PPARγ and Wnt/β-Catenin pathway in human breast cancer: expression pattern, molecular interaction and clinical/prognostic correlations

Tissue samples

Seventy breast cancer and 20 benign breast disease specimens from patients who underwent surgery between 2001 and 2003 at Department of General Surgery, Affiliated Hospital of Nantong University were formalin-fixed and paraffin-embedded for histopathologic diagnosis and immunohistochemical study. The TNM system of tumor staging and histological grade were performed according to the World Health Organization guidelines (Hartmann et al. 1981). The follow-up information was obtained from telephone contact and interview. Fresh samples were frozen in liquid nitrogen immediately after surgical removal and maintained at −80°C until use for western blot analysis. All human tissue was collected using protocols approved by the Ethics Committee of Affiliated Cancer Hospital of Nantong University. The clinical features of the patients, including age, histologic grade, tumor size, axillary lymph node status, TNM stage, and histology, are shown in Table 1. The median age of the patients was 49 years (range 29–90 years). The median 5-year follow-up time for the 70 patients was 56.5 months (range 6–79 months). The majority was T2 (n = 37, 53%) in size, poorly differentiated (n = 28, 40%) in histologic grade and 59% (n = 41) patients had no lymph node metastasis at the time of surgery. There were 21.4% TNM stage I, 51.4% TNM stage II and 27.2% TNM stage III tumors at the time of diagnosis. The majority (n = 45, 64%) was infiltrating ductal carcinoma, and the remaining 25 cases consisted of other types. Details are shown in Table 1.

Antibodies

The antibodies applied for the IHC in this study included: anti-PPARγ (sc-7196, 1:100, Santa Cruz, Biotechnology), anti-β-Catenin (9562, 1:100, cell signaling, technology), anti-Ki-67 (MIB-1, 1:100, Immunotech, Westbrook). Antibodies for western blotting and immunoprecipitation included: anti-β-Catenin (9562, 1:1,000, cell signaling, technology), anti-PPARγ (sc-7196, 1:800, Santa Cruz, Biotechnology) and anti-β-actin (sc-450, 1:3000, sigma).

Immunohistochemistry methods

Serial sections that were 4 μm thick were mounted on glass slides coated with 10% polylysine. Sections were dewaxed in xylene and rehydrated in graded ethanols. Endogenous peroxidase activity was blocked by immersion in 0.3% methanolic peroxide for 40 min. Immunoreactivity was enhanced by microwaving by incubating the tissue sections for 10 min in 0.1 M citrate buffer. Immunostaining were performed using the avidin biotin peroxidase complex method and antigen–antibody reactions were visualized with the chromogen diaminobenzadine.

Immunohistochemical evaluation

Three independent pathologists (SH, LC, and YW) evaluated the immunostaining results. PPARγ positive cells were counted by monitoring at least 300 cells from at least 10 randomly selected fields. We then calculated the percentage of antigen positive nuclei among the total number of nuclei counted to obtain the nuclear LI. To allow univariate and multivariate analyses, we divided patients into two groups according to PPARγ LIs average (17.1%): high expression group (LI ≥ 17.1%) and low expression group (LI < 17.1%).When evaluating the β-Catenin protein immunoreaction, staining was scored in a semi-quantitative fashion from 0 to 3 as described previously (Karatzas et al. 2000). When more than 90% of tumor cells uniformly exhibited the same intensity and pattern of immunostaining as the adjacent normal epithelial cells, the antigen expression was considered as normal or preserved and a score of 3 was assigned. A score of 2 was assigned for heterogeneous staining (i.e., mixed areas of positive and negative cells) and a score of 1 for altered cellular distribution of immunostaining (i.e., cytoplasmic or nuclear).Sections with less than 10% stained cells or with complete absence of staining were classified as negative(score 0) (Karayiannakis et al. 2001). A cut-off value of 10% or more positively stained nuclei in 10 high-power fields was used to define Ki-67 staining (Remmele et al. 1995). Immunohistochemical evaluation for ER, PR and Her-2 were finished by Department of Pathology, Affiliated Hospital of Nantong University.

Cell cultures

One normal human breast epithelial cell line HBL-100 and three human breast cancer cell lines: MCF-7(ER+), MDA-MB-231(ER−) and MDA-MB-435(ER−), which were gifted from Department of Oncology, Cancer Hospital of Fudan University were used in this study. All cell lines were maintained in RPMI 1640 (GibCo BRL, Grand Island, NY, USA) supplemented with 10% heat-inactivated fetal calf serum, 2 mM l-glutamine, and 100 U/mL penicillin–streptomycin mixture (GibCo BRL, Grand Island, NY, USA) at 37°C and 5% CO₂.

