Oncogene (2011) 30, 3625–3635 & 2011 Macmillan Publishers Limited All rights reserved 0950-9232/11 www.nature.com/onc ORIGINAL ARTICLE HGF-independent potentiation of EGFR action by c-Met

AM Dulak1, CT Gubish1, LP Stabile1, C Henry1 and JM Siegfried1,2

1Department of Pharmacology & Chemical Biology, University of Pittsburgh, Pittsburgh, PA, USA and 2University of Pittsburgh Cancer Institute, University of Pittsburgh, Pittsburgh, PA, USA

The c-Met receptor is a potential therapeutic target for non- Introduction small cell lung cancer (NSCLC). Signaling interactions between c-Met and the mutant epidermal growth factor Epidermal growth factor receptor (EGFR) is a receptor receptor (EGFR) have been studied extensively, but tyrosine kinase (RTK) that exhibits cellular action not signaling intermediates and biological consequences of only through ligands, such as EGF, transforming lateral signaling to c-Met in EGFR wild-type tumors are growth factor-alpha (TGF-a) and amphiregulin, but minimally understood. Our observations indicate that also through interactions with other RTKs and delayed c-Met activation in NSCLC cell lines is initiated G-protein coupled receptors (Daub et al., 1997; Knowlden by wild-type EGFR, the receptor most often found in et al., 2005; Liu et al., 2007; Thomas et al., 2006). EGFR NSCLC tumors. EGFR ligands induce accumulation of regulates well-characterized signaling cascades that activated c-Met, which begins at 8 h and continues for 48 h. modulate invasive growth, a program requiring precise This effect is accompanied by an increase in c-Met coordination of growth, motility and angiogenesis that is expression and phosphorylation of critical c-Met tyrosine unregulated in human cancers (Ritter and Arteaga, 2003). residues without activation of mitogen-activated protein Because of this, EGFR has been a strong candidate for kinase (MAPK) or Akt. Gene transcription is required for therapeutic targeting with tyrosine kinase inhibitors delayed c-Met activation; however, phosphorylation of (TKIs) such as gefitinib and erlotinib in many cancers, c-Met by EGFR occurs without production of hepatocyte including non-small cell lung cancer (NSCLC). However, growth factor (HGF) or another secreted factor, supporting despite preclinical evidence demonstrating high efficacy of a ligand-independent mechanism. Lateral signaling is EGFR inhibitors, these expectations were unmatched in blocked by two selective c-Met tyrosine kinase inhibitors clinical settings (Molina et al., 2006). One reason that (TKIs), PF2341066 and SU11274, or with gefitinib, an these EGFR inhibitors have not met expectations is due to EGFR TKI, suggesting kinase activity of both receptors is compensatory signaling through mechanisms involving required for this effect. Prolonged c-Src phosphorylation is additional RTKs. Specifically, there has been strong observed, and c-Src pathway is essential for EGFR evidence linking EGFR to activation of another promi- to c-Met communication. Pretreatment with pan-Src family nent RTK involved in invasive growth, c-Met (Engelman kinase inhibitors, PP2 and dasatinib, abolishes delayed et al., 2007). c-Met phosphorylation. A c-Src dominant-negative con- c-Met interacts with the hepatocyte growth factor struct reduces EGF-induced c-Met phosphorylation com- (HGF) ligand, also identified as scatter factor, in a pared with control, further confirming a c-Src requirement. paracrine, endocrine or autocrine manner (Nakamura Inhibition of c-Met with PF2341066 and siRNA decreases et al., 1986; Stoker et al., 1987; Bottaro et al., 1991; EGF-induced phenotypes of invasion by B86% and motility Naldini et al., 1991c). Canonical activation of c-Met by B81%, suggesting that a novel form of c-Met activation occurs when two c-Met monomers interact, inducing is utilized by EGFR to maximize these biological effects. trans-autophosphorylation on tyrosine residues Y1230, Combined targeting of c-Met and EGFR leads to increased Y1234 and Y1235 within the kinase domain. This xenograft antitumor activity, demonstrating that inhibition activation leads to subsequent phosphorylation on of downstream and lateral signaling from the EGFR-c-Src– tyrosine residues Y1349 and Y1356, which comprise c-Met axis might be effective in treatment of NSCLC. the multisubstrate docking site, and allows recruitment Oncogene (2011) 30, 3625–3635; doi:10.1038/onc.2011.84; of adapter molecules responsible for downstream signal- published online 21 March 2011 ing, such as Gab1 and c-Src (Naldini et al., 1991a, b; Ponzetto et al., 1994; Ma et al., 2003). Additionally, Keywords: c-Met; c-Src; EGFR; cross-talk phosphorylation of Y1365 regulates cell morphogenesis, and Y1003 is required for internalization and degra- dation of c-Met through c-Cbl interaction (Peschard et al., 2001). The HGF–c-Met interaction is critical in a Correspondence: Dr JM Siegfried, Department of Pharmacology & number of tightly controlled physiological processes, Chemical Biology, University of Pittsburgh, Suite 2.18, Hillman such as tissue regeneration, in contrast to uncontrolled Cancer Center, 5117 Centre Avenue, Pittsburgh, PA 15213, USA. E-mail: [email protected] HGF/c-Met activation in tumorigenesis (Trusolino Received 21 July 2010; revised 15 February 2011; accepted 17 February and Comoglio, 2002). Like EGFR, dysregulated HGF/ 2011; published online 21 March 2011 c-Met signaling has been implicated in many cancers, Potentiation of EGFR action by c-Met AM Dulak et al 3626 including NSCLC, and disruption of this pathway by Results c-Met-directed inhibitors decreases tumorigenic poten- tial (Stabile et al., 2004; Ma et al., 2005; Stabile et al., EGFR ligands induce prolonged phosphorylation and 2008). increased c-Met protein levels in NSCLC cells c-Met is frequently coexpressed with EGFR family It has been demonstrated that EGFR activation contri- members in human tumors, and it has been demon- butes to c-Met tyrosine phosphorylation in a variety of strated that these RTKs can signal to one another cell models, with varying kinetic parameters (Fischer et al., (Nakajima et al., 1999; Jo et al., 2000; Fischer et al., 2004; Xu et al., 2010). As NSCLC commonly express both 2004; Shattuck et al., 2008). Presnell et al. (1997) first receptors, we examined whether EGFR ligands induce c- showed that c-Met could be trans-activated by EGFR in Met phosphorylation in human NSCLC cell lines. Both rat liver epithelial cells constitutively expressing TGF-a. 201T and A549 cells were selected because of moderate Similar studies have since observed this signaling in expression levels of wild-type EGFR and wild-type c-Met, NSCLC tumor models where EGFR contains activating which is representative of the majority of NSCLC tumors mutations, such as L858R or E746-A750del (Guo et al., found in patients (Supplementary Figure S1). C-Met also 2008). Additionally, lateral signaling to c-Met from is not amplified in these cell lines (data not shown). EGFR has been identified in EGFR wild-type models, Analysis of activated c-Met was performed following and much of this data suggest that cross-communication stimulation with ligands EGF, TGF-a or HGF. Time- from EGFR to c-Met is dependent on EGFR or c-Met course analysis of EGF and TGF-a-treated 201T cells expression levels (Ponzetto et al., 1991; Bergstrom et al., revealed that c-Met phosphorylation first appears at 8 h, 2000). To further validate this mechanism, Xu et al. is sustained at maximal levels for 48 h, and decreases recently demonstrated that induction of c-Met phos- thereafter, whereas EGFR phosphorylation begins at phorylation required upregulation of c-Met through 5 min and is completed by 1 h (Figure 1a, Supplementary EGFR-activated hypoxia-inducible factor-1a in both Figure S2, and data not shown). In contrast, HGF EGFR wild-type and mutant cell lines (Xu et al., 2010). stimulation induced c-Met phosphorylation that begins Despite these increasing reports of EGFR–c-Met lateral within 5 min and quickly terminates. (Figure 1a). Similar signaling, the underlying mechanism of EGFR-induced results were obtained in A549 and H1435 cells (Supple- c-Met phosphorylation is not well understood, and the mentary Figure S3). EGFR ligand stimulation also extent of involvement of c-Met in carrying out EGFR- resulted in a 1.5- (EGF) to 2.0- (TGF-a) fold increase in initiated phenotypes has not been well documented. As total c-Met protein levels (Po0.0005) and a 2–4-fold c-Met signaling acts as an acquired resistance mechan- upregulation of c-Met transcript levels at 24 h (Po0.005) ism to EGFR TKI therapy, preclinical efforts are (Figure 1b and Supplementary Figure S4). ongoing to augment c-Met-targeted therapies with Preincubation with the pan-transcription inhibitor, EGFR inhibitors to facilitate tumor growth inhibition actinomycin D, before EGF administration abolished the (Engelman et al., 2007; Tang et al., 2008; Yano et al., increase in total c-Met protein and corresponding delayed 2008). phosphorylation of c-Met, while having no effect on direct We previously demonstrated that exogenous prosta- activation by HGF (Figure 1d). This demonstrates that a glandin E2 (PGE2) stimulated invasion in our NSCLC transcriptional component is required for EGFR trans- model system through an intricate signaling axis phosphorylation of c-Met and that the increase in c-Met requiring EGFR ligand production, c-Met and the Src expression is due to new c-Met transcription. family kinases (SFKs) (Siegfried et al., 2007). Here, we The individual c-Met tyrosine residues required for extend our studies to show that EGFR activation of activation and cell signaling are well characterized. c-Met in NSCLC cells is a delayed event that occurs not The initiating event for c-Met activation is phosphory- through HGF-ligand production, but through an intra- lation at residues Y1230/Y1234/Y1235, followed by cellular signaling cascade requiring gene transcription, residues required for cell signaling, such as Y1349, as well as tyrosine kinase activities of EGFR, the c-Src Y1365 and Y1003. To determine whether c-Met is and c-Met. We also found evidence for formation of being phosphorylated in response to EGFR ligands in a complex between c-Met and c-Src that was maximal at a manner similar to HGF, we measured individual 24 h after EGF stimulation. Moreover, c-Met pharma- phospho-tyrosine levels of c-Met. It was confirmed that cologic inhibition and siRNA knockdown reduced following EGF stimulation, all c-Met tyrosine residues EGF-induced invasion and motility, confirming that analyzed were phosphorylated and followed a pattern EGFR requires c-Met to maximize invasive pheno- similar to HGF stimulation, albeit at delayed time types. In combining EGFR (gefitinib) and c-Met points (Figure 1c). Together, these data suggest a novel (PF2341066) TKIs, we observed a statistically signifi- mode by which EGFR can regulate a prolonged, cant reduction in xenograft tumor growth of a NSCLC full activation of c-Met through phosphorylation that cell line expressing wild-type EGFR, compared with persists for long periods and increased c-Met protein. individual targeting. These results indicate that EGFR- induced, HGF ligand-independent delayed c-Met acti- vation requires c-Src and suggest that EGFR utilizes Kinase activities of both EGFR and c-Met are necessary lateral signaling to c-Met for maximal stimulation of for c-Met phosphorylation induced by EGF processes of invasive and motility that promote lung To ascertain the signaling cascade responsible for c-Met cancer. phosphorylation, use of the EGFR TKI, gefitinib, was

