Targeting Epithelial-to-Mesenchymal Transition with Met … · antibody; Ventana-Roche), p-Met Y1349 (130H2) rabbit mAb, p-Met Y1234/35 (D26) XP rabbit mAb (Cell Signal-ing Technology),

Cancer Therapy: Preclinical

Targeting Epithelial-to-Mesenchymal Transition with MetInhibitors Reverts Chemoresistance in Small Cell LungCancer

Israel Ca~nadas1,4, Federico Rojo1,7, �Alvaro Taus4, Oriol Arp��1,4, Montserrat Arum��-Ur��a5,6,Lara Pijuan5, Silvia Men�endez1,4, Sandra Zazo7, Manuel D�omine8, Marta Salido5, Sergi Mojal3,Antonio Garc��a de Herreros2,6, Ana Rovira1,4, Joan Albanell1,4,6, and Edurne Arriola1,4

AbstractPurpose: Met receptor phosphorylation is associated with poor prognosis in human small cell lung

cancer (SCLC). The aim of our work was to investigate the effects of hepatocyte growth factor (HGF)/

Met–mediated epithelial-to-mesenchymal transition (EMT) in SCLC and to evaluate the role of Met inhib-

ition in mesenchymal/chemorefractory SCLC models.

Experimental Design: SCLC models of HGF-induced EMT were evaluated in vitro and in vivo (subcu-

taneous xenografts in BALB/c nude mice) for chemosensitivity and response to Met inhibition with

PF-2341066 (crizotinib). Human SCLC samples at diagnosis (N¼ 87) and relapse (N¼ 5) were evaluated

by immunohistochemistry and immunofluorescence for EMT markers and Met status and these were

correlated with patient outcome.

Results: We identified that the activation of the Met receptor through HGF induced expression of

mesenchymal markers, an aggressive phenotype, and chemoresistance. Blockade of this process with the

Met inhibitor resensitized cells to chemotherapy in vitro and in vivo. Moreover, mesenchymal markers in

human SCLC specimens were associated with Met activation, predicted worse survival, and were upregu-

lated in chemorefractory disease.

Conclusion: These results provide novel evidence on an important role of Met-dependent EMT

in the adverse clinical behavior of SCLC and support clinical trials of Met inhibitors and chemotherapy

in this fatal disease. Clin Cancer Res; 20(4); 938–50. �2013 AACR.

IntroductionSmall cell lung cancer (SCLC) accounts for approxi-

mately 15% of lung cancers. Its tight association totobacco carcinogens’ exposure explains the high numberof genetic alterations described in this disease (1, 2). First-line treatment with platinum plus etoposide results in ahigh percentage of responses. Despite this, early chemor-efractory relapse causes short survival of patients withSCLC. So far, no trial evaluating targeted therapies inSCLC has shown benefit when compared with standard

chemotherapy. Our group has previously described anassociation between the Met pathway and outcome inpatients with SCLC (3), supporting a potential therapeu-tic interest of Met inhibitors in this disease.

Epithelial-to-mesenchymal transition (EMT) is a cellu-lar process characterized by loss of epithelial markersand acquisition of a mesenchymal phenotype, whichenables cancer cells to invade surrounding tissues andgenerate distant metastasis (4). Repression of E-cadherindefines the main hallmark of EMT. This process is initiat-ed by Snail1 transcriptional factor, a key inducer of thisconversion. Although Snail1 is induced at the early phasesof EMT and is necessary for this process, its expressionis not maintained in most mesenchymal cells; instead,E-cadherin silencing is dependent on other transcriptio-nal repressors induced by Snail1, such as Zeb1 and 2 (5).Mesenchymal features are associated with poor prognosisand chemoresistance in different tumor models (6–8).EMT can be induced by growth factors; among them,hepatocyte growth factor (HGF), the natural ligand of Mettyrosine kinase receptor, has been reported to be a potentinductor of EMT (9, 10).

PF-2341066 (crizotinib) is a small molecule that specif-ically inhibits Met and Alk kinases acting as an ATP

Authors' Affiliations: 1Cancer Research Program, 2Epithelial to Mesen-chymal Transition Laboratory, Cancer Research Program, IMIM; 3Consult-ing Service on Methodology for Biomedical Research; Departments of4Oncology and 5Pathology, Hospital del Mar; 6Universitat Pompeu Fabra,Barcelona; Departments of 7Pathology and 8Oncology, IIS-Fundaci�onJim�enez D��az, Madrid, Spain

Note: Supplementary data for this article are available at Clinical CancerResearch Online (http://clincancerres.aacrjournals.org/).

Corresponding Author: Edurne Arriola, Oncology Department, Hospitaldel Mar, Passeig Mar��tim, 25–29, 08003 Barcelona, Spain. Phone: 34-932-493-546; Fax: 34-932-483-366; E-mail: [email protected]

doi: 10.1158/1078-0432.CCR-13-1330

�2013 American Association for Cancer Research.

ClinicalCancer

Research

Clin Cancer Res; 20(4) February 15, 2014938

Research. on September 30, 2020. © 2014 American Association for Cancerclincancerres.aacrjournals.org Downloaded from

Published OnlineFirst November 27, 2013; DOI: 10.1158/1078-0432.CCR-13-1330

http://clincancerres.aacrjournals.org/

competitor. Crizotinib is approved for the treatment ofpatients with Alk-positive non–small cell lung cancer(NSCLC; ref. 11) and has also shown activity in tumorswith Met amplification (12).Selection of the correct patient population for molecular

treatment has become increasingly relevant for the impactof these targeted drugs in patients’ outcomes (13). This isparticularly challenging for biomarkers that may not be thedrivers of tumor progression, but play an important role intumor aggressiveness or resistance to therapies, such as Met(14). Experience with Met inhibitors in the treatment ofNSCLC illustrates the difficulty of selection of patientsaccording to biomarker status (i.e., Met expression levels)for the success of Met-targeting therapies (15).We hypothesized that Met activation by HGF may

cause EMT and, as a consequence, induce chemoresis-tance in SCLC. On the basis of this hypothesis, theobjectives of our work were (i) to explore in vitro andin vivo and in clinical specimens, the Met-induced EMTphenomena in SCLC; (ii) to test in vitro and in vivo, theeffects of Met inhibition to revert EMT; and (iii) to searchfor predictive biomarkers that discriminate patients withSCLC who would benefit from anti-Met therapies.

Materials and MethodsHGF-induced EMTWe seeded 6.7� 105 cells in 60-mm2 dishes with culture

medium containing 10% FBS. After 24 hours, HGF wasadded at 40 ng/mL. Fresh culture medium containing HGFwas added every 48 hours during 14 days. A morphologicstudy by lightmicroscopy and amolecular study byWesternblotting were performed at days 3, 6, and 10 to study ifcontinuous exposure to HGF was able to induce EMT inH69 cells.

Subcutaneous tumorigenesisFive-week-old male BALB/c nude mice were purchased

from Charles River Laboratories and hosted in the patho-

gen-free animal facility at the Barcelona Biomedical Re-search Park (PRBB). Animal treatments were done accord-ing to institution-approved protocols. Cells were resus-pended in sterile PBS with 50% Matrigel (BD Biosciences)and subcutaneously injected into the flank of mice. Tumorvolume was determined from caliper measurements oftumor length (L) and width (W) according to the formulaL�W2/2 three times a week. Tumors were allowed to growuntil the volume reached approximately 200 to 300 mm3.Mice were randomized to four groups with 10 mice in eachgroup. Treatment groups consisted of control (vehicle),etoposide (12mg/kg), PF-2341066 (100mg/kg), and com-bination. Etoposide was diluted in 200 mL 0.9% NaCland administered at a dose of 12 mg/kg in one dailyintraperitoneal injection on days 1 to 3 of treatment. PF-2341066 was diluted in water and administered daily byoral gavage at 100mg/kg for 14 days. This dose was selectedon the basis of previous literature (16), suggesting that atthis dose, Met is inhibited >90%, which potentially trans-lates in tumor responses in a Met-dependent model. Con-trol animals received 0.9% NaCl and water as vehicles.Mice were euthanized and tumors measured with calipers.Part of the tumors were fixed in 4% formalin and embedd-ed in paraffin for immunohistochemical studies.

