Natmedcmet by GnomeDX

Articles

EGFR and MET receptor tyrosine kinase–altered microRNA expression induces tumorigenesis and gefitinib resistance in lung cancers

npg

Michela Garofalo1,9, Giulia Romano2,9, Gianpiero Di Leva1,9, Gerard Nuovo1, Young-Jun Jeon1, Apollinaire Ngankeu1, Jin Sun1, Francesca Lovat1, Hansjuerg Alder1, Gerolama Condorelli3, Jeffrey A Engelman4, Mayumi Ono5, Jin Kyung Rho6, Luciano Cascione1,7, Stefano Volinia1, Kenneth P Nephew8 & Carlo M Croce1 The involvement of the MET oncogene in de novo and acquired resistance of non-small cell lung cancers (NSCLCs) to tyrosine kinase inhibitors (TKIs) has previously been reported, but the precise mechanism by which MET overexpression contributes to TKI-resistant NSCLC remains unclear. MicroRNAs (miRNAs) negatively regulate gene expression, and their dysregulation has been implicated in tumorigenesis. To understand their role in TKI-resistant NSCLCs, we examined changes in miRNA that are mediated by tyrosine kinase receptors. Here we report that miR-30b, miR-30c, miR-221 and miR-222 are modulated by both epidermal growth factor (EGF) and MET receptors, whereas miR-103 and miR-203 are controlled only by MET. We showed that these miRNAs have important roles in gefitinib-induced apoptosis and epithelial-mesenchymal transition of NSCLC cells in vitro and in vivo by inhibiting the expression of the genes encoding BCL2-like 11 (BIM), apoptotic peptidase activating factor 1 (APAF-1), protein kinase C « (PKC-«) and sarcoma viral oncogene homolog (SRC). These findings suggest that modulation of specific miRNAs may provide a therapeutic approach for the treatment of NSCLCs. MiRNAs repress gene expression by inhibiting mRNA translation or by promoting mRNA degradation and are considered to be master regulators of various processes, ranging from proliferation1 to apoptosis2. Both loss and gain of miRNA function contribute to cancer development through the upregulation and silencing, respectively, of different target genes. Altered miRNA expression in various human tumor types has been observed, and key roles for miRNAs in cancer pathogenesis and the response to therapy have been shown3,4. NSCLCs account for approximately 85% of all cases of lung cancer5. Although NSCLC is a remarkably heterogeneous disease that includes distinct morphological and molecular subtypes, activation of epidermal growth factor receptor (EGFR) and MET (the receptor tyrosine kinase (RTK) for hepatocyte growth factors) is common and is associated with stimulation of the rat sarcoma (RAS)-mitogen-activated protein kinase 1 (ERK) and the phosphoinositide-3-kinase (PI3K)-v-akt murine thymoma viral oncogene homolog 1 (AKT) axes, which leads to NSCLC cell proliferation, survival and invasion6. The TKIs gefitinib and erlotinib effectively target EGFR in individuals with NSCLC, but these therapeutic agents are ultimately limited by the emergence of mutations and other molecular mechanisms conferring drug resistance7.

MET protein expression and phosphorylation have been associated with both primary and acquired resistance to EGFR TKI therapy in NSCLC patients8,9, strongly implicating MET as an effective therapeutic target to overcome resistance to this important class of drugs in lung cancer10. Here we show that EGF and MET receptors, by modulating specific miRNAs, control gefitinib-induced apoptosis and NSCLC tumorigenesis. Our results are, to our knowledge, the first to identify EGF- and MET-receptor–regulated miRNAs representing oncogenic signaling networks in NSCLCs. RESULTS MiRNAs modulated by both EGFR and MET To identify EGFR- and MET-regulated miRNAs, we stably silenced EGFR and MET in Calu-1 cells from the American Type Culture Collection (ATCC) using shRNA lentiviral particles (Fig. 1a) and examined the global miRNA expression profiles. In EGFR- and METknockdown (EGFR-KD and MET-KD) Calu-1 cells, we identified 35 and 44 significantly (P < 0.05) dysregulated miRNAs, respectively (Fig. 1b and Supplementary Fig. 1a). MiRNAs with a greater than 1.5-fold (for EGFR) or a greater than 1.7-fold (for MET) change

1Department

of Molecular Virology, Immunology and Medical Genetics, Comprehensive Cancer Center, Ohio State University, Columbus, Ohio, USA. 2Istituto Di Ricovero e Cura a Carattere Scientifico Studio Diagnostica Nucleare (SDN), Naples, Italy. 3Department of Cellular and Molecular Biology and Pathology, Institute of Experimental Endocrinology and Oncology, Consiglio Nazionale delle Ricerche (IEOS-CNR), Faculty of Biotechnological Science, ‘Federico II’ University of Naples, Napoli, Italy. 4Massachusetts General Hospital Cancer Center and Department of Medicine, Harvard Medical School, Boston, Massachusetts, USA. 5Department of Pharmaceutical Oncology, Graduate School of Pharmaceutical Sciences, Kyushu University, Fukuoka, Japan. 6Department of Pulmonary and Critical Care Medicine, University of Ulsan, Seoul, Korea. 7Department of Clinical and Molecular Biomedicine, University of Catania, Catania, Italy. 8Department of Cellular and Integrated Physiology, Medical Sciences Program, Indiana University School of Medicine and the Indiana University Simon Cancer Center, Bloomington, Indiana, USA. 9These authors contributed equally to this work. Correspondence should be addressed to C.M.C. (carlo.croce@osumc.edu). Received 21 April; accepted 20 October; published online 11 December 2011; doi:10.1038/nm.2577

VOLUME 18 | NUMBER 1 | JANUARY 2012 nature medicine

articles miRNAs in EGFR-KD and MET-KD Calu-1 cells using quantitative RT-PCR (qRT-PCR) (Supplementary Fig. 1b) and northern blot (Fig. 1d) analyses.

are shown. After comparing these two lists of miRNAs, we found only eight that were regulated by both EGFR and MET (Fig. 1c): miR-21, miR-221 and miR-222, miR-30b and miR-30c, miR-29a and miR-29c and miR-100. We initially focused on miR-30b, miR-30c, miR-221 and miR-222, which were downregulated after both MET and EGFR silencing and which showed the highest fold changes in expression. We also investigated the two miRNAs that were most differentially induced after MET silencing, miR-103 and miR-203, based on recent evidence indicating MET overexpression in de novo and acquired resistance to TKIs8,9. We validated the expression of these six

