P-gp substrates, including paclitaxel (Taxol; Bristol-Myers Squibb, Wallingford, CT, USA) and epirubicin (Pharmorubicin; Pfizer, Inc., New York, NY, USA), and a non-P-gp substrate, epothilone (Ixempra; Bristol-Myers Squibb), were dissolved in either a special diluent (paclitaxel, epothilone) or in sterile distilled water (epirubicin). The substrates were stored frozen as stock solutions and thawed prior to use. The peak plasma concentrations (PPCs) of pacli-taxel, epirubicin and epothilone are 3.65 µg/ml, 0.4 µg/ml and 252 ng/ml, respectively. EGFR/HER2 tyrosine kinase inhibitor (lapatinib, Tykerb) and HER2 monoclonal antibodies (trastu-zumab, Herceptin) were purchased from GlaxoSmithKline, (GSK) (Brentford, UK) and Roche (Basel, Switzerland), and their PPCs are 2.43 µg/ml and 216 µg/ml, respectively. All other reagents were of analytical grade and obtained from commercial sources.

MCF7/Adr cells were cultured in RPMI-1640 medium (Gibco-BRL, Karlsruhe, Germany), and MCF7 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) (high glucose; Gibco-BRL) supplemented with 0.01 mg/ml bovine insulin (Sigma, St. Louis, MO, USA). All cell culture media contained 10% fetal bovine serum (FBS; PAA Laboratories, Linz, Austria), 100 U/ml penicillin and 100 µg/ml streptomycin. The cells were plated in 6-well plates at a density of 1×10⁶ cells/well and further incubated for 24 h at 37°C in a humidified atmosphere containing 5% CO₂. The medium was then removed and replaced with fresh medium containing paclitaxel, epirubicin and epothilone, respectively, with different PPCs (0.1, 1 and 10 PPC) for another 48 h. When lapatinib and trastuzumab were required, they were added into the incubation at the PPC for 1 h before addition of the chemotherapy drugs.

Total cellular RNA was extracted using the TriPure Isolation reagent (Sangon, Shanghai, China). The RNA samples were subjected to reverse transcription (RT) with 2 µg of RNA, Oligo (dT)₁₈, dNTP, and reaction buffer supplied with M-MlV reverse transcriptase (Promega, Madison, WI, USA). Real-time PCR reactions were then performed in 20 µl of solution with 2 µg of cDNA, 1 mM of each forward and reverse primer and 2X SYBR-Green Mix (Takara, Shuzo, Kyoto, Japan). Changes in the mRNA expression level were calculated following normalization to the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA level. Relative gene expression was determined by the fluorescence intensity ratio of the target gene to GAPDH. The primers used for the specific amplifications are shown in Table I.

The cells were collected and lysed in a modified RIPA buffer supplemented with 1 tablet/50 ml of Complete Protease Inhibitor Cocktail (Roche Molecular Biochemical, Indianapolis, IN, USA). Total cell lysate (50 µg of protein) was resolved using SDS-PAGE and electrophoretically transferred onto PVDF membranes (Millipore, Bedford, MA, USA). The membranes were blocked in 5% non-fat milk for 1 h at room temperature and then incubated overnight with the primary antibodies against EGFR (BD Biosciences, San Jose, CA, USA), HER2 and CD147 (both from Santa Cruz Biotechnology, Santa Cruz, CA, USA), MMP2 (Cell Signaling Technology, Inc., Danvers, MA, USA), MMP9 (Abcam, Cambridge, MA, USA) or β-actin (Sigma) at 4°C. The membranes were then incubated for 1 h at 4°C with the appropriate HRP-conjugated secondary antibodies (Invitrogen Life Technologies, Carlsbad, CA, USA). Detection of the protein expression levels using ECL (Pierce, Rockford, IL, USA) was performed according to the manufacturer's protocol.

A Boyden dual chamber assay was performed using Transwell chambers with 8-µm pore size membranes (BD Biosciences). A total of 5×10⁴ cells was suspended in serum-free media and added to the upper chamber, with serum-containing media added as the chemo-attractant in the lower chamber. The Transwell chamber membranes were coated with 40 µl of growth factor-reduced Matrigel (BD Biosciences). After 4 h at 37°C with 5% CO₂, the media and cells remaining in the upper chamber were removed using a cotton swab. The insert was fixed in methanol and stained using hematoxylin and eosin. The number of invading cells was counted in five random fields for each group, and the mean number was calculated.