Human samples and immunohistochemistryThe following antibodies were used for immunohisto-

chemical studies: Met (SP44) mouse mAb (monoclonalantibody; Ventana-Roche), p-Met Y1349 (130H2) rabbitmAb, p-Met Y1234/35 (D26) XP rabbit mAb (Cell Signal-ing Technology), E-cadherin (NCH-38) mouse mAb(Dako), NCAM (123C3)mousemAb (Dako), Snail1 (EC3)mousemAb, vimentin (V9)mousemAb (Dako), andCD31(SP38) rabbit mAb (Spring Bioscience).

The study population consisted of 87 patients diagnosedwith SCLC at any stage from whom we had clinical andfollow-up information. Specimens were retrospectivelyretrieved from Parc de Salut Mar Biobank (MARBiobanc,Barcelona, Spain) and Fundacion Jimenez Diaz Biobank(Madrid, Spain). This study was approved by the institu-tional review board of each participating center. Three-mmtissue sections from formalin-fixed and paraffin-embeddedsamples were obtained, mounted onto charged slides, andthen deparaffinized in xylene and hydrated.

After heat antigen retrieval was performed at a high-pH solution using PT Link platform (Dako), slidesfor immunohistochemistry were incubated with primaryantibody for 1 hour at a dilution of 1:1 for Met mAb,1:20 for p-Met Y1349, and 1:50 for p-Met Y1234/35,1:5 for Snail1, 1:100 for E-cadherin, 1:100 for NCAM(neuronal adhesion molecule), 1:100 for vimentin,and 1:200 for CD31. Then, sections were incubated withthe specific polymer EnVision FLEXþ (Dako) and re-vealed with 3,30-diaminobenzidine (DAB) as chromogen.In situ hybridization was done by deproteinization usingproteinase K digestion for 5 minutes at room tempera-ture and incubation with a set of specific digoxigenin-labeled probes for mRNA SPARC (agaaactgtggcagaggtga;

Translational RelevanceSmall cell lung cancer (SCLC) is a highly lethal di-

sease due to its chemorefractory nature after initialtreatment. No novel therapeutic strategy has been ableto improve outcome to date. Our work demonstratesthe existence of a group of patients with SCLC showingactivated Met and mesenchymal features that presentworse prognosis. Moreover, we observe upregulationof these features in relapsed disease. Our preclinicaldata show that Met mediates epithelial-to-mesenchy-mal transition, leading to chemoresistance and thatMet inhibition reverts chemoresistance. These findingsstrongly support the design of clinical trials with Metinhibitors plus chemotherapy in patients with SCLCwith Met activation and/or mesenchymal phenotype.

HGF/Met–Induced EMT in SCLC

www.aacrjournals.org Clin Cancer Res; 20(4) February 15, 2014 939




aagctccacctggactac at; ctgtgacctggacaatgaca; Cenbimo) at62�C for 1 hour. Reaction was revealed using an anti-digoxigenin rabbit mAb (9H27L19; Life Technologies),anti-rabbit EnVision FLEXþ, and DAB. Slides were coun-terstained with hematoxylin. Sections from same speci-mens above were incubated with normal mouse IgG2(immunoglobulin G; X0943; Dako) or normal rabbit Igfraction (X0903; Dako) instead of primary antibodies asnegative controls. Stainings were evaluated by twopathologists independently blinded to clinical informa-tion, on a light microscope (Olympus DX50; OlympusCorp.).

Met, p-Met, E-cadherin, and NCAM were scored whenany percentage of tumor cells was stained in themembrane.Snail1was evaluated in thenucleus of tumor cells. Vimentinand SPARCwere quantifiedwhen detected in the cytoplasmof tumor cells. A semiquantitative histoscore (H-score)was calculated, determined by estimation of the percent-age of tumor cells positively stained with low, medium, orhigh staining intensity for each marker. The final scorewas determined after applying a weighting factor to eachestimate. The formula used was: H-score ¼ (low%) þ 2 �(medium%) þ 3 � (high%), and the results ranged from0 to 300. The tumors in the present study were classifiedas p-MET–negative when the H-score was 0, versus p-Met–positive for any positive H-score. In our prior work, wedefined as negative, any case with H-score of 5 or less, butbecause only one tumor had this criteria in this series andthere was a continuous distribution of positive H-scores,we chose this new cutoff as per the statistical criteria.Measurement of microvascular density was determined asthe mean number of vessels by quantification of the CD31-expressing vascular structures using the ImageJ (NIH) soft-ware to recognize contiguous stained tubular vessels, asdescribed previously (17).Double immunofluorescence forcoexpression analysis was performed on tissue sectionsusing sameprimary antibodies and conditions as described.Snail1, p-Met, and vimentin were detected using appropri-ate Alexa Fluor 568 and 488 -conjugated goat anti-rabbitIgG and anti-mouse IgG antibodies (Life Technologies;diluted 1:700). Sections were counterstained with 40,6-diamidino-2-phenylindole dihydrochloride (DAPI; AbbottMolecular) to visualize cell nuclei. All incubations wereperformed at room temperature. Fluorescence assays wereperformed using a Dako Autostainer. Staining was evalu-ated by two investigators (F. Rojo and S. Zazo) using a CriNuance FX Multispectral Imaging System (PerkinElmer).

ResultsHGF-induced EMT in SCLC cells is reverted andprevented by PF-2341066

We exposed H69 SCLC cells to HGF (40 ng/mL) for10 days. At day 6, HGF-treated cells grew as single cellspartially attached to the plate (Fig. 1A, left). Because HGF-induced EMT is a dynamic process, we tested for EMTmarkers at various time points. Western blot analysisrevealed that HGF promoted a partial downregulation of

Met protein, stimulation of phosphorylated Met andERK1/2, increase in Snail1 at day 6, and downregulationof E-cadherin at day 10 (Fig. 1A, right, and SupplementaryFig. S1A), but not in NCAM (data not shown).

We assessed whether PF-2341066 prevented HGF-induced effects. PF-2341066 was added to fresh mediumupfront and every 48 hours along with HGF. As observedin Fig. 1A, the inhibitor prevented the disruption of cellclusters induced by HGF, as well as Met protein down-modulation, ERK1/2 activation, Snail1 increase, or E-cad-herin downregulation. We performed the experimentadding PF-2341066 at day 6, once the EMT phenotypewas detectable. The Met inhibitor reversed the HGF-induced morphologic (Fig. 1B, left) and molecular changes(Fig. 1B, right).

To further demonstrate that the observed effect was aresult of Met inhibition, we performed the same experi-ments with PHA-665752, which inhibits Met but not Alk,obtaining similar results (Supplementary Fig. S1B–S1D;ref. 18). Furthermore, we screened the cell lines used inthis study for ALK translocation by FISH and all werenegative (data not shown).

These data confirm that HGF, via Met activation, in-duced morphologic and molecular changes in H69 con-sistent with EMT, which are reversed or prevented by Metinhibition.

We stably knocked down the endogenous expression ofSnail1 in H69 cells (shSnailH69). We observed an almostcomplete downregulation of the protein compared withcells transfected with the control short hairpin RNA(shRNA; shctH69; Fig. 1C, left). We then exposed shctH69and shSnailH69 to HGF. We observed scattering of theshctH69 cells that grew partially attached to the plate(Fig. 1C, right) on day 6. In contrast, in shSnailH69 cells,no morphologic changes were observed under HGF expo-sure for 10 days. Detailed kinetic analysis showed thatHGF induces Snail1 expression at day 3 in shctH69 butdecayed at longer time points (day 6; Fig. 1C, left).

