Combined inhibition of AXL, Lyn and p130Cas kinases block migration of triple negative breast cancer cells

Blocking the migration of metastatic cancer cells is a major goal in the therapy of cancer. The receptor tyrosine kinase AXL is one of the main triggers for cancer cell migration in neoplasia of breast, colon, skin, thyroid and prostate. In our study we analyzed the effect of AXL inhibition on cell motility and viability in triple negative breast cancer cell lines overexpressing AXL. Thereby we reveal that the compound BMS777607, exhibiting the lowest IC50 values for inhibition of AXL kinase activity in the studied cell lines, attenuates cell motility to a lower extent than the kinase inhibitors MPCD84111 and SKI606. By analyzing the target kinases of MPCD84111 and SKI606 with kinase profiling assays we identified Lyn, a Src family kinase, as a target of both compounds. Knockdown of Lyn and the migration-related CRK-associated substrate (p130Cas), had a significant inhibitory effect on cell migration. Taken together, our findings highlight the importance of combinatorial or multikinase inhibition of non-receptor tyrosine kinases and AXL receptor tyrosine kinase in the therapy of triple negative breast cancer.


Introduction
The formation of metastasis resulting from migration of cells out of the primary tumor to invade distant sites in the organism is the number one reason for cancer patient mortality. [1][2][3] Cell migration is complex and it is precisely regulated by multiple factors such cell-cell and cell-substrate contacts but also by soluble mediators, the binding of which to their receptors on the cell surface triggers a network of interconnected signaling pathways. There is a plethora of mediators and effectors that represent potential targets for inhibitors with the capacity to influence cell migration. 4 The receptor tyrosine kinase (RTK) AXL (also known as UFO) is one of these targets and known to play an important role in cancer progression, invasion, metastasis, drug resistance and is correlated to patient mortality. 5,6 AXL belongs to the TAM (Tyro3, AXL, Mer) RTK family, 7 with all 3 members of the family being highly homologous in their extracellular and catalytic domains. 6 The activation of AXL occurs when the growth arrest specific 6 (Gas6) ligand binds to its extracellular domain, 8 which results in the activation of downstream signaling pathways such as mitogen activated protein kinases (MAPK), phosphatidylinositol 3-kinase (PI3K) RAC-a serine/threonineprotein kinase (AKT) and NF-KB (Nuclear Factor Kappa B). 9,10 AXL is frequently overexpressed in human cancers including lymphocytic leukemia, breast, colon, skin, thyroid and prostate cancer. [11][12][13][14][15][16][17] Down-regulation of AXL by siRNA attenuates the migration and invasion of breast, 18 ovarian, 19 hepatocellular carcinoma, 20 mesothelioma, 21 prostate 22 and pancreatic cancer. 23 Furthermore it has been shown that the inhibition of AXL results in the reduction of cancer progression and metastatic potential and induces apoptosis in cancer cells. 19,24 These observations suggest that AXL is an excellent target to counteract cancer and recently several AXL inhibitors have been described. 25 The aim of our study was to elucidate the impact of 3 tyrosine kinase inhibitors (TKI), namely SKI606, BMS777607 and MPCD84111 on Axl phosphorylation and the migration as well as viability of triple negative breast cancer (TNBC) cell lines.
SKI-606 (bosutinib) was described as a dual Src/Abl inhibitor exhibiting a significant effect on proliferation of colon cancer and chronic myelogenous leukemia. 26 It was also demonstrated that SKI606 inhibits AXL at 0.58 mmol/L concentrations and blocks migration and invasion of breast cancer cell lines. 12 BMS-777607 was described as a selective MET, RON (also known as MST1R) inhibitor but also shown to target AXL and Tyro3 kinase at low nanomolar concentrations. 27 BMS777607 inhibits AXL at 1.1 nM in a biochemical enzyme assay. 28 MPCD84111 is a quinolinyloxyphenylsulfonamides tyrosine kinase inhibitor (TKI) patented as an AXL inhibitor by the Max Planck Society under patent application example 12 from WO2011045084.
Our results prove that inhibition of migration by AXL inhibitors is independent form AXL kinase phosphorylation. Additionally, we highlight the importance of multikinase inhibition in the therapy of TNBC.

