Dasatinib blocks transcriptional and promigratory responses to transforming growth factor-beta in pancreatic adenocarcinoma cells through inhibition of Smad signalling: implications for in vivo mode of action

In order to test whether dasatinib can inhibit ALK5 function, we measured its effect on TGF-β1-induced C-terminal phosphorylation of Smad2 and Smad3 (p-Smad2C/3C), reflecting their state of activation. To this end, dasatinib dose-dependently inhibited TGF-β1-induced p-Smad2C and p-Smad3C in both Panc-1 and Colo-357 cells with the greatest effects seen at concentrations >0.1 μM (Fig. 1). A comparison of signal strengths between Panc-1 and Colo-357 cells after quantification by densitometry revealed that the dasatinib effect on both p-Smads was more pronounced in Panc-1 cells (Fig. 1).

The above data showed that TGF-β1-induced Smad2 and Smad3 activation was sensitive to dasatinib inhibition. Hence, we wondered whether this effect of dasatinib on the central mediators of TGF-β signalling would have consequences for Smad-mediated transcriptional activation of TGF-β1 target genes. To this end, cells were transiently transfected with the TGF-β reporter genes pCAGA₍₁₂₎-luc or 3TPlux. While pCAGA₍₁₂₎-luc is Smad-specific and responds to binding of Smad3/4 rather than Smad2/4 to the Smad binding elements (SBEs) in this plasmid [30], the 3TPlux plasmid contains promoter sequences from the plasminogen activator-inhibitor-1 gene and is responsive to both Smad and non-Smad pathways [31]. Treatment of Panc-1 cells with a combination of TGF-β1 and dasatinib strongly suppressed the response of both pCAGA₍₁₂₎-luc and 3TPlux to TGF-β1 induction in a dose-dependent manner (Fig. 2), and at a concentration of 10 μM was equally effective in suppressing reporter gene activity as the established ALK5 inhibitor SB431542 at 5 μM (Fig. 2). In contrast, the SRC inhibitor bosutinib (SKI-606, IC₅₀ = 1.2 nM) was unable to interfere with TGF-β transcriptional activity. Thus, dasatinib appears to be a powerful inhibitor of transcriptional responses to TGF-β in PDAC-derived cell lines.

Next, we asked whether dasatinib would also affect TGF-β1-induced upregulation of genes involved in EMT, cell migration, and invasion. Dasatinib at 10 μM effectively prevented the TGF-β1-induced downregulation of E-cadherin in Panc-1 cells and dramatically inhibited TGF-β1 induction of N-cadherin, vimentin, MMP2, MMP9, Slug and Snail in Colo-357 and Panc-1 cells as demonstrated by quantitative RT-PCR (qPCR, Fig. 3a) and immunoblotting (Fig. 3b). Although TGF-β induction of Snail protein in Colo-357 cells and N-cadherin protein in Panc-1 cells was only moderate, levels were clearly reduced upon dasatinib treatment (Fig. 3b). To show that the dasatinib-induced changes were caused by inhibition of ALK5 rather than SRC, we treated Colo-357 and Panc-1 cells, as above in the reporter gene assays (Fig. 2), with either SB431542 or bosutinib and repeated the qPCR analysis for two central regulators of TGF-β-induced EMT, Slug and vimentin (Fig. 3c). While SB431542 effectively abolished the TGF-β effect on Slug and vimentin mRNA levels in both cell lines, bosutinib lacked statistically significant effects (Fig. 3c).

Since this gene expression profile suggested that dasatinib might be able to interfere with TGF-β1-induced EMT and partial EMT has been shown to be associated with the aquisition of stem cell traits [32], we subjected Panc-1 cells to a long-term treatment with TGF-β1 according to a previously published protocol [32] to study the effect of dasatinib on the induction of stem cell gene expression. To this end, a 14-day exposure to TGF-β1 resulted in the upregulation of several pluripotency-associated genes: The TGF-β ligand superfamily members β_A activin and BMP2, the membrane-associated markers CD90, CD105 (also termed endoglin, a TGF-β coreceptor), and the transcriptional regulators OCT4 (the stem cell associated OCT4A isoform) and UTF1 (Fig. 4). Moreover, we observed downregulation of the adhesion molecule NCAM1 (Fig. 4a) and, surprisingly, of the xenobiotic transporter ATP-binding cassette sub-family G member 2 (ABCG2) (Fig. 4). Intriguingly, the TGF-β1 effect on all genes tested was effectively inhibited by concomitant treatment with dasatinib (Fig. 4a). To verify the possibility that dasatinib is able to block the TGF-β/EMT-induced generation of cells with stem cell character, we repeated the 14-day treatment of Panc-1 cells with TGF-β1 in the absence or presence of dasatinib, or SB431542 as control (long-term treatment of these cells with bosutinib was not possible due to its toxicity). Subsequently, we employed the colony formation assay (CFA) to assess the number of cells with stem cell features by virtue of their ability to generate from a single cell a microscopically discernible clone or colony [33, 34]. A quantitative analysis by counting revealed that both dasatinib, and SB431542 as control, strongly decreased the number of colonies/clones that arose from TGF-β1-treated cultures relative to untreated controls (Fig. 4b). Together, our data suggest that dasatinib interferes with TGF-β-induced EMT and stem cell generation and raise the possibility that it can block the generation of tumour-initiating cells in vivo.

