Differential cytoskeleton modeling by p40(ABL/BCR), p96(ABL/BCR)and BCR
The reciprocal ABL/BCR fusion proteins are BCR mutants. BCR has Rho-GEF and Rac-GAP activity, strongly suggesting a main role in the regulation of cytoskeleton modeling and cell motility by modulating the activity of the Rho subfamily of small GTPases, such as Rac, cdc42 and Rho. The basis for this is that Rho-GTPases regulate actin filament assembly. To determine how the BCR mutations related to t(9;22) interfere with its functionality we compared the phenotypes induced by p40(ABL/BCR) and p96(ABL/BCR) with that of wt BCR.
First, we investigated how the p40(ABL/BCR) and the p96(ABL/BCR) fusion proteins modulate the cytoskeleton in comparison to wt BCR. We used the bi-cistronic retroviral PINCO vector [17] for the expression of BCR, p40(ABL/BCR), p96(ABL/BCR) and p185(BCR/ABL) in Rat-1 cells, with the protein of interest under the control of the long terminal repeat (LTR) and the EGFP driven by a CMV promoter (Fig. 2A). For controls, the cells were infected with the empty PINCO vector, able to express only EGFP (mock). All retroviral constructs used in these experiments are shown in Fig. 2B. Expression of the transgenes were controlled by Western blotting (Fig 2C). Cytoskeleton modeling was detected by labeling the actin filaments with Cy3-conjugated phalloidin. The over-expression of BCR led to a Rho-like phenotype characterized by stress fibers and microspikes (Fig. 2D). In contrast, p96(ABL/BCR) induced a cdc42-like phenotype dominated by filopodia, whereas the p40(ABL/BCR)-related phenotype seemed to be a mixture between a Rac-like and a cdc42-like phenotype, due to the presence of both lamellopodia and filopodia (Fig. 2D). On the other hand, p185(BCR/ABL) induced a phenotype characterized by complete polarization of the cell and the formation in a high number of cells of one long, axon-like filopodium and few shorter neurite-like filopodia, conferring a neuron-like aspect to the cells attributable to strong Rho-activation [13] as already seen in PC12 cells [18]. This phenotype was influenced by the co-expression of either BCR or the ABL/BCRs. Both BCR and p40(ABL/BCR) abolished the polarization of the BCR/ABL-expressing cells which maintained the capacity to form the neurite-like filopodia. In contrast, BCR/ABL positive Rat-1 cells co-expressing p96(ABL/BCR) maintained a polarized phenotype with lamellopodia.
In summary, these data suggest that the ABL/BCR fusion proteins are able to modify the cytoskeleton of fibroblasts differently to wt BCR and BCR/ABL. Furthermore it seems that there is a complex functional interaction between the t(9;22)-associated fusion proteins.
Differential regulation of small Rho-like GTPases by p40(ABL/BCR), p96(ABL/BCR)and BCR
To determine the molecular mechanisms underlying the differences between BCR and its mutants regarding cytoskeleton modeling we looked for alterations in the regulation of small Rho-like GTPases. Therefore we assessed the activation status of endogenous Rho-like GTPases in Rat-1 cells expressing BCR and its mutants. As a control we used mock-transduced Rat-1 cells. Transduction efficiency was always more than 80%, as demonstrated by the percentage of EGFP-positive cells (data not shown).
As shown in Fig. 3, both p40(ABL/BCR) and p96(ABL/BCR) have lost the Rho-GEF function of BCR, because they were unable to activate Rho. The strong Rho-activation in BCR/ABL-expressing cells is most likely due to the ABL-kinase activity, given that the BCR/ABL fusion protein used in these assays lacks the Rho-GEF domain. Furthermore, both p40(ABL/BCR) and p96(ABL/BCR) have also lost the Rac-GAP function of BCR, as shown by their incapacity to suppress Rac-activation in Rat-1 cells (Fig. 3). In contrast, both have acquired the capacity to activate cdc42, whereas neither BCR nor BCR/ABL were able to activate cdc42 in these cells (Fig. 3)
Taken together, these data show that the ABL/BCR proteins present a pattern of small Rho GTPase activation which differs from that induced by wt BCR and BCR/ABL. In fact BCR can be considered a repressor of Rac and an activator of Rho that lacks any influence on cdc42, whereas the ABL/BCR proteins can be considered activators of cdc42 that have lost the capacity to activate Rho and to suppress Rac. In contrast, BCR/ABL is an activator of Rac and Rho, and has no influence on cdc42. These different activation patterns may account for the differences in cytoskeleton modeling.