Western blot analysis

Prior to immunoblotting, cells were washed three times with ice-cold PBS, resuspended in 2 × lysis buffer (50 mM Tris–HCl, 120 mM NaCl, 0.5% Nonidet P-40, 100 mM NaF, 200 μM Na₃VO₄, and protease inhibitor mixture) or frozen tissues were homogenized in lysis buffer (1% NP-40, 50 mmol/L Tris, pH 7.5, 5 mmol/L EDTA, 1% SDS, 1% sodium deoxycholate, 1% Triton X-100, 1 mmol/L PMSF, 10 mg/mL aprotinin, and 1 mg/mL leupeptin) and then incubated for 20 min at 4°C while rocking. Lysates were cleared by centrifugation (10 min × 12,000 rpm, 4°C). 50 μg of total protein was resolved by SDS-PAGE and transferred onto polyvinylidene difluoride membranes (Immbilon, Millipore). The membranes were first blocked with 5% nonfat dry milk and then incubated with the primary antibody described above for 2 h at room temperature. After three times of washes, filters were incubated with horseradish peroxidase-conjugated secondary human anti-mouse or anti-rabbit antibodies (1:1,000; pierce) for 1 h at room temperature according to the manufacturer’s instructions. Detection of immunocomplexes was performed with an enhanced chemiluminescence system (NEN Life Science Products, Boston, MA).

Immunoprecipitation

Nonconfluent cells were washed three times with ice-cold PBS or frozen tissues were homogenized with an electric homogenizer and then lysed in RIPA buffer (50 mM Tris pH 7.5, 1% Triton X-100, 0.5% deoxycholate, 10 mM EDTA, 150 mM NaCl, 50 mM NaF, 1 mM sodium vanadate, 1 μg/mL leupeptin and 0.1 μg/mL aprotinin), whole lysates were then centrifuged at 14,000×g for 10 min to remove cell debris. Samples were first precleared with a nonspecific IgG control. Precleared lysates were then incubated with 5 μl β-Catenin antibody and rotated at 4°C overnight followed by the addition of 40 μl protein G-sepharose slurry with rotation for 1 h. Then, remove the last supernatant and collect protein G beads which were washed with lysis buffer three times. Then add 50 μl of 2× loading buffer, boil at 95–100°C for 15 min to denature the protein and separate it from the protein G beads, then centrifuge and keep the supernatant where the protein is now. Finally, IP samples were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and analyzed by IB.

Statistical analysis

The chi-square (χ²) test and Fisher’s exact tests were used to compare the expression of all proteins as groups (low group vs. high group) with various clinic pathological parameters. The nonparametric Spearman’s rank correlation coefficient was applied to evaluate the strength of the relationship between PPARγ and β-Catenin expression. Survival analysis was undertaken using the Kaplan–Meier method and curves were compared using log-rank test. Cox’s proportional hazards model was used for multivariate analyses of the prognostic values. P < 0.05 was considered statistically significant. All computations were carried out using the SPSS 15.0 statistical program.

Expression of PPARγ, β-Catenin and Ki-67 in human breast cancer

Using IHC, we examined the distribution of PPARγ, β-Catenin and Ki-67 protein in paraffin-embedded mammary tissue sections screened from 20 benign breast diseases and 70 breast cancer patients. The results of these experiments indicated that mammary epithelial cells and myoepithelial cells from benign breast disease samples showed strong positive immunoreactivity for PPARγ in the nucleus (Fig. 1a), while breast carcinoma cells displayed weak (Fig. 1d) or no PPARγ staining (Fig. 1g). Moreover, it has been well documented that uniform, membranous β-Catenin staining localized at the intercellular borders was observed in normal mammary ducts and glands (Karayiannakis et al. 2001).Of the 20 benign breast diseases, we found β-Catenin not only located in the nuclei of myoepithelial and luminal epithelial cells, but also displayed moderate membrane staining in the non-tumorous ducts of breast benign disease tissues (Fig. 1b). As shown in Fig. 1e, increased membranous and cytoplasmic expression of β-Catenin was presented in the sample with TNM stage I. We also observed absence of β-Catenin staining in an infiltrating ductal carcinoma (TNM stage III), as shown in Fig. 1h. In order to investigate cellular proliferation rates of breast malignant tumor, we also assessed the expression of Ki-67 protein in all cases. The relevant results were exhibited in Fig. 1c, f, i. Ki-67 expression in benign tumor was lower than those of malignant tumor (P < 0.05). While among malignant tumors, staining intensity in TNM stage III was higher than that in stage I (P < 0.05).