Oncogene Potentiation of EGFR action by c-Met AM Dulak et al 3627 a Ligand Stimulation b Ligand Stimulation Time: 05’30’ 1h2h 4h 8h 24h 48h Time: 024h5’30’ 1h2h 4h 8h 48h

c-Met EGF pMet EGF β c-Met -Actin 1 0.7 0.6 0.9 0.7 1.2 5.5 17 14 1 1.1 1.1 1.2 1.0 0.9 1.2 1.6 1.5 TGF- pMet c-Met TGF- α c-Met β-Actin α IP: c-Met 1 0.5 0.5 1 0.7 0.8 6.6 5.6 4.1 1 0.9 1.5 1.3 1.8 1.8 2.2 2.2 1.2 Whole Cell Lysates HGF pMet c-Met HGF

c-Met β-Actin 1 10 13 20 25 12 7.7 4.1 0.8 10.9 0.9 1.3 1.3 1.3 1.2 1.2 0.7

cdEGF HGF Time (h): 0 2 4 8 24 48 0 1 2 4 8 24 pMet pMet (Y1234/35) c-Met 1 1.2 1.0 3.6 33 54 1 5.5 3.7 0.7 1.1 0.4 IP: c-Met 1 4.3 0.3 10 14 0.4 pMet (Y1349) c-Met 1 7.3 2.4 1.4 0.8 0.4 1540.7 0.8 0.8 33 β Total -Actin Lysate pMet (Y1003) 1 2.5 1.2 1.9 1.3 0.6

IP: c-Met 1 2.0 1.5 3.8 16 23 1 15 8.5 4.8 4.1 0.9 XX EGF X X HGF pMet (Y1365) X X X Act D 1 1.0 1.2 1.0 7.5 15 1 34 8.7 11 1.8 15