BMS777607, SKI606 and MPCD84111 exhibit different potential on AXL RTK inhibition
Overexpression of the RTK AXL and its implication in the migration of TNBC cell lines has already been demonstrated in publications and also by our studies (Fig. S2 and S3). 12,18,29 To determine the impact of 2 potential (BMS777607 and SKI606) and one patented (MPCD84111) AXL TKI on AXL phosphorylation we used 3 AXL-expressing TNBC cell lines. 12,30,31 The IC 50 values of AXL phosphorylation, were measured after 1 h of inhibitor treatment and subsequent Gas6 (250 ng/ml) stimulation. MPCD84111 displayed IC 50  In conclusion, the strongest inhibition of AXL phosphorylation was achieved by BMS777607 followed by MPCD84111 and SKI606.

The most potent AXL inhibitor BMS777607 does not inhibits significantly the migration and viability of TNBC cells
We determined the inhibitory effect of BMS777607, MPCD84111 and SKI606 on cell migration using Boyden chamber assays. The IC 50 value for MPCD84111 was 0.535 mM in MDA-MB-231 ( Fig. 2A), 2.9 mM in Hs578T  50 values we found that Hs578T cell were less sensitive to TKI treatment than MDA-MB-231 cells. Among the 3 analyzed AXL TKIs SKI606 had the lowest IC 50 value for inhibition of migration followed by MPCD84111 and BMS777607.
Additionally we analyzed the effect of AXL TKIs on cell viability. As illustrated in Figure 3, SKI606 and MPCD84111 had a significant inhibitory effect on Hs578T and BT549 cell viability. MPCD84111 exhibited an IC 50 value of 1.14 mM in Hs578T and 2.03 mM in BT549 ( Fig. 3D and G), SKI606 showed an IC 50 value of 2.85 mM in Hs578T and 1.27 mM in BT549 cells ( Fig. 3E and H) and BMS777607 had an IC 50 value of 9.43 mM in Hs578T cells and >10 mM in BT549 cells ( Fig. 3F and I). The MDA-MB-231 cell line was less sensitive to TKI treatment than Hs578T, with IC 50 values of 8.50 mM for MPCD84111 and 3.20 mM for SKI606, more than 10 mM for BMS777607 ( Fig. 3A-C).
The inhibitory concentrations for migration and viability were higher than the IC 50 value for inhibition of the AXL kinase activity, suggesting that the effect on migration and viability is not mediated through AXL RTK.

SKI606 and MPCD84111 display similarities in their target profile
Next, we wanted to understand why SKI606 and MPCD-84111 stronger block the migration of the TNBC cell lines than BMS777607. Therefore we performed a Cellular Target Profiling assay in Hs578T cell lysates to determine the targets of the compounds. The target spectrum of BMS777607 was reported by Schroeder et al., 28 therefore we focused on the characterization of SKI606 and MPCD84111.
The Cellular-Target-Profiling assay revealed that SKI606 mainly binds to the SRC-family kinases (Lyn, Src, Fyn and Yes), MAPK-family members, Ephrins and some of the PKC-family members. MPCD84111 displays a strong affinity for AXL, MET and Aurora B kinase and similarly to SKI606 binds to Lyn, Yes and Src ( Table 1). Between the 2 analyzed inhibitor SKI606 has with 40% more target than MPCD84111. This result is not surprising in the context that SKI606 potentially inhibits 148 kinases with a K d -value below 1 mM. 32 After comparing these data to the target profile of BMS777607, 28 we concluded that BMS777607 and MPCD84111 possess common targets such as Met and Aurora B. Binding to AXL kinase was confirmed for all 3 TKIs.
On the basis of Cellular Target Profiling assay and Ambit phage based competition binding assay data, 84 kinases were selected to further examine the compounds inhibitory potential in Kinase Profiler assay by Millipore, at 1 mM concentration. As expected from the results of the Cellular target profiling assay results, the Src family kinases and few other kinases (Abl, PTK5, EGFR, ErbB4, Lok, Tie2) had less than 50% activity after treatment with SKI606 and MPCD84111, whereas BMS777607 and MPCD84111 generally affected the Met-related kinases (Met, Ron) and the Aurora kinases (Aurora A/B/C) (Fig. 4). In conclusion, the profiling assays revealed that SKI606 and MPCD84111 commonly affect migration related kinases such as Src-family kinases, which might explain their significant effect on cell migration.