0.4

3.0 2.5 2.0 1.5 1.0 0.5 0

0.3 0.2 0.1 0

H 4 C 60 al u H -1 12 99 H 15 73 A5 49 H 29 2 H C C 82 7

Relative miR expression

11666 1 – 16 6–1662 1796– 122 06 16 –1 01 71 2

2.5 2.0 1.5 1.0 0.5 0

2 –1 ,

1, 7

1, 5

T W

–1 ,

71 2 1,

59 5–

12 2

0.4 0.3 0.2 0.1 0

H 4 C 60 al u H -1 12 99 H 15 73 A5 49 H 2 H 92 C C 82 7

Relative miR expression

1, 57 0

U T

51 7–

–1 ,

,5 7 W T

–1

27 M

–3 3 27

51 7

T W

Relative luciferase activity

Scr 203

65 6–

W T

miR-203 BS

2.5 2.0 1.5 1.0 0.5 0

0.30 0.24 0.18 0.12 0.06 0

H460 Scr 221 222 β-actin

miR-203 SRC

Scr 30b 30c BIM β-actin

Calu-1 Scr 103 PKC-ε β-actin

miR-221 miR-222 APAF-1

Scr

203 SRC β-actin

Calu-1

miR-30b miR-30c BIM

APAF-1 β-actin

BIM β-actin

1.5 1.0 0.5 0

Relative protein expression

Relative luciferase activity

16 27 1 – 16 7– 33 64 16 –1 23 67 0

npg

Scr 103

snRNA U6

APAF-1

Relative protein expression

miR-103 BS

miR-30c

An 0b ti30 c

MUT

TAA (2,411) TAA (2,060)(4,145) ATG (450) (5,537) ATG (198) SRC 3′ UTR PKC-ε 3′ UTR

miR-221 miR-222 miR-30b miR-30c miR-21 miR-29a miR-29c miR-100

H 4 C 60 al u H -1 12 9 H 9 15 73 A5 49 H 29 2 H C C 82 7

* * WT

miR-30b

ti-

MUT

Scr 30b 30c

H 1 12 9 H 9 15 73 A5 49 H 29 2 H C C 82 7

0.04

0.08

* *

140 120 100 80 60 40 20 0

0.12

miR-203

Sc r 2 An 21 ti22 2

8– 28

Relative luciferase activity

Scr 221 222

miR-30b and miR-30c BS

0.16

miR-103

ti-

4 29

16 4– 16 miR-221 and miR-222 BS

miR-222

Relative protein expression

Relative luciferase activity

100

(4,919) BIM 3′ UTR

P < 0.05

tr C sh 0

ATG (537)

ATG (289) TAA (705) (7,171) APAF 3′ UTR

Relative miR expression

TAA (4291)

H 46

0.2

0.4

0.6

MET relative expression

EGFR relative expression

0.8

1.4 1.2 1.0 0.8 0.6 0.4 0.2 0

Calu-1 shCtr shMET miR-221

Relative protein expression

1.0

ti3

β-actin

miR-548d miR-652 miR-30c miR-135b miR-30b let-7a miR-221 miR-29c let-7e miR-26b miR-222 miR-146a miR-200b miR-29a miR-20a miR-100 miR-365 miR-21 miR-29b miR-203 miR-103 miR-200c miR-328 miR-218

EGFR miRNA MET miRNA

β-actin

miR-361 miR-301 miR-324 miR-636 miR-30c miR-99b miR-15b miR-197 miR-125b miR-193a miR-30b miR-221 miR-182 miR-130b miR-222 miR-99a miR-21 miR-100 miR-210 miR-125a miR-29c miR-29a miR-200a miR-101

MET

3.00 2.00 1.00 0.00 –3.00 –2.00 –1.00

EGFR

shMET

Calu-1 shCtr shMET

shCtr

Relative miR expression

Calu-1 shCtr shEGFR

shEGFR

ti-

shCtr

Tyrosine-kinase–modulated miRNA targets As MET and EGFR RTKs have a key role in lung cancer tumorigenesis and progression11, we hypothesized that miR-103 and miR-203 (which are increased after MET knockdown) are tumor suppressors and that miR-221, miR-222, miR-30b and miR-30c (which are decreased after MET and EGFR silencing) are oncogenic. Our search for

miR-103 PKC-ε

Calu-1 shScr shMET APAF-1 BIM PKC-ε SRC GAPDH

Figure 1 TKI-regulated miRNA targets. (a) Downregulation of EGFR and MET proteins and mRNAs after EGFR and MET silencing. (b) Unsupervised hierarchical clustering in Calu-1 cells after knockdown of EGFR (shEGFR), MET (shMET) or shCtr (scrambled RNA). CTR, control. P < 0.05. (c) Intersection of miRNAs regulated by shEGFR and shMET. (d) Northern blots showing deregulated miRNAs after shMET. snRNA U6, loading control. (e) Luciferase report assays indicated direct interactions between the miRNAs and PRKCE (PKC-ε), SRC, APAF1 and BCL2L11 (BIM) 3′ UTRs. In SRC only, the site at 1,595–1,601 nt is implicated in binding with miR-203; deletion of the site at 1,706–1,712 nt did not rescue luciferase activity (Supplementary Fig. 2). WT, wild type; MUT, mutated; scr, scrambled. (f) Inverse correlation between miR-103, miR-203, miR-221, miR-222, miR-30b and miR-30c and target proteins in a panel of NSCLC cells. (g) MiR-221, miR-222, miR-30b and miR-30c overexpression decreased the concentration of APAF-1 and BIM proteins. (h) MiR-103 and miR-203 overexpression decreased the concentration of PKC-ε and SRC proteins. (i) Inhibitors of miR-221, miR-222, miR-30b and miR-30c increased APAF-1 and BIM expression. (j) shMET induced upregulation of APAF-1 and BIM and downregulation of SRC and PKC-ε. Results are representative of at least three independent experiments. Error bars, ± s.d *P < 0.001, **P< 0.05 by two-tailed Student’s t test.

nature medicine VOLUME 18 | NUMBER 1 | JANUARY 2012

Articles

miR-222

MET

miR103 and MET

miR-222 and MET

RGB

MET

miR-203 and MET

miR-30c and MET

RGB

miR-221

miR-222

0.1

0.1 High Low

High Low

0.8 2

0.6

R = 0.61

0.4 0.2 0

2 4 miR-103 expression

0.1 High Low 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

High Low

R 2= 0.58

2 4 miR-203 expression

npg

Figure 2 MET-miRNA coexpression analysis. (a) One hundred ten lung cancer tissues were analyzed for miR-103, miR-222, miR-203 and miR-30c expression by ISH and then for MET by IHC. The top row shows the miR-103 signal (blue), the MET signal (red) and the mixed signal, in which fluorescent yellow indicates miRNA and protein co-expression; there is a lack of miR-103 in the presence of MET expression. In the serial section of the same cancer (in the second row), miR-222, the MET image and the co-expression of miR-222 and MET are shown. Many cancer cells positive for miR-222 also express MET (yellow). The arrows in the left panel (in the third row) point to benign stromal cells that express miR-203 (blue) and not MET. The other images in the third row show the MET signal (red) and the mixed signal. The arrow in the left image in the fourth row points to cancer cells positive for miR-30c. The right images in each row show the RGB image of the ISH or IHC reaction. (b) Box plots showing miRNA expression in 40 individuals with lung cancer. Real-time PCR was used to classify tumors into two groups, EGFR-MET low and EGFR-MET high, using a round function with a cutoff of 0.5 (2 (−∆Ct)). *P < 0.0001 by Student’s t test. (c) XY scatter plots showing inverse correlation between MET and miR-103 and MET and miR-203. (d) MET and EGFR IHC on 40 lung tumor tissues. One representative case from 17 metastatic tumors expressing both MET and EGFR is shown. The large green arrows point to the tumor cells, and the small black arrows point to the stroma. Scale bars, 100 µm. EGFR