Treatment XX X DMSO c-Met

Figure 1 EGFR ligands induce c-Met activation. (a, b) 201T cells were serum deprived for 2 days before stimulation with EGF and TGF-a (10 nM) for 0–48 h, followed by immunoprecipitation (IP) with c-Met and western blotting with pY99, total c-Met and b-actin antibodies. (c) Specific c-Met phosphorylation sites were measured by c-Met IP, then immunoblotting with phospho-c-Met antibodies: Y1003, Y1234/35, Y1349 and Y1365. All blots were stripped and re-probed with a total c-Met antibody to confirm equal loading. (d) 201T cells were serum starved for 2 days, pretreated with actinomycin D (0.01 mg/ml), and treated with either EGF (10 nM) for 24 h or HGF (10 ng/ml) for 5 min. Cell lysates were analyzed for c-Met tyrosine phosphorylation and expression. employed to discern whether EGFR tyrosine kinase activity (Klinghoffer et al., 1999). This family, particularly was required. Pretreatment with gefitinib abolished tyrosine c-Src, has been identified as a key intermediary in phosphorylation of c-Met induced by EGF, while not regulating multiple steps for lateral RTK–RTK signal- inhibiting rapid c-Met tyrosine phosphorylation in response ing (Popsueva et al., 2003). Because of these data to HGF (Figures 2a and c). Additionally, gefitinib also and observations of c-Met tyrosine phosphorylation, we inhibited the accompanied increase in total c-Met protein hypothesized that the SFKs are critical signaling following EGF treatment (Figure 2a). No association was molecules in delayed EGFR activation of c-Met in observed between EGFR and c-Met (data not shown). NSCLC. To identify the presence of SFK members, we These results confirmed that induction of an active EGFR assessed protein expression in the lung cancer cell lines. by EGF was needed to facilitate c-Met phosphorylation. It was observed that c-Src was ubiquitously expressed We next addressed whether kinase activity of c-Met, across our NSCLC cells (Supplementary Figure S1). itself, is required for delayed c-Met phosphorylation. Activation of c-Src is known to be downstream To address this, 201T cells were pretreated with c-Met of EGFR (Goi et al., 2000). Therefore, we assessed TKIs, SU11274 and PF2341066, and c-Met phosphor- the response to EGF stimulation at delayed time ylation status was measured following EGF stimulation points corresponding with c-Met phosphorylation by (Figure 2b). These c-Met TKIs had no effect on EGFR measuring levels of activated c-Src in 201T cells. The autophosphorylation induced by EGF (Figure 2d). results suggest that EGF stimulation activated c-Src C-Met inhibitors ablated all c-Met phosphorylation in a biphasic and a continually increasing manner initiated by EGF, while having no effect on c-Met (Figure 3a). The initial activation of c-Src was present protein upregulation, demonstrating that c-Met is not 2 h post EGF stimulation and was subsequently reduced being utilized by EGFR as an adapter-like molecule, but to near baseline levels by 4 h, whereas the second wave rather delayed activation of c-Met kinase activity is used of greater activation was initiated 8 h post treatment and to relay the EGFR signal (Figure 2b). lasted 48 h. This pattern of c-Src phosphorylation suggests a temporal association between delayed c-Src activation and delayed c-Met activation following C-Src is a key biphasic mediator in signal transmittance EGF stimulation. PF2341066 treatment does not block from EGFR to c-Met delayed c-Src phosphorylation at 8–24 h, suggesting that The SFKs are a group of non-RTKs, including c-Src, all c-Src activation is upstream of c-Met (Supplementary which regulate downstream tumorigenic phenotypes Figure S5). C-Src has a central role in mediating

Oncogene Potentiation of EGFR action by c-Met AM Dulak et al 3628 pMet EGF c-Met IP: c-Met

Gefitinib c-Met Vehicle DMSO Gefitinib

pMet Total β Lysate -Actin

c-Met X X X EGF IP: c-Met X X X HGF c-Met X X X PF2341066 Total Lysate β-Actin X X X SU11274 Treatment X X X DMSO

pMet pEGFR c-Met

IP: c-Met EGFR X X HGF IP: EGFR X X X EGF X X EGF X X PF2341066 X X X Gefitinib X X SU11274 Treatment XX DMSO Treatment X X DMSO Figure 2 EGFR and c-Met kinase activity is required for EGF-induced c-Met phosphorylation. 201T cells were serum deprived for 2 days, pretreated with indicated inhibitors, and stimulated with EGF (10 nM) for 24 h or HGF (10 ng/ml) for 5 min. Cell lysates were analyzed for c-Met tyrosine phosphorylation and expression. (a) Pretreatment with gefitinib (100 nM,1mM,10mM) occurred for 2 h, and cells were stimulated with respective ligands. (b) After 2 h pretreatment with the c-Met inhibitors, SU11274 (1 mM) or PF2341066 (1 mM), cells were growth factor treated. (c) Cells were pretreated with gefitinib (10 mM), followed by stimulation with either HGF (10 ng/ml) or EGF (10 nM) for 5 min. Cell lysates were immunoprecipitated for c-Met and probed for pY99 and total c-Met. (d) 201T cells were pretreated with either SU11274 or PF2341066 (1 mM), followed by stimulation with EGF (10 nM) for 5 min. Cell lysates were immunoprecipitated for EGFR and probed for pY99 and total EGFR.

invasive phenotypes, and c-Src inhibition blocks all It has been hypothesized that c-Met dimerization EGF-induced migration of NSCLC cells at 48 h and activation can result from increased c-Met density (Supplementary Figure S6). (Ponzetto et al., 1991; Bergstrom et al., 2000). There- We explored whether inhibition of the SFKs could fore, we determined whether c-Src activation through block EGFR-induced c-Met phosphorylation through EGFR was required for elevated c-Met mRNA levels. the use of pan-SFK inhibitors, PP2 and dasatinib. Pre- Surprisingly, blockade of c-Src by PP2 had no signi- treatment with either PP2 or dasatinib blocked all EGF- ficant effect on EGF-induced c-Met transcript levels, induced c-Met phosphorylation while having minimal suggesting that c-Src is not upstream of the modest effect on increased total c-Met protein (Figure 3b). In increase in c-Met protein, but may be upstream of contrast, neither PP2 nor dasatinib showed any inhibi- other critical transcriptional events (Figures 3c and 4d). tory effect on c-Met autophosphorylation (Figure 3b), These experiments advocate that c-Src is a candidate demonstrating that the SFKs are an intermediary molecule that is likely involved at multiple points needed for signaling from EGFR to c-Met. To discern leading to initialization of EGFR–c-Met trans-activa- whether c-Src, specifically, is a key mediator, 201T cells tion, although the exact role remains unclear and were stably transfected with a dominant-negative c-Src participation of other SFKs in some of the effects construct that reduced c-Src activity in both dominant- cannot be excluded. negative c-Src construct-4 and dominant-negative c-Src construct-9 clones by 28 and 66%, respectively, com- pared to an empty vector (EV) plasmid as published HGF autocrine signaling is not responsible for previously (data not shown and Liu et al., 2007). EGFR-induced c-Met phosphorylation Stimulation of dominant-negative c-Src construct cells HGF autocrine loops have been observed in many tissue for 24 h with EGF resulted in a 2–3-fold decrease in total types; therefore, analysis of HGF production and c-Met phosphorylation compared with 201T EV cells secretion was performed at times corresponding with (Figure 3b). Through co-immunoprecipitation studies c-Met phosphorylation induced by EGF. Both 201T and following 24 h EGF stimulation, we found that there is A549 cells were treated with EGF under serum-free no evidence of a c-Src/c-Met or a c-Src/EGFR associa- conditions and tissue culture media was harvested for tion up to 24 h, suggesting that no persistent EGFR/ HGF enzyme-linked immunosorbent assay. No HGF c-Src/c-Met signaling complex is formed, and that c-Src secretion was detected at any time point, suggesting that activity at 8–48 h may indirectly induce c-Met phos- autocrine signaling is not occurring (Figure 4a). To phorylation (data not shown). This does not rule out the confirm this observation, mRNA was harvested from possibility that c-Src might directly phosphorylate c-Met 201T cells, following exposure to EGF, and HGF levels through a transient interaction. were measured. The results from the quantitative reverse