In contrast to wt BCR and p40(ABL/BCR), p96(ABL/BCR)inhibits and BCR/ABL increases the migration of hematopoietic progenitor cells in a three-dimensional stroma model
Cytoskeleton modeling plays an important role in the biology of hematopoietic precursors because it contributes to cell motility, defined by migration, adhesion and chemotaxis. Motility is a main feature of hematopoietic stem cells and more mature progenitors, as well as of functional white blood cells. Therefore, we investigated the effect of retroviral expression of BCR, p40(ABL/BCR), p96(ABL/BCR) and BCR/ABL on the motility of the murine IL-3-dependent 32D myeloid precursor cell line. We used 32D cells, because of their capacity to migrate in the three-dimensional stroma spheroids. Transduction efficiency was always more than 70%, as assessed by the percentage of GFP-positive cells (data not shown) and expression of the transgenes was controlled by Western blotting (Fig. 4A).
The effects on migration were assessed in a three-dimensional in vitro spheroid model based on murine M2-10B4 bone marrow stroma cells. It reproduces the stromal microenvironment in the bone marrow [19] and allows to study the functional consequences of the complex interaction between cell membrane, adhesion systems and cell signaling pathways. It has been previously shown that migration into the M2-10B4 spheroids is closely related to the presence of activated Rho in the progenitor cells [19]. Here we show that BCR/ABL enhances the capacity of the 32D cells to migrate into the spheroids, whereas BCR did not increase migration. In contrast to p40(ABL/BCR), which had no effect, the presence of p96(ABL/BCR) inhibited the migration of 32D cells into the spheroids (Fig. 4B).
Taken together, these data provide evidence that BCR, ABL/BCR and BCR/ABL confer to 32D cells a different migration potential in the three dimensional spheroid model.
In contrast to the ABL/BCR proteins, BCR increases and BCR/ABL diminishes the adhesion of hematopoietic progenitors to endothelium
Adhesion is a complex sequential process of capture, rolling and firm adhesion, mainly mediated by the P- and E-selectin on the endothelial cells and their binding proteins, such as PSGL-1, CD18, CD11a and Mac-1, on the leukocyte membrane. Adhesion can be modified by the activation pattern of Rho-like GTPases through an inside-out signaling mechanism [20]. Adhesion to endothelial cells under shear stress was assessed in a flow chamber model based on HUVEC cells. Here we investigated the capacity of 32D cells expressing wt BCR, p40(ABL/BCR), p96(ABL/BCR) or BCR/ABL to adhere to HUVEC cells in a flow chamber under different shear stress conditions (0.1 and 2 dyn/cm2). HUVEC cells were also stimulated by TNFα for the induction of selectin expression. The adhesion of the 32D cells to unstimulated HUVEC cells was not modified by the expression of the different transgenes, independently of shear stress. In contrast, BCR increased and p185(BCR/ABL) reduced the adhesion of 32D cells on TNFα-stimulated HUVEC under both shear conditions (Fig. 5A). 32D cells expressing p40(ABL/BCR) or p96(ABL/BCR) did not show significant modifications in adhesion to TNFα-stimulated HUVEC with respect to controls (Fig. 5A). These modifications in adhesion were independent of the expression levels of PSGL-1 and Mac-1, because these were not modified by the transgenes, as determined by comparison with mock-transduced controls (data not shown).
BCR and p96(ABL/BCR)reduce endothelial transmigration in an SDF-1α gradient of hematopoietic precursors
Endothelial transmigration, which is mainly mediated by integrin expression on the endothelial cells, was investigated in a HUVEC-based transwell assay with an SDF-1α chemotactic gradient. BCR and p96(ABL/BCR) reduced the chemotactic response of 32D cells, indicating a role of the Rho-GEF domain because BCR/ABL and p40(ABL/BCR) did not influence transmigration in comparison to controls. These effects were independent of the integrin ligand proteins. In fact, Icam-1 expression was not modified in 32D cells by the transgenes (data not shown). Furthermore, the expression of the SDF-1α receptor CXCR4 on the surface of 32D cells was not modified by the expression of the transgenes (data not shown).
Taken together, these data indicate that the modification of the activation pattern of Rho-like GTPases by BCR, p185(BCR/ABL) and the ABL/BCR fusion proteins, with the accompanying alterations in cytoskeleton modeling interferes with the motility of hematopoietic precursor cells.