ABL Tyrosine Kinase Inhibition Variable Effects on the Invasive Properties of Different Triple Negative Breast Cancer Cell Lines

The non-receptor tyrosine kinase ABL drives myeloid progenitor expansion in human chronic myeloid leukemia. ABL inhibition by the tyrosine kinase inhibitor nilotinib is a first-line treatment for this disease. Recently, ABL has also been implicated in the transforming properties of solid tumors, including triple negative (TN) breast cancer. TN breast cancers are highly metastatic and several cell lines derived from these tumors display high invasive activity in vitro. This feature is associated with the activation of actin-rich membrane structures called invadopodia that promote extracellular matrix degradation. Here, we investigated nilotinib effect on the invasive and migratory properties of different TN breast cancer cell lines. Nilotinib decreased both matrix degradation and invasion in the TN breast cancer cell lines MDA-MB 231 and MDA-MB 468. However, and unexpectedly, nilotinib increased by two-fold the invasive properties of the TN breast cancer cell line BT-549 and of Src-transformed fibroblasts. Both display much higher levels of ABL kinase activity compared to MDA-MB 231. Similar effects were obtained by siRNA-mediated down-regulation of ABL expression, confirming ABL central role in this process. ABL anti-tumor effect in BT-549 cells and Src-transformed fibroblasts was not dependent on EGF secretion, as recently reported in neck and squamous carcinoma cells. Rather, we identified the TRIO-RAC1 axis as an important downstream element of ABL activity in these cancer cells. In conclusion, the observation that TN breast cancer cell lines respond differently to ABL inhibitors could have implications for future therapies.


Introduction
Breast tumors are very heterogeneous and can be classified in three main groups based on their molecular profile: luminal cancers that express both estrogen and progesterone receptors; HER2-positive cancers that express the tyrosine kinase receptor ERBB2; and triple negative (TN) cancers in which none of these receptors is detected. TN breast cancers are the most Here, we investigated the effect of various ABL TKIs on the migratory and invasive properties of different TN breast cancer cell lines. We found that nilotinib inhibits the invasive properties of MDA-MB 231 and MDA-MB 468 cells, while it increases the invasiveness of BT-549 cells. Nilotinib also promoted invasiveness of Src-transformed fibroblasts. This result indicates that TN breast cancer cell lines respond differently to ABL inhibitors. We also found that the GEF TRIO and the GTPase RAC1 are potential downstream elements of ABL kinase negative regulation of invadopodia in TN breast cancer cells.

Gelatin degradation assay and invadopodia imaging
Gelatin degradation assays were performed as described previously [23]. Coverslips were ethanol-washed and coated with 50 μg/ml poly-D-lysine for 15 min, washed with PBS and crosslinked with 0.5% glutaraldehyde for 15 min. Coverslips were then inverted on a 20 μl drop of 1 mg/ml Oregon-Green 488-conjugated gelatin (Molecular Probes) for 20 min. After washing with PBS, coverslips were quenched with 5 mg/ml sodium borohydride for 5 min followed by washes with PBS. Finally, they were transferred in complete growth medium for 30 min-1 h before use. Cells were seeded and cultured on cross-linked gelatin for 3h in the absence (vehicle) or presence of inhibitors and then fixed for immunofluorescence studies. Wide-field imaging was done using Zeiss AxioimagerZ1 or Zeiss AxioimagerZ2 upright microscopes with Zeiss 40 X EC Plan NeofluaR 1.3 oil DIC and Zeiss 63X Plan-Apochromat 1.4 oil objectives. Image acquisition and quantification of gelatin degradation areas were carried out using Metamorph (Molecular Devices, Inc.).

Migration and invasion assays
Cell migration and invasion assays were performed in Boyden chambers. For migration assay, 20000 cells were added on the upper chamber containing medium with 0.5% FCS, while the lower chamber contained 10% FCS. For invasion assay, 50000 cells were added on the upper chamber containing 100 μl of 750 μg/ml Matrigel. Cells were fixed after 45 min (migration) or 24-48 h (invasion) in 3.7% paraformaldehyde solution containing 0.1% Triton X100 and 0.1% Hoechst 33342. Whole well images were acquired using a Zeiss Axiovert 200M or Leica DMIRE2 inverted microscope and a 10X EC Plan Neofluar 0.3 PH1 objective. Nuclei were counted in whole wells using the Metamorph software (Molecular Devices, Inc.).