Lyn and p130Cas affect the migration of triple negative breast cancer cells
Among the Src-family kinases targeted by SKI606 and MPCD84111, Lyn is the only one being significantly overexpressed in TNBC cell lines. 33 Based on these result we selected Lyn to analyze its function on the migration and viability of MDA-MB-231, Hs578T and BT549 cells. It is known that Lyn interacts 34 with one of the members of the focal adhesion complex, namely p130Cas. 35 Therefore we analyzed p130Cas function in the migration and viability of MDA-MB-231, Hs578T and BT549 cells. To evaluate a functional connection between the selected proteins and cell migration we used a siRNA knockdown. The siRNA knockdown efficiency was validated by Western blot analysis (Fig. 5C).
The knockdown of Lyn and p130Cas led to a significant decrease in the migration of all cell lines (Fig. 5A). In MDA-MB-231 and Hs578T cells, knockdown of Lyn had no significant impact on the cell viability while the knockdown of p130Cas showed a clear effect (Fig. 5B). In BT549 cells all siRNA treatments had a significant effect on cell viability (Fig. 5B). The results shown are means and standard deviation calculated from 3 independent experiments. The strongest inhibitory effect on AXL phosphorylation was achieved by BMS777607 followed by MPCD84111 and SKI606.

Effect of AXL receptor tyrosine kinase inhibitors on intracellular migration related pathways in triple negative breast cancer cells
In the next step we characterized the effect of AXL TKIs on migration related signaling pathways. Serum starved cells were treated with 1 mM of SKI606, BMS777607 or MPCD84111 for 1.5 or 48 h in serum free medium.
As illustrated in Figure 6, MPCD84111 caused an inhibition of phosphorylation of Lyn, Pyk2, FAK, p130Cas in all cell lines. There was no significant impact on phosphorylation of c-Src, showing that the inhibition of the FAK-p130Cas-complex is not dependent on phospho-c-Src in these cell lines. SKI606 had an inhibitory effect on the phosphorylation of Pyk2, FAK, p130Cas and Lyn comparable to that of MPCD84111. In contrast to MPCD84111, SKI606 inhibited c-Src Tyr 416. There were no significant changes in the phosphorylation of Lyn after 1.5 h of treatment with MPCD84111 and SKI606 in Hs578T and BT549 cells. BMS777607 led to a less significant inhibition of Lyn, Pyk2, FAK and p130Cas phosphorylation in all cell lines. At the same time BMS777607 was proven to inhibit c-Src in Hs578T cells after 1.5 h of treatment but finally the phosphorylation of c-Src recovered after 48 h (Fig. 6B).
These results are consistent with our findings that SKI606 and MPCD84111 inhibit cell migration due to inhibition of essential migration related kinases such as p130Cas or Lyn.

