Cetuximab Attenuates Metastasis and u-PAR Expression in Non–Small Cell Lung Cancer: u-PAR and E-Cadherin are Novel Biomarkers of Cetuximab Sensitivity Free

Lung cancer is one of the most prevalent cancers (1, 2), non–small cell lung cancer (NSCLC) accounting for 85% of cases (3). Epidermal growth factor (EGF) receptor (EGFR) is overexpressed in NSCLC (4) and correlates with tumor growth, metastasis, and poor prognosis (5). EGFR is the archetypal member of a superfamily of cell membrane receptors with intrinsic tyrosine kinase activity. The binding of EGFR ligands including EGF, transforming growth factor-α, and heparin-binding EGF initiates receptor dimerization, autophosphorylation of tyrosine residues within the cytoplasmic domain, and recruitment of adaptor proteins, activating signaling molecules downstream (6). Biological responses to EGFR activation include proliferation, differentiation, motility, and metabolism (7).

This implies a strong rationale for EGFR-targeting agents, such as Cetuximab (Erbitux), which has given promising results in colorectal and lung cancer (8, 9). Cetuximab is a chimeric mouse-human antibody targeting the extracellular domain of EGFR. Cetuximab inhibits the binding of activating ligands to the receptor, preventing ligand-dependent activation, and impedes downstream pathways, resulting in inhibition of cell cycle progression, growth, and angiogenesis (10, 11). In addition, the binding of Cetuximab initiates EGFR internalization and degradation (12). Although many in vitro studies have shown efficacy of Cetuximab in inhibiting cell growth and EGF-mediated signaling, studies investigating its effect on different steps of the metastatic cascade are rare. Moreover, the response rate to Cetuximab in the clinical setting, particularly as a monotherapy, despite promising results in colorectal cancer (8), is not clear yet in NSCLC (13). One possible reason is the lack of reliable biomarkers to identify patients who would best respond to Cetuximab. Interestingly, data on EGFR as an indicator of Cetuximab response are not convincing (9, 13).

The urokinase-type plasminogen activator (u-PA) and its receptor (u-PAR) are essential for proteolysis, migration, invasion, and metastasis (14). Activation of pro-u-PA leads to plasmin-dependent proteolysis of the extracellular matrix. u-PAR is often localized to focal adhesions, which modifies cell attachment, and stimulates migration (15). Extensive clinical studies have shown that expression of u-PAR, u-PA, and its specific inhibitor plasminogen activator inhibitor-1, are associated with tumor recurrence and poor survival (16). u-PAR has shown to interact with, or stimulate, a variety of signal transduction molecules, including integrins (17), tyrosine kinases (18), and serine/threonine kinases (19). EGFR has been shown to be a transducer of signals initiated by ligand-activated u-PAR, leading to cell migration (20). In addition, EGFR mediates an up-regulation of the u-PA system via mitogen-activated protein (MAP)/extracellular signal-regulated kinase (ERK) kinase/ERK pathway (21).

The present study was conducted to (a) determine whether u-PAR is regulated by Cetuximab treatment in NSCLC, and to identify molecular mechanisms mediating the inhibitory function of Cetuximab; (b) to determine whether Cetuximab inhibits different steps of metastasis in vitro and in vivo; (c) to determine whether molecular markers of metastasis, such as u-PAR, are indicators of Cetuximab resistance or sensitivity.

Our study reveals that inhibition of EGFR by Cetuximab decreases proliferation, invasion, and metastasis to distant organs in vivo. Furthermore, we show that Cetuximab decreases the binding of c-Jun and JunD to the u-PAR promoter, this leading to reduced u-PAR gene expression and invasion. Finally, we suggest E-cadherin and u-PAR as novel markers for Cetuximab sensitivity, a finding that could be applied to ongoing clinical trials on Cetuximab in NSCLC patients.

