The BCR-ABL associates with actin cytoskeleton, which is mediated by C-terminal FABD domain (McWhirter and Wang 1991; Van Etten et al. 1994). Among the BCR-ABL variants used in this study, we observed an exclusive cytoskeletal localization only in wildtype p190 and p210 BCR-ABL, kinase-dead BCR-ABL, and the BCR-ABL variant lacking the SH2 domain. In variants with deleted SH3 domain and those with removed or substituted Y177 (ΔY177, Y177F), the actin association was observed, but a significant moiety of the BCR-ABL showed also the retention in the ER-Golgi. Finally, no actin association in was found in ΔFABD, ΔIDR, BT and ΔAB variants, which instead showed cytosolic signal (Fig. 2B). This mislocalization appears due to the lack of actin binding domain in ΔFABD, or its defective conformation in ΔIDR-BCR-ABL lacking the region immediately adjacent to the FABD domain, consistent with the established role of FABD in actin localization (Hantschel et al., 2005). For BT and ΔAB variants, the absence of actin binding was consistent with previous results suggesting that these BCR regions also contribute to actin localization (McWhirter and Wang 1991). These regions also contribute to kinase activity of BCR-ABL, however kinase-dead BCR-ABL associates with actin, suggesting that these BCR regions promote actin localization independently of the BCR-ABL kinase activity, likely via correct oligomerization and formation of the BCR-ABL complexes.
All BCR-ABL variants with impaired cytoskeletal association were found autophosphorylated at Y412, suggesting that activation of the intrinsic kinase activity does not depend on cytoskeletal association of BCR-ABL. This is best exemplified by the ΔFABD-BCR-ABL, which showed Y412 autophosphorylation comparable to wildtype BCR-ABL, and also induced normal phosphorylation of SHIP2, STAT5, STAT1 and SHC1 (Fig. 4D). Interestingly, deletion of the IDR domain next to the FABD produced no negative effect on BCR-ABL autophosphorylation, but decreased SHIP2, STAT5, STAT1 and SHC1 activation. Because ΔIDR shows diffuse cytosolic localization similar to ΔFABD (Fig. 2), it is possible that impaired signaling of ΔIDR-BCR-ABL stems from altered conformation of the BCR-ABL protein rather than loss of proximity to the signaling mediators. Among the BCR-ABL core complex proteins, only CRKL binds the IDR domain directly (Sattler et al. 2002; Gregor et al. 2020). Both ΔIDR and ΔFABD showed impaired activation of CRKL, in contrast to BT and ΔAB, which activated CRKL to the levels comparable with wildtype BCR-ABL (Fig. 4E-G). The latter two variants lacked actin association completely (Fig. 2). This suggest that impaired CRKL activation in ΔIDR and ΔFABD variants is likely caused by the lack of binding motif (ΔIDR) or altered local topology of BCR-ABL (ΔFABD), but not a loss of proximity to the CRKL. Taken together, our data provide strong evidence that association of BCR-ABL with cytoskeleton is not necessary for integration of the core complex proteins, or activation of downstream signaling.
Individual deletions of the coiled-coil domain, ABL-binding region, SH3 or SH2 domain decreased BCR-ABL autophosphorylation at Y412 by approximately 59%, 47%, 27% and 70%, respectively (Fig. 4B-D), which is in agreement with the previously published evidence (Smith et al., 2003; Grebien et al. 2011). Deletions of particular domains did not affect phosphorylation of BCR-ABL substrates uniformly. For example, phosphorylation of SHIP2 and CRKL was not affected by SH2 deletion, while phosphorylation of STAT5, STAT1, and SHC1 decreased significantly (Fig. 4C, F), suggesting that both SHIP2 and CRKL associate well with BCR-ABL with decreased kinase activity. This was confirmed by co-immunoprecipitation experiments demonstrating normal association of SHIP2 and CRKL with ΔSH2-BCR-ABL (Fig. 7B, D). It was previously shown that SHIP2 and CRKL bind to kinase-dead or TKI- inhibited BCR-ABL (Gregor et al. 2020; Brehme et al. 2009). The lack of activation of some BCR-ABL substrates may be due to impaired association. For instance, the SHC1 phosphorylation was not decreased by the BT variant compared to the similar ΔAB, which showed significant decrease (Fig. 4B). Correspondingly, the co-immunoprecipitation experiments showed no association of SHC1 with ΔAB-BCR-ABL, while binding on the BT-BCR-ABL was not impaired (Fig. 7C). Altogether, our data show that structurally impaired BCR-ABL can recruit and activate its downstream signaling mediators.