Discussion
Triple negative breast cancer, which accounts up to 10% to 27% of breast cancer cases, is characterized by the absence of HER2, estrogen and progesterone receptor and by elevated phosphorylation of AXL, FAK, Met, Lyn, p130Cas and EGFR. 29,36,37 In this study we analyzed the efficacy of 3 AXL TKIs BMS777607, MPCD84111 and SKI606 exhibiting different selectivity profiles 28,32 on the migration and viability of AXL overexpressing TNBC cell lines, 30,31,38 namely MDA-MB-231, Hs578T and BT549.
Several studies demonstrated that the inhibition of AXL expression by AXL specific siRNAs attenuates the migration of TNBC cell lines. 12,18,39 Although these studies established the role of AXL protein in promoting TNBC cell migration, the phosphorylation status of AXL in breast cancer cell migration has not been clarified. In this study we elucidate the function of the AXL phosphorylation on the migration of TNBC cells. Therefore we used a selective AXL inhibitor BMS777607 and 2 AXL targeting multikinase inhibitor, SKI606 and MPCD84111 in our experiments. We show that in contrast to SKI606 and MPCD84111, the most specific and efficient AXL inhibitor BMS777607 was not capable to block the migration of TNBC cells. The inhibition values of migration were significantly higher than those determined for inhibition of AXL phosphorylation, suggesting that migration of the 3 studied TNBC cell lines does not depend on AXL phosporylation. Similar to our study Tan et al. analyzed the inhibitory potential of SKI606 on the migration of non small lung cancer cell lines. They showed that the inhibitory effect of SKI606 on migration was independent from AXL RTK. 40 This study is in line with our conclusion that efficiency of SKI606 and MPCD84111 is not correlated with the AXL phosphorylation inhibition. In contrast to our conclusion Zhang et al. correlated SKI606 migration inhibitory potential on TNBC cell with the compound effect on AXL phosphorlyation, which is contradictory because the IC 50 values for the inhibition of migration were much higher than the values for the inhibition of AXL phosphorylation. 12 Similar discrepancy between AXL and invasion inhibition by the AXL inhibitor R428 can be observed in the study of Holland et al., were the IC 50 value for invasion was 3000 nM whereas the IC 50 value for AXL phosphorylation inhibition was more than 200 times lower exhibiting a value of 14 nM. 41 Based on our results and on the publication of Vouriluoto et al. where it was shown that AXL kinase protein with gate keeper mutation has no effect on the migration of MDA-MB-231 cells, 39 we conclude that AXL might have a kinase independent function.
We can not exclude that AXL protein itself might play an important role as adhesion or scaffolding protein in cell migration as implicated by other publications by specific AXL siRNA knockdown experiments. 12,18 As a scaffolding protein AXL could have influence on multiple non-RTKs, serving as docking sites for as Lyn, Src, PI3K (p85), ZAP70, Syk, c-SRC, Lck, GRB2 and PLCg. 11,42 Antibodies against AXL and sAXL might interfere with the protein function of AXL shown by several publications. 19,43  However association and transactivation of AXL with EGFR and c-Met would limit the effects of targeted therapies. 