Amesenchymal cell type is generated in vitro fromSCLCcells treated with HGF

Upon prolonged treatment with HGF, we observed asmall pool of H69 cells with a mesenchymal (spindle-shaped) morphology that grew attached to the plates(Fig. 2A). This phenomenon was reproduced in six inde-pendent experiments. We isolated this subpopulation(H69-mesenchymal, H69M) and compared it with theparental H69 cells. H69M cells were phenotypically sta-ble, maintaining the same mesenchymal morphology inthe absence of HGF stimulation. A study of the karyotypeshowed the same chromosomal characteristics thanH69 cells ruling out a potential contamination by anothercell line (Supplementary Fig. S2A). As shown in Fig. 2B,left, H69M cells showed Met basal activation shown byphosphorylation of Tyr1349, and mesenchymal featurescharacterized by the upregulation of fibronectin anddownregulation of NCAM and E-cadherin. We then trea-ted the H69M cells with HGF for 15 minutes to assess if

Ca~nadas et al.

Clin Cancer Res; 20(4) February 15, 2014 Clinical Cancer Research940




they were still sensitive to HGF stimulation. We observeda higher increase in Met phosphorylation when comparedwith the increase observed in H69, and in the down-stream molecules GAB1 (pTyr 307) and ERK1/2, indicat-ing that these cells were more sensitive to exogenous HGFstimulation than parental H69.One interesting observation is that differences in the

intensity of baseline phosphorylation of Tyr1234/35 andTyr1349 and in response to HGF were found in theseexperiments. This could be due to the performance ofantibodies under our experimental conditions and/or tothe different role of each phosphorylation site in responseto HGF. Upon HGF binding, Met receptor is activated

through dimerization and phosphorylation of Tyr1234/35 on the catalytic domain. Subsequently, phosphorylationof docking site Tyr1349 occurs with recruitment of down-stream molecules as previously reported (19). Despite this,both were consistently upregulated upon HGF stimulationand expression decreased by Met inhibitors.

A global transcriptome analysis of H69 and H69Mshowed clear differences in mRNA expression. We found9,712 genes that were differentially expressed (GEO acces-sion number: GSE45120; Supplementary Fig. S2B). Accord-ing to Ingenuity Pathway Analysis, the top altered functionsin the comparison were "Cell movement," "Cell morphol-ogy," "Cellular growth and proliferation," and "Cell

A

B

Met

p-Met (Y1234/1235)

Tubulin

E-cadherin

Snail1

ERK1/2

p-ERK1/2

C HGF HGF+PF

Met

p-Met Tyr1234/1235

Tubulin

E-cadherin

Snail1

p-ERK1/2

+ PF (200 nmol/L)−+ −−−+ + HGF (40 ng/mL)+ + −−

Day 10Day 6

CTRL

HGF

6 days

HGF

+

PF 48 h

CTRL HGF 6 days

H69

shCT

H69

shSnail

C

Day 6 Day 10

CTRL

HGF

HGF

+

PF

p-Met Tyr1349

+ HGF+ −+ −+ −−shCT shSnail1 shCT shSnail1

3 days 6 days

Snail1

Tubulin

Figure 1. HGF stimulation inducesEMT in the H69 cell line and PF-2341066 prevents and reverts thisprocess. A, cell morphology bylight microscopy of H69 treatedwith HGF (40 ng/mL) or thecombination of HGF and PF-2341066 (200 nmol/L) during 14days. Photomicrographs at days 6and 10. Scale bars, 100 mm (left).Western blot analysis of H69 cellstreated with HGF (40 ng/mL) orHGF and PF-2341066 (200 nmol/L)during 6 and 10 days for p-Met(Tyr1234/1235, Tyr1349), Met, p-ERK1/2, Snail1, E-cadherin, andtubulin (right). B, cell morphologyby light microscopy of H69 treatedwith HGF (40 ng/mL) during 6 daysor with HGF and PF-2341066during 48 hours after 6 days of HGFstimulation. Photomicrographs atdays 6 and 8. Scale bars, 100 mm(left). Western blot analysis of H69cells treated for 8 days with HGF(40 ng/mL) or the combination ofHGF and PF-2341066 (200 nmol/L)for p-Met (Tyr1234/1235), Met, p-ERK1/2, Snail1, E-cadherin, andtubulin (right). C, Western blotanalysis showing stableknockdown of Snail1 in H69 cellstransduced with shRNA targetingSnail1 versus H69 cells transducedwith scrambled shRNA after 3 and6daysofHGFstimulation (left). Cellmorphology by light microscopy ofshCTH69 and shSnailH69 cellstreatedwith HGF (40 ng/mL) during10 days. Photomicrographs at day6. Scale bars, 100 mm (right).






− + − + − + − + HGF 40 ng/mL 15’

p-Met Tyr1349

p-Gab1 Tyr307

p-ERK1/2

Tubulin

p-Met Tyr1234/35

− − + + − − + + PF 200 nmol/L 3 hH69 H69M

E-cadherin

Fibronectin

NCAM

Snail1

A H69M

HGF 40 ng/mL 14 d

H69 H69adh

Isolation

Ce

llin

va

sio

n(%

ofco

ntr

ol)

H69 H69M0

50

100

150 CTRL

PF 100 nmol/L

*

*

*

* P < 0.01

B

Rela

tive

exp

ress

ion

SPARC FIBR02468

10

50,000

100,000

150,000C

Rela

tive

exp

r ess

ion

Vimentin HGF0

2

4

6

8

10100

150

200

250H69

H69M

HG

F (

pg/m

L)

H69 H69M0

50

100

150

200

250H69

H69M

Cell

gro

wth

(%

of contr

ol)

H69 H69M0

50

100

150 CTRL

PF 200 nmol/L* *

P < 0.05

Clo

nogenic

gro

wth

(% o

f contr

ol)

0

50

100

150 CTRL

PF 50 nmol/L

PF 200 nmol/L

P < 0.01

MetHGFGRB2SHCSOS1PXNSRCEGFRERBB2AREGEGFTGFAEREGCD44BMP1FASVEGFASPARCCAV1FN1ITGA5VIMZEB1ZEB2CDH2CDH11TGFB1ATMCHEK2CDKN1BCDKN1ABCL2L11CCND1MYC

D

HGF/Met pathway

EMT molecules

DNA repair/cell survival

H69_2 H69_1 H69_3 H69M_3 H69M_1 H69M_2

Color key

Row Z-score–1.5 0.5

Figure 2. HGF-induced EMT results in the appearance of amesenchymal subpopulation. A, cell morphology by lightmicroscopy of H69 cells treatedwith HGFduring 14 days showing the appearance of an adherent cell subpopulation. Scale bars, 100 mm. B, Western blot analysis of H69 and H69M cells treated withPF-2341066 (200 nmol/L) for 3 hours and stimulatedwithHGF (40 ng/mL) for 15minutes (left) and heatmapof altered genes inH69Mcells comparedwithH69(right). Horizontal rows, individual genes; vertical columns, individual samples. Color intensity means degree of gene expression modulation. Blue,upregulation; yellow, downregulation. Black, no change. C, qRT-PCR of selected genes. Expression values were normalized to the levels of a control RNA(RPLP0). Relative mRNA levels were normalized to the expression of the control sample (left); HGF production in H69 and H69M culture supernatantsmeasured by ELISA (right). D, cell growth of H69 andH69M cells treatedwith PF-2341066 (200 nmol/L) during 10 days. Cell number was assessed by ScepterAutomatedCell Counter. The number of viable cells in each treatmentwas plotted aspercentage of control (left); invasion assayofH69 andH69Mcells treatedwith PF-2341066 (100 nmol/L) during 24 hours. Each data point represents the mean � SD percentage cell invasion of three independent experiments,comparedwith controls at 100% (middle); clonogenic assay of H69Mcells culturedwith or without PF-2341066 at 50 or 200 nmol/L during 14 days. Each datapoint represents the mean � SD percentage clonogenic growth of three independent experiments, compared with controls at 100% (right).

Ca~nadas et al.





