Background
Despite significant progress in the biology of pancreatic ductal adenocarcinoma (PDAC), treatment options for affected patients are still very limited and far from being curative with the exception of surgery (R0 resection). Chemotherapy is complicated by the desmoplastic nature of the complex mircoenvironmental architecture of the tumour tissue which is known to favour metastasis and chemoresistance of tumour cells.
Due to its overexpression in PDAC and its various functions in driving tumour suppression, the non-receptor tyrosine kinase SRC represents a promising biological target in experimental and clinical approaches to treat PDAC [
1‐
3]. Dasatinib (N-(2-chloro-6-methyl- phenyl)-2-(6-(4-(2-hydroxyethyl)- piperazin-1-yl)-2-methylpyrimidin-4- ylamino)thiazole-5-carboxamide; BMS-354825, Sprycel), a tyrosine kinase inhibitor originally developed against BCR-ABL and SRC [
4] and currently used in the treatment of CML [
5] and Philadelphia chromosome-positive acute lymphoblastic leukemia (reviewed in [
6]), has also shown promise in the treatment of various epithelial tumours [
6] including pancreatic cancer. Dasatinib-mediated inhibition of SRC slowed tumour progression and metastasis of human PDAC cells in an orthotopic mouse model [
7], and stimulated migration, invasion, and apoptosis in human [
8] and murine [
9] PDAC cells. Moreover, dasatinib inhibited metastasis in a mouse model of PDAC but failed to suppress primary tumour growth and prolong survival [
7]. Along the same lines, a phase II study found that dasatinib as a single-agent did not have clinical activity as first-line therapy in patients with metastatic PDAC [
10].
Several preclinical studies have shown that dasatinib can potentiate the antitumoural action of various other anti-cancer drugs (reviewed in [
11]. For instance, dasatinib synergized with gemcitabine to induce anti-proliferation and apoptosis in the pancreatic cancer cell line MIA PaCa-2 by decreasing the levels of ALDH1A1, a marker of tumour-initiating/cancer stem cells [
12]. Interestingly, 2/8 patients with pancreatic cancer (both gemcitabine-refractory) who received both gemcitabine and dasatinib showed a partial response (stable disease ≥6 months) [
13]. When dasatinib was combined with gemcitabine
and erlotinib (an epidermal growth factor-receptor (EGF-R) inhibitor), it inhibited the growth of xenografts of both sensitive and resistant PDAC cells in vivo without increasing toxicity [
14]. More recently, concomitant targeting of SRC, EGF-R, and transforming growth factor (TGF)-β has been suggested as a novel therapeutic approach in pancreatic cancer [
15].
Although originally developed as an inhibitor of BCR-ABL and SRC [
16], dasatinib, in drug affinity chromatography experiments was shown to interact with over 40 kinases, including SRC family kinases (SFKs), receptor tyrosine kinases, serine/threonine kinases (STK), MAP kinases, and EphA2 [
17]. One of the STKs identified with this approach was the type I receptor for TGF-β (TβRI, also termed activin receptor-like kinase 5, ALK5) [
18]. TGF-β1 is a pleiotropic growth factor that controls several aspects of tumour cell behavior such as proliferation, angiogenesis, desmoplasia, cell migration/invasion, and metastasis. It has a central role in the initiation and progression of PDAC [
19] which is evident from the observation that its aberrant expression in advanced tumour stages is associated with decreased survival in PDAC patients [
20], and that the TGF-β1 signalling pathway is among the 12 core pathways that are genetically altered in 100 % of PDAC tumours [
21]. Besides ALK5, TGF-β1 requires a second membrane-bound STK receptor, designated type II (TβRII), for signal transmission into cells. Upon phosphorylation by TβRII, ALK5 initiates canonical Smad as well as non-Smad signalling pathways [
22] that together mediate the promigratory and proinvasive effects of TGF-β. For PDAC, this is evident from the Panc-1 orthotopic mouse model in which ectopic expression of kinase-active ALK5 (ALK5
T204D) strongly enhanced metastasis [
23] while pharmacologic inhibition of endogenous ALK5 suppressed it [
24]. Targeting ALK5 in vivo is therefore a feasible approach to the treatment of PDAC and other carcinomas.