Oncogene Potentiation of EGFR action by c-Met AM Dulak et al 3629 EGF Time (h): 0 2842448IgG pSrc (Y418) c-Src

IP: c-Src IgG 1 5.6 2.0 3.8 6.8 8.1

pMet pMet EV DN-Src 4 DN-Src 9 c-Met c-Met IP: c-Met IP: c-Met EGF: - ++--+ 1 13 4.0 33 30 0.7 1 14 2.8 14 15 0.4 pMet c-Met c-Met c-Met IP: c-Met Total Total β Lysate Lysate -Actin β -Actin 1 4.0 1 2.9 1 1.4 1 2.4 2.2 2.2 3.0 2.4 1 2.6 2.1 1.7 1.5 1.3 c-Met X X EGF X X EGF

Total β X X HGF X X HGF Lysate -Actin X X X PP2 X X X Dasatinib 1 1.3 1 1.2 1 1.3 Treatment XX X DMSO Treatment X XX DMSO

201T A549

Figure 3 C-Src mediates EGFR-induced c-Met phosphorylation. 201T cells were serum deprived for 2 days, followed by addition of EGF (10 nM) for 0–48 h. (a) Cell lysates were immunoprecipitated for total c-Src and analyzed for phospho-c-Src (Y418). (b) 201T cells were pretreated for 2 h with SFK inhibitors PP2 (500 nM) or dasatinib (50 nM) before 24 h EGF (10 nM) or 5 min HGF (10 ng/ml) stimulation, or were stably transfected with either a dominant-negative c-Src construct (DN-Src) or empty vector (EV) and stimulated with EGF. Cell lysates were prepared and analyzed for phospho- and total c-Met levels. (c) A549 and 201T cells were pretreated for 2 h with PP2 (500 nM) before 24 h EGF stimulation, mRNA was harvested, and subjected to c-Met quantitative reverse transcriptase-PCR using b-Gus as an internal control. ***Po0.0005; **Po0.005; *Po0.05 Student’s t-test.

transcriptase–PCR assay detected no presence of the to tumor progression. PF2341066 treatment signifi- HGF transcript produced in unstimulated or stimulated cantly reduced EGF-induced 201T and A549 cellular NSCLC cells. (Figure 4b). It was also observed that invasion by 86% at 48 h (Figure 5a). Cell motility media harvested from 24 h EGF-treated 201T cells did through EGFR, as assessed by 48 h wound-healing not induce c-Met phosphorylation on serum-deprived assay, was also inhibited with PF2341066 by 81% and 201T cells up to 48 h, suggesting that no secreted factor 57% in 201T and A549 cells, respectively (Figure 5b). is responsible for c-Met activation (data not shown). To confirm these findings, invasion and wound- A neutralizing antibody to HGF did not block healing experiments were repeated with c-Met specific EGF-induced phosphorylation of c-Met (Figure 4c). siRNA. Knockdown of c-Met by siRNA resulted in a On the basis of these data, HGF is not involved in 92% specific reduction of c-Met compared with the non- EGF-initiated c-Met lateral signaling. Taken together, targeting siRNA (Figure 5d). This knockdown resulted our findings favor an intracellular lateral signaling in similar results to PF2341066 in EGF-induced wound- cascade from EGFR to c-Met that involves transcrip- healing and invasion experiments, where a 77 and 53% tion of an as-yet unidentified protein that initiates reduction in EGF-induced invasion and wound healing delayed c-Met phosphorylation (Figure 4d). was observed in 201T cells, respectively (Figure 5c). Cell proliferation in response to EGFR ligand only increases cell number by about 25% at 24 h, when maximal C-Met inhibition abrogates EGF-induced invasive growth migration is inhibited. Knockdown of c-Met does phenotypes inhibit this modest growth response, so it is possible It is well documented that both EGFR and c-Met that c-Met does participate in some EGFR-induced initiate signaling cascades for invasive growth, and growth responses (data not shown). Despite c-Met downstream signaling from each RTK involves similar being phosphorylated at all tyrosines and required for molecules (Trusolino and Comoglio, 2002). On the basis EGFR phenotypes, delayed c-Met activation by EGF of our observations of delayed c-Met activation by was not associated with mitogen-activated protein EGF, we hypothesized that c-Met mediates at least part kinase (MAPK) and Akt phosphorylation, suggesting of the EGFR-induced invasion and motility that leads that these normal elements of c-Met signaling are not

Oncogene Potentiation of EGFR action by c-Met AM Dulak et al 3630 HGF mRNA

MODEL Rapid Delayed Clustering? pMet EGF TGF-α c-Met c-Met EGFRTranscription c-Met IP: c-Met 1 0.9 5.4 5.6 8.1 1.5 P c-Met P β-Actin Total Lysate 10.2 1.8 2.1 1.5 1.4 P c-Src ynthesis X X EGF rotein S X X HGF P ew XXX HGF-Nab N

Treatment X X X IgG Control Gene Transcription Invasion Motility

Figure 4 EGFR activation of c-Met does not require HGF production or secretion. A549 and 201T cells were serum starved for 2 days before stimulation with EGF (10 nM). (a) Tissue culture media was harvested for 0–24 h time points and subjected to HGF enzyme-linked immunosorbent assay with fibroblast-conditioned media (FCM) as a positive control. (b) Cells were treated, mRNA was harvested, and subjected to HGF quantitative reverse transcriptase-PCR using b-Gus as an internal control. Normal lung fibroblast (NLF) mRNA was utilized as a positive control, whereas the breast cancer cell line, MCF-7, was used as a negative control for HGF expression. (c) EGF (10 nM) or HGF (10 ng/ml) was pretreated for 2 h with HGF-neutralizing antibody or IgG control (300 ng/ml) before addition to 201T cells for 24 h. Cell lysates were prepared and analyzed for phospho- and total c-Met levels. (d) Proposed model of delayed c-Met activation. In NSCLC cells, EGFR ligands induce a delayed accumulation of tyrosine- phosphorylated c-Met receptors, as well as an increase in total c-Met levels. EGFR ligands activate EGFR tyrosine kinase activity, which leads to rapid c-Src signaling and new gene transcription that is required for EGFR–c-Met lateral communication. At delayed time points, activated c-Src continues to accumulate, and c-Met is phosphorylated to maximize EGFR-activated phenotypes of invasion and motility. Production of unidentified proteins might cooperate with c-Src to cause c-Met phosphorylation. This could occur by promoting the clustering of c-Met molecules.