Signaling." Within the "top-ten" genes that were overex-pressed in H69M we found important players in EMT suchas SPARC, caveolin, integrin a-5 (fibronectin receptor),fibronectin 1, and vimentin (20, 21). HGF/MET pathwaygenes such as HGF, GRB2, or paxillin were also overex-pressed in H69M. Figure 2B, right, shows differentiallyexpressed genes involved in EMT, DNA repair and survival,and those associated with the MET pathway. The upregula-tion in SPARC, fibronectin, vimentin, and HGF was con-firmed by quantitative real-time PCR (qRT-PCR; Fig. 2C,left). HGF overexpression levels in H69M cells were furtherconfirmed at the protein level by ELISA for secreted HGFin the culturemedium(Fig. 2C, right). These results confirmthe mesenchymal nature of H69M and suggest that upre-gulation of EMTmarkers in these cells could be caused by anautocrine loop responsible for Met basal activation. Metexpression in each H69M replicate was, however, variable.This observation may be related to small differences in Metexpression in the original cell cultures, which possiblyinfluence RNA levels. Having showed a distinct EMT phe-notype between these two cell lines and the variable Metexpression, we then addressed the role of MET in EMTphenotype in further experiments.To elucidate if the molecular changes observed in H69M

affected cell behavior, we compared growth kinetics andinvasion capabilities between H69 and H69M. H69Mshowed a slightly but nonsignificant increased proliferationrate, and higher invasiveness (Fig. 2D, left and middle)when compared with H69. Similarly to H69, H69M wassensitive to the addition of PF-2341066 that blocked Metphosphorylation at basal and HGF-stimulated conditions(Fig. 2B, left). This inhibition translated in a significantdecrease in proliferation, invasion, and clonogenic growth.The reduction on clonogenic growth was statistically signi-ficant in a dose-dependent manner, achieving a magnitudeof almost 50% (Fig. 2D). Interestingly, PF-2341066 didnot reverse the spindle-like morphology of H69M at lightmicroscopy in the clonogenic assay (data not shown).

HGF-mediated EMT enhances tumorigenesis in SCLCcellsH69 and H69M cells were injected in the flanks of

immunodeficient mice (n ¼ 10) to establish tumor xeno-grafts. We observed that tumors generated by H69M werelarger (median volume, 1.5� 103mm3) comparedwith thetumors generated by H69 (median volume, 0.57 � 103

mm3; Fig. 3A). Moreover, H69Mweremore invasive locally(surrounding soft tissue infiltration), showing foci of infil-trating cells in 9 of 10 mice compared with 1 of 10 mice forH69 cells (Fig. 3B). To study whether the results occurredin other subpopulations, we performed an in vivo experi-ment with a different mesenchymal subpopulation deriv-ed from H69 obtained by prolonged exposure to HGF(H69M2). This cell line presented a mesenchymal pheno-type by Western blotting and secreted higher levels ofHGF compared with H69. We confirmed the increasedtumorigenic capacity of this mesenchymal subpopulation(Supplementary Fig. S3).

We then analyzed H69M xenografts compared withthose generated by H69 cells. Histologic characteristics ofboth tumor xenografts were similar, suggesting that H69Mwere able to recapitulate the original round-small cellmorphology of the parental H69 (Fig. 3C, hematoxylin andeosin).

Immunohistochemical studies revealed that H69M cellspresented increased Met phosphorylation, expression ofSnail1, vimentin, and SPARC and decreased expression ofE-cadherin and NCAM. We also found increased vascular-ization of H69M-derived tumors (Fig. 3C). Multiplexingassays with Snail1 and p-Met antibodies demonstrated thatSnail1 was expressed in cells with upregulated Met activity(Fig. 3D).

We repeated these assays using a lower number of cells,100 (N ¼ 5) and 500 (N ¼ 5). For 100 cells, we observedtumor formation in 5 of 5 mice with H69M injection andin 3 of 5 mice for H69 (median tumor volume, 165 vs. 79mm3, respectively). When injecting 500 cells, 5 of 5 micewith H69M cells formed tumors, whereas 3 of 5 mice did itfor H69 cells (median tumor volume, 384 vs. 82 mm3,respectively).

In addition, we performed an in vivo assay, injecting 106

H69M cells in the flank of mice and treating with controlvehicle and PF-2341066 100 mg/kg from day 0 (N ¼ 5 percondition). The Met inhibitor significantly decreased tu-mor growth (final median volume, 1,047 mm3 for untreat-ed vs. 550 mm3 for treated mice; P < 0.05; SupplementaryFig. S4A). We also found a significant decrease in localinvasion in the treated mice. Immunohistochemical ana-lysis of the xenograft confirmed that treatment withPF-2341066 significantly decreased p-Met, mesenchymalmarker expression, and angiogenesis and increased epithe-lial marker expression (Supplementary Fig. S4B).

These results suggest that Met activation by HGF inducesa mesenchymal transformation that translates in a moretumorigenic and aggressive behavior that can be blockedwith a Met inhibitor.

HGF-mediated EMT in SCLC induces chemoresistancein vitro and in vivo and MET inhibition sensitizes cellsto chemotherapy

Because EMT has been associated with apoptosis resis-tance in other cellular systems (22), we investigated if themesenchymal phenotype caused increased chemoresis-tance in our model. We treated H69 andH69Mwith etopo-side, a standard chemotherapeutic agent used in SCLC.MTS assays demonstrated that H69M were chemoresistant,compared with H69 (Supplementary Fig. S5A).

To elucidate if Met inhibition could revert chemoresis-tance, we performed a clonogenic assay with H69M cellswith etoposide, PF-2341066, and the combination ofboth. We observed the highest inhibition rate with thecombination of etoposide and PF-2341066, suggestingthat Met inhibition by PF-2341066 conferred chemosen-sitivity (Supplementary Fig. S5B, left).

We confirmed this hypothesis in mouse heterotopictumors. We xenografted H69M cells and treated mice with






etoposide, PF-2341066, or the combination. We did notobserve a significant effect on tumor growth by any of thetwo drugs alone; however, the combination produced asignificant decrease in tumor growth compared with eitheragent alone (P < 0.05, Fig. 4A).

Of note, inhibition of p-Met by PF-2341066 of mesen-chymal cells did not affect the expression of EMT markersin short term assays performed following 3 hours of drugexposure (Western blotting; Fig. 2B and SupplementaryFig. S3A). However, longer time exposure experimentsin vitro (i.e., cell growth 8 days, invasive capacity 24 hours,and clonogenic growth lasting 14 days) demonstrated thecellular effects of Met inhibition (Fig. 2D). These effectsand their correlation with molecular changes were morerobustly demonstrated in vivo (>20-day experiments), sug-

gesting that HGF-induced EMT is a dynamic process andinactivation of p-Met does not elicit phenotypic EMTchanges until later time points (Fig. 4C).

We then evaluated effects of the different treatmentstrategies in H69-derived xenografts. Tumors originatedfrom H69 were sensitive to etoposide; addition of PF didnot significantly modify the response (Fig. 4B).

To further confirm the capacity of PF-2341066 of con-ferring chemosensitivity, we used the same experimentalapproach with another SCLC cell line, H841. This cellline was sensitive to HGF stimulation showing increasedexpression of p-MET (Supplementary Fig. S5C, left), and amesenchymal phenotype. It secreted higher levels of HGFwhen compared with H69 (Supplementary Fig. S5C, top,right). Sequencing and FISH studies revealed no mutations

Snail1 + p-Met + DAPI p-Met DAPI

H69 H69M

H&

E

CD

31

Snail1

p-M

et

SP

AR

CV

imentin

NC

AM

H69 H69M

0

0

0

0

0

H69 H69M

P = 0.001

P = 0.003

P = 0.001

H-s

core

in

tum

or

ce

llsH

-score

in

tum

or

ce

lls

P = 0.001

E-c

ad

he

rin

H-s

core

in

tum

or

ce

lls

H69 H69M

H69 H69M

H69 H69M

CD

31-s

tru

ctu

res

H69M H69

H69 H69M

H69 H69MNu

mb

er

of

inva

siv

e fo

ci

in c

om

ple

te s

ectio

n P < 0.001

H69 H69MDays

Tum

or

volu

me (

mm

3)