MET

miR-30c

MET

MET expression

miR-203

miR-30c

* 1

c Lung tumors

miR-30b

MET expression

miR-103

Relative expression (log)

mRNA targets using the TargetScan and PicTar computational tools revealed that the 3′ untranslated regions (3′ UTRs) of human APAF1, BCL2L11 (also known as BIM), PRKCE (also known as PKC-ε ) and SRC contain evolutionarily conserved binding sites specific for miR-221 and miR-222, miR-30b and miR-30c, miR-103 and miR-203, respectively (Supplementary Fig. 2a). We focused on these genes based on their role in TKI sensitivity (BCL2L11 (BIM) and APAF1)12,13 or TKI resistance (SRC)14 or in the negative allo steric modulation of EGFR signaling (PRKCE (PKC-ε))15. To investigate whether the miRNAs directly interact with these four putative target genes, we cotransfected pGL3 3′ UTR luciferase reporter vectors with synthetic miR-103, miR-203, miR-221, miR-222, miR-30b and miR-30c. A decrease in luciferase activity indicated direct interactions between the miRNAs and the PRKCE (PKC-ε), SRC, APAF1 and BCL2L11 (BIM) 3′ UTRs (Fig. 1e), and target gene repression was rescued by mutations or deletions in the complementary seed sites (Fig. 1e and Supplementary Fig. 2a). A western blot analysis showed an inverse correlation (P < 0.05) between miR-221, miR-222, miR-103, miR-203, miR-30b and miR-30c expression and the amount of target protein in an NSCLC cell panel (Supplementary Fig. 2b,c), which was confirmed by determining the Pearson correlation coefficients (Fig. 1f and Supplementary Fig. 2d). Results from the immunoblot analysis fully agreed with data obtained using reporter gene assays. Ectopic expression of miR-221, miR-222, miR-30b and miR-30c in H460 cells markedly decreased BIM and APAF-1 expression, and enforced expression of miR-103 and miR-203

clearly reduced the concentrations of PKC-ε and SRC protein (Fig. 1g,h). Conversely, knockdown of miR-221, miR-222, miR-30b and miR-30c increased the concentrations of APAF-1 and BIM protein (Fig. 1i). As MET knockdown Calu-1 cells showed an increase of APAF-1 and BIM concentrations and a decrease of PKC-ε and SRC concentrations (Fig. 1j), enforced expression of miR-221,miR-222, miR-30b and miR-30c in MET knockdown Calu-1 cells strongly reduced APAF-1 and BIM expression (Supplementary Fig. 3a), whereas miR-103 and miR-203 knockdown increased SRC and PKC-ε expression (Supplementary Fig. 3b). Collectively, these data indicate a direct correlation between change in expression of PKC-ε, SRC, APAF-1 and BIM proteins and these specific miRNAs after MET silencing in NSCLC cells (Fig. 1g–j and Supplementary Fig. 3a,b). Detection of PKC-ε, SRC, APAF-1 and BIM proteins in vivo in 110 lung cancer specimens (Supplementary Table 1) using miRNA in situ hybridization (ISH) followed by immunohistochemistry (IHC) showed a more significant negative correlation between these proteins and miR-103, miR-203, miR-221, miR-222, miR-30b and miR-30c in human tumors (Supplementary Fig. 4b). We observed an inverse correlation between miR-203 and SRC expression, miR-30c and BIM expression, miR-103 and PKC-ε expression and miR-222 and APAF-1 expression in the majority of the lung cancer tissues (Supplementary Fig. 4a,b). In addition, we saw MET overexpression in 52% (57/110) of the same 110 lung tumor samples (shown using miRNA ISH and MET IHC; Supplementary Fig. 5a), and we saw low miR-103 and miR-203 expression and high miR-222

VOLUME 18 | NUMBER 1 | JANUARY 2012 nature medicine

NT Gefitinib 5 µM Gefitinib 10 µM

1.0 0.8

0.4 0.2 0 A549

c Gefitinib 5 µM Gefitinib 10 µM

0.6

Calu-1

npg

PC9

+ –

HCC827

1.6

1.2

0.8 0.4 0 Calu-1

A549

PC9

HCC827

– –

+ –

Percent of control

0.1

0.8

0.6 0.4 0.2 0 Calu-1

A549

PC9

– –

– +

+ –

HCC827

d – +

BIM

APAF-1

p-ERK1/2

ERKs total

β-actin

Gefitinib concentation (µM)

NT Gefitinib 5 µM Gefitinib 10 µM

1.0

HCC827

Calu-1 – +

0.1

HCC827 – gefitinib HCC827 + gefitinib

Gefitinib concentation (µM)

NT Gefitinib 5 µM Gefitinib 10 µM

PC9 – –

Gefitinib concentation (µM)

Relative miR-30c expression

Relative miR-30b expression

0.1

120 100 80 60 40 20 0

Relative miR-222 expression

0.1

Gefitinib concentation (µM)

PC9 – gefitinib PC9 + gefitinib

Relative expression

A549 – gefitinib A549 + gefitinib

120 100 80 60 40 20 0

NT Gefitinib 5 µM Gefitinib 10 µM

2.0 1.5

1.0 0.5 0 Calu-1

A549

PC9

HCC827

HCC827 GR

2.0

– Gefitinib + Gefitinib

1.6 1.2 0.8 0.4 0

1.6 Relative expression

Calu-1 – gefitinib Calu-1 + gefitinib

120 100 80 60 40 20 0

Percent of control

120 100 80 60 40 20 0

Relative miR-221 expression

Percent of control

articles

1.2

miR-221

miR-222

miR-30b

miR-30c

PC9 GR – Gefitinib + Gefitinib

0.8

Figure 3 Gefitinib downregulates miR-221, miR-222, miR-30b and miR-30c. (a) Calu-1, A549, 0.4 PC9 and HCC827 cells were treated with increasing concentrations of gefitinib. Cell viability, relative to untreated controls, was measured after 24 h. Each data point represents the mean 0 ± s.d. of five wells. (b) qRT-PCR showing miR-30b, miR-30c, miR-221 and miR-222 downregulation miR-221 miR-222 miR-30b miR-30c only in PC9 and HCC827 gefitinib-sensitive cells and not in Calu-1 and A549 gefitinib-resistant cells after treatment with 5 µM or 10 µM gefitinib. NT, non-treated cells. (c) PC9, Calu-1 and HCC827 cells were treated for 24 h with 5 µM or 10 µM gefitinib. An increase of BIM and APAF-1 expression and a decrease of the phosphorylation of the ERKs were observed only in the HCC827 and PC9 gefitinib-sensitive cells but not in the Calu-1 gefitinib-resistant cells. β-actin was used as a loading control. (d) qRT-PCR showing that miR-221, miR-222, miR-30b and miR-30c expression did not decrease in HCC827 GR and PC9 GR cells (cells with acquired gefitinib resistance) exposed to 10 µM gefitinib for 24 h. All quantitative data were generated from a minimum of three replicates. Error bars, ± s.d. A two tailed Student’s t test was used to determine the P values. *P < 0.001, **P < 0.05.

and miR-30c expression in tumors overexpressing MET (Fig. 2a and Supplementary Fig. 5a); conversely, we observed high miR-103 and miR-203 and low miR-222 and miR-30c expression in tumors without MET expression. Notably, the majority of tumors overexpressing MET had accompanying metastases (Supplementary Fig. 5b), indicating that MET-regulated miRNAs have a role in the metastatic spread of lung cancer cells. We extended our analysis to 40 independent lung tumors with an annotated clinical history (Supplementary Table 2), which we divided into two groups of ‘low’ and ‘high’ MET and EGFR expression based on qRT-PCR analyses (Fig. 2 and Supplementary Fig. 5c). An analysis of variance confirmed that the miRNAs (miR-30b and miR-30c and miR-221 and miR-222) were differentially expressed between the low and high groups, whereas using a Pearson coefficient, we identified an inverse correlation between MET and miR-103 and MET and miR-203 (Fig. 2b,c). We confirmed the qRT-PCR results using an IHC analysis for MET and EGFR (Supplementary Fig. 5d). In addition, we observed MET overexpression in tumors that had distant metastases compared to nonmetastatic tumors, but there was no correlation between metastases and EGFR expression in these 40 lung cancers (Fig. 2d and Supplementary Fig. 5e).