Given the ability of dasatinib to block TGF-β1-induced EMT and cell motility-associated gene expression as well as Smad activation and transcriptional activity in conjunction with the requirement for Smad3 for TGF-β1-dependent migration of PDAC cells [35], we hypothesized that dasatinib also interferes with TGF-β1-induced cell motility in PDAC cells. To this end, dasatinib at the 10 μM concentration abolished both basal and TGF-β1-stimulated cell migration in both Panc-1 and Colo-357 cells (Fig. 5a). To determine whether this antimigratory effect of dasatinib was dose-dependent and to be able to calculate the concentration of half-maximal inhibition, we serially diluted the dasatinib and repeated the migration assays with lower concentrations. Interestingly, a concentration as low as 0.01 μM afforded an approximately 50 % reduction in migration activity in both cell types (Fig. 5b, magenta curves). With respect to inhibition of TGF-β1-stimulated migration, the dasatinib effect was comparable to that of SB431542 affording complete inhibition (Fig. 5c), however, dasatinib unlike SB431542 also suppressed the increase in basal migratory activity which in contrast to Colo-357 cells was quite high in Panc-1 cells (Fig. 5c). Interestingly, we have shown earlier that the high basal migration (which precedes TGF-β1-stimulated migration) was SRC-dependent [36], suggesting that in Panc-1 cells dasatinib acts as a dual inhibitor of ALK5 and SRC-mediated migration. Together, these results clearly show that dasatinib, in a very sensitive and effective fashion, blocked TGF-β1-induced migration/invasion in PDAC cells in vitro. This data together with those presented in Figs. 1, 2, and 3 strongly support our contention that dasatinib represents a powerful inhibitor of TGF-β1-induced cell motility by inhibiting ALK5 rather than SRC or other SFKs.

As shown previously, the experimental SRC inhibitors PP2 and PP1 strongly interfered with the high constitutive migratory activity conferred on these cells by ectopic expression of ALK5^T204D, a kinase-active version of ALK5 that harbors an activating point mutation in its glycine- and serine-rich (GS) domain [37] and does not rely on ligand binding, complex formation and phosphorylation by the TβRII kinase in order to signal. To test whether dasatinib was capable of mimicking the PP1/PP2 effect, we subjected Panc-1-ALK5^T204D cells (one of several individual and previously characterised clones [23]) to the RTCA migration assay. Notably, both dasatinib (Fig. 6a) and SB431542 (Fig. 6b) strongly reduced both migration directed by the exogenously expressed ALK5^T204D receptors and the extra portion induced by stimulation of endogenous (non-mutant) receptors in response to their activation by recombinant TGF-β1 ligand. These data are consistent with the results from dasatinib inhibition of ligand-induced cell migration in wild-type cells (see Fig. 5) and confirm that dasatinib targets the ALK5 kinase function or events downstream thereof for inhibition.

TGF-β1 was obtained from ReliaTech (Wolfenbüttel, Germany). Dasatinib was provided by Bristol-Myers Squibb (UK) and bosutinib (SKI-606) was kindly donated by Pfizer (USA). Stocks of these compounds were prepared in dimethyl sulfoxide (DMSO) which at the concentration used (0.1 %) had no measurable effect on cellular activities (data not shown).

The PDAC cell lines Panc-1 and Colo-357 were maintained in standard culture medium consisting of RPMI 1640 supplemented with 10 % fetal calf serum (FCS), L-glutamine, sodium pyruvate and 50 units/ml penicillin and streptomycin. The generation and characterization of individual cell clones from stable retroviral transduction of Panc-1 cells with a kinase-active mutant of ALK5 (T204D mutation) was described in detail in a previous publication [23]. These cells were cultured in the presence of 700 μg/ml geneticin (Life Technologies, Darmstadt, Germany). For stimulation experiments, cells were pretreated with DMSO or inhibitors (Dasatinib, SB431542) for 30–60 min before the addition of TGF-β1. The long-term treatment of Panc-1 cells with TGF-β1 was performed in standard culture medium with 2 % FCS according to a previously published protocol [32].

Inhibitor-treated cells were lysed in PhosphoSafe buffer (Merck) and the protein concentrations determined with the DC protein assay (Bio-Rad, München, Germany). Equal amounts of cellular proteins were fractionated by SDS-PAGE, transferred to PVDF membrane and immunoblotted as described in detail earlier [26]. The antibodies used were: β-actin (Sigma-Aldrich, #A1978), N-cadherin (BD Transduction Lab. #610920), anti-phospho-Smad2(Ser465/467) and anti-phospho-Smad3(Ser423/425), both from Cell Signalling Technology (Frankfurt/Main, Germany), Smad2 (Epitomics, Burlingame, CA, #1736-1), Smad3 (Abcam, Cambridge, UK, #ab40854), Snail (Cell Signalling Technology, #4719), and vimentin (Sigma-Aldrich, #V6630). In some experiments, the intensities of bands were quantified by densitometry using NIH image J.