re-activated by c-Met at delayed time points (Figure 5e). antitumor effects, athymic nude mice bearing 201T flank These results indicate that EGFR utilizes lateral acti- tumors were treated with either gefitinib, PF2341066, vation of c-Met to maximize cell motility and invasion combination or vehicle for 5 days per week for 3 weeks. through non-classical, delayed signaling mechanisms. PF2341066 alone at a dose of 50 mg/kg had no significant effect on inhibiting tumor xenograft growth, whereas gefitinib significantly reduced tumor volume Combinational targeting of c-Met and EGFR have by 51%. Combining TKI therapies produced a 66% enhanced antitumor activity in a xenograft model of decrease in tumor volume; a significantly greater NSCLC effect compared with PF2341066 and gefitinib groups There is increasing evidence that c-Met compensatory individually (Figure 6a). EGFR and c-Met targets signaling acts as a resistance mechanism against EGFR were confirmed in treatment groups by immunohisto- TKIs. In particular, Engelman, et al. has established chemistry (Supplementary Figure S7). Ki-67 staining that c-Met signaling was a driver of acquired EGFR of tumors at 28 days revealed that the combination TKI resistance in EGFR mutant lung cancer cells group had a significantly greater effect compared (Engelman et al., 2007). NSCLC with wild-type with individual PF2341066 and gefitinib treatments EGFR are also inherently resistant to EGFR TKIs. (Figure 6b). Although minimal inhibition of xenograft The identification of delayed c-Met activation in tumor volume was observed with PF2341066 treatment NSCLC cells with wild-type EGFR, as in the case of a alone, a significant decrease in Ki-67 staining was majority of NSCLC patients, provides a rationale for measured (Figures 6a and b). This is likely due to combining therapies to improve response to EGFR stromal tissue comprising 40% of tumor volume in the TKIs. To address whether combinational targeting PF2341066-treated group. Stromal cells appear not to of EGFR and c-Met pathways leads to enhanced be actively proliferating, as shown in Figure 6b. Because

Oncogene Potentiation of EGFR action by c-Met AM Dulak et al 3631 201T A549 c-Met siRNA Invasion Motility

DMSO PF2341066 Vehicle 4h EGF 24h EGF 5’ HGF 4h EGF 24h EGF Vehicle 5’ HGF Vehicle 5’ EGF

siRNA: pMAPK Mock NT c-Met c-Met Total MAPK c-Src

β-Actin pAkt

Total Akt

β-Actin

Figure 5 EGFR relies on c-Met for maximal induction of EGFR phenotypes. 201T and A549 cells were serum deprived for 24 h before all 48-h phenotypic assays stimulated with EGF (50 ng/ml) or HGF (50 ng/ml). For experiments involving PF2341066 (1 mM), cells were pretreated with the c-Met inhibitor or dimethylsulphoxide for 2 h. For c-Met siRNA transfection experiments, 201T cells were plated and subjected to non-targeting siRNA or c-Met siRNA. Following transfection, cells were serum deprived in 24-well matrigel invasion chambers or 12-well plates for wound-healing assays. (a) Cells were plated in matrigel invasion chambers with growth factors added to the lower chamber only. Invading cells of three independent experiments were counted. (b) Cells were grown on 12-well plates before serum starvation and wounding. Addition of ligands followed c-Met inhibitor pretreatment or siRNA knockdown. Wounds were imaged at 0 and 48 h by  10 light microscopy. Migrating cells were measured by comparing 48 h wound size with initial wound size and expressed as percent wound closure. Mean of three independent samples per treatment group. (c) Invasion and wound-healing assays were repeated in 201T cells with c-Met or non-targeting siRNA knockdown, as described previously. Mean of three independent samples per treatment group. (d) 201T cells were mock transfected or treated with either non-targeting siRNA (NT) or c-Met siRNA for 8 h. After 48 h, cell lysates were prepared and analyzed for total c-Met levels, c-Src and b-actin. (e) 201T cells were serum deprived for 2 days before pretreatment with PF2341066 (1 mM), followed by stimulation with either EGF (10 nM) or HGF (10 ng/ml). In separate experiments, cell lysates were analyzed for MAPK and Akt phosphorylation and expression. ***Po0.0005; **Po0.005; *Po0.05 Student’s t-test.

remodeling of tumor architecture follows PF2341066 c-Met. Here, EGFR ligands induce prolonged c-Met treatment, volume measurements likely underestimate tyrosine phosphorylation after a pronounced delay that the true antitumor effect. is not attributable to an HGF-ligand autocrine produc- tion. Additionally, we have ruled out the secretion of other factors that might trigger c-Met phosphorylation Discussion because conditioned media experiments could not repli- cate the EGF-induced effect. Although increased c-Met We present findings characterizing a novel mode of transcription was observed following EGF treatment of delayed lateral communication from EGFR to c-Met NSCLC cells, along with a modest increase in total that is dependent on c-Src and mediates invasion and c-Met protein, this increase in c-Met protein levels cell movement. These findings provide an alternative appears not to be solely responsible for the delayed working model of c-Met activation by EGFR in NSCLC c-Met phosphorylation because it is independent of c-Src cell lines expressing wild-type, non-amplified EGFR and activity. On the basis of a recent report, the rise in

Oncogene Potentiation of EGFR action by c-Met AM Dulak et al 3632

Figure 6 Simultaneous targeting of EGFR and c-Met causes enhanced 201T xenograft tumor growth inhibition. (a) 201T cells (2  106) were injected into nude mice on day 0. On day 10, tumors were measured and experimental treatments were initiated. PF2341066 (50 mg/kg), gefitinib (150 mg/kg) and vehicle control (0.9% saline/1% Tween-80) were administered daily by oral gavage until day 28. (b) Representative Ki-67-stained sections of xenograft tumor imaged at  20 magnification. Quantitation of Ki-67 staining was performed by counting 5 fields at  40 magnification. n ¼ 5; ***Po0.0005; **Po0.005; *Po0.05; Student’s t-test with Welch’s correction. Everything is compared with the vehicle, except where indicated.

protein level may be attributed to induction of new A possible explanation for this is that EGFR utilizes c-Met transcripts through EGFR activation of the c-Met as a second-in-line molecule not to initiate hypoxia-inducible factor-1a pathway (Xu et al., 2010). response, but to amplify the complex phenotypic Moreover, studies in thyroid carcinoma and human programs following early activation by EGFR. We also gastric carcinoma cells suggested that increased c-Met anticipate that an alternative c-Met cascade is being levels alone might lead to receptor activation, but activated at delayed time points, as MAPK and Akt are conclude that there was a requirement of additional not phosphorylated at time points corresponding to factors to achieve phosphorylation (Ponzetto et al., 1991; c-Met phosphorylation. This mode of signaling might Bergstrom et al., 2000). Although it has been shown that occur through intracellular relocalization of a distinct c-Src can signal to hypoxia-inducible factor-1a, c-Src is population of c-Met molecules that do not associate minimally involved in our model because PP2 had with the Gab1 adapter protein, which is known to be almost no effect on c-Met protein and mRNA levels required for prolonged MAPK activation (Cai et al., following EGF stimulation (Karni et al., 2002). This 2002). Early, but not yet reproduced confocal micro- suggests that in the NSCLC cell studies, c-Src-dependent scopy studies, point to an intracellular compartmenta- c-Met phosphorylation is a distinct event from c-Met lization of c-Met molecules 24 h following EGF upregulation. We cannot, however, rule out that de novo addition. Sequestering of c-Met might stabilize c-Met c-Met molecules are also contributing to this novel and allow for an altered mode of signaling. Future pathway in some capacity. This is the first time c-Src has studies will aim to measure the levels of effectors been implicated as a critical intermediary linking EGFR responsible for the late-acting signaling from c-Met, activation to c-Met phosphorylation. such as E-cadherin, focal adhesion kinase and other Activation of invasion and motility are hallmarks potential candidates downstream of c-Met that regulate of EGFR signaling that drive lung tumor progres- invasive phenotypes (Kim et al., 2009). sion (Ritter & Arteaga, 2003). We showed that EGFR Our results suggest a model for lateral signaling from activation of invasive phenotypes rely largely on EGFR to c-Met that involves intracellular intermediates c-Met signaling through an HGF-independent pathway. and requires both gene transcription and extended c-Src