0 5 10 15 20 25 300

500

1,000

1,500

2,000

2,500H69H69M

P < 0.05

BA

C

D

P = 0.004

Snail1

H69M H69

P = 0.001

H69 H69M

H-s

core

in

tum

or

ce

llsH

-score

in

tum

or

ce

llsH

-score

in

tum

or

ce

lls

H69 H69M

H69 H69M

H69 H69M

H69 H69M

P = 0.012

200

150

100

50

0

250200150100500

200

150

100

50

0

1086420

504030

20100

200

150

100

50

0

706050403020100

10

8

6

4

2

0

Figure 3. H69M cells are more tumorigenic and maintain mesenchymal features in vivo. A, photograph of a representative mouse with H69 and H69Mcells subcutaneously inoculated at right and left flanks, respectively. Photograph of the excised tumors from the same animal (left) and tumorvolume measurement after 25 days of injection (right). Each data point represents mean � SD tumor volumes. B, hematoxylin and eosin (H&E) stainedsections from heterotopic H69M xenografts revealed infiltration of adjacent soft tissues by tumor cells compared with H69 tumors. Dashed blacklines indicate the edge of the tumor in H69 and invasion of muscle and adipose tissues in H69M. Scale bar, 100 mm. Quantification of the number ofinvasive foci in complete H&E-stained section for each group (mean � SEM) and significance are displayed. C, staining for H&E, p-Met, E-cadherin,NCAM, Snail, vimentin, SPARC, and CD31 in H69 and H69M tumors. Scale bars, 75 mm. Quantification of expression for each marker betweenH69 and H69M tumors (mean H-score � SEM) and significance are displayed. D, representative image of immunofluorescence staining in H69Mtumor for Snail (red), p-Met (green), and DAPI (blue). Snail is detected in the nuclei (pink, red þ blue) and p-Met in the membrane and cytoplasm (green).Scale bars, 15 mm.

Ca~nadas et al.





Control PF Etoposide Combination

Sn

ail1

SP

AR

CV

ime

ntin

NC

AM

CD

31

CD

31- s

tru

ctu

res

p-M

et

E-c

ad

he

rin

H-s

core

in

tum

or

ce

lls

A B

Days Posttreatment0 5 10 15 20 25 30 35

0

500

1,000

1,500

2,000 CTRLEtoposide12 mg/kgPF100 mg/kgCombination

P < 0.05

Days Posttreatment

Tu

mo

rvo

lum

e (

mm

3)

Tu

mo

rvo

lum

e (

mm

3)

7 10 13 15 17 20 22 230

500

1,000

1,500

2,000

2,500 CTRLEtoposide 12 mg/kgPF 100 mg/kgCombination

P < 0.05

C

0

5

0

5

0

5

0

5

0

1 2 3 4

1 2 3 4

1 2 3 4

1 2 3 4

1 2 3 4

1 2 3 4

Control PF Etoposido PF+Etoposido

CTRL PF Et Comb

CTRL PF Et Comb

CTRL PF Et Comb

CTRL PF Et Comb

CTRL PF Et Comb

CTRL PF Et Comb

CTRL PF Et Comb

200175150125100

755025

0

150

125

100

75

50

25

0

H-s

core

in

tum

or

ce

llsH

-score

in

tum

or

ce

llsH

-score

in

tum

or

ce

llsH

-score

in

tum

or

ce

llsH

-score

in

tum

or

ce

lls

100

75

50

25

0

25

20

15

10

5

0

40

30

20

10

0

150

125

100

75

50

25

0

50

40

30

20

10

0

CTRL Et PF Comb CTRL Et PF Comb

H69M H69

CTRL Et PF Comb

**

**

**

**

**

**

**

*

*

Figure 4. PF-2341066 increases sensitivity to etoposide (Et) in H69M-derived tumors. A and B, photograph of representative mice of each group oftreatment with H69M (A) or H69 cells (B) subcutaneously inoculated at right flanks, and photograph of the excised tumors from the sameanimal. Graphs show changes in tumor volume of H69M (A) and H69 (B) xenograft treated with etoposide alone, PF-2341066 alone, or thecombination of both treatments. Dose schedules were etoposide (12 mg/kg/i.p. on days 1 to 3 of treatment), PF-2341066 (100 mg/kg/oralgavage daily), and the combination of both. Each data point represents the mean � SD tumor volume of each group of mice comparedwith control. C, p-Met, E-cadherin, NCAM, Snail, vimentin, SPARC, and CD31 staining in H69M tumors in control, PF 100 mg/kg, etoposide12 mg/kg, and combination conditions of treatment. Met inhibition by PF alone or in combination with etoposide induces epithelialphenotype in tumor cells. Scale bars, 75 mm. Quantification of expression for each marker in the different treatment conditions (mean H-score �SEM) and significance are displayed. �, P < 0.05 (PF arm compared with the combination). ��, P < 0.05 (etoposide arm compared with thecombination).






or amplification in MET (data not shown). In clonogenicassays, H841 and H69M cells presented a similar sensitiv-ity to PF-2341066. It was also sensitive to etoposide,although the inhibitory effect was much greater when bothdrugs were combined (Supplementary Fig. S5C, bottom,right). Accordingly, PF-2341066 or etoposide alone didnot significantly inhibit tumor xenograft growth whencompared with the control group (Supplementary Fig.S5D). However, the combined therapy produced a signif-icant decrease in tumor growth with respect to either groupalone (P < 0.05; Supplementary Fig. S6A), suggesting thatMet inhibition increased sensitivity of etoposide in chemo-refractory H841-derived tumors (Supplementary Fig. S5D).The similar phenotypic response to chemotherapy andMet inhibition of H69M and H841 cell lines might berelated to the presence of a mesenchymal phenotype inboth cell lines inbasal conditions aswell as a commoneffectof exogenously added HGF in terms of induction of p-Metand downstream molecules p-GAB and p-ERK. However,differences in baseline phosphorylation of Tyr1234/35and Tyr1349 donot seem to account for the effects observedin these mesenchymal-like cell lines (baseline phosphory-lation is not present in H841). Furthermore, no commongenetic alterations in the Met gene were detected. Ofnote, we tested another mesenchymal-like SCLC cell line,SHP77, which was chemoresistant, did not express Met orp-Met with HGF addition, and was unresponsive toPF-2341066 (data not shown), suggesting that the mesen-chymal phenotype per se is not sufficient to predict re-sponse to anti-Met therapy.

Immunohistochemical studies of H841-derived tumorsin the different treatment groups, confirmed previouslyobserved downregulation of p-Met and mesenchymalmarkers in xenografts treated with PF-2341066 (Supple-mentary Fig. S6B). Finally, in the H841-derived xenograftexperiment, in the combined treatment group, we con-tinued the experiment with 6 mice for 12 additional days,withdrawing the drugs in three cases. When the treatmentwas discontinued, the tumors significantly regrew com-pared with those that were kept treated (SupplementaryFig. S6C).

Mesenchymal phenotype in human SCLC is associatedwith a worse outcome

We retrospectively analyzed 87 SCLC specimens obtain-ed at diagnosis between 2002 and 2012 with sufficientmaterial for biomarker analysis and available clinical infor-mation. This series did not include patients from ourprevious publication (3). Supplementary Table S1 showsthe patients’ characteristics. All the patients received stan-dard first-line chemotherapy with platinum and etoposideand concomitant radiotherapy in cases if indicated. Meanfollow-up time was 14 months (0.2–105 months). Wedetermined Snail1, vimentin, SPARC, NCAM, Met, p-Met,and E-cadherin expression in these specimens. Figure 5Aand Supplementary Table S2 show the pattern of expressionof each biomarker. As observed, most of the cases werenegative (no expression) for Snail1, vimentin, SPARC, and

p-Met. There was not a relationship among the magnitude(intensity) of the signal in the positive cases (Supplemen-tary Table S3). As no cutoff with clinical validity has beenestablished for these biomarkers, we used the median asan arbitrary cutoff point. p-Met was expressed in 34% ofpatients, confirming our previous results. As dichotomousvariables, we found direct association of p-Met expressionwith Snail1, vimentin, and Met and inverse correlationwith E-cadherin (all P values < 0.001; SupplementaryTable S4). Therewas a significant direct association betweenSnail1, vimentin, and SPARC and an inverse correlationbetween each of them and E-cadherin (all P values < 0.001).We found no association between Met and p-Met withclinical characteristics (Performance Status, stage).