Materials, antibodies, cell lines, plasmids. Media/fetal bovine serum (FBS) were purchased from Invitrogen/Life Technologies (Karlsruhe/Germany) and Sigma (Taufkirchen/Germany). The following antibodies were from Santa Cruz: c-Jun (sc-822×), JunD (sc-74×), Fra-1 (sc-22794×), Fra-2 (sc-604×), c-Fos (sc-52×), FosB (sc-7203×), b-actin (sc-1616-R), unspecific IgG, and E-cadherin (sc-8426); EGFR (2232) was from Cell Signalling. The E-cadherin plasmid was kindly provided by Prof. B. Gumbiner, University of Virginia, Charlottesville, VA. EGFR wt and mutant plasmids were gifts from Prof. J. Settleman, Harvard Medical School, Charlestown, MA. Oligonucleotides were supplied by Metabion. Transwell chambers (1 cm²/12 μm) were from Costar. Matrigel was purchased from BD Biosciences. All cell lines were obtained from American Type Culture Collection. Fertilized specific pathogen-free eggs were from Charles River.

3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay. Two thousand cells were seeded in a 96-well plate in 100-μL 10% FBS medium. Twenty-four hours later, Cetuximab was applied at concentrations between 0.3 to 4 μmol/L. The growth inhibitory effect was evaluated using 20 μL per well CellTiter96 AQueousOneSolution (Promega). Absorbance (490 nm) was measured using an automated reader.

Matrigel invasion. A549 cells (1 × 10⁵), starved in serum-free medium (24 h), were plated on transwell chambers precoated with 20 μg Matrigel. Cetuximab (4 μmol/L) and EGF (100 nmol/L) were applied to bottom and top chambers. Ten percent FBS medium in the lower chamber served as chemoattractant. After 24 h, noninvading cells were removed with cotton swabs. Invaded cells were trypsinized and counted using the ATP-luminescence–based motility-invasion assay (22).

ELISA. Cells were washed with PBS and lysed in extraction-buffer (Biosource). Protein concentration was determined by bicinchoninic acid (Pierce). u-PAR protein was assayed using the Imubind-u-PAR-ELISA kit (AmericanDiagnostica).

Reporter assay. Cells were transiently transfected with pGL3-basic, −395/+51 u-PAR-reporter plasmid, using Effectene (Qiagen). Reporter assays were performed using the Dual-luciferase assay-system (Promega), normalized for transfection efficiency by cotransfected Renilla-luciferase.

Real-time PCR. Total RNA was isolated with RNeasy-kit (Qiagen), 1 μg of total RNA was reverse transcribed by random hexamer primer, using SuperScript-II reverse transcriptase (Invitrogen). u-PAR and EGFR-mRNA were quantified by TaqMan qRT-PCR with specific primer probes (Applied Biosystems) and normalized to β-Actin (Human ACTB) using the 2^−ΔΔCT method. EGFR gene copy number was determined as described (23). All qPCRs were performed in triplicate.

Nuclear extracts, electrophoretic mobility shift assay, supershift, and chromatin immunoprecipitation assay. Nuclear extracts, electrophoretic mobility shift assays, and Supershift using u-PAR promoter sequence GTGATCACAACTCCATGAGTCAGGGCCGAG-3′ including activator protein (AP-1; −190/−171), and chromatin immunoprecipitation (ChIP) analyses were performed as described (22). Briefly, cells were plated and starved for 48 h. After Cetuximab pretreatment (4 h) and 1-h EGF stimulation, cells were fixed (1%formaldehyde/10 min), lysed, and the genomic DNA sheared to 300 bp-1.5 kb. ChIP with 2 μg specific (p-c-Jun, JunD) and nonspecific IgG was done overnight using aliquots of precleared lysates. The primer set for final PCR was used as described (22). A control primer set amplifying a region 2,000 bp upstream of the u-PAR transcription start site, lacking AP-1site, was applied for negative control amplification and did not show enrichment.