All animal experiments were performed under the approval of the Animal Experimentation Committee of Fudan University, Shanghai, China. BALB/c female nude mice (age, 4–6 weeks; weight, 18–20 g) were housed under a 12-h light/12-h dark cycle with free access to food and water in sterile conditions. The mice were fed a sterilized murine diet and water. The mice (n=6 per group) were subcutaneously injected with MCF7/Adr cells (1×10⁷ cells/mouse in 200 µl of 10 mM phosphate-buffered saline, pH 7.2) in their second mammary fat pads. When the tumors grew to 1 cm³, the mice were injected with paclitaxel (18 mg/kg), paclitaxel (18 mg/kg) + trastuzumab (first time: 36.4 mg/kg, second time: 18.2 mg/kg), or 0.9% normal saline (0.1 ml) as a control, once per week for 4 weeks. The mice were sacrificed during the 5th week.

After approval from our Institutional Review Board, we selected 65 cases of local advanced breast cancer from a cohort of women who had undergone a core-needle biopsy (pathology-confirmed breast invasive ductal carcinoma) and neo-adjuvant chemotherapy (paclitaxel 175 mg/m² + epirubicin 75 mg/m², once every 3 weeks for 4 cycles) at our institution between 2009 and 2011. All patients had received a radical mastectomy after 4 cycles of chemotherapy. According to the Japanese Breast Cancer Society (JBCS) criteria for the histological evaluation of the therapeutic response after neo-adjuvant chemotherapy (8), the cohort was divided into three groups, including pathological complete response (PCR, n=13), partial response (n=27) and no response (n=25). Tissue specimens before and after chemotherapy were conserved for testing.

Paraffin-embedded tissue samples from 18 murine tumors and 65 mammary carcinomas were prepared. The slides were dehydrated in xylene and graded alcohols. Antigen retrieval was performed (EGFR, proteinase K at room temperature for 20 min; HER2 and P-gp, 0.01 M citrate buffer at pH 6.0 at 95°C for 20 min). Then, the slides were incubated with diluted primary antibody (anti-EGFR, 1:100; anti-HER2, 1:400; anti-P-gp, 1:50) for 12 h, followed by incubation with a biotinylated secondary antibody for 1 h, peroxidase-labelled streptavidin for 15 min (LSAB-2 system; Dako, Glostrup, Denmark), and diamino-benzidine and hydrogen peroxide chromogen substrate plus diaminobenzidine enhancer (Dako) for 10 min. The slides were counterstained with Mayer's hematoxylin. The analysis of EGFR, HER2 and P-gp protein expression was performed using an EGFR rabbit polyclonal Ab (BD Biosciences), a HER2 rabbit polyclonal Ab (Santa Cruz Biotechnology) and a P-gp mouse monoclonal Ab (Calbiochem, San Diego, CA, USA). For the statistical analysis, according to the study by Han et al (9), the total staining for EGFR, HER2 and P-gp was scored based on the intensity and percentage of cells with EGFR, HER2 and P-gp cytoplasmic staining using the following scale: score 0, negative staining for all of the tumor cells; score 1, negative/weak staining for all of the tumor cells or moderate staining in <10% of the cells; score 2, moderate staining in >30% but <70% of the tumor cells or strong staining within 10% of the tumor cells; score 3, moderate staining in >70%, or strong staining in >30% but <70% of all the tumor cells; and score 4, strong staining in >70% of all the tumor cells. The reproducibility of EGFR, HER2 and P-gp staining was examined between two laboratories (by two independent pathologists).

Statistical analysis was carried out using SPSS 19.0 software. All experiments were repeated at least three times, and the results are presented as the mean ± standard errors (SEM). The differences were analyzed by t-test. Pearson's correlation coefficients were used to determine whether two prognosis-related factors correlated with each other. Kaplan-Meier survival analysis was used to estimate the prognostic relevance. P<0.05 was considered to indicate a statistically significant difference (two-tailed).

Following treatment with the non-P-gp substrate epothilone, the production of CD147 and MMP2/9 in both the MCF7/Adr and MCF7 cell lines remained unchanged (P>0.05) (Fig. 1A). However, incubation with the P-gp substrates (paclitaxel and epirubicin) resulted in a marked increase in CD147 and MMP2/9 at both the transcription and protein level in the MCF7/Adr cells (P<0.05), while the MCF7 cells did not show a similar response (Fig. 1B and C). Transwell chamber assays showed that MCF7/Adr cells exposed to P-gp substrates displayed a substantially higher invasive ability compared with the corresponding controls (P<0.05) (Fig. 1D).