nature medicine VOLUME 18 | NUMBER 1 | JANUARY 2012

Tyrosine-kinase–regulated miRNAs control gefitinib sensitivity Having shown that EGFR regulates miR-221, miR-222, miR-30b and miR-30c, we investigated a role for these miRNAs in gefitinibinduced apoptosis in NSCLCs with wild-type EGFR (Calu-1 and A549 cells) compared to those with EGFR that has exon 19 deletions (PC9 and HCC827 cells). Calu-1 and A549 cells were completely resistant to all concentrations of gefitinib tested (up to 20 µM); in contrast, the growth of PC9 and HCC827 EGFR mutant cells was significantly inhibited, even at low doses (0.1 µM) of gefitinib (Fig. 3a), which is in agreement with results from a previous study16. Notably, after gefitinib treatment, we observed marked miR-30b, miR-30c, miR-221 and miR-222 downregulation and increased amounts of BIM and APAF-1 protein, only in PC9 and HCC827 gefitinib-sensitive cells (Fig. 3b,c). In addition, as previously shown16, the concentration of phosphorylated ERKs was markedly lower in HCC827 and PC9 cells, but not in Calu-1 cells, compared to untreated cells (Fig. 3c). To directly assess the relevance of miR-30b, miR-30c, miR-221 and miR-222 in gefitinib-induced apoptosis, we analyzed the expression of these miRNAs in NSCLC cells with acquired gefitinib resistance, obtained after long-term exposure to increasing drug concentrations8,17: PC9 gefitinib-resistant (PC9 GR) cells with an EGFR Thr790

PC9

* Scr 30b/c 221/222

600,000

200,000 0

Caspase-3/7activity

200 0

14 21 Days

800,000 600,000

PC9

Caspase-3/7 activity

Anti-Scr Anti-30b/c Anti-221/222

160,000 120,000 80,000

Caspase-3/7 activity

siScr siBIM siAPAF-1

400

200

0 0

21 28 35

Days

203 – gef 203 + gef

150

Anti-221 – gef Anti-221 + gef

600

100

400

50 0

P = 0.0002 P < 0.001 P < 0.05 EV and 30b/c BIM WT and 30b/c BIM MUT and 30b/c

Anti-30 – gef Anti-30 + gef

600

14 21 Days

21 Days

Anti-Ctr – Gef

200 7

14 21 Days

Anti-30c

+ Gef

– Gef

+ Gef

Anti-221 – Gef

+ Gef

40,000 0

PC9 GR

A549

Days

200,000

100

400,000

103 – gef 103 + gef

5 0 10 15 Gefitinib concentration (µM)

10 5 15 0 Gefitinib concentration (µM)

Anti-Ctr – gef Anti-Ctr + gef

600 400

Gefitinib concentration (µM)

HCC827 GR

Anti-Scr Anti-30b/c Anti-221/222

siScr siBIM siAPAF-1

Calu-1

Anti-Scr Anti-30b/c Anti-221/222

HCC827

400,000

150

Ctr – gef Ctr + gef

800,000

Tumor volume mm

GAPDH

250 200 150 100 50 0

Percent of control Percent of control Percent of control Percent of control

npg

–

CleavedPARP Flag

0 5 10 15 Gefitinib concentration (µM) 120 100 80 60 40 20 0

APAF-1

0 5 10 15 Gefitinib concentration (µM) 120 100 80 60 40 20 0

–

Scr 30b/c 221/222

0 5 10 15 Gefitinib concentration (µM) 120 100 80 60 40 20 0

BIM +

PARP

0 5 10 15 Gefitinib concentration (µM) 120 100 80 60 40 20 0

EV Gefitinib –

Tumor volume mm

HCC827

Tumor volume mm

120 100 80 60 40 20 0

Caspase-3/7 activity

Percent of control

Articles

5 10 15 0 Gefitinib concentration (µM)

200,000 160,000

A549

120,000 80,000 40,000 0

Empty virus

P < 0.0001 P < 0.0001 EV and 221/222 P < 0.0001

– Gef

+ Gef

103 – Gef

+ Gef

203 – Gef

+ Gef

APAF-1 WT and 221/222 APAF-1 MUT and 221/222

5 10 15 0 Gefitinib concentration (µM)

Figure 4 MiR-30b, miR-30c, miR-221, miR-222, miR-103 and miR-203 regulate gefitinib sensitivity. (a) Sensitive (HCC827 and PC9) and resistant (HCC827GR, PC9GR and Calu-1) cells treated with increasing concentrations of gefitinib. Each data point represents the mean ± s.d of six wells. (b) A western blot showing an increase in gefitinib-induced cleaved PARP fragments after overexpression of BIM and APAF-1 Flag-tagged plasmids and after treatment with gefitinib (15 µM) in A549 cells. EV, empty virus. (c) Silencing of BIM (siBIM) and APAF-1 (siAPAF-1) in HCC827 and PC9 cells reduces the response to gefitinib. (d) Overexpression of BIM and APAF-1 complementary DNAs insensitive to miR-30b, miR-30c, miR-221 and miR-222 induces gefitinib sensitivity in A549 cells. SiScr, SRC siRNA. (e,f) Overexpression of miR-103 and miR-203 and silencing of miR-30c and miR-222 increase gefitinib sensitivity in vivo. (e,f) Growth curve of engrafted tumors (e) and comparison of engrafted tumors (f) in nude mice injected with A549 cells stably infected with inhibitors of control miRNAs (ctr), miR-30c or miR-221 and with miR-103 and miR-203 or an empty virus as a control. The images show average-sized tumors from among five tumors from each category. In a, and c–e, error bars, ± s.d. *P < 0.001, **P < 0.05 by two-tailed Student’s t test. Gef, gefitinib.

alteration and HCC827 gefitinib-resistant (HCC827 GR) cells with MET amplification. In contrast to the gefitinib-responsive parental cells, we did not observe lower expression of miR-30b, miR-30c, miR-221 and miR-222 or modulation of their relative targets after treatment with gefitinib (Fig. 3d and Supplementary Fig. 6a). Of note, miR-30c, miR-221 and miR-222 overexpression in gefitinibsensitive HCC827 and PC9 cells rendered these cells less responsive to treatment with gefitinib compared to parental PC9 and HCC827 cells (Fig. 4a and Supplementary Fig. 7a), and knockdown of miR-30b, miR-30c, miR-221 and miR-222 led to increased gefitinib sensitivity in Calu-1, HCC827 GR and PC9 GR cells (Fig. 4a and Supplementary Fig. 7b), indicating that these miRNAs are key modulators of TKI resistance.

To investigate the contribution of APAF-1 and BIM downregulation mediated by miR-30b, miR-30c, miR-221 and miR-222 to the cellular TKI response, we overexpressed APAF-1 and BIM in A549 gefitinib-resistant cells. We observed gefitinib-induced poly-(ADPribose) polymerase (PARP) cleavage in cells overexpressing BIM and APAF-1 but not in cells transfected with an empty vector plasmid (Fig. 4b). Conversely, the response to gefitinib was reduced by BIM and APAF-1 silencing in gefitinib-sensitive HCC827 and PC9 cells (Fig. 4c). We cloned wild-type and mutated 3′ UTRs of BIM and APAF-1 (which we used for luciferase assays; Fig. 1e) downstream of BIM and APAF-1 coding sequences and performed caspase-3/7 and viability assays. We observed no increase in cell death after treatment with gefitinib of A549 cells cotransfected with wild-type