Reverse phase protein array. Cells were lysed with reverse phase protein array (RPPA) lysis buffer [150 mmol/L NaCl, 50 mmol/L HEPES (pH 7.4), 1.5 mmol/L MgCl₂, 1 mmol/L EGTA, 100 mmol/L NaF, 10 mmol/L NaPPi, 10%glycerol, 1%TritonX100, Complete Protease Inhibitor tablet (Roche), 1 mmol/L Na₃VO₄] and lysates were prepared and serially diluted for slide printing as previously described (24). Slides were stained with validated primary antibodies, and the signal was amplified using a DakoCytomation-catalyzed system. A secondary antibody (anti-mouse or anti-rabbit) was used as a starting point for amplification of the signal. For signal detection, diaminobenzidine was incubated for 2 min. Slides were scanned on HP-Scanjet-8200. The intensity of each spot was determined by the MicroVigene software (Vigene Tech). Protein amounts were determined by calculating the EC₅₀ of individual dilution curves for each sample and interpolating each sample from a supercurve constructed for each protein in a script written in R (25). Samples were adjusted for loading by expanding upon a method similar to that of Nishizuka and colleagues, in which all proteins were used for adjustment (26). Normalized values were mean centered for supervised and unsupervised hierarchical clustering. To generate heat maps, the Treeview (Stanford University) and X-cluster software were used (24).

Chicken embryo metastasis assay. This was performed as described (27), with some modifications: 2.0 × 10⁶ cells were placed on the chorioallantoic membrane (CAM) of 10-d-old chicken embryos. Eggs were incubated for another 8 d, after which extraembryonic tumors were excised/weighed. Metastasis was determined by harvesting lungs and liver, and processing the tissue for human DNA by quantitative alu-PCR as described (28). Antibodies were directly applied or i.v. injected on day 11/14/16 of embryonic development.

Patient tumors and immunohistochemistry. Paraffin-embedded specimens from the first 25 NSCLC-patients of the Gemtax-IV trial were collected after informed consent and verification by a pathologist. Gemtax-IV is an ongoing phaseII/III trial on gemcitabine-based chemotherapy combined with Cetuximab in stage III/IV-NSCLC patients. Tissues were processed and analyzed for E-cadherin, u-PAR, EGFR by immunohistochemistry as described (29). Slides were scored blinded by a pathologist. Scores 0 to 3 were assigned according to the percentage of positive cells (0, 0%; 1, <25%; 2, 25–50%; 3, >50%) and staining intensity (0, 0; 1, 1+; 2, 2+; 3, 3+). Both scores were multiplied to give an overall score of 0 to 9. The overall score was assigned to one of the following groups: 0, negative; 1 to 2, weak; 3 to 6, moderate; 8 to 9, strong expression.

Statistical analysis. This was performed using StatView5.0 (Macintosh/Windows). Values were expressed as mean ± SD. Associations between groups were calculated with the Student's t test unless otherwise indicated (Pearson's correlation or Fisher's exact test). P value of ≤0.05 was defined as significant, 0.05 < P ≤ 0.1 as a trend.

Effects of Cetuximab on the growth and invasion of human NSCLC cells. Seven human NSCLC cell lines were analyzed for their sensitivity to Cetuximab (Fig. 1A). H1395 and Calu3 cells (adenocarcinomas) displayed the highest susceptibility to growth inhibition by 4 μmol/L Cetuximab. A moderate response was observed in A427 and A549 (squamous carcinoma) cells, whereas LXF289, H1299 (adenocarcinoma), and H460 (large cell type) cells showed no change in proliferation. All seven cell lines express wt-EGFR, their protein levels did not differ/change drastically, and EGFR-protein expression or changes did not correlate to Cetuximab response. We did not find any correlation between EGFR gene copy number, or mRNA, or introduced EGFR-mutations (see Discussion), to drug response (Fig. 1B).

Next, to determine whether Cetuximab affects the invasion of NSCLC cells in vitro, we evaluated the invasive capacity of A549 cells (Transwell in vitro invasion assay). Cetuximab/EGF-treated cells displayed a significantly (P = 0.03) reduced ability to invade Matrigel, compared with EGF- and control-untreated cells (Fig. 1C). Reduced invasion was detected after 24-hour Cetuximab treatment, in contrast to reduced proliferation, which was detected after a longer incubation period of the same Cetuximab concentration (6 days). These results indicate that Cetuximab inhibits proliferation and invasion independently in NSCLC, suggesting that genes involved in invasion are the initial targets, followed by proproliferative genes.