Like SRC, TGF-β/ALK5 signalling is currently targeted in the experimental and clinical treatment of various tumours. Given i) the interaction of dasatinib with ALK5 [
18,
25], ii) the structural similarity of dasatinib with the experimental SRC inhibitors PP2 and PP1, and iii) the ability of PP2 and PP1 to effectively inhibit the ALK5 kinase activity as well as TGF-β1-induced prooncogenic responses [
26,
27], we hypothesized that dasatinib should be able to block TGF-β1 signalling towards migratory, invasive and prometastatic outcomes. That dasatinib may possess potential efficacy against profibrotic TGF-β signalling in vivo was suggested by preclinical studies, in which dasatinib treatment of scleroderma and normal fibroblasts led to decreased production of extracellular matrix proteins [
28]. In light of the clinical use and efficacy of dasatinib, it is mandatory to understand its molecular mode of action in vivo including possible side-effects, regardless of whether they are adverse or beneficial for the patients.
To investigate the effect of dasatinib on TGF-β/ALK5 signalling in PDAC, we employed two TGF-β sensitive cell lines (Panc-1, Colo-357) that have been used in orthotopic mouse models of PDAC for evaluation of TGF-β antitumour activity in vivo [
23,
29]. Using impedance-based real-time measurement of cell migration, we show here that dasatinib strongly and dose-dependently inhibited TGF-β1-induced migratory responses
in vitro. As a result of dasatinib inhibition, activation of Smad2/3, upregulation of TGF-β transcriptional reporter genes as well as EMT/migration/invasion-associated gene expression in PDAC-derived cell lines was compromised. These results have implications for the use of dasatinib in experimental therapeutics, providing a molecular explanation for the reduction in the cells’ migratory response and an in vivo correlate for its anti-metastatic action. Furthermore, our data suggest that this agent based on its dual effect on protumourigenic TGF-β and SRC signalling pathways may be useful in the treatment of late stage metastatic disease in PDAC.
Discussion
The tyrosine kinase inhibitor dasatinib at low concentrations (IC <1.0 nM) potently inhibits ABL and SFKs. At higher concentrations it also inhibits other TKs and STKs such as ALK5, p38 MAPK, AKT, and FAK, all of which have been implicated in TGF-β signalling either in the canonical branch (ALK5), as downstream mediators of Smads (AKT, FAK, p38 MAPK) or in the non-canonical branch (p38 MAPK, FAK) [
22]. ALK5 was originally identified by drug affinity chromatography to interact directly with dasatinib [
18]. Later, this observation was confirmed in docking studies using the previously reported crystal structure of the ALK5 cytoplasmic domain. Here, the binding of dasatinib to ALK5 presented with the highest score when compared to the structurally related bosutinib, the commercially available ALK5 inhibitor LY-364947, and dorsomorphin [
25].
Likewise, we have shown previously that PP2 and PP1, which share structural similarity with dasatinib, effectively attenuated various oncogenic responses to TGF-β1 in malignant PDAC cell lines and strongly inhibited ALK5 kinase activity in
in vitro kinase assays [
26]. In the present study, we have tested in the Panc-1 and Colo-357 cell lines the prediction that dasatinib can mimic the action of PP2 and PP1 on TGF-β signalling. We found that dasatinib effectively blocked TGF-β1-dependent reporter gene activity, Smad2/3 activation and cell migration. The inhibitory effect of dasatinib (IC
50 = 0.8 nM for SRC in cell-free assays) on ligand-induced reporter gene activity was not dependent on SRC since it was not mimicked by another SRC inhibitor, bosutinib (IC
50 = 1.2 nM, [
38]). Moreover, dasatinib strongly inhibited migration driven by the ALK5
T204D mutant (Fig.
6). The inhibitory effect of dasatinib on TGF-β1/ALK5-mediated cell motility corresponded well with the potency of this agent to inhibit various TGF-β1-regulated marker genes involved in EMT, migration/invasion, and a cancer stem cell phenotype. Consistent with this, qPCR assays revealed that dasatinib suppressed TGF-β1 induction of MMP2, MMP9, N-cadherin, vimentin, Snail and Slug in Panc-1 and Colo-357 cells. In Panc-1 cells, dasatinib also attenuated downregulation of the adhesion molecules E-cadherin and NCAM1, together providing a molecular explanation for the anti-migratory/anti-invasive effect of dasatinib. Moreover, dasatinib effectively prevented the TGF-β1-induced upregulation of several stem cell-associated genes, some of which are either members of the TGF-β superfamily of ligands (β
A activin, BMP2) or function as a co-receptor for TGF-β (CD105/endoglin) and strongly decreased the number of colony-forming units derived from long-term TGF-β1-treated cells assumed to derive from cancer stem cells. From these data, we conclude that dasatinib targets ALK5 for inhibition of TGF-β-induced cell motility
in vitro and likely also other EMT-associated changes such as cancer stem cell differentiation.