VOLUME 18 | NUMBER 1 | JANUARY 2012 nature medicine

npg

Calu-1 Scr miR

miR-103

miR-203

A549 miR-103

Scr miR

miR-203

0h siPKC-ε

siScr

siSRC 24 h

2,000

Scr miR 103

siScr siPKC-ε siSRC

203

** * Scr

A549

nature medicine VOLUME 18 | NUMBER 1 | JANUARY 2012

3 10

Percent of cells

C SR

G0-G1

90 80 70 60 50 40 30 20 10 0

-ε

A N

C PK

iR rs

3′ UTRs for BIM and APAF-1 and with miR-30b, miR-30c, miR-221 and miR-222; conversely, miRNA binding-site mutations or deletions restored the apoptotic response to gefitinib, suggesting that the effects of both APAF-1 and BIM on gefitinib sensitivity were directly related to knockdown of these proteins mediated by miR-30b, miR-30c, miR-221 and miR-222 (Fig. 4d and Supplementary Fig. 7c). Because MET overexpression is associated with gefitinib resistance and because miR-30b, miR-30c, miR-221 and miR-222 are also regulated by MET, we hypothesized that MET could mediate resistance to gefitinib treatment through the regulation of these miRNAs. Thus, we speculated that the simultaneous inhibition of MET and EGFR could overcome gefitinib resistance in NSCLCs. In support of this hypo thesis, we observed downregulation of miR-30b, miR-30c, miR-221 and miR-222 in Calu-1 and A549 cells overexpressing MET (data not shown) after MET knockdown or treatment with the MET inhibitor SU11274 (Supplementary Fig. 8a,b). In addition, we observed increased caspase-3/7 activity and decreased cell viability in SU11274treated Calu-1 and MET-KD Calu-1 cells that we exposed to different concentrations of gefitinib (Supplementary Fig. 8c,d). Taken together, these results suggest that MET overexpression induces resistance to gefitinib treatment in TKI-resistant Calu-1 cells through the upregulation of miR-30b, miR-30c, miR-221 and miR-222 and that inhibition of both EGFR and MET is needed to shut down these miRNAs and their survival effects. Other miRNAs commonly deregulated by EGFR and MET, including miR-21, miR-29a, miR-29c and miR-100 (Fig. 1c), were downregulated in HCC827 and PC9 cells treated with gefitinib (Supplementary Fig. 9a). Of note, we did not observe downregulation of miR-21, miR-29a, miR-29c and miR-100 in HCC827 GR and PC9 GR cells after gefitinib treatment (Supplementary Fig. 9b); however, enforced expression of miR-21, miR-29a, miR-29c and miR-100 increased gefitinib resistance in HCC827 and PC9 cells (Supplementary Fig. 10a). We therefore concluded that EGFR and MET control oncogenic signaling networks through common miRNAs. Recently, researchers from another study reported that miR-21 is downregulated after gefitinib treatment of NSCLC cells with EGFR

203

Calu-1

G0-G1

90 80 70 60 50 40 30 20 10 0

103

r Sc

500

100 80 60 40 20 0

1,500

Migration distance

Absorbance at 595 nm

Figure 5 MiR-103 and miR-203 inhibit the migration and proliferation of NSCLCs. (a) Representative images of cells that migrated through the filter and that were stained with crystal violet. Scale bar, 40 µm. The results are means ± s.d. n = 3 experiments. *P < 0.001. (b) Representative photographs of scratched areas of the confluent monolayer of A549 cells transfected with miR-103, miR-203 or control miRNA (Scr miR) at 0 h and 24 h after wounding with a pipet tip. Scale bar, 500 µm. The magnification is the same for all the panels *P < 0.00001, **P < 0.001 relative to miRNA scrambled transfected cells. (c) Flow cytometric distributions of Calu-1 and A549 cells transfected with control miRNAs, miR-103 (103), miR-203 (203), control siRNA (Scr siRNA) and siRNAs of PKC-ε (siPKC-ε) and SRC (siSRC). The effect of miR-203 on the cell cycle is slightly stronger than that of miR-103, as assessed by the ratio between the G0-G1 and S phases. All quantitative values show mean ± s.d. n = 5. A two-tailed Student’s t test was used to determine the P values for the G0-G1:S ratios. *P < 0.00001, **P < 0.005 compared to scrambled miRNA. Scr miR, scrambled miR; siScr, siRNA scrambled.

Percent of cells

articles

activating mutations18. For this reason, we further investigated whether miR-21 knockdown by oligonucleotide inhibitors of miRNAs could restore gefitinib sensitivity in NSCLC cells with de novo or acquired resistance. MiR-21 knockdown increased sensitivity to gefitinib-induced apoptosis in A549, HCC827GR and PC9GR cells, suggesting that this miRNA has a major role in the EGFR-MET signaling pathway (Supplementary Fig. 11a,b). Next, we focused on miR-103 and miR-203, which are strongly downregulated in MET-expressing Calu-1 cells (Fig. 1d). As expected, treatment of Calu-1 cells with SU11274 increased (P < 0.05) the expression of miR-103 and miR-203 (Supplementary Fig. 12a). Their targets, SRC and PKC-ε, exert pro-survival effects and contribute to gefitinib resistance by activating the AKT and ERK signaling pathways14,15. Accordingly, overexpression of miR-103 and miR-203 in A549 cells was associated with reduced phosphorylation of AKT and its substrate glycogen synthase kinase 3 β (GSK3β) and reduced phosphorylation of the ERKs (Supplementary Fig. 13a). Researchers previously reported that MET induces gefitinib resistance through persistent PI3K-AKT and ERK signaling activation8, and our results indicate that MET overexpression controls gefitinib resistance through activation of the AKT-ERK pathways and is mediated, at least in part, by miR-103 and miR-203. In agreement with the results from the previous study8, enforced expression of miR-103 or miR-203 or silencing of PKC-ε and SRC increased the sensitivity of Calu-1 cells to gefitinib (as assessed by caspase-3/7 and viability assays; Supplementary Fig. 13b,c). Notably, we found that miR-103 and miR-203 expression decreased and SRC and PKC-ε expression consequently increased in HCC827 gefitinib-resistant cells with acquired MET amplification and gefitinib resistance8 compared to the HCC827 parental cells, corroborating our hypothesis that MET controls the response to TKIs, at least in part through miR-103, miR-203 and their respective targets (Supplementary Fig. 13d,e). Finally, to analyze sensitivity to gefitinib in vivo, we stably transfected A549 cells with GFP lentivirus constructs containing either full-length miR-103 or miR-203, or anti–miR-221 and anti–miR-30c.

Vimentin Zeb1

Snail

Snail F-actin DAPI Vimentin

Calu-1 Snail Fibronectin Vimentin Zeb1 N-cadherin GAPDH

npg

Vim F-actin DAPI

miR-103

P= 0.00064 P= 0.00034 P= P= 0.008 P= P = 0.0064 0.0045 0.0077

Fibronectin Fibr F-actin DAPI

Ctr miR

e P= 0.00036 P= 0.0012

miR-203

E-cadherin GAPDH

miR-203

shCtr shMET

E-cadh F-actin DAPI

miR-103

Slug

Relative expression Eca d N he -c ad rin P= 1.26 × 10–5 he rin Ze b1 Ze b2 Sl ug Sn ai T l Vi wis m Fi e t br nt on in ec tin