44,45 To comprehend the efficiency of SKI606 and MPCD84111 on the migration of TNBC cells we examined the targets of SKI606 and MPCD84111. Based on our experiments we concluded that SKI606 has 40% more cellular targets than MPCD84111 but still both compounds commonly affect the migration related Src family kinases ( Table 1 and Fig 4). This could explain their significant inhibitory effect on the migration of TNBC cell lines. Among the Src family kinases targeted by SKI606 and MPCD84111 (Yes, Fyn, Hck, Fgr, c-Src, Blk, Lyn), we analyzed the role of Lyn. Lyn has been described as an important target in haematopoietic and solid tumors including TNBC, and was proven to be one of the most overexpressed kinase in TNBC cell lines. 29,33 Lyn plays an important role in the migration of cancer cells and was shown that can interact with p130Cas, one of the member of the focal adhesion complex. 34 Association of p130Cas with motility, invasion and survival of TNBC cells has been shown before by several publications. 29,46,47 Therefore additionally we analyzed the function of p130Cas in the migration and viability of TNBC cells. We proved that the siRNA-mediated knockdown of Lyn and p130Cas attenuated the migration of all 3 TNBC cell lines. Consequently our results underline the importance of Lyn and p130Cas in the migration of TNBC cell lines. At the same time the AXL TKIs SKI606 and MPCD84111 inhibit the phosphorylation of Pyk2, FAK, p130Cas and Lyn all 3 TNBC cell lines, while BMS777607 had no effect on these signaling molecules. This might explain why BMS777607, being the most efficient AXL TKI published until now, does not inhibit migration of TNBC cells. Despite other reports we show that Src inhibition attenuates breast cancer cell migration, 48,49 our study proves that the strong inhibitory effect on migration exerted by MPCD84111 does not depend on the inhibition of c-Src phosphorylation. Thus, the migration of TNBC cell lines described here does not depend on the activity of c-Src kinase but on other tyrosine kinases such as p130Cas and Lyn.
We also demonstrate that viability inhibition is not correlated with the migration attenuation, as the growth inhibitory concentrations of the investigated inhibitors (MPCD84111, BMS777607 and SKI606) were significantly higher than the IC 50 of migration inhibition. Thus we exclude viability inhibition being responsible for the reduction of cell migration in TNBC cells by the analyzed AXL inhibitors.
The present study reveals that efficacy of SKI606 and MPCD84111 to block TNBC cell migration is independent from AXL kinase phosphorylation. Although we cannot exclude AXL important role in modulation of tumor associated vasculature and immune cell function 43 and modulation of natural killer cells by the TAM receptors in vivo. 50 Our data show that SKI606 and MPCD84111 inhibit breast cancer cell migration through migration related kinases, namely Lyn and p130Cas. However tumor cell migration can not be blocked completely by inhibition of Lyn or p130Cas (Fig. 5). Our results underline previous findings that TNBC cell lines are not addicted to a single signaling pathway. 29 Based on these findings we conclude that the use of multikinase inhibitors such as SKI606 and MPCD84111 which target migration related kinases might improve the treatment of TNBC.
Short interfering RNA transfections AXL, Lyn and BCAR1 short interfering RNAs were purchased from Ambion (Invitrogen). The sequence for AXL 1845 siRNA was sense: 5 0 GGAACUGCAUGCUGAAUGAtt3 0 ; antisense: 5 0 UCAUUCAGCAUGCAGUUCCtg3 0 , for AXL 1847 siRNA was sense: 5 0 CAGCGAGAUUUAUGACUAUtt3 0 ;  Tyrosine kinase inhibitors SKI606 and MPCD84111 were obtained from Vichem Chemie (Budapest, Hungary). MPCD84111 is patented under application example 12 from WO2011045084. BMS777607 was a kind gift from LDC Discovery Center GmbH (Dortmund, Germany). All inhibitors were dissolved in DMSO and stored at room temperature in 10 mM stock solution.