We show that deletion or substitution of Y177 upregulates the BCR-ABL-mediated activation of the RAS-ERK pathway. Both Y177F and ΔY177 associate predominantly with the ER-Golgi compared to wildtype BCR-ABL, which localizes mainly to F-actin (Fig. 2). This suggests that altered localization can bring BCR-ABL to proximity with different substrates, leading to alternative mode of ERK activation. As ΔSH3-BCR-ABL also localizes to ER-Golgi, but does not upregulate ERK, the localization alone is unlikely to be the only cause of increased ERK activation by the Y177F- and ΔY177-BCR-ABL. The Y177F substitution or deletion disrupts GRB2 binding and the stronger interaction with SHC1, observed in Y177F- and ΔY177-BCR-ABL (Fig. 5F), suggests that Y177 and ΔY177 utilize SHC1 instead of GRB2 to activate the RAS-ERK pathway. The BT construct does not upregulate pSHC1 (Fig. 5B) and interaction of this construct with SHC1 is not changed (Fig. 5F, 7C), suggesting that BT construct utilizes different way of upregulating ERK that is SHC1 independent.
The precise role of Y177 in BCR-ABL oncogenic signaling remains controversial. It was shown that Y177F-BCR-ABL is not capable to transform Rat-1 fibroblasts (Pendergast et al. 1993), but can transform primary B-lymphocytes and the cytokine-dependent hematopoietic cell lines ex vivo (Cortez et al. 1995; Sattler et al. 2002). Mice with Y177F-BCR-ABL did not develop CML, but succumbed to the B- and T-lymphoid leukemias with prolonged latency (He et al. 2002; Zhang et al. 2001; Million and Van Etten 2000). This demonstrates that Y177 is not crucial for lymphocyte transformation by BCR-ABL; its absence changes the leukemia phenotype and extends disease latency. Our data demonstrate that the ability of Y177F-BCR-ABL to cause leukemia may stem from the ability to bind and activate the downstream signaling effectors, such as RAS-ERK.
In this article, we evaluated the contribution of the BCR-ABL structural domains to localization, downstream signaling and composition of the BCR-ABL core complex. We found that deletions of individual BCR-ABL domains altered but not abolished the BCR-ABL autophosphorylation, and activation of its signaling effectors. With the exception of the BT-BCR-ABL, the sedimentation of the signaling complexes of the truncated BCR-ABL variants in native cell lysates was comparable to wildtype BCR-ABL (Fig. 6). Moreover, in co-immunoprecipitation experiments (Fig. 7), the successful interaction was confirmed in 10 out of the 12 interactions probed, i.e. between four truncated BCR-ABL variants and three members of BCR-ABL core complex. Together, the data demonstrate that the BCR-ABL variants can integrate a fully functional signaling complexes, despite removal of the structural domains, and impaired subcellular localization.
Several studies have suggested a novel therapy approach to cure CML by targeting the residual cancer cells by molecules disrupting the BCR-ABL signaling complex. Here we show that the BCR-ABL complexes are very plastic, suggesting that multiple protein-protein interactions would have to be targeted at a time to achieve complete silencing of the BCR-ABL signaling. Instead, degradation of the BCR-ABL appears to be a better alternative, for instance by proteolytic-targeting chimera (PROTAC). Recently, PROTAC was developed against BCR-ABL, and showed specific degradation of BCR-ABL and its clinically relevant kinase domain mutants, and induction of apoptosis in human primary CML cells (Burslem et al. 2019).