Background
Epithelial to mesenchymal transition (EMT) is a biological process in polarized epithelial cells, which occurs in various physiological and pathological conditions [
1]. Complete EMT is characterized by spindle-like cell morphology, loss of epithelial cellular markers such as E-cadherin, and gain of mesenchymal phenotype by expressing filament proteins including vimentin and α-smooth muscle actin [
1,
2]. Cells undergoing EMT are highly mobile and invasive [
2,
3]. During embryonic development, EMT enables cells to migrate or invade into neighboring tissues and maturate or differentiate into specialized cells [
1,
2]. In epithelial malignant progression, EMT has emerged as a critical player in regulating cancer cell invasive phenotype [
4,
5]. Acquiring EMT is a critical step for cancer cells to dissociate from a primary tumor mass and subsequently migrate and invade adjacent tissues for remote metastasis [
4,
5]. Recently, EMT has been linked with cancer stem-like phenotype in certain epithelia tumors [
6,
7]. As demonstrated, breast cancer cells express several cellular markers that resemble the stem-like phenotype during their progression towards EMT [
6,
7]. These observations highlight the importance of cellular EMT program in tumorigenic progression of cancer cells.
Development of EMT in cancer cells is regulated and precisely controlled at different cellular levels [
4,
5]. Various proteins such as receptor tyrosine kinases (RTK) [
8‐
10], cytokine receptors [
11,
12], intracellular signaling molecules [
13,
14], and transcriptional factors [
15,
16] are involved in cellular EMT program. At the signaling level, RTK-mediated activation of extracellular signal-regulated kinase (Erk1/2) has been implicated as a critical pathway for initiation of EMT [
13,
17,
18]. Transforming growth factor (TGF)-β1-stimulated TGF-β receptor I/II and Smad signaling also play a pivotal role in induction of EMT [
11,
19]. Additional pathways such as Wnt-β-catenin signaling also have been implicated in EMT [
20]. Convincing evidence indicates that signals coordinated among different pathways such as the RTK-Erk1/2 and TGF-β1-Smad pathways maximize trans-differentiation of epithelial tumor cells towards EMT [
1,
2]. Moreover, such coordination raises the possibility that a converging signal for diverse pathways may exist, and may act as a central determinant controlling cellular EMT program.
Human 90 kDa ribosomal S6 kinases (RSK) belong to a family of Ser/Thr kinases with two unique functional kinase domains [
21]. The family consists of four isoforms (RSK1-4), of which RSK1 and RSK2 are currently under intensive investigation for their roles in cellular signaling [
21‐
23]. In quiescent cells, RSK forms a protein-protein complex with Erk1/2 [
24] and is considered to be a downstream signaling molecule of the Ras-Erk1/2 pathway [
21]. Activation of RSK is featured by phosphorylation, dissociation from Erk1/2, and subsequent nuclear translocation [
21]. Various extracellular factors including growth factors, cytokines, chemokines, peptide hormones, and neurotransmitters are known to directly activate RSK [
21]. RSK phosphorylation occurs at multiple Ser and Thr residues through sequential steps by various kinases such as Erk1/2 [
21‐
24]. Activated RSK phosphorylates many cytosolic and nuclear targets such as FLNA, BAD, DAPK, p27
KIP1, and transcription factors including CREB, NF-κB, and NFAT3 [
21‐
25]. Recently, RSK has emerged as a major player in the control of epithelial cell phenotype and motility [
22]. RSK is indicated as a principal effector of the Ras-Erk1/2 pathway for eliciting a coordinated promotile/invasive program and phenotype in epithelial cells [
22]. A genome-wide RNAi screen also has found that multiple proteins in various pathways depend on RSK for cellular migration [
23]. These discoveries indicate that activation of RSK could be an essential convergent point for regulating cellular phenotypic changes and motile/invasive activities.
The present study sought to identify the major signaling molecule(s) responsible for EMT induced by macrophage-stimulating protein (MSP) [
26], also known as hepatocyte growth factor (HGF)-like protein [
27]. MSP is a serum-derived growth factor that specifically binds and activates the RON receptor tyrosine kinase [
28,
29], a member of the MET proto-oncogene family [
27]. Previous studies have observed that RON-mediated activation of the Ras-Erk1/2 pathway is critically important in transducing signals leading to EMT [
30,
31]. However, the downstream signaling molecule(s) that controls RON-mediated EMT is unknown. To facilitate this study, Martin-Darby canine kidney (MDCK) cells expressing human RON, which is known to show complete EMT [
30,
31] was used as a model and a cell-shape based screen using various small chemical inhibitors was applied. By analyzing potential signaling proteins that are involved in MSP-induced EMT-like activities, we discovered that RSK2 is a principle effector molecule responsible for MSP-induced EMT in MDCK and human cancer cells. Evidence also indicates that RSK2 is responsible for TGF-β1-induced EMT.
Discussion
The purpose of this study is to identify the major signaling molecule(s) that controls MSP-induced EMT in epithelial cells. Altered RON expression and activation contribute to malignant progression of various epithelial cancers [
30,
42]. RON is overexpressed in various types of primary cancer samples including those from colon, breast, and pancreas [
42]. Aberrant RON activation also causes increased tumor cell proliferation, matrix invasion, and drug resistance [
42]. Currently, the role of MSP and RON in regulating EMT under physiological conditions is largely unknown. In contrast, MSP-induced RON activation or RON overexpression have been shown to induce EMT in various cancer cells including colon, breast, and pancreas [
30,
31,
43‐
45]. The changes to mesenchymal phenotype in RON-activated tumor cells have been considered as a molecular basis for increased tumor malignancy including cell migration, matrix invasion, and distance metastasis [
42]. Several upstream signaling proteins such as Erk1/2 have been implicated in MSP-induced EMT [
30,
31]; however, the major effector molecule(s) that transduces RON signals leading to EMT is still unknown. Intracellular proteins such as β-catenin and NF-κB have been identified as effector molecules in MSP-induced EMT [
45‐
47]. Nevertheless, their significance is often limited to particular cell models. Thus, identification of the major signaling molecule(s) is important not only for an understanding of the cellular mechanisms of EMT, but also for the development of potential therapies that target cancer cell migration and invasion.
Results from this study indicate that RSK2 is a major determinant bridging RON signaling to EMT. This conclusion is supported by the following evidence. First, inhibition of RSK, as indicated in the cell-shape based screen by using specific RSK inhibitor SL0101, completely prevented MSP-induced spindle-like morphology. Inhibitors that target other proteins such as NF-κB, Stat3, and hedgehog, except CP-1 and PD98059, only showed moderate effect. This indicates that RSK activation is essential in MSP-induced spindle-like morphology. Second, MSP-induced RON activation dissociated RSK2 from Erk1/2, and caused RSK2 phosphorylation and subsequent nuclear translocation. These data suggest that MSP is a strong RSK activation inducer, which is mediated by RON transduced signals. Third, RSK2 phosphorylation relied on the RON-Erk1/2 pathways. Inhibition of RON or Erk1/2 by their corresponding small chemical inhibitors prevented MSP-induced RSK2 phosphorylation. These data also established that RSK is a downstream molecule in the MSP-RON-Erk1/2 axis. Fourth, inhibition of RSK2 by SL0101 blocked MSP-induced spindle-like changes, which is evident by the redistribution of β-catenin to the membrane and reorganization of f-actin to original epithelial morphology. Moreover, in SL0101 treated cells, epithelial morphology was completely restored with re-expression of E-cadherin and claudin-1, reduction of vimentin expression, and minimized transcription repressor Snail expression. Fifth, SL0101 prevention of RSK2 activation decreased MSP and TGF-β1-induced cell migration. As shown in the wound healing assay, RON-mediated cell migration was dramatically reduced upon inhibition of RSK2 by SL0101. Finally, RSK2 overexpression led to EMT-like phenotypes in colon HT-29 cancer cells that express extremely low levels of RSK2. Moreover, specific siRNA-mediated RSK2 knockdown prevented MSP and TGF-β1-induced EMT-like activity in pancreatic cancer L3.6pl cells. Considering these factors, we concluded that SRK2 is the major effector molecule in RON-mediated EMT.
In reviewing cellular mechanisms underlying EMT in different types of epithelial and cancerous cells, it is apparent that various proteins belonging to multiple signaling pathways are involved in regulating EMT [
4,
5]. The identified proteins include Erk1/2, PI-3 kinase, AKT, p38, β-catenin, NF-κB, Stat3, Smad, and others [
11‐
20]. The typical example is the Erk1/2-mediated signaling event that leads to EMT [
17,
22]. Specifically, Erk2 but not Erk1 has been found to be critical in EMT induction, which is mediated by DEF motif-dependent signaling events [
17]. Currently, the signaling proteins participated in EMT represent at least seven different signaling pathways. The involvement of such diverse signaling proteins suggests the possible existence of a central signaling molecule that acts as a switch for initiation of EMT in epithelial cells. In supporting this notion, recent studies has shown that RSK acts as a principal effector molecule in coordinating cellular EMT program in epithelial cells [
22]. Genome-wide RNAi screen also has discovered that multiple proteins in a broad range of pathways depend on RSK for induction of cellular migration program [
23]. We observed that RSK2 activation is critical in controlling EMT in MDCK and cancer cells mediated by MSP. Moreover, RSK2 is also required for TGF-β1-induced EMT. The involvement of RSK2 in two different signaling pathways suggests that RSK2 acts as a potential central molecule in regulating EMT and cell migration. In other words, RSK2 activation acts as the convergent point for both RON-Erk1/2 and TGF-β receptor I/II-Smad pathways leading to complete EMT.
The importance of RSK2 in RON signaling also establishes a critical link to other signaling molecules observed in MSP-induced EMT and cell migration. Activation of Erk1/2 is required for MSP-induced EMT [
30,
31]. As a downstream molecule of the Erk1/2 pathway, RSK2 transduces MSP-induced and Erk1/2 mediated signal for EMT as demonstrated in this study. In breast cancer cells, NF-κB activation is implicated in RON-mediated cellular motility [
47]. RSK is known to activate NF-κB by phosphorylating NF-κB inhibitor IκBα and inducing its degradation [
48]. This finding suggests that the observed NF-κB activity in MSP-stimulated breast cancer cells could be channeled through RON-activated RSK2. In colon cancer cells stimulated by MSP, increased β-catenin accumulation contributes to spindle-like morphologies with increased migration [
35]. RSK2 activation is known to increase steady-state of β-catenin through phosphorylation and inhibition of a β-catenin regulator GSK-3β [
49]. These activities imply that the RON-mediated inhibition of GSK-3β [
35] could be caused by MSP-induced RSK2 activation. The role of MSP-activated AKT activity in cell migration is another example [
34]. Currently, evidence of direct RSK activation by AKT is not available. In contrast, studies have indicated that RSK is a mediator of growth factor-induced activation of PI-3 kinase and AKT in epithelial cells [
50]. Thus, it is likely that MSP-induced AKT activation is mediated by RSK. Such activation facilitates AKT in regulating MSP-induced cell migration. Considering all these facts, we reasoned that RSK is centered in MSP-induced and RON-mediated EMT with increased cell migration.
Studies sing pancreatic L3.6pl and colon HT-29 cells provide additional evidence showing the importance of RSK2 in MSP-induced EMT-like activity. First, we confirmed results derived from the MDCK cell model and demonstrated that RSK2 but not RSK1 is selectively involved in regulating RON-mediated EMT and associated cell migration. In the L3.6pl cell model, only RSK2 specific siRNA prevented MSP-induced EMT and cell migration. Second, we demonstrated that MSP-induced EMT-like phenotype is dependent on RSK2 expression and activation. In L3.6pl cells that express regular levels of RSK1 and RSK2, MSP induces EMT-like phenotypes featured by elongated cell morphology, reduced E-cadherin expression, and increased vimentin expression (Figure
6). In contrast, these activities were not observed in HT-29 cells that express minimal levels of RSK1 and RSK2. HT-29 cells express both RON and oncogenic variant RON160 and both regulate HT-29 cell growth [
51]. However, MSP fails to induce EMT and migration in HT-29 cells, which provides indirect evidence indicating the role of RSK2 in MSP-induced EMT and cell migration. Rescue experiments by pRSK2 cDNA transfection confirmed this theory. As shown in Figure
6C, RSK2-transfected HT-29 cells underwent spindle-like morphological changes with diminished E-cadherin and increased vimentin expression. Additional evidence supporting this notion comes from studies using RSK2-specific siRNA. Knockdown of RSK2 expression significantly inhibited MSP-induced L3.6pl cell migration (Figure
7), which reaffirms the importance of RSK2 in MSP-induced EMT. The final observation is that the effect of RSK2 on EMT is not limited to MSP. TGF-β1-induced EMT and cell migration also were affected by inhibition of RSK2. HT-29 cells with minimal RSK2 expression did not respond to TGF-β1. Spindle-like morphology was only seen when RSK2 is overexpressed. Western blot analysis of E-cadherin and vimentin expression in RSK2 deficient and transfected HT-29 cells confirmed that this is the case. RSK2 siRNA based analysis of cell migration further demonstrated that knockdown of RSK2 expression significantly impairs TGF-β1-induced L3.6pl cell migration.
Authors' contributions
QM performed the majority of biochemical analysis and biological experiments. SG and SSP did cellular immunofluorescent studies. HPY worked on anti-RON antibody production and characterization, RWZ, YQZ, and MHW participated in the design of the study and drafted the manuscript. All authors have read and approved the final manuscript.