Although it has long been speculated that mammalian stem cells undergo asymmetric and symmetric division, whether this actually occurs is unclear. In addition, whether microenvironments that influence stem cells to differentiate or self-renew do so by shifting the balance of asymmetric and symmetric divisions and whether the balance can also be subverted by oncogenes are fundamental questions that need to be addressed to better understand the basis of stem cell-fate decisions and oncogenic transformation.
Using a transgenic Notch reporter mouse, in which the GFP status of a cell acts as a sensor for the differentiated state, we have imaged how hematopoietic precursors divide in response to different stimuli during growth, differentiation, and oncogenic transformation. Specifically we show that the balance between asymmetric and symmetric divisions is not hardwired in precursors but instead is responsive to extrinsic and intrinsic cues. Precursors placed in a prodifferentiation environment preferentially divide through asymmetric or symmetric commitment division, whereas those placed in a prorenewal environment primarily divide through symmetric renewal divisions. Importantly, our data show that oncogenes also have the ability to influence a cell’s choice between symmetric or asymmetric division. In this context, we find that, although BCR-ABL predominantly influences a cell’s rate of division and death, NUP98-HOXA9 predominantly influences the normal balance of symmetric and asymmetric division. This suggests that different oncogenes subvert different aspects of cellular function, and an oncoprotein’s mode of action may dictate the type of leukemia it generates.
Although it is known that precursor cells in invertebrates divide by asymmetric and symmetric cell division, whether this occurs in mammalian cells, particularly in the hematopoietic system, was less clear. The fact that GFP expression correlates with an undifferentiated state in the TNR mice allowed us to develop a system to track the different outcomes of cell division in real time, thereby allowing screening and assessment of different stimuli or microenvironments without the need for any micromanipulation of progeny cells. Using this approach, we observed that, when hematopoietic precursors were placed on 7F2 cells, which induce differentiation, the cells underwent asymmetric division twice as often as symmetric renewal divisions. By contrast, when placed on OP9 cells, which promote the maintenance and growth of immature cells, the cells divided through symmetric renewal 2.5 times more often than asymmetric division. Interestingly, the rates of growth and death were similar on the two environments. These results suggest that one way by which the microenvironment might influence growth or differentiation is to alter the balance of symmetric and asymmetric division of hematopoietic precursors.
Previous pioneering studies in the hematopoietic system that have focused on the question of whether hematopoietic cells can undergo asymmetric division have primarily used a clone-splitting technique (Brummendorf et al., 1998
; Leary et al., 1985
; Suda et al., 1984
). In this approach, single hematopoietic cells are cultured, their immediate progeny physically separated, and the fates of these progeny tracked by assays that measure the rate of proliferation or colony formation. This approach revealed that in cultured human cord blood hematopoietic progenitors 3%–17% of the divisions were asymmetric and that the division pattern was not influenced by exposure to different cytokines (Mayani et al., 1993
). A similar lack of influence of cytokines was observed in CD34+
cells from human fetal liver where the rate of cell division was used as to track whether paired daughter cells behaved differently (Huang et al., 1999
). These data indicate that the decision between different types of division may not be controlled by extrinsic regulatory molecules but rather by intrinsic mechanisms. However, experiments using more purified mouse CD34-KLS cells (Osawa et al., 1996
) have shown that culture with cytokines such as SCF and IL3, which induce greater levels of differentiation, led to asymmetric divisions in 52%–62% of the cells, whereas culture with cytokines such as SCF and Tpo, which tend to preserve undifferentiated cells, led to asymmetric division in only 17% of the cells (Takano et al., 2004
). These data suggest that cytokines that decrease differentiation do so by decreasing asymmetric division and are consistent with our findings, which show that distinct microenvironments do lead to a corresponding change in the balance of hematopoietic precursor divisions. Our work also provides a possible mechanism underlying the choice between symmetric and asymmetric fates by showing that asymmetric divisions correspond to asymmetric distribution and segregation of Numb. Although our data also show that increased Numb can contribute to specification of the differentiated daughter cell, it is likely that multiple other molecules are involved in this asymmetric division event, and a complete elucidation of the relevant pathways will be an important area of future work. In this regard, it is important to point out that it has recently been reported that primitive human hematopoietic cells asymmetrically segregate proteins such as CD63 and CD71 during mitosis (Beckmann et al., 2007
). Together with our work, this suggests that the intrinsic segregation of a variety of protein determinants likely plays a role in fate specification of both mouse and human HSCs.
Using our imaging system, we have also found that oncogenes can subvert the normal balance of asymmetric and symmetric division. Interestingly, although the oncoprotein BCR-ABL had a clear impact on increasing the rate of cell division and decreasing the rate of cell death, it did not affect the choice between asymmetric and symmetric division compared to control conditions. In contrast, NUP98-HOXA9 led to a significant increase in symmetric renewal divisions but did not affect the rates of division and cell death. The differential impact of the two oncogenes may be important because BCR-ABL is mostly associated with CML, which is an indolent disease with intact differentiation. In contrast, NUP98-HOXA9 is often associated with AML and blast crisis CML, cancers that display an aggressive growth of more immature cells and reduced differentiation. It is important to note that, in mouse models, although the use of BCR-ABL recapitulates CML-like disease, NUP98-HOXA9 leads to a myelo-proliferative disease that progresses later to clonal AML. In addition, NUP98-HOXA9 can cooperate with BCR-ABL to lead to blast crisis phase of CML (Dash et al., 2002
; Mayotte et al., 2002
). It is possible that it is the differential impact of these oncogenes on the balance of asymmetric and symmetric divisions that defines the ability of BCR-ABL to induce only a chronic and differentiated disease and NUP98-HOXA9 to induce a more aggressive and undifferentiated disease.
Our work provides an important mammalian complement to studies of Drosophila
neuroblasts that have examined the consequences of mutations of genes that are involved in asymmetric renewal on tumor formation. Neuroblasts are normally polarized along their apical-basal axis, and the apical cell becomes another neuroblast while the basal daughter becomes the ganglion mother cell that goes on to make differentiated cells. Mutations of the gene brat, which normally segregates asymetrically to specify the differentiated fate, lead to unregulated growth of two uncommitted daughter cells and subsequent tumorigenesis (Betschinger et al., 2006
; Lee et al., 2006
). In addition, altering asymmetric division of neuroblasts by mutating the asymmetrically distributed protein Numb, Miranda, or prospero leads to overgrowth of neuroblasts and subsequent formation of tumors (Caussinus and Gonzalez, 2005
). These data suggest that, at least in flies, subversion of the balance between symmetric and asymmetric division is an important component of tumor growth. Our data indicate that such subversion of a normal cell-division pattern also in fact occurs during mammalian oncogenesis. However, by comparing the effects of different oncogenes, our work also suggests that the alteration of asymmetry may be specific to some oncogenes and not others, and raises the possibility that whether an oncogene alters asymmetry may dictate the type of leukemia it generates. Perhaps one of the more exciting implications of our findings is that molecules that alter the choice between symmetric and asymmetric division could be used to inhibit or slow the aggressive growth of acute leukemias. Only further work along these lines will reveal whether altering the aberrant symmetry of transformed cells can be a new target for cancer therapy.