To explore the significance of the two molecules in human breast cancer progression and further characterize the relationship between PPARγ and β-Catenin, we continued to investigate their abundance by western blot analysis. First, we investigated the basal abundance of PPARγ and β-Catenin in one normal human breast epithelial cell line HBL-100 and three human breast cancer cell lines: MCF-7(ER+), MDA-MB-231(ER−) and MDA-MB-435(ER−). As shown in Fig. 2a, the different expression levels of PPARγ and β-Catenin were observed in all four cell lines. The normal breast epithelial cell line HBL-100 expressed high level of PPARγ and moderate level of β-Catenin compared with the other three cancer cells. As we expected, the relative abundance of PPARγ and β-Catenin appeared to exhibit an inverse correlation in these three cancer cells. Consistent with previous studies, MCF-7 cell lines (ER+) displayed highest abundance of PPARγ and low expression level of β-Catenin among these tumor cell lines. On the contrary, lowest PPARγ activity and highest β-Catenin expression were obtained in MDA-MB-435 cells (ER−).To be continued, in five paired frozen breast tumor tissues and para-cancerous tissues, the presence of an inverse correlation between PPARγ and β-Catenin were also obtained by western blot analysis (Fig. 2b). The PPARγ expression was dramatically decreased in four of five tumors compared with para-cancerous tissues. On the other hand, these same tumor samples had higher expression levels of β-Catenin than para-cancerous tissues, though not quite obvious.

The clinical relevance of PPARγ

To gain better insight into the clinical relevance of PPARγ, we investigated the relationship of PPARγ expression and established clinical factors. With a cut-off of average LIs 17.1%, PPARγ immunostaining was low in 66% (46 of 70) of the breast cancer patients. PPARγ expression was inversely associated with histologic grade (P = 0.012), tumor size (P = 0.007), axillary lymph node status (P = 0.044) and TNM stage (P = 0.026), whereas no association was seen between PPARγ and age (P = 0.513), histology (P = 0.764). The detailed data are shown in Table 1. Next we examined the relationship of PPARγ and pathologic prognostic parameters. As shown in Table 2, an inverse correlation was found between PPARγ and β-Catenin immunoactivity. It appeared that tumors with PPARγ low immunoactivity were more likely to exhibit abnormal β-Catenin expression (P = 0.023). By employing the nonparametric Spearman’s rank test, we also obtained the Spearman’s correlation coefficient for the relationship between them (R = −0.398, P = 0.001). PPARγ was also associated with ER status (P = 0.012) and preferred to display low expression level in ER negative tissues. Not surprisingly, tumors with weak PPARγ immunoactivity showed markedly increased Ki-67 LIs (P = 0.006). This inverse link could be explained as one possible reason for the poor prognosis of breast cancer with high proliferative activity. Whereas no significant correlation was seen between PPARγ expression and either PR (P = 0.099) or HER-2 (P = 0.175), which did not reach statistical significance. Taken together, these data demonstrated that PPARγ expression was correlated with histologic grade, tumor size, axillary lymph node status, clinical grade (TNM stage), abnormal β-Catenin expression rate, ER and Ki-67.

Physical interaction between PPARγ and β-Catenin in human breast carcinoma

Consequently, in order to investigate whether there was also a physical interaction between PPARγ and β-Catenin in human breast cancer, we performed co-immunoprecipitation in three breast cancer cell lines and freshly frozen materials from patients underwent surgery. The experiments shown in Fig. 3a indicated that PPARγ interacted with β-Catenin in all three human breast cancer cell lines. However, their effect of interaction was not at the same level. MCF-7 cells showed the strongest interaction in all three cell lines. However, there was no significant difference in the binding of two molecules in the other two cell lines MDA-MB-231 and MDA-MB-435. In the following experiments, we tested the PPARγ/β-Catenin complex formation in breast freshly frozen tissues (Fig. 3b). Indeed, consistent with the findings of cells, also the interaction exited in normal breast tissue, benign tumor and malignant tumors. Whereas, we detected that interaction between PPARγ and β-Catenin reduced in TNM stage III tumor compared with stage I tumor. This suggested that the alleviated expression of PPARγ in the advanced tumor might contribute to the reduction of the binding complex. Because the number of samples analyzed here was small, the relationship between the interaction and the clinical parameters of patients could not be made.

Correlation between PPARγ and patients’ survival

We carried out Kaplan–Meier analysis to study the correlation between PPARγ and patients’ survival. Survival analysis was restricted to 70 patients with follow-up data and results of PPARγ by IHC. By using Kaplan–Meier analysis, patients with high expression level of PPARγ were significantly associated with long overall survival (P = 0.001, Fig. 4). Multivariate analysis using the Cox’s proportional hazards model showed that PPARγ protein was an independent prognostic indicator for patients’ overall survival (P = 0.005, Table 3).