The altered invasion of lung cancer cells after Cetuximab treatment prompted the question as to the potential target molecules in this context. Jo and colleagues (30) suggested that EGFR and u-PAR are engaged in the same multiprotein assembly on the cell surface. Moreover, it is known that EGF stimulates u-PAR gene expression (31). Therefore, we performed Matrigel invasion with both scrambled and si-u-PAR–transfected A549 cells. Knockdown of u-PAR in A549 severely impaired EGF-induced invasion. This suggests that u-PAR is required for optimal invasion in cultured NSCLC (Fig. 1D).

Cetuximab treatment inhibits EGF-induced u-PAR gene expression. Because u-PAR was seen to be essential for basal and EGF-induced NSCLC-invasion, and EGF is reported to stimulate u-PAR gene expression (30, 31), we sought to investigate whether EGFR blockade by Cetuximab counters promoter activity and expression of u-PAR. Luciferase reporter assays were performed on H1395 cells by transfecting a reporter plasmid containing the −395/+51 region of the u-PAR promoter, and treating with EGF, Cetuximab/EGF, or Cetuximab alone. As shown in Fig. 2A, the higher activity of the u-PAR promoter in EGF-treated samples was significantly inhibited (P = 0.001) in Cetuximab-pretreated cells to levels similar to non–EGF-stimulated cells. Furthermore, a ∼3-fold increase in u-PAR mRNA was detected after 12/24-h EGF stimulation, which was efficiently down-regulated in cells pretreated with Cetuximab (Fig. 2B). Also, significantly lower u-PAR protein (ELISA) was detected in Cetuximab treated samples (Fig. 2B,, bottom). Another Cetuximab-sensitive cell line, A549, showed similar results for u-PAR mRNA and protein (Fig. 2C). In contrast, the resistant H1299 and H460 cells did not respond to EGF stimulation and/or Cetuximab regarding u-PAR (Supplementary Fig. S1). These observations show the ability of Cetuximab to inhibit the EGF-induced transcriptional up-regulation of u-PAR gene expression in drug-sensitive NSCLC cell lines.

Downstream EGFR pathways affected in Cetuximab-treated NSCLC cells. To delineate signaling mediators of Cetuximab, we applied the RPPA method to implicate the activity of molecules downstream of EGFR after Cetuximab inhibition. RPPA-based quantification of protein phosphorylation in starved, EGF-stimulated, and drug-treated cells, respectively, revealed that the phosphorylation of molecules of the MAP kinase (MAPK) pathway were decreased in Cetuximab pretreated samples, compared with untreated and EGF-stimulated cells (Fig. 3A). Inhibition was strongly pronounced in Cetuximab-sensitive H1395, and less in Cetuximab-resistant H460 cells (Fig. 3B). Also, the phosphorylation of molecules such as HER2 (Fig. 3C), another ErbB family member, and S6 kinase (Fig. 3D), which is involved in protein synthesis, was strongly inhibited after Cetuximab treatment, and the pattern of inhibition was related to drug sensitivity. This analysis suggested that MEK-MAPK is the major pathway affected after Cetuximab treatment because no significant changes were detected in the phosphorylation of the proximal portion of the PI3K-Akt pathway. Although pS6 is frequently downstream of the PI3K/Akt, the RAS, MAPK, p90RSK cascade can increase pS6 through effects on the TSC2-tumor suppressor gene.

Cetuximab inhibits EGF-activated AP-1 (c-Jun and JunD) binding to the u-PAR promoter in vitro and in vivo. MAPKs, specifically ERK, c-Jun-NH₂-kinase, and c-Fos–regulating kinase, regulate the activity of AP-1 transcription factors by modulating both their expression and transactivating capacity (32). The proximal −395/+51 region of the u-PAR promoter contains three major cis-elements required for an induction of the promoter upon various stimuli, an AP-1 consensus motif at −190/−171, region −152/−135 bound with Sp1/Sp3/an AP2α-like factor, and a proximal nuclear factor-κB (NF-κB) region (−52/−23; refs. 22, 33). As a possible consequence of inhibition of MEK/MAPK signaling, we analyzed which of these transcription factors and binding sites are mediating EGF-induced, and Cetuximab-inhibited, u-PAR gene expression, using electrophoretic mobility shift assay and supershift analysis. As shown in Fig. 4A, mutation of the AP-1 (−190/−171) site within the u-PAR promoter construct abolished EGF-induced reporter activity (P = 0.002). Electrophoretic mobility shift assay analyzes revealed higher AP-1 binding in EGF-stimulated samples, whereas little change was detected in Sp1/Sp3/AP2α-like and NF-κB complexes binding to the other u-PAR promoter motifs (data not shown). Furthermore, increased binding of AP-1 was evident in EGF-stimulated samples, followed by a prominent decrease of binding in Cetuximab pretreated samples (Fig. 4B).

