the development of treatment related acute myelogenous leukemia (tAML) in patients who are treated with DNA damaging agents followed by myelosuppressive agents which impair the relative fitness of HPC.
tAML is a one of the most devastating side effects of cancer therapy and occurs in 2–15% of patients treated for cancer (3). Clinical chemotherapeutics which act by inhibition of topoisomerase II (e.g. etoposide and daunomycin) are known to induce MLL gene rearrangements associated with tAML. Depending on the treatment protocol, these drugs may be followed by other myelosuppressive drugs (e.g. purine anti-metabolites like 6-mercaptopurine (or historically, 6-thioguanine (6TG)) and/or anti-folates like methotrexate) which have also been associated with tAML (4). Thus, it is possible that leukemia may be initiated by a topoisomerase inhibitor and promoted by subsequent therapy. We hypothesized that the administration of a clinically relevant chemotherapeutic would promote leukemia development from oncogene harboring HPC in part by impairing the relative fitness of normal HPC.
We used 6TG to model prolonged myelosuppression after chromosomal translocation resulting from DNA damaging chemotherapy. To demonstrate impaired HPC fitness induced by 6TG, mice were treated with 6TG (administered in the drinking water) for 4 weeks, followed by bone marrow (BM) harvest and assessment of HPC fitness by colony forming unit (CFU) assay. Indeed, the numbers of CFU-GEMM and CFU-GM were reduced in the mice which were treated with 6TG as compared to the untreated mice (Supplemental Figure 1). We next sought to develop a model of leukemogenesis in which a small fraction of transplanted cells harbored MLL-AF9, and recipient mice were treated with 6TG. To do so, BM from young MLL-AF9 knock-in mice (5) was harvested and mixed at a 1:9 ratio with competitor BM from wild type (WT) mice that express green fluorescent protein (GFP) and then transplanted into sub-lethally irradiated recipient mice (Fig 1A). In this model, all donor cells are CD45.2+ which distinguishes them from recipient cells which are CD45.1+. Thus, the origination of any given cell can be determined using flow cytometry. Recipient animals were left untreated for 3 weeks to allow hematopoiesis to recover. Peripheral blood was then analyzed by complete blood count (CBC) and flow cytometry, and animals were divided to receive 6TG or no treatment. One group of animals was transplanted with WT BM without MLL-AF9+ BM and was treated with 6-TG to control for toxic effects of the medication, consisting primarily of mild myelosuppression (Supplemental Figure 2). Peripheral blood was analyzed every 2 weeks by CBC and flow cytometry. The percentages of GFPNeg donor cells (pre-leukemic MLL-AF9+) were consistently and substantially higher in the animals that received WT competitor BM and treated with 6TG as compared to the animals that received WT BM and were left untreated (ANOVA p<0.05; Figure 1B). MLL-AF9 knock-in mice develop myeloproliferative disease (MPD) prior to succumbing to leukemia (5). We used total white blood cell counts to monitor recipient mice for the development of MPD, which we defined as WBC of 25,000/μl on 2 consecutive CBCs. Consistent with the increase in the percentages of pre-leukemic MLL-AF9+ cells, the 6TG treated recipients of WT competitor BM cells developed MPD earlier than those recipients of competitor BM which were untreated (p=0.04; Figure 1C).
In order to address the possibility that the relative fitness of normal HPC may play a role in the expansion of oncogene bearing, pre-leukemic populations, we included recipient groups transplanted with BM from mice with mutated, non-functional hypoxanthine phosphoribosyltransferase (HPRT) (6). HPRT is required to activate 6TG into its cytotoxic metabolite, so HPC from these mice are resistant to treatment with 6TG (as evidenced by a lack of reduction in CFU formation after treatment with 6TG (Supplementary Figure 1)). When treated with 6TG, recipients of HPRTmt BM competitors did not have as rapid a rise in the percentages of pre-leukemia cells (ANOVA p<0.05; Figure 1B) and the development of MPD was significantly delayed as compared to recipients of WT competitor cells (p<0.006; Figure 1C). Together, these data indicate that 6TG promotes the development of MPD by reducing the relative fitness of HPC and providing a relative advantage for oncogene bearing cells.
In this model, the development of MPD heralded the onset of symptomatic leukemia. In 2 independent experiments, the overall survival of 6TG treated recipients of WT competitor BM cells was shorter than that of recipients of competitor BM which were left untreated (p<0.0001; Figure 1D). Consistently, though, in 6TG treated recipients, the presence of HPRTmt HPC prolonged overall survival compared to WT HPC (p<0.0001), suggesting that the presence of relatively fit HPC can, to some extent, slow leukemia development. Nonetheless, the recipients of HPRTmt competitor BM cells succumbed to leukemia more rapidly and at greater penetrance than untreated recipients of WT or HPRTmt BM (p<0.0001), indicating that impaired HPC fitness is only one aspect of leukemia promotion by 6TG. Purine analogs are known to be immunosuppressive, thus the differences in the penetrance and rate of leukemia development among the experimental groups could be the result of differences in the presence of an intact immune system. To control for this possibility, we bred the HPRTmt mice with mice mutant in the alpha T-cell receptor (TCR), which are immunocompromised because they have essentially no functional T-cells (7). The BM from the resulting double mutant mice (HPRTmtTCRmt) was then used as competition with pre-leukemic MLL-AF9+ cells as before, and the recipients were treated with 6TG. These chimeric recipients were thus immunocompromised, but not due to 6TG suppression of T-cell function. Survival in these recipients was significantly longer in the 6TG treated recipients of HPRTmtTCRmt BM, as compared to the 6TG treated recipients of WT competitors (p=0.003), indicating that suppression of T-cell function does not alter the kinetics of leukemia promotion by 6TG in this model. Still, the effect of 6TG on other aspects of the immune system cannot be excluded from these experiments.