Biochemical assays
Immunoprecipitation experiments, western blotting and kinase assays were performed as described in [22,2]. Briefly, cells were rinsed twice in PBS and scraped in 2X RIPA lysis buffer or in lysis buffer containing 1% Triton X-100 or 1% NP40, 10mM Tris-HCl (pH 7.5), 150 mM NaCl, 5 mM EDTA, 75U/ml aprotinin, and 1 mM vanadate. Immunoprecipitations were done using 800μg-1mg protein extracts and specific antibodies. 20 to 50 μg proteins were used as whole cell lysate (WCL). Immunoprecipitates and WCL were separated on 5-10% SDS-PAGE gels and transferred to Immobilon membranes. Detection was performed using the ECL System.

Variable effects of ABL kinase inhibition on the invasive properties of TN breast cancer cell lines
We first determined whether ABL kinase inhibitors currently used in the clinic affect the invasive properties of three TN breast cancer cell lines (MDA-MB 231, BT-549, and MDA-MB 468) and of transformed cells the reference model cell line 3T3 SrcY527F. To this aim, we plated cells on Oregon Green 488-conjugated gelatin-coated coverslips and then assessed invadopodia activity in each cell line after incubation with DMSO (negative control), imatinib, nilotinib, dasatinib or the SRC-like inhibitor SU6656 (positive control) by quantifying the dark areas over the green background. SU6656 and dasatinib inhibited the invasive activity of all cell lines, whereas imatinib and nilotinib had an inhibitory effect only in MDA-MB 231 and MDA-MB 468 cells (Fig. 1A, B and S1 Fig. A, B). Unexpectedly, both inhibitors increased by 2-fold gelatin degradation in BT-549 (Fig. 1A, B) and 3T3 SrcY527F cells (S1 Fig. A, B). Both effects were dose-dependent (Fig. 1C). Nilotinib-mediated degradation increase was associated with profound remodeling of the actin cytoskeleton as indicated by the appearance of increased number of F-actin punctae, indicative of invadopodia in BT-549 ( Fig. 1A) and many podosome rosettes in 3T3 SrcY527F cells (S1 Fig. A). These data suggest that imatinib and nilotinib increase matrix degradation by inducing invadopodia, in agreement with a recent report showing that the ABL kinases negatively regulate invadopodia/podosomes in 3T3 SrcY527F cells [12]. As imatinib and nilotinib can also target other TKs [24], we confirmed the involvement of ABL in this process by down-regulating ABL expression using specific siRNAs. ABL depletion resulted in increased matrix degradation in BT-549 cells and reduced degradation in MDA-MB  (Fig. 1D, E). This confirmed the central role of ABL in this cellular process, although a contribution of ARG cannot be excluded. In breast cancer, the ability of cells to form invadopodia is tightly linked to their invasive potential. Accordingly, incubation with nilotinib increased by 2.5-fold invasion of BT-549 ( Fig. 2A) and per cell (in μm 2 ; mean ± SEM). *p<0.05 (D) BT-549 cells were transfected twice at an interval of 24h control siRNA (Ctrl) or ABL siRNA. After 48h, cells were plated on Oregon Green 488 gelatin-coated coverslips for 3h. After fixation, the degradation area per cell (in μm 2 ; mean ± SEM) was quantified. *p<0.05, NS: non-significant. (E) BT-549 and MDA-MB 231 cells transfected twice with control (Ctrl) or ABL siRNAs as in (D) were lysed and immunoprecipitated with an anti-ABL antibody. Densitometry quantification normalized to tubulin is shown.
doi:10.1371/journal.pone.0118854.g001  (Fig. 2B). Conversely, nilotinib strongly reduced migration of MDA-MB 231 cells, but not of BT-549 (Fig. 2C) and 3T3 SrcY527F cells (S2 Fig. B), suggesting that ABL does not regulate tumor cell motility on its own. This was further confirmed by ABL expression down regulation using specific siRNAs (Fig. 3A, B and C). The HB-EGF autocrine loop is not involved in ABL negative regulation of BT-549 cell invasion We next investigated the underlying mechanism of ABL anti-invasive activity in BT-549 and 3T3 SrcY527F cells. These two cell lines display the highest ABL kinase activity levels among the cells used in this study ( Fig. 4A and S2 Fig. D). Nilotinib did not have any effect on SFK activity in BT-549 cells (Fig. 4B) and 3T3 SrcY527F (S2 Fig. E), suggesting that ABL kinase inhibition did not have any indirect effect on SFK activity, differently from what reported in [12]. Accordingly, nilotinib did not have any appreciable effect on the global tyrosine phosphorylation level in BT549 (Fig. 5E) and 3T3 SrcY527F cells (S2 Fig. C). These findings exclude a were used for in vitro kinase assays to test ABL kinase activity. Immunoprecipitates were incubated with GST-Crk as a substrate and P 32 -labeled ATP. P 32 incorporation by GST-Crk was detected by autoradiography. The Abl content of immunoprecipitates and densitometry quantification normalized to Abl are also shown. (B) Triton X-100 lysates of BT-549 and MDA-MB 231 cells incubated with DMSO (-) or 100nM nilotinib (+) were immunoprecipitated with an anti-CST1 antibody that recognizes SRC, FYN and YES. In vitro kinase assay to test SRC activity was performed using enolase as substrate and P 32 -labeled ATP. P 32 incorporation by enolase was detected by autoradiography. Src and tubulin are used as loading control. Densitometry quantification normalized to tubulin is shown.  potential negative ABL feedback loop towards SFK activity. The situation was different in MDA-MB 231 where Abl inhibition by nilotinib reduced SFK activity and this might explain why Abl inhibition strongly impacted on matrix degradation in this case (Fig. 4B). A recent study described an HB-EGF-dependent autocrine loop involved in invadopodia activity and regulated by ABL kinase activities [12]. As this ABL-dependent feedback loop was observed only in serum-starved conditions, we quantified gelatin degradation after serum starvation and incubation, or not, with EGF and/or nilotinib in Src-transformed fibroblasts. Gelatin degradation after serum starvation was similarly increased by incubation with EGF and/or nilotinib (S3 Fig. A, B), as described in [12]. Conversely, EGF had no effect on BT-549 cells (Fig. 5A, B), although the EGF receptor was well expressed and activated as indicated by the increased levels of tyrosine phosphorylation and phosphorylated MAPKs upon incubation with EGF (Fig. 5E). In MDA MB231 cells, that expressed low amount of the receptor (Fig. 5E) we observed no significant effect of EGF on gelatin degradation (Fig. 5C, D) and as described in [12].