To investigate the AP-1 family members affected in binding to this promoter region by Cetuximab, we performed supershift assays using anti–p-c-Jun, JunB, JunD, c-Fos, FosB, Fra-1, and Fra-2 antibodies (normal IgG as a control). As shown in Fig. 4C, c-Jun and JunD were identified as the major components of AP-1 family complexes bound to this region upon EGF-stimulation, and inhibited by Cetuximab. Additionally, we performed ChIP with c-Jun and JunD antibodies to analyze binding to the endogenous u-PAR promoter in vivo, in mock-, EGF-, or Cetuximab/EGF-treated cells. Quantitative PCR amplification of the immunoprecipitated DNA using primers specific for region −190/−171 showed a clear increase of c-Jun and JunD binding in EGF-treated samples (Fig. 4D). As compared with electrophoretic mobility shift assay, JunD-binding was more prominent than c-Jun in the EGF-stimulated samples. However, Cetuximab pretreatment abolished binding of both transcription factors in vivo (Fig. 4D). Taken together, these data indicate that EGF-induced u-PAR gene expression is mediated by AP-1, specifically c-Jun and especially Jun-D, which can be abolished by Cetuximab treatment.

Cetuximab reduces the potential of NSCLC cells to proliferate and metastasize in vivo. We used the chicken embryo chorioallantoic membrane (CAM) model to study the effect of Cetuximab on the tumorigenicity of NSCLC cells in vivo, and their ability to metastasize to distant organs. For testing tumorigenicity, H1395 cells were inoculated on the CAM of 10-day-old chicken embryos, and 100 μg per egg Cetuximab, or equal amounts of control-IgG, were added on top of the CAM, thrice during 8-day incubation. All Cetuximab-treated embryos were healthy at the end of the experiment. We detected a clear reduction in tumor size (P = 0.095) formed by H1395 cells on the upper CAM after Cetuximab treatment, in contrast to IgG-treated eggs (Fig. 5A). Similar to our in vitro results, we detected a significant reduction in u-PAR protein in vivo in drug- versus IgG-treated tumors (P = 0.006; Fig. 5B). To determine the effect on the metastatic capacity of these cells by Cetuximab, lungs and livers were harvested from the embryos, and the metastasized cells detected by real-time alu-PCR. We failed to detect any metastasized cells in Cetuximab- and IgG-treated samples (data not shown), which could be due to the inability of H1395 cells to metastasize in this particular model. Therefore, we allowed A549 cells to grow on the CAM of 10-day-old chicken embryos for 2 days, then applied Cetuximab (100 μg per egg) or IgG systemically by i.v. injection thrice during an 8-day incubation period. We observed a significant decrease in metastasized A549 cells in the Cetuximab-treated group. In 7 of 10 IgG-treated eggs, a range of 92 to 6 cells (29 ± 33) in the liver, and in 8 of 10 IgG-treated eggs, a range of 51 to 5 cells (12 ± 15) in lungs were detected (Fig. 5C and D), whereas in 11 Cetuximab-treated samples, only 2 contained metastasized human cells in liver (P = 0.029) and 3 in lung (73, 7 and 5, 55, 14 cells, respectively; P = 0.029), indicating that metastasis of A549 cells is significantly impaired after drug treatment in vivo. These observations show the ability of Cetuximab to block primary tumor development, and show for first time that Cetuximab inhibits metastasis to distant organs of NSCLC cells in vivo.