3.0 2.5 2.0 1.5 1.0 0.5 0

E-cadherin Ctr miR

Snail Fibronectin

shMET

203

N-cadherin DAPI

103

N-cadherin

Scr

Vim F-actin DAPI

M ET

Vimentin

Snail F-actin DAPI

shCtr

Snail

shMET

a shCtr

Articles

EGFR

MET

Figure 6 MET induces epithelial-mesenchymal transition. (a) Morphological changes of Calu-1 cells after MET knockdown. Scale bar, 20 µm. (b) Immunofluorescence of Snail, vimentin and PKC-ε N-cadherin in Calu-1 shCtr and Calu-1 shMET cells. Snail expression is strong and nuclear in 103 Dicer Calu-1 shCtr cells and is weaker and cytoplasmic in Calu-1 shMET cells. Scale bars, 20 µm. 203 SRC (c) Western blots showing the fibronectin, vimentin and Snail downregulation and the upregulation miR-21 Mesenchymal-epithelial of E-cadherin after MET knockdown in Calu-1 cells. Loading control, GAPDH. (d) qRT-PCR transition showing the expression of epithelial and mesenchymal markers in Calu-1 shCtr and Calu-1 shMET 30b/c 221/222 30b/c 221/222 PTEN cells. (e) Immunofluorescence showing that fibronectin, Snail and vimentin expression decreases after miR-103 or miR-203 overexpression in Calu-1 cells. Scale bar, 20 µm. Ctr miR, control BIM APAF-1 BIM APAF-1 miRNA. (f) Immunofluorescence showing the increased E-cadherin signal after miR-103– or miR-203–enforced expression in Calu-1 cells. Scale bar, 40 µm. (g) Immunoblot showing the Gefitinib resistance Gefitinib resistance downregulation of mesenchymal markers after miR-103 or miR-203 overexpression. (h) We propose EMT a model in which MET downregulates miR-103 and miR-203, which in turn, upregulate PKC-ε, Dicer and SRC, inducing gefitinib resistance and epithelial-mesenchymal transition. MET also induces miR-30b, miR-30c, miR-221, miR-222 and miR-21 upregulation and the consequent gefitinib resistance through BIM, APAF-1 and PTEN downregulation. EGFR increases miR-221, miR-222, miR-30b and miR-30c expression. Shown in red are the upregulated miRNAs, and shown in green are the downregulated miRNAs. Results (b–g) are representative of at least four independent experiments. P values, two-tailed Student’s t test. Error bars, ± s.d.

Overexpression of miR-103 and miR-203 or knockdown of miR-221 and miR-30c resulted in marked inhibition of tumor growth and increased sensitivity to gefitinib-induced apoptosis in nude mice after 2 weeks of treatment (Fig. 4e,f and Supplementary Fig. 14a). We confirmed the downregulation of miR-221 and miR-222 and the upregulation of miR-103 and miR-203 in the xenograft tumors by qRT-PCR (Supplementary Fig. 14b). MiR-103 and miR-203 reduce NSCLC cell migration The anti-proliferative activity of miR-103 and miR-203 in various cancers has been previously reported 19–21. To further investigate the functional role of miR-103 and miR-203 in NSCLC tumorigenesis, we assessed the effects of miR-103 and miR-203 gain of function and the loss of PKC-ε and SRC on cell migration and cell cycle kinetics. Migration was reduced by about 60% compared to controls in cells with increased miR-103 and miR-203 expression or decreased PRKCE (PKC-ε) and SRC expression (Fig. 5a). We further confirmed these results using a wound-healing assay (Fig. 5b). In addition, A549 and Calu-1 cells transfected with miR-103, miR-203 or PRKCE (PKC-ε), and SRC siRNAs showed an increased G1 cell fraction and a corresponding decreased number of cells in the S and G2-M phases, with miR-203 and SRC siRNA having a slightly stronger effect as compared to miR-103 and PRKCE (PKC-ε) siRNA (Fig. 5c).

MiR-103 and miR-203 promote the mesenchymal-to-epithelial An association between the epithelial-mesenchymal transition (EMT) and the development of chemoresistance, including resistance to EGFR-targeted therapy, that leads to recurrence of disease and meta stasis has recently been reported22. Although identifying the molecular events underlying EMT is an area under intense investigation, what triggers the onset of the EMT in tumor cells is unknown. We observed a change in cellular shape of Calu-1 cells from a fibroblastoid morphology to an epithelial polarized phenotype after knockout of MET (Fig. 6a), prompting us to further investigate whether this morphological change could be a result of a mesenchymal-toepithelial transition. We assessed the expression of key EMT-associated markers and observed, in Calu-1 MET knockdown cells compared to Calu-1 Sh controls (which encode a scrambled shRNA sequence that does not lead to the specific degradation of any known cellular mRNA), decreased expression of mesenchymal markers and increased E-cadherin expression (Fig. 6b–d), strongly indicating reversion of Calu-1 cells back to an epithelial phenotype after MET knockdown. Notably, in MET-KD cells, Snail protein expression was lower than in cells without MET knockdown, was localized to the cytoplasm (Fig. 6b) and the protein itself was presumably nonfunctional23. It is notable that we did not observe a morphological change in EGFR-KD Calu-1 cells, in which miR-200c (refs. 24,25), miR-103 and miR-203 were not upregulated, as was the case in MET knockdown cells

VOLUME 18 | NUMBER 1 | JANUARY 2012 nature medicine

npg

articles (Supplementary Fig. 1a). To test whether miR-103 and miR-203 were involved in the mesenchymal-epithelial transition, we overexpressed these miRNAs in Calu-1 cells and observed downregulation of several mesenchymal markers and increased E-cadherin expression, indicating a role for these miRNAs in the mesenchymal-epithelial transition (Fig. 6e–g and Supplementary Fig. 15a,b). In addition, silencing of PRKCE (PKC-ε) and SRC in Calu-1 cells increased the amount of E-cadherin and decreased the levels of SNAIL, ZEB1 (encoding zinc finger E-box binding 1), ZEB2 (encoding zinc finger E-box binding 2), vimentin and fibronectin mRNA compared to cells transfected with a siRNA control (Supplementary Fig. 15c). Because it was recently reported that miR-103 targets Dicer26, we investigated the effects of Dicer knockdown on tumorigenesis and on gefitinib-induced apoptosis of NSCLCs. Notably, near complete Dicer knockdown reduced not only gefitinib resistance but also the migration and expression of the mesenchymal markers of NSCLC cells, suggesting that miR-103 could also be involved in the mesenchymal-epithelial transition pro cess through Dicer downregulation (Supplementary Results and Supplementary Fig. 16). In conclusion, by regulating the expression of specific miRNAs, MET orchestrates the convergence of several EMT-associated pathways, including the Dicer, SRC, PKC-ε and AKT pathways, supporting the possibility that MET targeting could be a strategy to control EMT and NSCLC progression. DISCUSSION A strong correlation between activating mutations in the EGFR tyrosine kinase domain and the response to erlotinib and gefitinib has been reported in several trials27. Different mechanisms of resistance to EGFR TKIs have also been described. The EGFR T790M mutation17, which reduces the affinity of the EGFR to gefitinib and erlotinib, and MET gene amplification produce acquired resistance to anti-EGFR agents8. The results of this study provide evidence that EGFR and MET receptor tyrosine kinases, through regulation of expression of specific miRNAs, control the metastatic behavior and gefitinib resistance of NSCLCs. In a previous study, we identified MET as a regulator of miR-221 and miR-222 expression3. To better understand the pathway(s) involved in NSCLC tumorigenesis and drug resistance, we investigated miRNAs modulated by EGFR and MET tyrosine kinases. We focused on miR-30b, miR-30c, miR-221 and miR-222, which are regulated by both EGFR and MET, and miR-103 and miR-203, which are regulated by MET only. We show that gefitinib treatment triggers programmed cell death through the downregulation of miR-30b, miR-30c, miR-221 and miR-222 and the consequent upregulation of APAF-1 and BIM in gefitinib-sensitive HCC827 and PC9 cells. We also show that gefitinib treatment does not decrease miR-30b, miR-30c, miR-221 and miR-222 expression in gefitinib-resistant Calu-1, A549 and HCC827 GR cells as a result of MET overexpression. Therefore, EGFR inhibition alone in cells overexpressing MET is not sufficient to induce the downregulation of these miRNAs and, accordingly, cell death. We show that gefitinib resistance could be overcome by MET inhibitors, which downregulate miR-30b, miR-30c, miR-221 and miR-222 and sensitize NSCLCs to gefitinib or by anti–miR-221 and anti–miR-30c, which strongly increase gefitinib sensitivity in vitro and in xenograft mouse models in vivo. Taken together, these results indicate that the modulation of specific miRNAs, such as miR-30b, miR-30c, miR-221 and miR-222, could have therapeutic applications to sensibilize lung tumors to TKI therapy. PTEN loss, by partially uncoupling mutant EGFR from downstream signaling and by activating EGFR, contributes to erlotinib resistance28. Previously, we found PTEN to be a miR-221 and miR-222

