Cetuximab, an anti-EGFR human-mouse chimeric monoclonal antibody (mAb), was purchased from Merck Serono (Darmstadt, Germany). SCH-772984, a selective inhibitor of extracellular signal regulated kinase (Erk1/2) that displays behaviors of both type I and II kinase inhibitors, was purchased from Selleckchem. SCH-772984 was dissolved in dimethylsulfoxide (DMSO) and maintained as a concentrated stock at −20°C. Working concentrations were diluted in culture medium just before each experiment.

We received approval for the present study from the Clinical Research Ethics Committee, and tumor specimens were collected after obtaining patient informed consent in accordance with the institutional guidelines. We selected 274 samples from 148 patients (78 men and 70 women) with primary colorectal carcinoma who underwent surgery at the Department of General Surgery, Peking University First Hospital from January 2006 to June 2009, including 148 colorectal carcinoma and 126 normal adjacent mucosa (at least 2 cm distant from the tumor). The patients were selected according to TNM stage; the case number of each stage did not differ greatly. None of the patients had received chemotherapy and radiotherapy before surgery. Two pathologists reviewed all of the histological slides.

The human colon cancer cell lines and a human colon normal mucosa cell line were obtained from the Colon Cancer Institute of Peking University First Hospital (Beijing, China), including SW480, HT29, Rko, Caco-2, Lovo and FHC cell lines. These cells were routinely cultured in Dulbecco's modified Eagle's medium [DMEM containing 10% fetal bovine serum (FBS)] (both from Gibco) and penicillin/streptomycin at 37°C in a 5% CO₂-humidified atmosphere.

Cancer cells were seeded into 96-well plates and treated with different concentrations of SCH-772984 (range, 5–90 mg/ml), cetuximab (range 5–700 mg/ml) alone for 48 h or with cetuximab (100 mg/ml) for a period of time (6 h-10 days). Cell proliferation was measured with MTT assay. The IC₅₀ value and proliferation rate were calculated according to optical density from a microplate reader. Each test was performed in quadruplicate.

SW480 and Caco-2 cells were seeded in 25 cm² flasks, treated for 72 h and stained with Annexin V-fluorescein isothiocyanate (FITC) and 7-amino-actinomycin D (7-AAD) in the dark. Viable (7-AAD-negative) and dead (7-AAD-positive) cell populations were quantitated by flow cytometry using a flow cytometer (BD Influx).

The small inhibitor duplex RNAs (siRNAs) (On-target paxillin) PXN-siRNAs (human: SR303929) were purchased from OriGene. OriGene synthesized three siRNA duplexes targeting human paxillin mRNA (PXN-siRNA). The targeting sequences were siRNA1, UGACGAAAGAGAAGCCUAAGCGGAA; siRNA2, UGAACGCUGUACAGCAUAACCCGCC; and siRNA3, GACAAUGCCAGCAUAAAUCCAUCCA. In the present study, siRNA1 was used since it effectively inhibited PXN expression in our preliminary experiments. The siCONTROL non-targeting siRNA (human, SR30004) was used as a negative control (NC). Cells were transfected with 40 nmol/l siRNAs using Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer's instructions. The day before the transfection, the cells were plated in 25 cm² flasks at 50–70% confluency in medium supplemented with 5% FBS. Cells were harvested 72 h after the transfection. Efficiency of gene silencing was detected by western blotting.

The human PXN-cDNA was purchased from OriGene (RC213811). It was cloned into vector Pcmv6-Entry (cat. #PS1000-001) and was verified by full sequencing (NM_002859). This human PXN cDNA was derived from single colony E. coli cultures and purified through the OriGene ion-exchange plasmid purification system (PowerPrep. HP Midiprep kits with Prefilters NP 100024). The non-targeting plasmid was used as a negative control. The cells were transfected with 1.5 mg/ml cDNAs using Lipofectamine 2000 reagent following the manufacturer's protocols. The day before the transfection, the cells were plated in 25 cm² flasks at 50–70% confluency in medium supplemented with 10% FBS. They were harvested 72 h after the transfection. Efficiency of gene silencing was detected by western blotting. When transfecting PXN-cDNA and PXN-siRNA at the same time, Lipofectamine 2000 reagent and 30 pmol of siRNA/1 mg of DNA was used.