RAC1 and TRIO may act downstream of ABL kinases to control cell invasion in BT-549 cells and Src-transformed fibroblasts
Incubation with nilotinib also increased the area of BT-549 (Fig. 6A) and 3T3 SrcY527F cells (S4 Fig. A) compared to cells treated with DMSO alone or to that observed in MDA-MB 231 cells (Fig. 6B). This observation is consistent with a previous report showing the increase of Abl-/-fibroblast area upon spreading and implicating the small GTPase RAC1 [15]. We thus investigated whether RAC1 could be involved in ABL anti-invasive activity. Like with nilotinib, incubation with NSC 23766 (a RAC1 inhibitor) also increased the cell area (not shown) and gelatin degradation in BT-549 and 3T3 SrcY527F cells (Fig. 6C, D and S4 Fig. B) but not in MDA-MB 231 cells (Fig. 6D). This effect was RAC-specific because the RHO GTPase inhibitor C3 did not have any effect (Fig. 6C and S4 Fig. B). To determine how RAC affects ABL antiinvasive signaling, we first investigated the role of the GEF TRIO because it controls RAC1 activity in invadopodia disassembly [14]. TRIO inhibition by sub-optimal doses (due to toxicity) of the pharmacological inhibitor ITX3 [18] significantly increased gelatin degradation by BT-549 and 3T3 SrcY527F cells compared to cells incubated with DMSO (Fig. 6C, D and S4  Fig. B). We further confirmed TRIO role by showing that TRIO down-regulation with two different shRNAs (shTrio1 and shTrio2) increased gelatin degradation by BT-549 (Fig. 7A, B and C) and 3T3 SrcY527F cells (S4 Fig. C, D and E), while it reduced gelatin degradation by MDA-MB 231 cells (Fig. 7D, E and F). Moreover, TRIO down-regulation strongly promoted the formation of podosome rosettes in 3T3 SrcY527F cells (S4 Fig. C). These findings suggest that ABL anti-invasive activity is mediated through a TRIO/RAC signaling pathway. To confirm this hypothesis, we investigated whether constitutive RAC1 activation could bypass nilotinib inhibition of ABL anti-invasive activity in 3T3 SrcY527F and BT-549 cells. Indeed, after incubation with nilotinib, the degradation areas of BT-549 cells transfected with mCherry-Trio were significantly smaller than in neighboring non-transfected cells (Ctrl) (Fig. 7G, H). Similar results were obtained in cells transfected with mCherry-GEFD1, a construct that includes only the RAC1-specific GEF domain of TRIO, demonstrating that this effect is specifically linked to RAC activation (Fig. 7G, H). Altogether, these results suggest that in BT-549 and 3T3 SrcY527F cells, ABL anti-invasive activity is mediated via a TRIO/RAC signaling pathway.