We then assessed coexpression of p-Met staining withSnail1 and vimentin in human samples. Figure 5B demon-strates that both markers are coexpressed and tumor cellswith Snail1 and vimentin expression also presented Metphosphorylation. The percentage of tumor cells with coex-pression for Snail1 and vimentin, Snail1 and p-MET, andp-MET and vimentin was estimated in a subset of speci-mens (N ¼ 25) by double immunofluorescence. The meanof Snail1-positive cells coexpressing vimentin was 73%(�20% SD), Snail1-positive cells coexpressing p-MET was74% (�17% SD). and p-MET–positive cells coexpressingvimentin was 80% (�13% SD).

For survival analysis, we first evaluated the associationof classic prognostic factors for patients with SCLC. Asexpected, poor performance status (�2) and stage IV diseasewere associated with worse outcome (P < 0.001). We firstperformed a univariate analysis of each biomarker withsurvival illustrated in Fig. 6A. All biomarkers were asso-ciated with outcome except for p-Met and E-cadherin thatshowed a trend toward statistical significance. In our pre-vious work (3), p-Met was significantly (P < 0.001) asso-ciated with survival in univariate and multivariate analysis.In the present study, p-Met was of borderline significancein the univariate analysis and significantly associatedwith survival in multivariate analysis. The difference in theP values between the two studies by univariate analysismay be related to relatively small sample size (albeitlarge for a SCLC study), or minor differences in the cutoffsused for positivity due to intrinsic differences in both studypopulations. However, the independent significance ob-served in both studies by multivariate analysis stronglysupports the role of p-Met expression in SCLC outcomes.In the multivariate analysis, Snail1, vimentin, and SPARCexpression also correlated with poor prognosis (all P values< 0.05). In contrast, E-cadherin expression was a marker ofbetter outcome (P < 0.05; Supplementary Table S5 andSupplementary Fig. S7).

Mesenchymal features are enhanced in relapseddiseasein human SCLC

We prospectively obtained paired samples (biopsies orcytology block) from 11 patients before first-line treatmentand at first relapse/progression.We sought to compare EMTmarkers and Met status in these samples to validate our

Ca~nadas et al.





preclinical hypothesis. Only five paired cases were adequatefor immunohistochemical analysis. Globally, we observedupregulation of mesenchymal markers and p-Met in thespecimens obtained at relapse (Supplementary Table S6and Fig. 6B).

DiscussionHere, we demonstrate using preclinical models that the

activation of the HGF/Met pathway induces EMT pheno-type in SCLC cells and generates a mesenchymal popula-tion, more tumorigenic and chemoresistant than the paren-tal cells. More importantly, Met-specific inhibition, viamodulation of thesemesenchymal biomarkers reverses thismesenchymal transition and subsequently increases che-mosensitivity in SCLCmodels. Of translational relevance, amesenchymal phenotype is associated with Met activation

in the same tumor cells. Importantly, these mesenchymalfeatures and Met phosphorylation are predictive of poorsurvival in patients with SCLC and are upregulated inchemoresistant or relapsed tumors. Globally, these resultssuggest a potential therapeutic role for PF-2341066 incombination with chemotherapy in the treatment of Met-activated SCLC.

SCLC is a highly lethal disease. No improvements inthe treatment outcome have occurred for many years,leaving "old agent" chemotherapy and radiotherapy asthe backbone treatment for these patients. Moreover,resistance to available therapy is a typical feature of thistumor at relapse, a phenomenon that has been elusive totackle, to date. We have previously observed that Metactivation is an independent prognostic factor in SCLC(3), suggesting a potential role for Met-targeted the-rapies in this disease. HGF has been previously associated

A Snail1 SPARCVimentinE-cadherinp-MetMet

Ex

pre

ss

ion

Lo

w/n

o e

xp

ressio

n

B

Snail1 + Vimentin + DAPI Snail1 Vimentin DAPI

Snail1 + p-Met + DAPI Snail1 p-Met DAPI

p-Met + Vimentin + DAPI p-Met VIM DAPI

Figure 5. Met and EMT biomarker expression in SCLC. A, representative images from positive and negative human SCLC cases for Met, p-Met,E-cadherin, Snail, vimentin, and SPARC. Scale bars, 75 mm. B, representative images of immunofluorescence staining in three SCLC human tumors forSnail (red), vimentin (green), and DAPI (blue); Snail (red), p-Met (green), and DAPI (blue); and p-Met (red), vimentin (green), and DAPI (blue). Colocalizationof Snail and vimentin, Snail and p-Met, and p-Met and vimentin is observed. Scale bars, 15 mm.






with poor prognosis and tumor burden in patients withSCLC (23, 24). Because Met is a well-known inducer ofEMT in other models, we hypothesized that this transi-tion, when induced by HGF, could be responsible forresistance to chemotherapy. We confirmed this hypo-thesis using H69 cells treated with HGF. Upon HGF-chronic exposure, we isolated a mesenchymal chemore-sistant and highly tumorigenic subpopulation mimickingchemoresistant SCLC. Interestingly, these cells showedan increased basal Met phosphorylation when comparedwith the parental H69 along with upregulation of mes-enchymal markers in vitro and in vivo and increasedangiogenesis. The role of HGF as an inducer of EMT(10, 25, 26) and angiogenesis (27, 28) has been reportedpreviously.

These cells endogenously expressed HGF, leading to anautocrine loop and constitutive activation of the pathway.This has been demonstrated in other tumor models (29,30). The existence of autocrine loops has been shown to becommon in mesenchymal cells participating in the stabili-zation of their phenotype (5).

PF-2341066, now crizotinib, is a dual Met/Alk inhibitorthat has demonstrated a clear benefit in the treatment ofpatients with Alk-positive NSCLC (11). This drug is orallydelivered and has a good safety profile. In our study, Metinhibition by PF-2341066 decreased the expression ofSnail1, SPARC, and vimentin in SCLC xenografts. Althoughthis was not sufficient to arrest tumor growth in vivo, therewas a potent effect of PF-2341066 in resensitizing cells tochemotherapy, specifically to etoposide. In addition, in our

A Snail1

Time (mo)

P = 0.022

Su

rviv

al

(%)

Negative Positive

Vimentin

P = 0.003

Su

rviv

al

(%)

Negative Positive

SPARC

P = 0.021

Su

rviv

al

(%)

Negative Positive

p-Met

P = 0.066

Su

rviv

al

(%)

Negative Positive

Snail1-Vimentin-SPARC

P = 0.002

Su

rviv

al

(%)

All negative Any positive

E-cadherin

P = 0.179

Su

rviv

al

(%)

NegativePositive

Snail1 SPARCVimentinE-cadherinp-MetMetB

Dia

gn

osis

Rela

pseP

ati

en

t #

5

0 12 24 36 48 60 0 12 24 36 48 60 0 12 24 36 48 60

0 12 24 36 48 60 0 12 24 36 48 60 0 12 24 36 48 60

100

80

60

40

20

0

100

80

60

40

20

0

100

80

60

40

20

0

100

80

60

40

20

0

100

80

60

40

20

0

100

80

60

40

20

0

Time (mo) Time (mo)

Time (mo)Time (mo)Time (mo)

Figure 6. Met and EMT markers: association with survival in SCLC and changes after treatment. A, Kaplan–Meier curves for overall survival univariateanalysis according to biomarker status. B, paired samples from SCLC patient #5 at diagnosis and relapse time points, stained for Met, p-Met,E-cadherin, Snail, vimentin (VIM), and SPARC. Upregulation of mesenchymal markers and downregulation in E-cadherin expression were detectedin relapse sample. Scale bars, 75 mm.