SW480, SW620, HT29, Rko, Caco-2 and Lovo cells were seeded into 25 cm² flasks and treated for 72 h. Total proteins were obtained by a protein extraction kit (KeyGen Biotech, Nanjing, China), including phosphorylated proteins. The protein concentration was estimated by a modified Bradford assay (KeyGen Biotech). Immunoreactive bands were visualized by enhanced chemiluminescence (ECL; GE Healthcare/Amersham). Antibodies against p-Erk, cleaved caspase-3 and GAPDH (HRP-conjugated) were purchased from Cell Signaling Technology. Antibodies against PXN (Tyr-118) were purchased from Abcam (Hong Kong, China). Secondary antibodies coupled to horseradish peroxidase were purchased from Signalway Antibody (USA). The antibodies were diluted according to the instructions before the experiment. Each experiment was carried out for several times.

The immunohistochemical staining was performed on paraffin sections (4-µm) using a streptavidin-peroxidase immunostaining kit. After deparaffinization with xylene and rehydration through a graded ethanol series, the sections were subjected to microwave antigen retrieval with an ethylene-diaminetetraacetic acid (EDTA) buffer solution (1 mmol/l, pH 8.0) at 98°C for 10 min to unmask antigenic epitopes. Endogenous peroxidase activity was blocked with 0.3% hydrogen peroxide for 15 min at room temperature. After blocking non-specific binding sites with a blocking serum for 30 min at room temperature, the sections were incubated with a mouse monoclonal anti-PXN antibody (1:100; Abcam) at 4°C overnight. Visualization of the antibody-enzyme complex was achieved with 3,3-diaminobenzidine tetrahydrochloride (DAB). The sections were counterstained with Mayer's hematoxylin. Appropriate positive controls were included, and as a negative control, normal serum and PBS were substituted for the primary antibody.

To quantify PXN protein expression, both the extent and intensity of immunoreactivity were assessed and scored. In the present study, the scores of the extent of immunoreactivity ranged from 0 to 3 and were determined according to the percentage of cells that exhibited positive staining in each microscopic field of view (0, <25%; 1, 25–50%; 2, >50–75%; and 3, >75–100%). The scores of IHC intensity were as follows: 0, negative staining; 1, weak staining; 2, moderate staining; and 3, strong staining. By multiplying the scores for extent and intensity, a total score ranging from 0 to 9 was achieved. The expression level of PXN was considered high when the total score was ≥4 and low when the total score was <4. The results were assessed by two independent observers who did not have any knowledge of the clinicopathological features of the patients. If the results differed, then they consulted and reached a final consensus.

The statistical analyses of data were carried out using SPSS 17.0 statistical software (SPSS, Inc., USA). The relationship between PXN expression and clinicopathological characteristics was analyzed by the χ² test. All P-values represent two-sided tests of statistical significance with P-value <0.05.

The expression of PXN was examined in 148 colorectal adenocarcinoma samples and 126 normal adjacent mucosa samples through an immunohistochemical staining approach. A total of 101 of the 148 (68.2%) colorectal adonocarcinoma samples and 56 of the 126 (44.4%) normal tissues were PXN-positive (P<0.001) (Table I). PXN protein was mainly expressed in the cytoplasm and was found to be overexpressed in the tumors compared with the normal adjacent tissues (Fig. 1A). This result was further confirmed in 60 colorectal carcinoma samples with paired normal adjacent tissues by western blotting. In 43 of 60 (71.7%) cases, PXN expression levels in the tumor tissues (T) were at least 2-fold higher than levels in the normal adjacent mucosa (N) (Fig. 1B).