Oncogene Potentiation of EGFR action by c-Met AM Dulak et al 3633 activation that is upstream of c-Met. The fully phos- both receptors as well as trans-activation from EGFR phorylated c-Met molecules that accumulate at delayed to c-Met. time points do not activate common downstream In summary, we present the findings identifying a mediators, such as MAPK and Akt. The observed signaling requirement for c-Met to achieve maximal temporal association between c-Met and c-Src activa- EGFR-induced invasion and motility, which occurs tion suggests that they might cooperate and lead to through delayed events involving c-Src. Delayed c-Met prolonged activation of regulatory molecules responsi- signaling occurs through a non-canonical pathway. ble for invasion and cell motility downstream of both Targeting c-Met in addition to EGFR leads to an c-Met and c-Src (Figure 4d). We hypothesize that after enhanced antitumor effect in a wild-type, non-amplified EGFR stimulation, c-Src activates a signaling cascade EGFR and c-Met cell line. The findings presented in this that leads to transcription of an unknown intermediate study provide insight regarding the manner by which the accompanying the c-Met protein increase. This inter- EGFR–c-Src–c-Met axis regulates lung tumor progres- mediary might also be required for late c-Src activation. sion through varied temporal modulation. Future It does not appear that a complex between c-Met and studies will seek to clarify the exact role of c-Src, and c-Src forms at these late time points. Having ruled which other potential signaling intermediaries are active out secreted factors, a probable scenario for late c-Src in EGFR–c-Met lateral signaling, as well as to identify activation involves the production of molecules that what effector molecules downstream of delayed c-Met shift c-Src towards increased phosphorylation status, signaling regulate invasion and motility in NSCLC in such as increased association with Src homology-2 response to EGFR ligands. domain-containing proteins or increased phosphatase activity to remove inhibitory phosphorylation. It is possible that increased integrin production or clustering Materials and methods might account for both c-Met and c-Src events as both kinases demonstrate increased phosphorylation in re- Reagents and cell culture sponse to integrin interaction with the substratum NSCLC cell lines 201T were established in our laboratory (Wang et al., 1996; Danilkovitch-Miagkova et al., from primary tissue, and A549 and H1435 were obtained from 2000; Moro et al., 2002). The c-Met and c-Src events American Type Culture Collection (Manassas, VA, USA) and might be linear or parallel with one another and future maintained at 37 1Cin5%CO2 (Siegfried et al., 1999). All cells studies will aim to determine whether integrins have a were grown in basal medium eagle with 10% fetal bovine role in delayed c-Met phosphorylation by EGFR ligand serum and 2 mML-Glutamine. Rabbit polyclonal anti-pY1349- (Pawson and Scott, 1997). Met, anti-pY1365-Met, anti-Y1003-Met and anti-pY418-Src, as well as mouse monoclonal anti-c-Src were obtained from An alternate model for c-Met activation was con- Invitrogen (Carlsbad, CA, USA). Rabbit anti-c-Met (C-28), sidered, where c-Src might directly phosphorylate mouse anti-c-Src (B-12) and mouse anti-pY99 were purchased c-Met. Here, we showed that although delayed c-Src from Santa Cruz Biotechnology (Santa Cruz, CA, USA). activation is coupled with delayed c-Met phosphoryla- Mouse anti-c-Met, rabbit anti-EGFR, rabbit anti-Y1234/35- tion, no stabilized association between c-Src and c-Met Met, and anti-rabbit and anti-mouse-HRP-linked secondary was observed. These data conflict with such a model, antibodies were obtained from Cell Signaling (Beverly, MA, and suggest that c-Src is not directly phosphorylating USA). Anti-EGFR mouse antibody was purchased from BD c-Met, implying an additional mediator is needed for Transduction Laboratories (San Jose, CA, USA). Dasatinib increased c-Met phosphorylation. It is still possible that and gefitinib were from Chemitek (Indianapolis, IN, USA). c-Src might regulate more than one event in this Actinomycin D, SU11274 and PP2 were purchased from Calbiochem (San Diego, CA, USA). The c-Met and non- pathway due to prolonged c-Src phosphorylation. targeting siRNA ON-TARGETplus SMARTpools were Clinical responses to EGFR TKI in NSCLC patients purchased from Dharmacon (Lafayette, CO, USA). EGF, have produced variable results (Molina et al., 2006). TGF-a, HGF, HGF-neutralizing antibody and non-immune In East Asian populations, where there is a high rate of IgG control were purchased from R&D Systems (Minneapolis, EGFR mutation (B40%), EGFR-targeted therapies MN, USA). give a good response. In contrast, patients from western countries tend to have a low frequency of EGFR Cell treatments, immunoprecipitation, and immunoblotting mutations (3–8% of all lung tumors) and respond Cells were washed with phosphate-buffered saline and lysed poorly to EGFR TKIs (Wu et al., 2008). In studies with NP-40 buffer, as described previously (Stabile et al., reported here, we focused our efforts on EGFR–c-Met 2008). The insoluble fraction was cleared by centrifugation at communication in NSCLC cell lines expressing wild- 10 000 Â g for 15 min at 4 1C. Protein concentrations were type EGFR and c-Met. Through identification of a measured by Bradford assay (Pierce). For immunoprecipita- dual-regulation paradigm of tumorigenic signaling from tion, equal protein was subjected to preclearing and immuno- EGFR to c-Met, we further show that combining precipitation with antibodies overnight at 4 1C, then incubated EGFR and c-Met TKIs provided an enhanced inhibi- with Protein A-Agarose Beads (Thermo Fisher Scientific, Rockford, IL, USA) for 2 h at 4 1C. Immunoblotting was tion of tumor growth and a concurrent reduction in performed as described previously (Stabile et al., 2008). Ki-67 staining in 201T xenograft tumors. These results For re-probing, blots were stripped using IgG elution suggest a rationale for simultaneous targeting of EGFR buffer (Thermo Fisher Scientific). Blots were quantified by and c-Met to short-circuit signaling cascades responsible densitometry and ImageJ analysis (National Institutes of for invasive growth initiated and maintained through Health, Bethesda, MD, USA). For immunoprecipitations, all