Gordian and colleagues [
25] studied the combined effects of dasatinib and TGF-β on A549 NSCLC cells. When combined with TGF-β1 stimulation, dasatinib induced apoptosis in EGF-R mutant cells along with upregulation of pro-apoptotic BIM protein. Unfortunately, cell motility responses were not analysed in this study. For analysis of Smad2 and Smad3 activation by TGF-β1, these authors used a single concentration of 100 nM dasatinib rather than providing a dose–response curve with higher concentrations. With this dose they observed an increase in the levels of p-Smad3C compared to levels seen with TGF-β1 alone. Interestingly, we also observed in PDAC cells a tendency for an increase or at least a lack of effect in the concentration range of 10–100 nM, which was particularly evident in the Colo-357 cell line (see Fig.
1).
Several drugs have been developed for blocking TGF-β signalling that are in various stages of experimental and clinical evaluation. Strategies for inhibition of TGF-β signalling include blocking i) the binding of TGF-β to its receptors, ii) the ALK5 kinase using small molecules that act as competitive inhibitors for its ATP-binding site (SB431542, SB505124, SD-093, SD-208, LY580276, and Y2109761, a novel TβRI and TβRII dual inhibitor), and iii) Smad intracellular signal transduction via the Smad3 inhibitor SIS3 [
39]. However, although many of these drugs show promise in pre-clinical studies, the dual role of TGF-β in tumour progression requires a deeper understanding of the TGF-β signalling crosstalk with other pathways in order to design successful therapeutic approaches and protect the patients from undesired side effects.
Taken together, our results clearly show that dasatinib can block the TGF-β1-dependent cell motility at concentrations as low as 0.01 μM. However, this extremely sensitive and potent effect of dasatinib, particularly in Panc-1 cells (see Fig.
5), may have resulted from combined inhibition of ALK5
and SRC since we have previously shown that SRC contributed to TGF-β1-mediated cell migration without affecting TGF-β1/ALK5-induced activation of Smad2 and Smad3. This differential contribution of SRC may explain why dasatinib was able to block cell migration at much lower concentrations than phosphorylation of Smad2C/3C (compare Figs.
1 and
5). The higher efficacy of dasatinib in blocking TGF-β1-induced cell motility may be due to co-inhibition of p38 MAPK, the activation by TGF-β1 of which is SRC-dependent [
40] and required for cell migration in Panc-1 cells (H.U., unpublished observation). Experiments are currently underway to solve this issue. That dasatinib likely acted on SRC in the migration assays was evident from its suppressing effect on basal migration activity in Panc-1 cells which was previously shown to be SRC-dependent [
36]. Notably, TGF-β stimulation of p38 MAPK activation and cell invasion requires SRC to phosphorylate TβRII [
40]. Because this event is upstream of ALK5 activation, SRC is unlikely to be involved in ALK5
T204D-mediated cell migration consistent with the very similar migration curves of dasatinib and SB431542-treated cells.
In the light of our results, the possibility remains that dasatinib’s therapeutic effects result to a large extent from inhibition of TGF-β signalling rather than signalling by its
bona fide target SRC. That dasatinib can indeed compromise undesired effects of TGF-β in vivo was suggested by preclinical studies in an
in vitro model of a fibrosing disorder in which dasatinib treatment of scleroderma and normal fibroblasts led to decreased production of extracellular matrix proteins [
28]. The suppressive effect of dasatinib on TGF-β signalling is likely mediated by inhibition of the ALK5 kinase activity and may be exploited therapeutically in more advanced PDAC (when TGF-β1 expression and SRC activity are high) to synergistically reduce invasion, metastasis and eventually cancer stem cell formation. Consistent with this, concomitant targeting of EGF-R, TGF-β and SRC has been suggested as a novel therapeutic approach in pancreatic cancer [
15].
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
TB, BR, and RK carried out the majority of the experimental work, including immunoblot analyses, reporter gene and cell migration assays, and helped in design of the study, interpretation of the data and statistical analysis. HU and FG were the principal designer of the study, performed data analysis and drafted the manuscript. DR, HB and HL participated in coordination of the study and critically reviewed and communicated the manuscript. All authors read and approved the final manuscript.