Discussion
It would be tempting to extend the application of ABL-targeting therapies to other cancers given their success in CML. However, this study is in agreement with a growing list of reports showing contrasting roles of the ABL kinases in solid tumors [11,12]. We found that ABL kinase inhibition by nilotinib or imatinib had opposite effects in BT-549 cells and MDA-MB 231 or MDA-MB 468 cells. The effect in BT-549 cells is similar to that in the mouse 4T1 breast cancer cell line, which is considered to be equivalent to human TN breast cancer cell lines in some reports [25]. How to explain the different outcomes of ABL inhibition in these different TN breast cancer cell lines? A recent study on the PKCα-PI3K signaling axis in the regulation of cell invasion retained our attention [26]. The authors demonstrated that PKCα has opposite effects on invadopodia formation, depending on the cell PI3K status: loss of PKCα inhibited invadopodia formation in cells harboring wild type PI3K signaling components, while it increased invadopodia formation in cells with deregulated PI3K signaling due to activating PI3K somatic mutations or PTEN inactivation. Particularly, the invasive activity of BT-549 cells, which do not express PTEN, was negatively regulated by PKCα [26]. On the other hand, PI3K/ PTEN signaling is not mutated in MDA-MB 231 cells [27]. It would be interesting to further investigate whether and how ABL kinases might affect PI3K/PKCα signaling to control invadopodia formation.
We then found that TRIO and RAC1 might be downstream components of the signaling pathway whereby ABL negatively regulates cell invasion in BT-549 cells. This is in agreement with previous reports showing a link between ABL kinases and TRIO or RAC1 [15,16] and with a recent study demonstrating a negative regulatory role of the TRIO-RAC1 axis on invadopodia activity of breast carcinoma cells [14]. What remains to be determined is how ABL is connected to TRIO. We could not identify TRIO as a direct ABL substrate because we did not detect tyrosine phosphorylation of TRIO in our assays (data not shown). While we cannot formally exclude TRIO regulation by tyrosine phosphorylation, ABL could regulate TRIO indirectly via PKCα-or PI3K-dependent mechanisms.
EGF is involved in this ABL negative function in head and neck squamous cell carcinoma. Conversely, we could not confirm such a mechanism in the TN breast cancer cell lines used in this study [12]. Other secreted factors could be involved, such as TGFβ, which is regulated by the ABL kinases in 4T1 breast cancer cells [11]. Netrin-1, which is specifically secreted by BT-549 cells, could also be an interesting candidate [28] because its receptor DCC can signal to TRIO [17] and is involved in invadopodia-driven basement membrane transmigration in vivo [29].
In conclusion, we found that, in vitro, ABL kinase role in the invasive activity of TN breast cancer cells varies in different cell lines. Consistent with our in vitro observation, dasatinib has limited activity, as single-agent, in unselected patients with metastatic breast cancer. Similarly, a recent phase II study in which patients were selected based on dasatinib predictive gene signatures defined in preclinical models did not report any satisfactory tumor response except in one patient [30]. Therefore, there is a clear need to identify novel and reliable markers for patient selection. Unraveling the molecular cause of the heterogeneous ABL effect on the invasive behavior of breast tumor cells might help identifying reliable markers for patient selection.