Oncogene Potentiation of EGFR action by c-Met AM Dulak et al 3634 phosphoproteins were normalized to respective total immuno- In vivo tumor xenograft model precipitated protein, and whole cell lysate proteins utilized Female athymic nude-Foxn1nu mice were obtained from b-actin for quantitation. Harlan (Somerville, NJ, USA). 201T cells were suspended in serum-free phosphate-buffered saline supplemented with 50% Matrigel (BD Biosciences). Cells (2  106) were injected in the Human HGF enzyme-linked immunosorbent assay hind flank region of each mouse and allowed to develop. Ten Cell culture media was harvested and analyzed in duplicate by days after tumor implantation, the mice were divided into four Quantikine human HGF enzyme-linked immunosorbent assay treatment groups (10 mice per group) as follows: (a) placebo, (R&D Systems), according to the manufacturer’s instructions. (b) PF2341066, (c) gefitnib and (d) PF2341066 plus gefitnib. Conditioned media from fibroblasts was used as a positive Treatment with PF2341066 (50 mg/kg) or vehicle control control for HGF. (0.9% saline/1% Tween-80) was administered daily by oral gavage and gefitnib (150 mg/kg) was administered twice a week Quantitative real time–PCR by oral gavage for 3 weeks. Tumor size was measured weekly Total RNA was extracted using the RNAeasy kit (Qiagen, and reported as an average relative tumor volume calculated as Valencia, CA, USA) according to manufacturer’s directions. (length  width  height  p)/2(mm3). At the end of the treat- cDNA was synthesized by a single reverse transcription ment period, animals were sacrificed; tumors were removed reaction in a thermocycler. The primers and probes used for and fixed in 10% buffered formalin for immunohistochemical detection of HGF, c-Met and b-glucuronidase are provided in analysis. Animal care was in strict compliance with the insti- Supplementary data. tutional guidelines established by the University of Pittsburgh.

Wound-healing assay Statistical analysis Cells were plated on 12-well plates, grown to a confluent All values are expressed as the mean±s.e.m. and were monolayer, and serum-deprived for 24 h. Wounds were created obtained using Student’s t-test, except for the tumor xenograft and washed twice with 1  phosphate-buffered saline. All experiment, for which an unpaired t-test with Welch’s treatments were applied every 24 h over the experimental time correction was utilized. Significance tests were performed with course. Three wells per experimental treatment were examined a two-sided significance level of 0.05. at  10 magnification by light microscope. Migration distance was assessed by measuring the ability of cells to close the wound area. Abbreviations

Matrigel invasion assay EGF, epidermal growth factor; EGFR, epidermal growth factor Growth factor-reduced matrigel-coated transwell chambers receptor; HGF, hepatocyte growth factor; NSCLC, non-small (BD Biosciences) were activated in serum-free media at 37 1C cell lung cancer; SFK, Src family kinase; TKI, tyrosine kinase for 2 h. NSCLC cells (1  104) were serum deprived for 24 h inhibitor; TGF-a, transforming growth factor-a. and plated on transwell chambers in 1% charcoal-stripped fetal bovine serum containing media in the top chamber with indicated inhibitor treatments. In the bottom chamber, growth Conflict of interest factors and inhibitors were refreshed every 24 h. Non-invading cells in the top chamber were removed by cotton swab, and The authors declare no conflict of interest. invading cells were fixed and stained using Diff-Quik staining solutions, according to the manufacturer’s instructions (VWR International, Radnor, PA, USA). The number of invading Acknowledgements cells was counted at  10 magnification. This work was supported by NIH grant R01 CA79882 to JMS siRNA transfection from the National Cancer Institute and by a Predoctoral 201T cells were transfected with 50 pmol of either c-Met or Fellowship in Pharmacology by Pharmaceutical Research and non-targeting siRNA pools using Oligofectamine (Invitrogen) Manufacturers of America Foundation awarded to Austin M for 8 h. Transfection medium was replaced overnight followed Dulak. We would like to thank Pfizer for kindly providing by experimental treatments. PF2341066.

References

Bergstrom JD, Westermark B, Heldin NE. (2000). Epidermal growth phosphorylation requires tyrosine kinase activity of both the receptor factor receptor signaling activates met in human anaplastic thyroid and c-Src. J Biol Chem 275: 14783–14786. carcinoma cells. Exp Cell Res 259: 293–299. Daub H, Wallasch C, Lankenau A, Herrlich A, Ullrich A. (1997). Bottaro DP, Rubin JS, Faletto DL, Chan AM, Kmiecik TE, Vande Signal characteristics of G protein-transactivated EGF receptor. Woude GF et al. (1991). Identification of the hepatocyte growth Embo J 16: 7032–7044. factor receptor as the c-met proto-oncogene product. Science 251: Engelman JA, Zejnullahu K, Mitsudomi T, Song Y, Hyland C, 802–804. Park JO et al. (2007). MET amplification leads to gefitinib resistance Cai T, Nishida K, Hirano T, Khavari PA. (2002). Gab1 and SHP-2 in lung cancer by activating ERBB3 signaling. Science 316: 1039–1043. promote Ras/MAPK regulation of epidermal growth and differen- Fischer OM, Giordano S, Comoglio PM, Ullrich A. (2004). Reactive tiation. J Cell Biol 159: 103–112. oxygen species mediate Met receptor transactivation by G protein- Danilkovitch-Miagkova A, Angeloni D, Skeel A, Donley S, Lerman M, coupled receptors and the epidermal growth factor receptor in Leonard EJ. (2000). Integrin-mediated RON growth factor receptor human carcinoma cells. J Biol Chem 279: 28970–28978.