E-cadherin and u-PAR: potential markers of Cetuximab sensitivity. Finally, we attempted to identify invasion-related molecules potentially predicting Cetuximab response or resistance. E-cadherin cell adhesion protein is an important regulator of morphogenesis and tissue regeneration in normal epithelium, and a decrease in E-cadherin supports the metastasis of cancer cells (34). These data prompted us to ask whether E-cadherin expression could be a marker of sensitivity to Cetuximab in NSCLC. Basal expression of E-cadherin was analyzed by Western Blot in all NSCLC cell lines used in this study (Fig. 6A). We observed significantly higher E-cadherin expression in cell lines sensitive to Cetuximab, compared with resistant cell lines (P = 0.04). Furthermore, in Cetuximab-treated cells, we observed an increase of E-cadherin in RPPA (Fig. 3A).

Next, we asked whether overexpression of E-cadherin would resensitize resistant cells to Cetuximab. H460-resistant cells were stably transfected to overexpress E-cadherin, increased E-cadherin expression in transfected cells being confirmed by immunofluorescence (Supplementary Fig. S2). Clone-E3 displayed the highest E-cadherin expression after stable transfection, this clone showing a significantly improved response to the drug in terms of growth inhibition compared with vector control (P = 0.01; Fig. 6B). This analysis revealed that E-cadherin re-expression in H460 cells improves response to Cetuximab.

Moreover, significantly higher amounts of u-PAR protein were detected in resistant compared with sensitive cell lines (P = 0.02; Fig. 6C,, bottom), implicating u-PAR as another marker of drug sensitivity. In addition, u-PAR mRNA in these cells showed a similar pattern (Fig. 6C,, top). To determine whether decreasing u-PAR in Cetuximab-resistant cells would alter their sensitivity toward Cetuximab, H1299 and LXF289 cells, being resistant to Cetuximab, were transfected with siRNA against u-PAR. Efficient knockdown of u-PAR in these cells is shown in Supplementary Fig. S3. We observed a significant (P = 0.001/P < 0.001 for H1299/LXF289) increase in Cetuximab-induced growth inhibition after si-u-PAR treatment, compared with cells transfected with scrambled siRNA (Fig. 6D). This result for the first time implicates u-PAR as a marker for Cetuximab sensitivity.

To support this hypothesis in vivo, we stained and evaluated u-PAR and E-cadherin protein, as well as EGFR, in 25 NSCLC patients who had undergone chemotherapeutic treatment (similar in all patients) in combination with Cetuximab in an ongoing clinical study (Gemtax-IV). After staining for E-cadherin, u-PAR, and EGFR, semiquantitative immunohistochemistry scores were compared with the clinical response of the patients. As a result, 50% (3 of 6) of the patients with partial response showed high E-cadherin together with moderate/low u-PAR gene expression. In contrast, 63% (5 of 8), a majority of the patients with progressive disease, were characterized by low E-cadherin and high u-PAR protein (Supplementary Fig. S4). Patients with stable disease showed a mixed situation, the majority (55%; 6 of 11) staining moderately or high for the favorable E-cadherin, and also moderately or high for the unfavorable u-PAR. Due to the low case numbers, statistical significance could not be obtained. Nevertheless, data do suggest that therapy response of patients is more likely in cases with high E-cadherin/low u-PAR, and that progression under therapy is more common in patients with low E-cadherin/high u-PAR. EGFR staining did not show an association with drug response, which is in agreement with our observations on the cell lines.

Our present findings explain mechanisms contributing to Cetuximab-induced suppression of cell proliferation and invasion/metastasis-related processes in NSCLC. We report the novel finding that Cetuximab reduces the potential of NSCLC to metastasize to distant organs in vivo. We further show the ability of Cetuximab to inhibit EGFR-downstream signaling via especially MAPK and AP-1, leading to the down-regulation of u-PAR gene expression, which is one of the most relevant molecules for invasion and metastasis. Finally, for NSCLC, we suggest E-cadherin and, for the first time, u-PAR to be surrogate markers for Cetuximab-sensitivity, which might advance to clinical predictors of response to this drug in NSCLC, after further validation in larger patient sets.

In agreement with previous studies, our initial observations have shown that Cetuximab reduces both the growth and invasion of lung cancer cells (10, 11). Some authors reported that EGFR can induce the u-PA/u-PAR/integrin/fibronectin induced growth pathway (20, 21). Nicholl and colleagues (35) reported that blocking EGFR inhibits both u-PA–dependent cell proliferation and migration. In extension of this, our results revealed that in NSCLC Cetuximab inhibits EGFR activity, in particular by reducing extracellular signal regulated kinase ERK(1/2)-phosphorylation, this being associated with low u-PAR- mRNA and protein, and silencing of MEK/MAPK signaling.

In the context of proliferation/migration, it is well-known that MAPKs are essential to phosphorylate and activate AP-1 family members (32, 36). In our observations, Cetuximab treatment inhibited the EGF-induced phosphorylation of MEK and ERK1/2 in the sensitive cell line H1395, followed by reduced binding of AP-1 transcription factors to the corresponding cis-element in the u-PAR-basal promoter. In glioma cells, it has been shown earlier that this AP-1 site is important for the up-regulation of u-PAR gene transcription in response to phorbol 12-myristate 13-acetate, and Fra-1, c-Fos, c-Jun, and JunD are the specific AP-1 family members with elevated binding to the u-PAR promoter (37). For NSCLC, we observed that EGF stimulation recruited especially c-Jun and JunD to motif −190/−171 of the u-PAR promoter as determined by supershift and ChIP assays. Furthermore, we observed significantly altered binding of c-Jun and JunD to the endogenous u-PAR promoter upon Cetuximab/EGF treatment. Interestingly, there was a difference in binding to the u-PAR promoter corresponding oligonucleotide (supershift) and the endogenous promoter (ChIP): JunD binding was relatively modest in vitro, but there was a highly significant binding affinity in vivo. In contrast, p-c-Jun binding was higher in vitro, and modest in vivo. This discrepancy might be due to technical differences between the assays, or differences in the complex interplay of factors between the in vitro and in vivo situation. Based on the results seen for the endogenous promoter (ChIP), JunD seems to be the major component bound to u-PAR-motif −190/−171 in vivo, in NSCLC cells. In support of this observation, Eickelberg and colleagues (38) showed that JunD is the main AP-1 family member mediating many biological phenomena, including growth factor-induced proliferation of lung fibroblasts.

The mutual regulation of EGFR/u-PAR and their participation in proliferation and metastasis (20, 21) led us to examine the potential of Cetuximab to inhibit growth and metastasis of NSCLC cells in vivo. Our results unveiled that upon Cetuximab-treatment tumors formed from non-metastatic H1395 cells were considerably smaller than the tumors in the IgG control group, this being paralleled by reduced amounts of u-PAR in the same tumors. In highly metastatic A549 cells, metastasis was clearly suppressed by Cetuximab in vivo. Thus, Cetuximab adds to the compounds able to inhibit primary tumor growth and metastasis. For example, Papo and colleagues (39) observed that host defense–derived cytolytic cationic polypeptides kill cancer cells via cytoplasmic membrane depolarization, and the resulting membrane disruption can provide a therapeutic means of inhibiting both tumor growth and metastasis of various cancers.

Some reports have highlighted EGFR mutations as playing a role in predicting response to oral tyrosine kinase inhibitors of EGFR (40, 41). It is still unclear whether EGFR mutations or EGFR overexpression play a role in Cetuximab sensitivity. There are contradictory data, some studies correlating EGFR mutations to sensitivity (42, 43), some showing that activating mutations do not predict response to Cetuximab monotherapy (8). Furthermore, several authors failed to identify EGFR overexpression as a reliable indicator for drug response (9, 44). In our present study, we overexpressed wt and mutant (deletion in 747_753insS and mutation in L858R) EGFR in H460 cells, known for a heterozygous deletion of one of its EGFR allele (45), and did not observe any correlation of these variables with Cetuximab sensitivity (data not shown). Furthermore, we did not detect an association of EGFR amplification with sensitivity, although Hirsch and colleagues (46) in a first clinical study observed an association of EGFR amplification by fluorescence in situ hybridization with prognosis in NSCLC. This might be due to either differences between in vitro and in vivo situations, or in methodology. Nevertheless, our data support the notions that (a) measuring EGFR by itself may not be sufficient to implicate Cetuximab sensitivity, and (b) that Cetuximab might be a valuable therapeutic strategy independent of EGFR mutations. Recently, a novel mechanism of interaction of EGFR and an active-type glucose transporter family-sodium/glucose cotransporter was shown to prevent autophagic death of tumor cells (47). It will be important to determine if Cetuximab is able to interfere with this interaction as well, and to induce autophagy in tumor cells. This might be an additional mechanism by which Cetuximab could function as a therapeutic tool, independent of EGFR mutations, or of the level of EGFR expression.

Recently, Moustafa and colleagues (48) showed that blocking EGFR by an antibody (LA1) resulted in cell differentiation and up-regulation of E-cadherin in invasive lung tumors. Our data support this observation because higher levels of E-cadherin were measured by RPPA in drug-treated samples in sensitive cells (Fig. 3A). Even more, we were able to improve drug response when introducing E-cadherin into resistant, E-cadherin–low H460 cells. Our results are consistent with reports of Black and colleagues (49) who observed that silencing E-cadherin in urothelial carcinoma cells significantly reduced responsiveness to Cetuximab. Similarly, Witta and colleagues (50) observed that NSCLC cell lines sensitive to EGFR-tyrosine kinase inhibitors (MS-275) express E-cadherin, whereas resistant lines do not. Here, also from resected NSCLC tumors we report that E-cadherin protein might become a clinically important marker for predicting Cetuximab responsiveness, given further validation.

This is the very first study to implicate u-PAR as a predictor of Cetuximab resistance (Fig. 6C). We showed that total basal u-PAR protein expression is significantly associated with sensitivity of NSCLC cell lines to Cetuximab. Interestingly, the amount of u-PAR analyzed separately on the cell surface did not show a clear correlation to drug resistance (Supplementary Fig. S5), implicating that drug response may not depend on a specific membrane expression or interaction, or that different molecular conditions of membrane-bound u-PAR might be important. Liu and colleagues (20) demostrated that u-PAR overexpression in the presence of normal EGFR levels leads to an activation of a ligand-independent EGFR signal in vivo, suggesting a possible mechanism for acquired resistance to anti-EGFR therapy. For EGFR activation, these authors highlight the role of domain1 of the u-PAR molecule, and it is interesting to speculate that an alteration of domain1, despite of high membrane-u-PAR, might still render NSCLC cells sensitive to Cetuximab, as observed for Calu3 in our study. Clearly, future investigations on this issue are important.

Second, we successfully showed that reducing endogenous u-PAR by siRNA knockdown can rescue sensitivity to Cetuximab. Furthermore, our analysis of preliminary patient samples revealed that the majority of patients with progression under Cetuximab-treatment show high u-PAR staining. Thus, u-PAR might become a valuable biomarker for predicting response to Cetuximab when treating NSCLC patients. Our data moreover show that the combination of high u-PAR with low E-cadherin might be a useful marker combination to define Cetuximab-resistant patients. The patient data are preliminary and need to be confirmed in a statistically robust set of NSCLC patients in a larger clinical study; hence, such efforts are currently ongoing in our group.

Taken together, this is the first study to show that Cetuximab negatively regulates u-PAR via JunD and c-Jun, and induces E-cadherin, these molecules being implicated in tumor growth and metastasis. The findings elucidate mechanisms by which Cetuximab inhibits metastasis in lung cancer. Furthermore, E-cadherin and especially u-PAR are novel and highly interesting potential markers to be considered to aid in decisions for Cetuximab therapy of NSCLC patients.

No potential conflicts of interest were disclosed.

Grant support: Merck KGaA, Darmstadt, Germany. H. Allgayer is supported by Alfried-Krupp-von-Bohlen-und-Halbach Foundation, Essen; Dr. Hella-Buehler Foundation, Heidelberg; Hector Foundation, Weinheim; the FRONTIER-program, Heidelberg-University Excellence Initiative; and Dr. Ingrid-zu-Solms Foundation, Frankfurt/Germany.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Annette Gruber for excellent technical assistance and the Drs. C. Schumann, C. Eschbach, A. von-Bierbauer, and M. Wittek (Gemtax-IV study) for patient samples.