Ca~nadas et al.





patient population, there was a strong and independentcorrelation of this mesenchymal phenotype with Met acti-vation and poor patient outcome. Without the data ofuntreated patients’ arm, it is not possible to specify whetherthese biomarkers are predictive or prognostic in this disease.However, 85% of patients in this series showed response tofirst-line treatment, and this was not associated with bio-marker status (data not shown).The mechanistic association between Met activation

and mesenchymal features was supported by the coex-pression of p-Met, Snail1, and vimentin in the sametumor cells and by the changes in EMT markers upontreatment with crizotinib. Upregulation of mesenchymalbiomarkers and Met activation have both been associatedseparately with prognosis in several tumor types (6, 31–33) and coexistence of both markers with stemness (34).One could hypothesize that the presence of a subsetof cells with these markers in SCLC at diagnosis couldpredict the enrichment of this chemoresistant populationat progression after chemotherapy and, therefore, explainthe poor prognosis of these patients in our study. Thiswould be consistent with the greater antitumor effectobserved when PF-2341066 is added upfront (Supple-mentary Fig. S4A) when compared with in vivo experi-ments using established H69M tumor xenografts (i.e.,average tumor volume �200 mm3). In the latter model,PF-2341066 alone did not significantly affect tumorgrowth albeit EMT markers and vasculogenesis weremodified in the tumor specimens assayed at the timeanimals were sacrificed (Fig. 4). In contrast, PF-2341066significantly reduced tumor growth and mesenchymalbiomarkers when drug exposure was initiated beforetumor xenografts were established (Supplementary Fig.S4). A possible explanation, beyond a potential greatereffect of drug just based on lower tumor volume, is thata higher proportion of tumor cells at earlier timepoints is more mesenchymal than at later time points.Regardless of this, our data suggest that the inhibition ofEMT and vasculogenesis markers plays a role in thesensitizing effects of PF-2341066 toward etoposide andwe propose that the main role of Met inhibition wouldbe to enhance chemotherapy activity instead of a role asa single-drug strategy.However, the selection of patients who may benefit from

this drug is crucial. Some researchers postulate that a selec-tion according to Met expression by immunohistochemis-try would be a good criterion for further evaluation of Metinhibitors (i.e., onartuzumab; ClinicalTrials.gov identifier:NCT01456325).In our experience, Met levels (neither by mRNA or pro-

tein quantification) do not reflect Met activity and do notpredict for Met inhibitor sensitivity; accordingly, Met phos-phorylation would be a superior biomarker for pathwayactivation. However, p-Met antibodies have poor per-formance due to lack of sensitivity and specificity on routineclinical samples.We previously reported that Met mutations or poly-

morphisms could be a marker of response to Met inhi-

bitors. However, in the present work, we report that aMet-dependent cell line, H841, without any known Metmutation or polymorphism, but enriched with mesen-chymal features, was also sensitive to Met inhibition. Thisnovel observation suggests the potential benefit of Metinhibitors for patients with SCLC with mesenchymalfeatures in the tumor.

We initially focused on Snail1 as it is the best-character-ized inducer of EMT and its role in our SCLC models wasconfirmed by shRNA experiments. Because our ultimategoal was to explore an EMT phenotype of potential clinicalinterest, as assayed by robust epithelial and mesenchymalmarkers, but not to study EMT at the transcriptional level,we did not assay additional EMT transcriptional repressorssuch as Zeb1, Zeb2, Twist1, or Slug.

Our work suggests that the selection of patients accord-ing to mesenchymal biomarkers (SPARC, vimentin, andE-cadherin) in combination with Met expression is a goodalternative for selecting patients with SCLC for clinicaltrials of Met inhibitors plus chemotherapy.

Disclosure of Potential Conflicts of InterestE. Arriola is a consultant/advisory board member of Pfizer, Inc. No

potential conflicts of interest were disclosed by the other authors.

Authors' ContributionsConception and design: I. Ca~nadas, F. Rojo, M. D�omine, A.G. de Herreros,E. ArriolaDevelopment of methodology: I. Ca~nadas, F. Rojo, O. Arp��, S. Zazo,M. D�omine, S. Mojal, E. ArriolaAcquisitionofdata (provided animals, acquired andmanagedpatients,

provided facilities, etc.): I. Ca~nadas, F. Rojo, �A. Taus, O. Arp��, M. Arum��-Ur��a, L. Pijuan, S. Men�endez, S. Zazo, M. D�omine, E. ArriolaAnalysis and interpretation of data (e.g., statistical analysis, biosta-tistics, computational analysis): I. Ca~nadas, F. Rojo, M. Arum��-Ur��a,S. Men�endez, M. D�omine, M. Salido, S. Mojal, A.G. de Herreros, A. Rovira,J. Albanell, E. ArriolaWriting, review, and/or revision of the manuscript: I. Ca~nadas, F. Rojo,S. Zazo, M. D�omine, A.G. de Herreros, A. Rovira, J. Albanell, E. ArriolaAdministrative, technical, or material support (i.e., reporting or orga-

nizingdata, constructingdatabases): I. Ca~nadas, F. Rojo, �A. Taus, O. Arp��,L. Pijuan, E. ArriolaStudy supervision: I. Ca~nadas, F. Rojo, M. D�omine, A. Rovira, J. Albanell,E. Arriola

AcknowledgmentsThe authors thank Lara Nonell and Eulalia Puigdecanet for the support in

the performance of microarray experiments, Raul Pe~na and Rosa Vi~nas forexperimental support, and Xavier Villanueva for his kind collaboration insamplemanagement. The authors also thank Pfizer for generously providingPHA-665752 and PF-2341066.

Grant SupportThis workwas supported by RD12/0036/0051, RD09/0076/0036, RD09/

0076/0101, RD06/0020/0040, PI12/00680, PI12/01552, PI09/01594,SAF2010-16089, 2009 SGR 321, and the "Xarxa de Bancs de tumors"sponsored by Pla Director d’Oncologia de Catalunya (XBTC). J. Albanelland F. Rojo are recipients of the intensification program ISCIII/FEDER. Theauthors thank Fundaci�o Cellex (Barcelona) for a generous donation to theHospital del Mar Medical Oncology Service.

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

ReceivedMay 21, 2013; revised November 6, 2013; accepted November 7,2013; published OnlineFirst November 27, 2013.






References1. Pleasance ED, Stephens PJ, O'Meara S, McBride DJ, Meynert A,

Jones D, et al. A small-cell lung cancer genome with complex signa-tures of tobacco exposure. Nature 2010;463:184–90.

2. Arriola E, Canadas I, Arumi M, Rojo F, Rovira A, Albanell J. Geneticchanges in small cell lung carcinoma. Clin Transl Oncol 2008;10:189–97.

3. Arriola E, Canadas I, Arumi-Uria M, Domine M, Lopez-Vilarino JA, ArpiO, et al. MET phosphorylation predicts poor outcome in small celllung carcinoma and its inhibition blocks HGF-induced effects inMET mutant cell lines. Br J Cancer 2011;105:814–23.

4. Craene BD, Berx G. Regulatory networks defining EMT during cancerinitiation and progression. Nat Rev Cancer 2013;13:97–110.

5. Garcia de Herreros A, Baulida J. Cooperation, amplification, and feed-back in epithelial-mesenchymal transition. Biochim Biophys Acta2012;1825:223–8.

6. Soltermann A, Tischler V, Arbogast S, Braun J, Probst-Hensch N,WederW, et al. Prognostic significance of epithelial-mesenchymal andmesenchymal-epithelial transition protein expression in non-small celllung cancer. Clin Cancer Res 2008;14:7430–7.

7. Vendrell JA, Thollet A, Nguyen NT, Ghayad SE, Vinot S, Bieche I, et al.ZNF217 is a marker of poor prognosis in breast cancer that drivesepithelial-mesenchymal transition and invasion. Cancer Res 2012;72:3593–606.

8. YangMH,WuMZ, Chiou SH, Chen PM, Chang SY, Liu CJ, et al. Directregulation of TWIST by HIF-1alpha promotes metastasis. Nat Cell Biol2008;10:295–305.

9. Ogunwobi OO, Liu C. Hepatocyte growth factor upregulation pro-motes carcinogenesis and epithelial-mesenchymal transition in hepa-tocellular carcinoma via Akt andCOX-2 pathways. Clin ExpMetastasis2011;28:721–31.

10. Grotegut S, von Schweinitz D, Christofori G, Lehembre F. Hepatocytegrowth factor induces cell scattering through MAPK/Egr-1-mediatedupregulation of Snail. EMBO J 2006;25:3534–45.

11. Kwak EL, Bang YJ, Camidge DR, Shaw AT, Solomon B,Maki RG, et al.Anaplastic lymphoma kinase inhibition in non-small-cell lung cancer.N Engl J Med 2010;363:1693–703.

12. Ou SH, Bazhenova L, Camidge DR, Solomon BJ, Herman J, Kain T,et al. Rapid and dramatic radiographic and clinical response to anALK inhibitor (crizotinib, PF02341066) in an ALK translocation-positive patient with non-small cell lung cancer. J Thorac Oncol2010;5:2044–6.

13. Mok TS, Wu YL, Thongprasert S, Yang CH, Chu DT, Saijo N, et al.Gefitinib or carboplatin-paclitaxel in pulmonary adenocarcinoma.N Engl J Med 2009;361:947–57.

14. Comoglio PM, Giordano S, Trusolino L. Drug development of METinhibitors: targeting oncogene addiction and expedience. Nat RevDrug Discov 2008;7:504–16.

15. Sequist LV, vonPawel J,GarmeyEG,AkerleyWL,BruggerW,Ferrari D,et al. Randomized phase II study of erlotinib plus tivantinib versuserlotinib plus placebo in previously treated non-small-cell lung cancer.J Clin Oncol 2011;29:3307–15.

16. Yamazaki S, Vicini P, Shen Z, Zou HY, Lee J, Li Q, et al. Phar-macokinetic/pharmacodynamic modeling of crizotinib for anaplas-tic lymphoma kinase inhibition and antitumor efficacy in humantumor xenograft mouse models. J Pharmacol Exp Ther 2011;340:549–57.

17. Moncho-Amor V, Ibanez de Caceres I, Bandres E,Martinez-Poveda B,Orgaz JL, Sanchez-Perez I, et al. DUSP1/MKP1 promotes angiogen-esis, invasion andmetastasis in non-small-cell lung cancer. Oncogene2011;30:668–78.

18. Christensen JG, SchreckR, Burrows J, Kuruganti P, Chan E, LeP, et al.A selective small molecule inhibitor of c-Met kinase inhibits c-Met-dependent phenotypes in vitro and exhibits cytoreductive antitumoractivity in vivo. Cancer Res 2003;63:7345–55.

19. Ponzetto C, Bardelli A, Zhen Z, Maina F, dalla Zonca P, Giordano S,et al. A multifunctional docking site mediates signaling and transfor-mation by the hepatocyte growth factor/scatter factor receptor family.Cell 1994;77:261–71.

20. Fenouille N, Tichet M, Dufies M, Pottier A, Mogha A, Soo JK, et al. Theepithelial-mesenchymal transition (EMT) regulatory factor SLUG(SNAI2) is a downstream target of SPARC and AKT in promotingmelanoma cell invasion. PLoS ONE 2012;7:e40378.

21. Huang C, Qiu Z, Wang L, Peng Z, Jia Z, Logsdon CD, et al. A novelFoxM1-caveolin signaling pathway promotes pancreatic cancer inva-sion and metastasis. Cancer Res 2011;72:655–65.

22. Antoon JW, Lai R, Struckhoff AP, Nitschke AM, Elliott S, Martin EC,et al. Altered death receptor signaling promotes epithelial-to-mesen-chymal transition and acquired chemoresistance. Sci Rep 2012;2:539.

23. Bharti A, Ma PC, Maulik G, Singh R, Khan E, Skarin AT, et al. Hap-toglobin alpha-subunit and hepatocyte growth factor can potentiallyserve as serum tumor biomarkers in small cell lung cancer. AnticancerRes 2004;24:1031–8.

24. Takigawa N, Segawa Y, Maeda Y, Takata I, Fujimoto N. Serumhepatocyte growth factor/scatter factor levels in small cell lung cancerpatients. Lung Cancer 1997;17:211–8.

25. Toiyama Y, Yasuda H, Saigusa S, Matushita K, Fujikawa H, TanakaK, et al. Co-expression of hepatocyte growth factor and c-Metpredicts peritoneal dissemination established by autocrine hepa-tocyte growth factor/c-Met signaling in gastric cancer. Int J Cancer2011;130:2912–21.

26. LuKV,ChangJP,ParachoniakCA,PandikaMM,AghiMK,MeyronetD,et al. VEGF inhibits tumor cell invasion and mesenchymal transitionthrough a MET/VEGFR2 complex. Cancer Cell 2012;22:21–35.

27. Shojaei F, Lee JH, Simmons BH, Wong A, Esparza CO, Plumlee PA,et al. HGF/c-Met acts as an alternative angiogenic pathway in suni-tinib-resistant tumors. Cancer Res 2010;70:10090–100.

28. Hoot KE, Oka M, Han G, Bottinger E, Zhang Q, Wang XJ. HGFupregulation contributes to angiogenesis in mice with keratinocyte-specific Smad2 deletion. J Clin Invest 2010;120:3606–16.

29. Kentsis A, Reed C, Rice KL, Sanda T, Rodig SJ, Tholouli E, et al.Autocrine activation of the MET receptor tyrosine kinase in acutemyeloid leukemia. Nat Med 2012;18:1118–22.

30. Xie Q, Bradley R, Kang L, Koeman J, Ascierto ML, Worschech A, et al.Hepatocyte growth factor (HGF) autocrine activation predicts sensi-tivity to MET inhibition in glioblastoma. Proc Natl Acad Sci U S A2012;109:570–5.

31. Merikallio H, Turpeenniemi-Hujanen T, Paakko P, Makitaro R, Riitta K,Salo S, et al. Snail promotes an invasive phenotype in lung carcinoma.Respir Res 2012;13:104.

32. Stein GY, Yosef N, Reichman H, Horev J, Laser-Azogui A, Berens A,et al. Met kinetic signature derived from the response to HGF/SF in acellular model predicts breast cancer patient survival. PLoS ONE2012;7:e45969.

33. Knight JF, LesurfR, ZhaoH,PinnaduwageD,DavisRR,SalehSM, et al.Met synergizes with p53 loss to inducemammary tumors that possessfeatures of claudin-low breast cancer. Proc Natl Acad Sci U S A2013;110:E1301–10.

34. De Bacco F, Casanova E, Medico E, Pellegatta S, Orzan F, Albano R,et al. TheMET oncogene is a functional marker of a glioblastoma stemcell subtype. Cancer Res 2012;72:4537–50.

Ca~nadas et al.





2014;20:938-950. Published OnlineFirst November 27, 2013.Clin Cancer Res Israel Cañadas, Federico Rojo, Álvaro Taus, et al. Reverts Chemoresistance in Small Cell Lung CancerTargeting Epithelial-to-Mesenchymal Transition with Met Inhibitors

Updated version

10.1158/1078-0432.CCR-13-1330doi:

Access the most recent version of this article at:

Material

Supplementary

http://clincancerres.aacrjournals.org/content/suppl/2013/12/02/1078-0432.CCR-13-1330.DC1

Access the most recent supplemental material at:

Cited articles

http://clincancerres.aacrjournals.org/content/20/4/938.full#ref-list-1

This article cites 34 articles, 9 of which you can access for free at:

Citing articles

http://clincancerres.aacrjournals.org/content/20/4/938.full#related-urls

This article has been cited by 5 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://clincancerres.aacrjournals.org/content/20/4/938To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/lookup/doi/10.1158/1078-0432.CCR-13-1330

http://clincancerres.aacrjournals.org/content/suppl/2013/12/02/1078-0432.CCR-13-1330.DC1

http://clincancerres.aacrjournals.org/content/20/4/938.full#ref-list-1

http://clincancerres.aacrjournals.org/content/20/4/938.full#related-urls

http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://clincancerres.aacrjournals.org/content/20/4/938


Documents

Targeting Epithelial-to-Mesenchymal Transition with Met … · antibody; Ventana-Roche), p-Met Y1349 (130H2) rabbit mAb, p-Met Y1234/35 (D26) XP rabbit mAb (Cell Signal-ing Technology),