The procedure and the conditions for real-time quantitative RT-PCR (qPCR) which was performed on an I-cycler with IQ software (Bio-Rad) were published earlier [23]. Primers were generally chosen to span exon-intron boundaries. However because OCT4 is devoid of introns, amplification of its mRNA required the removal of residual genomic DNA from the isolated RNA. Therefore, RNA was prepared with the NucleoSpin® RNA isolation kit involving digestion with RNase-free DNase I (Macherey-Nagel, Düren, Germany). Sequence information for amplification primers is provided in Additional file 1 Table S1 and in Ref. [41]. All values for the genes of interest were normalised to those for β-actin and TBP, and relative gene expression was calculated by the 2^-ΔΔCt method.

For reporter gene assays, Panc-1 cells were seeded in 96-well plates and cotransfected on the next day (4 h, serum-free) with LipofectAmine 2000 (Life Technologies), either one of the two TGF-β-responsive promoter-firefly luciferase reporter plasmids 3TPlux and pCAGA₍₁₂₎-luc (kindly provided by Drs. J. Massagué and S. Dooley, respectively), and the Renilla luciferase encoding vector pRL-TK-luc (Promega, Heidelberg, Germany). Each well received the same total amount of DNA. Following removal of the transfection mixture, cells were allowed to recover overnight in standard growth medium and on the following day were treated with TGF-β1 [5 ng/ml] and inhibitors or vehicle (added 1 h prior to the addition of TGF-β1) for a total of 24 h. Cells were then lysed in Glo lysis buffer (Promega) and luciferase activities determined with the Dual Luciferase Assay System (Promega). In all reporter gene assays, the data for each condition were derived from six parallel wells and were corrected for transfection efficiency with Renilla luciferase activity.

We applied the xCELLigence RTCA technology (OLS, Bremen, Germany) which represents a non-invasive and label-free approach for continuous (real-time) monitoring of cell migration and invasion on a cell culture level (for review see [42]. Considering that growth factors or their inhibitors may exert maximal effects at different time points after treatment, it is useful to monitor cell behaviour continuously and over a prolonged period of time. The precise determination of peak migratory/invasive activity of a given cell population may aid in deciphering the underlying mechanism of action of receptor kinases and their inhibitors.

Migration experiments were performed using modified 16-well plates (CIM-Plates 16, OLS ). The setup of the experimental device was described in detail earlier [27]. To begin an experiment, overnight serum-starved cells (30,000–60,000 per well) were mixed and preincubated with vehicle or dasatinib in 100 μl of culture medium containing 1 % FCS. After 30 min, TGF–β1 was added to a final concentration of 5 ng/ml and cells were seeded in the upper chambers. FCS, TGF-β1 and inhibitors had been added before to the lower chambers at the same concentrations (chemokinesis). After cell addition, CIM-Plates 16 remained at room temperature in the laminar flow hood for 30 min to allow cells to settle onto the membrane. PBS was added to the empty space surrounding the wells in order to prevent interference from evaporation. Each condition was performed in quadruplicate with a programmed signal detection every 15 min for a total of 8–24 h depending on the cell type. Data acquisition and analysis was performed with the RTCA software (version 1.2, OLS).

The CFAs are commonly used as survival assays to assess the cytotoxic effects of a compound on cell lines and to evaluate the efficacy of anticancer therapeutics. More recently, CFAs became an interesting method in cancer stem cell biology due to their potential to assess the capacity of cells to produce progeny. The assay relies on the assumption that only a cell which is largely undifferentiated can give rise to a countable colony. Consequently, the cell of origin must have features of a stem cell or progenitor cell. The CFA assay was carried out according to published protocols [33, 34]. Briefly, after treatment of Panc-1 cells with TGF-β ± dasatinib, cells were detached with Accutase, resuspended in culture medium and centrifuged for 5 min at 400 g. Subsequently, supernatants were discarded and cell pellets were resuspended in 1 ml of culture medium. Fourhundred cells each were seeded in three wells of a 6-well culture plate for each condition and incubated for14 days. After incubation, medium was aspirated, colonies were washed once in PBS, and fixed with 1 ml 4.5 % (w/v) paraformaldehyde for 15 min. Colonies were then stained with 0.1 % crystal violet for 1 h and finally washed twice in ddH₂O and air-dried overnight. Colonies consisting of at least 50 cells were counted.

Statistical significance was calculated from at least three independent experiments using the unpaired Student’s t-test. Data were considered significant at p < 0.05. The levels of significance are as follows: *, p < 0.05; **, p ≤ 0.01; ***, p ≤ 0.001.