nature medicine VOLUME 18 | NUMBER 1 | JANUARY 2012

target3, making it reasonable to speculate that these two miRNAs have a role in the gefitinib resistance of NSCLC cells, not only through APAF-1 but also through PTEN regulation. Notably, overexpression of another miRNA targeting PTEN29,30, miR-21, induced gefitinib resistance in HCC827 and PC9 gefitinib-sensitive cells. We also show that miR-103 and miR-203, which are upregulated after MET silencing or treatment with the MET inhibitor SU11274, induce apoptosis in gefitinib-resistant NSCLCs, reduce mesenchymal markers and increase epithelial cell junction proteins compared to wild-type Calu-1 cells by downregulating the expression of PKC-ε, SRC and Dicer. EMT has recently been shown to have a role in acquired resistance to gefitinib in A549 cells, indicating that mesenchymal status is related to the ‘inherent resistance’ to gefitinib or erlotinib in NSCLCs31. As proposed in the model in Figure 6h, MET expression downregulates miR-103 and miR-203 and upregulates miR-221, miR-222, miR-30b and miR-30c, inducing gefitinib resistance, and epithelial-mesenchymal transition in NSCLCs. The identification of prognostic and predictive factors associated with sensitivity or resistance to anti-EGFR agents is a key area of investigation, and aberrant key signaling proteins, including RAS-MEK, AKT–mammalian target of rapamycin (mTOR) and MET kinase, have recently been recognized as key targets of investigation8. As activation or amplification of MET signaling contributes to TKI resistance through multiple independent mechanisms and leads to the rapid evolution of drug resistance, stratifying NSCLCs based on MET expression or MET-regulated miRNAs may allow for individualization of treatment. Such a strategy has the potential to increase treatment efficacy by eliminating unnecessary side effects of a particular therapeutic regimen in NSCLC patients who would not benefit from that specific regimen. In addition, our clinical validation studies on lung tumor specimens reveal that MET overexpression and the consequent absence of miR-103 and miR-203 can be used to identify primary lung tumors with metastatic capacity. Indeed, reduced expression of miR-103 and miR-203 could be predictive of more aggressive, early metastatic tumors. Finally, miRNAs combined with TKIs might provide a new strategy to treat NSCLCs in the future. Methods Methods and any associated references are available in the online version of the paper at http://www.nature.com/naturemedicine/. Note: Supplementary information is available on the Nature Medicine website. Acknowledgments We thank K. Huebner and S. Lutz for revisions to the paper and P. Fadda and S. Miller for qRT-PCR assistance. We also thank K. Sergott and Ventana Medical Systems for supplying the immunohistochemistry reagents used in this study. M. Nuovo (The Ohio State University Medical Center) kindly provided the photomicroscopy work. We are grateful for research support from The Ohio State University Targeted Investment in Excellence Award, the US National Institutes of Health grant CA113001 and the Kimmel Scholar Award (M.G.). AUTHOR CONTRIBUTIONS M.G. designed the research. M.G., G.R., G.D.L., Y.-J.J., A.N. and F.L. performed the research. G.D.L. and J.S. conducted the mouse experiments. G.N. performed the IHC and ISH experiments. H.A. performed the microarray experiments and analyses. K.P.N. and G.C. provided discussions and advice. J.A.E., M.O. and J.K.R. provided cells with acquired gefitinib resistance. S.V. and L.C. performed bioinformatics analyses. M.G. and C.M.C. wrote the paper. COMPETING FINANCIAL INTERESTS The authors declare no competing financial interests. Published online at http://www.nature.com/naturemedicine/. Reprints and permissions information is available online at http://www.nature.com/ reprints/index.html.

1. Wu, N. et al. MiR-373, a novel regulator of PPP6C, functions as an oncogene in hepatocellular carcinoma. FEBS J. 278, 2044–2054 (2011). 2. Chan, J.A., Krichevsky, A.M. & Kosik, K.S. MicroRNA-21 is an antiapoptotic factor in human glioblastoma cells. Cancer Res. 65, 6029–6033 (2005). 3. Garofalo, M. et al. MiR-221&222 regulate TRAIL resistance and enhance tumorigenicity through PTEN and TIMP3 downregulation. Cancer Cell 16, 498–509 (2009). 4. Cochrane, D.R., Spoleastra Howe, E.N., Nordeen, S.K. & Richer, J.K. MicroRNA200c mitigates invasiveness and restores sensitivity to microtubule-targeting chemotherapeutic agents. Mol. Cancer Ther. 8, 1055–1066 (2009). 5. Molina, J.R., Yang, P., Cassivi, S.D., Schild, S.E. & Adjei, A.A. Non-small cell lung cancer: epidemiology, risk factors, treatment, and survivorship. Mayo Clin. Proc. 83, 584–594 (2008). 6. Sekido, Y. Genomic abnormalities and signal transduction dysregulation in malignant mesothelioma cells. Cancer Sci. 101, 1–6 (2010). 7. Politi, K., Fan, P.D., Shen, R., Zakowski, M. & Varmus, H. Erlotinib resistance in mouse models of epidermal growth factor receptor–induced lung adenocarcinoma. Dis. Model Mech. 3, 111–119 (2010). 8. Engelman, J.A. et al. MET amplification leads to gefitinib resistance in lung cancer by activating ERBB3 signaling. Science 316, 1039–1043 (2007). 9. Benedettini, E. et al. Met activation in non-small cell lung cancer is associated with de novo resistance to EGFR inhibitors and the development of brain metastasis. Am. J. Pathol. 177, 415–423 (2010). 10. Agarwal, S. et al. Association of constitutively activated hepatocyte growth factor receptor (Met) with resistance to a dual EGFR/Her2 inhibitor in non-small-cell lung cancer cells. Br. J. Cancer 100, 941–949 (2009). 11. Stabile, L.P. et al. Targeting of both the c-Met and EGFR pathways results in additive inhibition of lung tumorigenesis in transgenic mice. Cancers (Basel). 2, 2153–2170 (2010). 12. Cragg, M.S., Kuroda, J., Puthalakath, H., Huang, D.C. & Strasser, A. Gefitinibinduced killing of NSCLC cell lines expressing mutant EGFR requires BIM and can be enhanced by BH3 mimetics. PLoS Med. 4, 1681–1689 (2007). 13. Deng, W.G., Knon, J., Ekmekcioglu, S., Poindexter, N.J. & Grimm, E.A. IL-24 gene transfer sensitizes melanoma cells to erlotinib through modulation of the APAF-1 and Akt signaling pathways. Melanoma Res. 21, 44–56 (2010). 14. Qin, B. et al. Activated Src and Ras induce gefitinib resistance by activation of signaling pathways downstream of epidermal growth factor receptor in human gallbladder adenocarcinoma cells. Cancer Chemother. Pharmacol. 58, 577–584 (2006).