Cellular profiling assay
The assay was performed by Kinaxo on s578T cell lysates. The assay combines different technology platforms, such as the stable isotope labeling by amino acids (SILAC) in cell culture or isobaric tags for relative and absolute quatification (ITRAQ). These technologies are used together with affinity-based separation methods with data showing the binding affinities of the compound to the corresponding targets. The assay is described in detail in the publication of Contradt et al., published in 2011. 51 Kinase profiler assay The assay was performed according to the manufacturer instruction by Millipore on 84 kinases with 1 mM compound concentration.

Cell viability and migration assay
We measured cell viability using a luciferase-coupled ATP quantitation assay (CellTiter-Glo; Promega, Cat no: G7571). Cells were dispensed at 1000 cells/100 ml/well in tissue-culture treated 96-well white bottom assay plates (PerkinElmer). After 24 h incubation the cells were treated with the indicated inhibitor concentration from the inhibitor dilution, except the siRNA treated cells. After inhibitor addition, plates were incubated for 72 h at 7 C in humidified atmosphere containing 5% CO 2 .
At the end of the incubation period, CellTiter-Glo reagent was added, to the contents of the plates, mixed by an orbital shaker for 2 min and incubated for 15 min. The luminescent signal was recorded using a Microplate Luminometer LB96V (Berthold Technologies). The experiments were performed at least 3 times.
The motility of the cells was determined by Boyden chamber assay. 50,000 cells were plated on top of a membrane with 8 mm pores (BD Transduction Laboratories, Heidelberg, Germany) in 300 ml of medium with 0.1% of FCS. As chemoattractant 1% FCS in 700 ml of medium was used. After incubation for 3.5 h at 37 C the cells remaining in the insert were removed with a cotton swab and the cells on the bottom of the filter were fixed with 0.05% Crystal violet/20% methanol (Sigma, Cat no: 32213-2). The cells which had traversed the membrane were counted and images were taken with 10 x objectives under brightfield illumination using a charge-coupled device cameramounted on a Axiovert 300 Microscope (Zeiss) and analyzed with MetaVue software. At least 5 random fields were counted per well.

Western blot analysis
Starved cells were incubated with 1 mM of indicated inhibitor concentration for 1.5 or 48 h 100 micrograms of total protein lysate were denatured with sample buffer containing SDS and 2mercaptoethanol at 95 C for 5 min and electrophoretically separated on a 4% to 12% polyacrylamide gel, transferred to a nitrocellulose membrane, and blocked with NET-gelatine. Antibodies raised against p-Lyn (Tyr507) # 2731, Lyn # 2732, p-Src (Tyr 416) #2101, p-Pyk (Tyr 580) #3291, p FAK (Tyr576/577) # 3281, p p130Cas (Tyr 410) #4011, p130 Cas # 13846 were obtained from Cell Signaling Technologies. Anti AXL antibody # sc-1096 was purchased from Santa Cruz. Anti-tubulin #T1926 was used from Sigma-Aldrich. The antibodies were used at 1:2,000 dilutions with overnight incubation at 4 C in NET-gelatine. The secondary antibodies HRP-coupled goat anti-rabbit (Bio-Rad) and goat anti-mouse (Sigma) were used at 1:10,000 dilutions in NET-gelatine. Detection of bound antibodies was performed by using an ECL substrate reaction (Perkin Elmer, Cat no: NEL103001EA) and exposed to Kodak X-Omat (Amersham Biosciences) film.

Immunoprecipitation
The homemade anti AXL capture antibody was pre-coupled to 40 ml A-Sepharose beads (GE Healthcare) in lysis buffer for 1 hour. After incubation were washed 3 times with lysis buffer. Cell lysates and pre-coupled antibody-beads were incubated at 4 C for 16 hours. The precipitates were washed 3 times with 1 ml lysis buffer, suspended in Laemmli buffer and boiled for 10 minutes. The AXL protein phosphorylation was analyzed by western blot using the homemade anti-p-Tyr clone 4G10 antibody. The total level of AXL was detected by using the anti-AXL Figure 6. Effect of AXL RTK inhibitors on intracellular signaling pathways. Immunoblots of lysates of MDA-MB-231 (A), Hs578T (B) and BT549 (C) starved cells treated with 1 mM of MPCD84111, SKI606 or BMS777607 for 1.5 or 48 h using antibodies against phospho-Pyk2 (pY580), phospho-FAK (pY576/577), phospho-p130Cas (pY410), phospho-Lyn (pY507), phospho-Src (pY416) and Tubulin (represents the loading control).

Statistical and data analysis
Assays were performed as biological triplicates, comparing inhibitor treated cell to DMSO control. IC 50 values were determined from dose response curve generated by XLfit5.1.0 software (IDBS). For statistical analysis Dunnett's Multiple Comparison Test was performed using GraphPad Prism 5 (GraphPad Software, Inc..). At values p ˂ 0.0001 differences were considered statistically significant.

Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.