Follow-up results in a total of 148 patients were analyzed. The median duration of follow-up was 48.6 months (range 1–75 months), and a total of 63 patients died during the follow-up period. Univariate analyses revealed that the 3- and 5-year survival rates of the PXN-negative group were higher (71.8 and 65.3%) than those of the positive group (52.5 and 35.4%, log-rank testing, P=0.004, Fig. 1C). The survival rate was correlated with the expression of PXN.

The expression of PXN was assessed in five different colon cancer cell lines and a normal mucosal cell line by western blotting. Compared with FHC, a normal human colon cell line, the 5 colon cancer cell lines showed a higher level of PXN. Among the Caco-2, Lovo, SW480, HT-29 and Rko cells, PXN was expressed at the highest level in the SW480 cells, secondly in the Lovo, HT-29 and Rko cells, and least in the Caco-2 cells (Fig. 2). Therefore, we chose SW480 and Caco-2 cells to perform subsequent experiments.

The positive rate of PXN was higher in the colorectal adenocarcinoma samples than that in the normal colorectal mucosa samples (68.2 vs. 44.4%, P<0.001, Table I). The rate of recurrence in the colorectal patients was higher in the PXN-positive group (27.7%) than that in the PXN-negative group (12.8%, P=0.032; Table I). PXN positivity was closely related to TNM stage (86.4, 74.6, 61.5 and 50.0%, P=0.023, Table I) and distant metastasis (88 vs. 64.2%, P=0.014; Table I).

SW480 and Caco-2 cells were treated with increasing concentrations of cetuximab (5–700 mg/ml). The percentage of proliferation in the SW480 cells slightly increased at first and then slightly decreased (Fig. 3A). This showed that the SW480 cells were not sensitive to cetuximab. Compared with the SW480 cells, the Caco-2 cells were relatively sensitive to cetuximab, and the IC₅₀ value was ~189.3 mg/ml (Fig. 3A).

To explore how PXN and p-Erk change with the treatment of cetuximab in SW480 and Caco-2 cells, the cells were treated for a period of time (from 6 h to 10 days) at a concentration of 100 mg/ml (Fig. 3B). As shown in Fig. 3C, the expression of PXN declined at first, and then increased gradually and stabilized in the SW480 cells, which was very consistent with the proliferation rate in Fig. 3B, the lowest point of which was at ~24 h. In contrast to the SW480 cells, PXN expression in the Caco-2 cells did not significantly change. These data demonstrated that the high expression level of PXN may be one of the reasons for the insensitivity to cetuximab of SW480 cells, which warrants further experiments for confirmation. Moreover, the expression changes in p-Erk in both the SW480 and Caco-2 cells (Fig. 3C) were consistent with the proliferation rates in Fig. 3B.

To clarify the effect of an Erk1/2 inhibitor on SW480 and Caco-2 cells, the cells were treated with increasing concentrations of SCH-772984 (5–90 mg/ml) for 48 h. SCH-772984 treatment of the SW480 and Caco-2 cells induced a dose-dependent inhibition of cell growth with an IC₅₀ value of ~47.5 and 56.4 mg/ml, respectively (Fig. 3D), which demonstrated that both SW480 and Caco-2 cells were sensitive to SCH-772984.

To determine whether PXN activation could be one of the reasons for cetuximab resistance in SW480 cells, a further investigation was carried out to ascertain whether reduction in PXN expression restores cetuximab sensitivity. First of all, we compared the effect of transfection between siRNA1 and siRNA2. As shown in Fig. 4A, transfection with siRNA1 for 72 h significantly reduced PXN protein expression in the SW480 cells, and thus, siRNA1 (PXN-siRNA) was chosen to carry out the following transfection study.