15. Weisheit, S., Schafer, C., Mertens, C., Berndt, A. & Liebmann, C. PKCε acts as negative allosteric modulator of EGF receptor signaling. Cell. Signal. 23, 436–448 (2011). 16. Turke, A.B. et al. Preexistence and clonal selection of MET amplification in EGFR mutant NSCLC. Cancer Cell 17, 77–88 (2010). 17. Yoshida, T. et al. Effects of Src inhibitors on cell growth and epidermal growth factor receptor and MET signaling in gefitinib-resistant non-small cell lung cancer cells with acquired MET amplification. Cancer Sci. 101, 167–172 (2010). 18. Seike, M. et al. MiR-21 is an EGFR-regulated anti-apoptotic factor in lung cancer in never-smokers. Proc. Natl. Acad. Sci. USA 106, 12085–12090 (2009). 19. Lee, K.H. et al. Epigenetic silencing of microRNA miR-107 regulates cyclin-dependent kinase 6 expression in pancreatic cancer. Pancreatology 9, 293–301 (2009). 20. Viticchiè, G. et al. MiR-203 controls proliferation, migration and invasive potential of prostate cancer cell lines. Cell Cycle 10, 1121–1131 (2011). 21. Yuan, Y. et al. MicroRNA-203 inhibits cell proliferation by repressing ∆Np63 expression in human esophageal squamous cell carcinoma. BMC Cancer 7, 11–57 (2011). 22. Sabbah, M. et al. Molecular signature and therapeutic perspective of the epithelialto-mesenchymal transitions in epithelial cancers. Drug Resist. Updat. 11, 123–151 (2008). 23. Wu, Y. & Zhou, B.P. Snail: more than EMT. Cell Adhes. Migr. 4, 199–203 (2010). 24. Wellner, U. et al. The EMT-activator ZEB1 promotes tumorigenicity by repressing stemness-inhibiting microRNAs. Nat. Cell Biol. 11, 1487–1495 (2009). 25. Chang, C.J. et al. p53 regulates epithelial-mesenchymal transition and stem cell properties through modulating miRNAs. Nat. Cell Biol. 13, 317–323 (2011). 26. Martello, G. et al. A microRNA targeting dicer for metastasis control. Cell 141, 1195–1207 (2010). 27. Maemondo, M. et al. Gefitinib or chemotherapy for non-small-cell lung cancer with mutated EGFR. N. Engl. J. Med. 25, 2380–2388 (2010). 28. Sos, M.L. et al. PTEN loss contributes to erlotinib resistance in EGFR-mutant lung cancer by activation of Akt and EGFR. Cancer Res. 15, 3256–3261 (2009). 29. Meng, F. et al. MicroRNA-21 regulates expression of the PTEN tumor suppressor gene in human hepatocellular cancer. Gastroenterology 133, 647–658 (2007). 30. Zhang, J.G. et al. MicroRNA-21 (miR-21) represses tumor suppressor PTEN and promotes growth and invasion in non-small cell lung cancer (NSCLC). Clin. Chim. Acta 411, 846–852 (2010). 31. Rho, J.K. et al. Epithelial to mesenchymal transition derived from repeated exposure to gefitinib determines the sensitivity to EGFR inhibitors in A549, a non-small cell lung cancer cell line. Lung Cancer 63, 219–226 (2009).

npg

Articles

VOLUME 18 | NUMBER 1 | JANUARY 2012 nature medicine

ONLINE METHODS

In vivo experiments. A549 cells were stably infected with a control miRNA, miR-103 and miR-203 or with control inhibitor of miRNA or a lentiviral inhibitor of miR-221 and miR-30c (SBI). We injected 5 × 106 viable cells subcutaneously into the right flanks of 6-week-old male nude mice (Charles River Breeding Laboratories). Treatment was started 7 d after tumor cell inoculation. Gefitinib was administered Monday through Friday for 2 weeks as an oral gavage at concentrations of 200 mg per kg of body weight in 1% Tween 80 (Sigma) in sterile Milli-Q water (the vehicle control was 0.5% Tween 80 in sterile Milli-Q water). Tumor size was assessed twice per week using a digital caliper. Tumor volumes were determined by measuring the length (l) and the width (w) of the tumor and calculating the volume (V = lw2/2). We killed the mice 35 days after injection. Statistical significance between the control and treated mice was evaluated using a Student’s t test. Mouse experiments were conducted after approval by the institutional animal care and use committee at Ohio State University. Migration assay. Transwell insert chambers with an 8-µm porous membrane (Greiner Bio One) were used for the assay. Cells were washed three times with PBS and added to the top chamber in serum-free medium. The bottom chamber was filled with medium containing 10% FBS. Cells were incubated for 24 h at 37 °C in a 5% CO2 humidified incubator. To quantify migrating cells, cells in the top chamber were removed by using a cotton-tipped swab, and the migrated cells were fixed in PBS, 25% glutaraldehyde and stained with crystal violet stain,

visualized under a phase-contrast microscope and photographed. Crystalviolet–stained cells were then solubilized in acetic acid and methanol (1:1), and absorbance was measured at 595 nm. Immunofluorescence. Cells were grown on Lab-Tek II CC2 chamber slides (Nunc), fixed with 4% paraformaldehyde and permeabilized with 0.2% Triton X-100/PBS before blocking with 10% sheep serum (Caltag Laboratories). All the primary antibodies were from Abcam. Secondary antibodies were goat antibodies to mouse or rabbit coupled to Alexa 488 (Invitrogen). F-actin was stained by using a phalloidin reagent (Invitrogen). Cell nuclei were visualized with DAPI (Sigma). Slides were mounted with SlowFade Gold Antifade reagent (Invitrogen). Cell death and cell proliferation quantification. For detection of caspase3/7 activity, cells were cultured in 96-well plates, in triplicate, treated with 5 µM, 10 µM or 15 µM gefitinib and analyzed using a Caspase-Glo 3/7 Assay kit (Promega) according to the manufacturer’s instructions. Continuous variables are expressed as means ± s.d. Cell viability was examined with 3-(4,5dimethylthiazol-2-yl)-2,5-dipheniltetrazolium bromide (MTS)-Cell Titer 96 AQueous One Solution Cell Proliferation Assay (Promega) according to the manufacturer’s protocol. Metabolically active cells were detected by adding 20 µl of MTS to each well. After 1 h of incubation, the plates were analyzed in a Multilabel Counter (Bio-Rad Laboratories). Statistical analyses. Student’s t tests, one-way analysis of variance and Fisher’s exact tests were used to determine statistical significance. A Pearson correlation coefficient was calculated to test the inverse relation between miR-103, miR-203, miR-221, miR-222, miR-30b and miR-30c and their putative targets, and between MET and miR-103 and miR-203. Statistical significance for all the tests, assessed by calculating the P values, was defined as P < 0.05. Additional methods. Detailed methodology is described in the Supplementary Methods.

npg

TaqMan Array MicroRNA Cards. For RNA extraction, see the Supplementary Methods. The TaqMan Array Human MicroRNA Card (Applied Biosystem) Set v3.0 is a two-card set containing a total of 384 TaqMan MicroRNA Assays per card that enables accurate quantification of 754 human miRNAs. Included on each array are three TaqMan MicroRNA Assays as endogenous controls to aid in data normalization and one TaqMan MicroRNA Assay not related to human as a negative control. An additional preamplification step was enabled by using Megaplex PreAmp Primers, Human Pool Set v3.0 for situations where sensitivity is of the utmost importance or where the sample is limiting.

doi:10.1038/nm.2577

nature medicine