We next sought to determine if a specific population of MLL-AF9+ HPC maintained a proliferative/survival advantage in the context of 6TG treatment, contributing to the increase in pre-leukemia burden in these mice. To do so, 3 mice from each group were euthanized at 8 weeks after transplantation (5 weeks of treatment) and the BM was analyzed by flow cytometry for different populations enriched for early hematopoietic progenitors including hematopoietic stem cells (HSC) and multipotent progenitors (MPP), common myeloid progenitors (CMP) and granulocyte-macrophage progenitors (GMP) (Supplemental Figure 2). Without treatment, MLL-AF9+ HPC have an advantage over WT HPC, such that the majority of phenotypic HPC are derived from the MLL-AF9+ BM even though they were transplanted as a small minority (Figure 2A). Based on the data from peripheral blood at the 3 week time-point (Figure 1B), this advantage is primarily in the reconstitution period after transplantation. However, in the context of 6TG treatment, pre-leukemic MLL-AF9+ HPC appear to have a greater advantage, particularly at the level of MPP (Figure 2A), suggesting that treatment with 6TG may promote the progression of leukemia by increasing the relative selective advantage of oncogene bearing HPC over normal HPC. When HPC from 6TG treated recipients of HPRTmt BM were analyzed by flow cytometry, a dramatic difference in the percentages of non-oncogene bearing cells was seen as compared to all other groups at the level of the HSC, CMP and GMP (Figure 2A). This suggests that these HPRTmt HPC (GFPneg CD45.2+ cells) limit the dominance of MLL-AF9+ HPC specifically in the presence of 6TG. Surprisingly, the relative advantage of HPRTmt MPP cells in the context of 6TG treatment was not demonstrated, perhaps suggesting that in this model, this is a population in which MLL-AF9+ progenitors possess unique fitness which cannot be overcome, even by cells which should not suffer functional impairment by 6TG.
Despite being transplanted in the minority (only 10% of cells), Mll-AF9+ HPC immediately contribute to hematopoiesis to a greater extent than WT or HPRTmt HPC indicating a relative advantage over WT and HPRTmt BM cells, at least in a reconstitution setting. When recipients of WT competitors are treated with 6TG, though, the advantage conferred by MLL-AF9 extends beyond the reconstitution phase, suggesting that MLL-AF9+ cells may be more resistant to 6TG as compared to WT HPC, providing these cells a greater advantage over WT cells when recipients are treated with 6TG. To test this possibility, whole BM or lineage depleted BM was cultured in cytokine containing media with or without 6TG. Surprisingly, proliferation of Mll-AF9+ HPC was suppressed to the same extent as that of WT HPC cells (Figure 2B). Furthermore, when cultured cells were assessed for apoptosis, MLL-AF9+ and WT HPC had similar percentages of early and late apoptotic cells as measured by flow cytometry for annexin V and propidium iodide (PI) exclusion (Figure 2C), indicating that the cultured MLL-AF9+ cells are not inherently resistant to the cytotoxic effects of 6TG. The lack of protection is somewhat surprising, given that MLL-AF9+ cells have been shown to exhibit a diminished p53 response to DNA damage (8).
These data provide clinical relevance to an under-appreciated aspect of leukemogenesis. That is, the relative fitness of non-oncogene initiated HPC influences the development and progression of leukemia. While our previous reports highlight the biologic importance of these findings (1, 2), we have now paired clinically relevant HPC impairment with leukemia initiation to emphasize the clinical relevance of this phenomenon. Future studies are designed to test other medications and oncogenes, and to further elucidate the mechanisms of the observed phenomena. Of course, prolonged treatment with myelosuppressive medications, including 6TG, is beneficial for the vast majority of patients for whom they are prescribed. However, this comes with some risk of impairing the fitness of normal HPC and allowing oncogene initiated HPC a competitive advantage and the development of tAML, compounding the potential of these drugs to induce leukemia. These risks should be carefully considered in the development of treatment protocols in which DNA damaging agents may be given prior to prolonged myelosuppression.
This work was supported by grants from the Brent Eley Foundation (CP), The Cancer League of Colorado (CP), The National Institutes of Health (P20 CA103680-CP; RO1 CA109657-JD), and the Leukemia and Lymphoma Society (108692-JD). MLL-AF9+ and HPRTmt mice were kindly provided by Drs. J.H. Kersey and H.A. Jinnah, respectively.
Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)
Conflict of Interest:
The authors declare no conflict of interest.