Inhibitors of EGFR and its family members are often used in inhibiting tumor cell proliferation. However, no report has evaluated EGFR/HER2 inhibition for the treatment of reversing the drug-resistance of MDR breast cancer.

To ascertain whether EGFR/HER2 is associated with P-gp substrate-induced CD147 and MMP2/9 upregulation in MDR cells, EGFR and HER2 expression levels were measured in the MCF7/Adr and MCF7 cells after drug treatment. As expected, the results from real-time PCR showed large increases in EGFR and HER2 mRNA only in the MCF7/Adr cells treated with the P-gp substrates. Western blot analysis demonstrated the corresponding alterations at the protein level (Fig. 2A–C), consistent with the enhanced production of CD147 and MMP2/9. There was no similar change in the other groups.

Interestingly, when MCF7/Adr cells were pretreated with lapatinib, an EGFR/HER2 tyrosine kinase inhibitor, or trastu-zumab, an HER2 monoclonal antibody, the enhancement of CD147 and MMP2/9 evoked by the P-gp substrates was largely inhibited (P<0.05), as confirmed using real-time PCR and western blot analysis (Fig. 2D). Additionally, the transcription and protein expression of EGFR and HER2 were also down-regulated (P<0.05) (Fig. 2D). The Transwell chamber assays showed that only the MCF7/Adr cells that were pretreated with lapatinib or trastuzumab demonstrated significantly decreased invasive activity (P<0.05) (Fig. 2E).

The expression levels of EGFR, HER2 and P-gp were detected using IHC in Balb/c mice bearing MCF7/Adr breast carcinomas (Fig. 3A). Similarly, the upregulation of EGFR, HER2 and P-gp was found only in the group treated with paclitaxel (P<0.05) (Table II). In vitro, the inhibition of EGFR/HER2 could interrupt the stimulation of CD147, MMP2 and MMP9 evoked by P-gp substrates in the MDR cells. As expected, the expression of EGFR and P-gp was also clearly downregulated in the group co-treated with paclitaxel and trastuzumab (P<0.05). However, the decrease in HER2 expression was not significant (P>0.05) (Table II).

The treatment of MDR breast cancer with P-gp substrates could adversely affect the function of P-gp and inhibit the degradation of EGFR, which increases the expression of CD147 and MMPs and promotes tumor cell invasion/metastasis. To further validate this hypothesis, 65 local advanced breast cancer (LABC) cases undergoing P-gp substrate chemotherapy (paclitaxel and epirubicin) were chosen for the present study. The expression levels of EGFR, HER2 and P-gp were compared via IHC (Fig. 3B). According to the chemo-response, the patients were divided into 3 groups: pathological complete response (PCR; n=13), partial response (n=27) and no response (n=25). Upregulation of EGFR and P-gp after chemotherapy was only observed in the no response group (P<0.05) (Table III). Furthermore, we analyzed the relationship between chemo-response and various clinicopathological factors of the 65 cases: estrogen receptor (ER), progestrone receptor (PR) and Ki67. The no response group exhibited increasing lymph node metastasis (P<0.01) (Table IV). However, in the partial response and PCR groups, there were fewer lymph node metastases. There was no statistical correlation between EGFR and ER, PR and Ki67, and HER2 and P-gp (P>0.05) (Table IV). In the PCR group, due to a lack of tumor tissue after chemotherapy, the differential comparison before and after chemotherapy could not be conducted. The above results revealed that P-gp substrate chemotherapeutic drugs can promote the malignant properties of MDR breast cancer in the clinic, and EGFR and HER2 may be involved in the progression.

A survival analysis was conducted in 65 cases. Kaplan-Meier plots for OS and DFS stratified by EGFR and P-gp expression levels before chemotherapy are shown in Fig. 4A–C. There was no correlation between HER2 and OS/DFS (data not shown). In the total cases, we found that higher EGFR expression was associated with shorter OS time (P<0.01) (Fig. 4A). Furthermore, we separated the patients according to P-gp-positive or -negative status. In the P-gp-positive group (n=43), EGFR expression levels were associated with shorter OS (P<0.01) and shorter DFS times (P<0.05) (Fig. 4B); however, in the P-gp-negative group (n=22), there was no correlation between EGFR and OS/DFS times (P>0.05) (Fig. 4C). The results revealed that EGFR is a poor prognostic factor in MDR breast cancer patients, especially in P-gp-positive tumors.

The survival analyses also revealed a trend between poorer chemo-response and shorter OS and DFS, although this finding was not statistically significant (Fig. 4D), most likely due to the small number of cases.