Although single-agent cetuximab did not significantly affect SW480 apoptosis, cetuximab treatment in combination with PXN silencing showed a statistically significant apoptotic effect on SW480 cells (Fig. 4B). The expression of apoptotic protein cleaved caspase-3 was consistent with the apoptotic rate as shown in Fig. 4B. PXN silencing also inhibited p-Erk activation in the SW480 cells as shown by downregulation of p-Erk levels (Fig. 4B), which may be the mechanism through which SW480 cells regained sensitivity to cetuximab.

To further evaluate whether PXN activation confers resistance to cetuximab in colorectal cancer cells, the Caco-2 cell line, confirmed to have low expression of PXN, was used. After transfection with PXN-cDNA for 72 h, PXN protein expression was significantly increased in the Caco-2 cells (Fig. 4A). As illustrated in Fig. 4C, the overexpression of PXN effectively suppressed apoptosis in the Caco-2 cells with or without cetuximab treatment. Corresponding to the apoptotic data, the expression of apoptotic protein cleaved caspase-3 was downregulated with the decrease in the apoptotic rate (Fig. 4C). PXN overexpression also promoted the activation of p-Erk in the Caco-2 cells as shown by upregulation of p-Erk levels (Fig. 4C). These data further demonstrated that PXN confers resistance to cetuximab and knockdown of PXN improves sensitivity to cetuximab in colorectal cancers by downregulating the expression of p-Erk.

To further evaluate the role of PXN activation in cetuximab resistance, co-transfection with PXN-siRNA and PXN-cDNA in SW480 and Caco-2 cells was performed. Compared with transfection with PXN-siRNA alone, the apoptotic rate was greatly reduced after co-transfection with PXN-siRNA and PXN-cDNA for 72 h in the SW480 cells (Fig. 5A). The expression of apoptotic protein cleaved caspase-3 increased corresponding to the apoptotic rate (Fig. 5A). Moreover, the expression of PXN increased as expected after co-transfection, compared with transfection with PXN-siRNA alone and activation of PXN also upregulated p-Erk levels (Fig. 5A). Similarly, after co-transfection with PXN-siRNA and PXN-cDNA for 72 h in the Caco-2 cells, the apoptotic rate was greatly reduced, compared with transfection with PXN-cDNA alone (Fig. 5B). Not surprisingly, the expression of apoptotic protein cleaved caspase-3 increased, and activation of PXN and p-Erk was inhibited as shown by downregulation of p-Erk levels (Fig. 5B).

From the experiments above, we found that the expression levels of protein PXN and p-Erk were very synchronous. To explore the relationship between PXN and p-Erk and further evaluate the role of PXN activation, SW480 and Caco-2 cells were treated with cetuximab and SCH-772984, a selective Erk1/2 inhibitor. SCH-772984 treatment of SW480 and Caco-2 cells induced a dose-dependent inhibition of cell growth with an IC₅₀ value of ~47.5 and 56.4 mg/ml, respectively (Fig. 3D). Although cetuximab treatment had little effect on cell growth in the SW480 cells, the combined treatment with SCH-772984 restored the sensitivity of SW480 cells to cetuximab (Fig. 5A). Moreover, as illustrated in Fig. 5A, the percentage of apoptotic cells also obviously increased, compared with the group treated with cetuximab or SCH-772984 alone. Meanwhile, the expression levels of PXN and p-Erk were suppressed significantly in both cell lines (Fig. 5A and B). Based on the outcomes above, another test was performed to make the entire experiment more rigorous. After transfection with PXN-cDNA in the SW480 and Caco-2 cells for 24 h, the cells were treated with cetuximab and SCH-772984 for another 48 h. The results in Fig. 5A and B showed that the percentage of apoptotic cells decreased significantly, compared with the group that was treated only with cetuximab or SCH-772984. Correspondingly, the downregulation of cleaved caspase-3 and upregulation of both PXN and p-Erk expression levels are illustrated in Fig. 5A and B. In summary, knockdown of PXN restored or improved sensitivity to cetuximab by downregu-lating the expression level of p-Erk.