Oncogene Potentiation of EGFR action by c-Met AM Dulak et al 3635 Goi T, Shipitsin M, Lu Z, Foster DA, Klinz SG, Feig LA. (2000). An transformation by the hepatocyte growth factor/scatter factor EGF receptor/Ral-GTPase signaling cascade regulates c-Src activity receptor family. Cell 77: 261–271. and substrate specificity. Embo J 19: 623–630. Ponzetto C, Giordano S, Peverali F, Della Valle G, Abate ML, Vaula Guo A, Villen J, Kornhauser J, Lee KA, Stokes MP, Rikova K et al. G et al. (1991). c-met is amplified but not mutated in a cell line with (2008). Signaling networks assembled by oncogenic EGFR and an activated met tyrosine kinase. Oncogene 6: 553–559. c-Met. Proc Natl Acad Sci USA 105: 692–697. Popsueva A, Poteryaev D, Arighi E, Meng X, Angers-Loustau A, Jo M, Stolz DB, Esplen JE, Dorko K, Michalopoulos GK, Strom SC. Kaplan D et al. (2003). GDNF promotes tubulogenesis of (2000). Cross-talk between epidermal growth factor receptor and GFRalpha1-expressing MDCK cells by Src-mediated phosphory- c-Met signal pathways in transformed cells. J Biol Chem 275: lation of Met receptor tyrosine kinase. J Cell Biol 161: 119–129. 8806–8811. Presnell SC, Stolz DB, Mars WM, Jo M, Michalopoulos GK, Karni R, Dor Y, Keshet E, Meyuhas O, Levitzki A. (2002). Activated Strom SC. (1997). Modifications of the hepatocyte growth pp60c-Src leads to elevated hypoxia-inducible factor (HIF)-1alpha factor/c-met pathway by constitutive expression of transforming expression under normoxia. J Biol Chem 277: 42919–42925. growth factor-alpha in rat liver epithelial cells. Mol Carcinog 18: Kim LC, Song L, Haura EB. (2009). Src kinases as therapeutic targets 244–255. for cancer. Nat Rev Clin Oncol 6: 587–595. Ritter CA, Arteaga CL. (2003). The epidermal growth factor receptor- Klinghoffer RA, Sachsenmaier C, Cooper JA, Soriano P. (1999). Src tyrosine kinase: a promising therapeutic target in solid tumors. family kinases are required for integrin but not PDGFR signal Semin Oncol 30: 3–11. transduction. Embo J 18: 2459–2471. Shattuck DL, Miller JK, Carraway III KL, Sweeney C. (2008). Met Knowlden JM, Hutcheson IR, Barrow D, Gee JM, Nicholson RI. receptor contributes to trastuzumab resistance of Her2-overexpres- (2005). Insulin-like growth factor-I receptor signaling in tamoxifen- sing breast cancer cells. Cancer Res 68: 1471–1477. resistant breast cancer: a supporting role to the epidermal growth Siegfried JM, Gubish CT, Rothstein ME, Queiroz de Oliveira PE, factor receptor. Endocrinology 146: 4609–4618. Stabile LP. (2007). Signaling pathways involved in cyclooxygenase-2 Liu X, Carlisle DL, Swick MC, Gaither-Davis A, Grandis JR, induction by hepatocyte growth factor in non small-cell lung cancer. Siegfried JM. (2007). Gastrin-releasing peptide activates Akt Mol Pharmacol 72: 769–779. through the epidermal growth factor receptor pathway and Siegfried JM, Krishnamachary N, Gaither Davis A, Gubish C, Hunt abrogates the effect of gefitinib. Exp Cell Res 313: 1361–1372. JD, Shriver SP. (1999). Evidence for autocrine actions of Ma PC, Jagadeeswaran R, Jagadeesh S, Tretiakova MS, Nallasura V, neuromedin B and gastrin-releasing peptide in non-small cell lung Fox EA et al. (2005). Functional expression and mutations of c-Met cancer. Pulm Pharmacol Ther 12: 291–302. and its therapeutic inhibition with SU11274 and small interfering Stabile LP, Lyker JS, Huang L, Siegfried JM. (2004). Inhibition of RNA in non-small cell lung cancer. Cancer Res 65: 1479–1488. human non-small cell lung tumors by a c-Met antisense/U6 Ma PC, Maulik G, Christensen J, Salgia R. (2003). Met: structure, expression plasmid strategy. Gene Ther 11: 325–335. functions and potential for therapeutic inhibition. Cancer Meta- Stabile LP, Rothstein ME, Keohavong P, Jin J, Yin J, Land SR et al. stasis Rev 22: 309–325. (2008). Therapeutic targeting of human hepatocyte growth factor Molina JR, Adjei AA, Jett JR. (2006). Advances in chemotherapy of with a single neutralizing monoclonal antibody reduces lung non-small cell lung cancer. Chest 130: 1211–1219. tumorigenesis. Mol Cancer Ther 7: 1913–1922. Moro L, Dolce L, Cabodi S, Bergatto E, Boeri Erba E, Smeriglio M Stoker M, Gherardi E, Perryman M, Gray J. (1987). Scatter factor is a et al. (2002). Integrin-induced epidermal growth factor (EGF) receptor fibroblast-derived modulator of epithelial cell mobility. Nature 327: activation requires c-Src and p130Cas and leads to phosphorylation of 239–242. specific EGF receptor tyrosines. JBiolChem277: 9405–9414. Tang Z, Du R, Jiang S, Wu C, Barkauskas DS, Richey J et al. Nakajima M, Sawada H, Yamada Y, Watanabe A, Tatsumi M, (2008). Dual MET-EGFR combinatorial inhibition against Yamashita J et al. (1999). The prognostic significance of amplifica- T790M-EGFR-mediated erlotinib-resistant lung cancer. Br J tion and overexpression of c-met and c-erb B-2 in human gastric Cancer 99: 911–922. carcinomas. Cancer 85: 1894–1902. Thomas SM, Bhola NE, Zhang Q, Contrucci SC, Wentzel AL, Nakamura T, Teramoto H, Ichihara A. (1986). Purification and Freilino ML et al. (2006). Cross-talk between G protein-coupled characterization of a growth factor from rat platelets for mature receptor and epidermal growth factor receptor signaling pathways parenchymal hepatocytes in primary cultures. Proc Natl Acad Sci contributes to growth and invasion of head and neck squamous cell USA 83: 6489–6493. carcinoma. Cancer Res 66: 11831–11839. Naldini L, Vigna E, Ferracini R, Longati P, Gandino L, Prat M et al. Trusolino L, Comoglio PM. (2002). Scatter-factor and semaphorin (1991a). The tyrosine kinase encoded by the MET proto-oncogene is receptors: cell signalling for invasive growth. Nat Rev Cancer 2: activated by autophosphorylation. Mol Cell Biol 11: 1793–1803. 289–300. Naldini L, Vigna E, Narsimhan RP, Gaudino G, Zarnegar R, Wang R, Kobayashi R, Bishop JM. (1996). Cellular adherence elicits Michalopoulos GK et al. (1991b). Hepatocyte growth factor ligand-independent activation of the Met cell-surface receptor. Proc (HGF) stimulates the tyrosine kinase activity of the receptor Natl Acad Sci USA 93: 8425–8430. encoded by the proto-oncogene c-MET. Oncogene 6: 501–504. Wu CC, Hsu HY, Liu HP, Chang JW, Chen YT, Hsieh WY et al. Naldini L, Weidner KM, Vigna E, Gaudino G, Bardelli A, Ponzetto C (2008). Reversed mutation rates of KRAS and EGFR genes in et al. (1991c). Scatter factor and hepatocyte growth factor are adenocarcinoma of the lung in Taiwan and their implications. indistinguishable ligands for the MET receptor. Embo J 10: 2867–2878. Cancer 113: 3199–3208. Pawson T, Scott JD. (1997). Signaling through scaffold, anchoring, Xu L, Nilsson MB, Saintigny P, Cascone T, Herynk MH, Du Z et al. and adaptor proteins. Science 278: 2075–2080. (2010). Epidermal growth factor receptor regulates MET levels and Peschard P, Fournier TM, Lamorte L, Naujokas MA, Band H, invasiveness through hypoxia-inducible factor-1a in non-small cell Langdon WY et al. (2001). Mutation of the c-Cbl TKB domain lung cancer cells. Oncogene 29: 2616–2627. binding site on the Met receptor tyrosine kinase converts it into a Yano S, Wang W, Li Q, Matsumoto K, Sakurama H, Nakamura T transforming protein. Mol Cell 8: 995–1004. et al. (2008). Hepatocyte growth factor induces gefitinib resistance Ponzetto C, Bardelli A, Zhen Z, Maina F, dalla Zonca P, Giordano S of lung adenocarcinoma with epidermal growth factor receptor- et al. (1994). A multifunctional docking site mediates signaling and activating mutations. Cancer Res 68: 9479–9487.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene